Marleen de Bruijne Marleen de Bruijne

Marleen de Bruijne / Publications&abstracts

 

Papers in international journals

 
Martine Loeve, Tim Rosenow, Vladlena Gorbunova, Wim C J. Hop, Harm A W M. Tiddens, and Marleen de Bruijne. Reversibility of trapped air on chest computed tomography in cystic fibrosis patients. European Journal of Radiology, 84(6):1184-1190, 2015. [ link ]
To investigate changes in trapped air volume and distribution over time and compare computed tomography (CT) with pulmonary function tests for determining trapped air.Thirty children contributed two CTs and pulmonary function tests over 2 years. Localized changes in trapped air on CT were assessed using image analysis software, by deforming the CT at timepoint 2 to match timepoint 1, and measuring the volume of stable (TAstable), disappeared (TAdisappeared) and new (TAnew) trapped air as a proportion of total lung volume. We used the difference between total lung capacity measured by plethysmography and helium dilution, residual volume to total lung capacity ratio, forced expiratory flow at 75% of vital capacity, and maximum mid-expiratory flow as pulmonary function test markers of trapped air. Statistical analysis included Wilcoxon's signed rank test and Spearman correlation coefficients.Median (range) age at baseline was 11.9 (5-17) years. Median (range) of trapped air was 9.5 (2-33)% at timepoint 1 and 9.0 (0-25)% at timepoint 2 (p=0.49). Median (range) TAstable, TAdisappeared and TAnew were respectively 3.0 (0-12)%, 5.0 (1-22)% and 7.0 (0-20)%. Trapped air on CT correlated statistically significantly with all pulmonary function measures (p<0.01), other than residual volume to total lung capacity ratio (p=0.37).Trapped air on CT did not significantly progress over 2 years, may have a substantial stable component, and is significantly correlated with pulmonary function markers.

 
Mathilde M.W. Wille, Laura H. Thomsen, Jens Petersen, Marleen de Bruijne, Asger Dirksen, Jesper H. Pedersen, and SaherB. Shaker. Visual assessment of early emphysema and interstitial abnormalities on ct is useful in lung cancer risk analysis. European Radiology, pages 1-8, 2015. [ link ]
Keywords: Computed tomography; Lung cancer; Emphysema; Interstitial abnormalities; Comorbidity
 
A.G. van Opbroek, M.A. Ikram, M.W. Vernooij, and M. de Bruijne. Transfer learning improves supervised image segmentation across imaging protocols. IEEE Transactions on Medical Imaging, 34(5):1018-1030, 2015. [ link ]
The variation between images obtained with different scanners or different imaging protocols presents a major challenge in automatic segmentation of biomedical images. This variation especially hampers the application of otherwise successful supervised-learning techniques which, in order to perform well, often require a large amount of labeled training data that is exactly representative of the target data. We therefore propose to use transfer learning for image segmentation. Transfer-learning techniques can cope with differences in distributions between training and target data, and therefore may improve performance over supervised learning for segmentation across scanners and scan protocols. We present four transfer classifiers that can train a classification scheme with only a small amount of representative training data, in addition to a larger amount of other training data with slightly different characteristics. The performance of the four transfer classifiers was compared to that of standard supervised classification on two MRI brain-segmentation tasks with multi-site data: white matter, gray matter, and CSF segmentation; and white-matter- /MS-lesion segmentation. The experiments showed that when there is only a small amount of representative training data available, transfer learning can greatly outperform common supervised-learning approaches, minimizing classification errors by up to 60%.

 
Tim Rosenow, Merel C J. Oudraad, Conor P. Murray, Lidija Turkovic, Wieying Kuo, Marleen de Bruijne, Sarath C. Ranganathan, Harm A W M. Tiddens, Stephen M. Stick, and AREST CF. PRAGMA-CF: a quantitative structural lung disease CT outcome in young children with cystic fibrosis. Am. J. Respir. Crit. Care Med., Mar 2015.
Chest computed tomography (CT) is the gold standard for demonstrating cystic fibrosis (CF) airways disease. However, there are no standardised outcome measures appropriate for children under 6 years.We developed the Perth-Rotterdam Annotated Grid Morphometric Analysis for CF (PRAGMA-CF), a quantitative measure of airways disease, and compared it to the commonly used CF-CT scoring method.CT scans from the Australian Respiratory Early Surveillance Team for CF (AREST CF) cohort in Western Australia were included. PRAGMA-CF was performed by annotating a grid overlaid on ten axial slices for the presence of bronchiectasis, mucous plugging or other airway abnormalities (inspiratory scans) and trapped air (expiratory scans). The separate proportions of total disease (%Dis), bronchiectasis (%Bx) and trapped air (%TA) were determined. Thirty scans were used for observer reliability, and thirty paired scans obtained at 1 and 3-years old were used for comparison with a validated standard and biological plausibility.Intraobserver, intraclass correlation coefficients (95% confidence interval) for %Dis, %Bx and %TA were 0.93 (0.86 - 0.97), 0.93 (0.85 - 0.96) and 0.96 (0.91 - 0.98), respectively. The change in %Dis (p = 0.004) and %Bx (p = 0.001) with PRAGMA-CF was related to neutrophil elastase (NE) presence at age 3, whereas only the change in bronchiectasis score was related to NE (p < 0.001) with CF-CT. Sample size calculations for various effect sizes are presented.PRAGMA-CF is a sensitive and reproducible outcome measure for assessing the extent of lung disease in very young children with CF.

 
Arna van Engelen, Anouk C. van Dijk, Martine T.B. Truijman, Ronald van ’t Klooster, Annegreet van Opbroek, Aad van der Lugt, Wiro J. Niessen, M. Eline Kooi, and Marleen de Bruijne. Multi-center MRI carotid plaque component segmentation using feature normalization and transfer learning. IEEE Transactions on Medical Imaging, 2014. In Press. [ link ]
Automated segmentation of plaque components in carotid artery MRI is important to enable large studies on plaque vulnerability, and for incorporating plaque composition as an imaging biomarker in clinical practice. Especially supervised classification techniques, which learn from labeled examples, have shown good performance. However, a disadvantage of supervised methods is their reduced performance on data different from the training data, for example on images acquired with different scanners. Reducing the amount of manual annotations required for each new dataset will facilitate widespread implementation of supervised methods. In this paper we segment carotid plaque components of clinical interest (fibrous tissue, lipid tissue, calcification and intraplaque hemorrhage) in a multicenter MRI study. We perform voxelwise tissue classification by traditional same-center training, and compare results with two approaches that use little or no annotated same-center data. These approaches additionally use an annotated set of differentcenter data. We evaluate 1) a non-linear feature normalization approach, and 2) two transfer-learning algorithms that use same and different-center data with different weights. Results showed that the best results were obtained for a combination of feature normalization and transfer learning. While for the other approaches significant differences in voxelwise or mean volume errors were found compared with the reference samecenter training, the proposed approach did not yield significant differences from that reference. We conclude that both extensive feature normalization and transfer learning can be valuable for the development of supervised methods that perform well on different types of datasets.

 
J. Petersen, M.M.W. Wille, L.L. Rakêt, A. Feragen, J.H. Pedersen, M. Nielsen, A. Dirksen, and M. de Bruijne. Effect of inspiration on airway dimensions measured in maximal inspiration CT images of subjects without airflow limitation. European Radiology, 24(9):2319-2325, 2014. [ link ]
 
A. Feragen, J. Petersen, M. Owen, P. Lo, L.H. Thomsen, M.M.W. Wille, A. Dirksen, and M. de Bruijne. Geodesic atlas-based labeling of anatomical trees: Application and evaluation on airways extracted from CT. IEEE Transactions on Medical Imaging, 2014. In press. [ link ]
 
A van Engelen, T Wannarong, G Parraga, W.J. Niessen, A Fenster, J.D. Spence, and M. de Bruijne. Three-dimensional carotid ultrasound plaque texture predicts vascular events. Stroke, 45:2695-2701, 2014. [ link ]
Background and Purpose—Carotid ultrasound atherosclerosis measurements, including those of the arterial wall and plaque, provide a way to monitor patients at risk of vascular events. Our objective was to examine carotid ultrasound plaque texture measurements and the change in carotid plaque texture during 1 year in patients at risk of events and to compare these with measurements of plaque volume and other risk factors as predictors of vascular events.

Methods—We evaluated 298 patients with carotid atherosclerosis using 3-dimensional (3D) ultrasound at baseline and after 1 year and measured carotid plaque volume and 376 measures of plaque texture. Patients were followed up to 5 years (median [range], 3.12 [0.77–4.66]) for myocardial infarction, transient ischemic attack, and stroke. Sparse Cox regression was used to select the most predictive plaque texture measurements in independent training sets using a 10-fold cross-validation, repeated 5×, to ensure unbiased results.

Results—Receiver operator curves and Kaplan–Meier analysis showed that changes in texture and total plaque volume combined provided the best predictor of vascular events. In multivariate Cox regression, changes in plaque texture (median hazard ratio, 1.4; P<0.001) and total plaque volume (median hazard ratio, 1.5 per 100 mm3; P<0.001) were both significant predictors, whereas the Framingham risk score was not.

Conclusions—Changes in both plaque texture and volume are strongly predictive of vascular events. In high-risk patients, 3D ultrasound plaque measurements should be considered for vascular event risk prediction.

 
Hakim C. Achterberg, Fedde van der Lijn, Tom den Heijer, Meike W. Vernooij, M Arfan Ikram, Wiro J. Niessen, and Marleen de Bruijne. Hippocampal shape is predictive for the development of dementia in a normal, elderly population. Human Brain Mapping, 35(5):2359-2371, 2014. [ link ]
Previous studies have shown that hippocampal volume is an early marker for dementia. We investigated whether hippocampal shape characteristics extracted from MRI scans are predictive for the development of dementia during follow up in subjects who were nondemented at baseline. Furthermore, we assessed whether hippocampal shape provides additional predictive value independent of hippocampal volume. Five hundred eleven brain MRI scans from elderly nondemented participants of a prospective population-based imaging study were used. During the 10-year follow-up period, 52 of these subjects developed dementia. For training and evaluation independent of age and gender, a subset of 50 cases and 150 matched controls was selected. The hippocampus was segmented using an automated method. From the segmentation, the volume was determined and a statistical shape model was constructed. We trained a classifier to distinguish between subjects who developed dementia and subjects who stayed cognitively healthy. For all subjects the a posteriori probability to develop dementia was estimated using the classifier in a cross-validation experiment. The area under the ROC curve for volume, shape, and the combination of both were, respectively, 0.724, 0.743, and 0.766. A logistic regression model showed that adding shape to a model using volume corrected for age and gender increased the global model-fit significantly (P = 0.0063). We conclude that hippocampal shape derived from MRI scans is predictive for dementia before clinical symptoms arise, independent of age and gender. Furthermore, the results suggest that hippocampal shape provides additional predictive value over hippocampal volume and that combining shape and volume leads to better prediction. Hum Brain Mapp, 2013. © 2013 Wiley Periodicals, Inc.

 
A van Engelen, W.J. Niessen, S. Klein, H.C. Groen, H.J.M. Verhagen, J.J. Wentzel, A van der Lugt, and M. de Bruijne. Atherosclerotic plaque component segmentation in combined MRI and CTA data incorporating class label uncertainty. PLoS ONE, 2014. [ link ]
Atherosclerotic plaque composition can indicate plaque vulnerability. We segment atherosclerotic plaque components from the carotid artery on a combination of in vivo MRI and CT-angiography (CTA) data using supervised voxelwise classification. In contrast to previous studies the ground truth for training is directly obtained from 3D registration with histology for fibrous and lipid-rich necrotic tissue, and with CT for calcification. This registration does, however, not provide accurate voxelwise correspondence. We therefore evaluate three approaches that incorporate uncertainty in the ground truth used for training: I) soft labels are created by Gaussian blurring of the original binary histology segmentations to reduce weights at the boundaries between components, and are weighted by the estimated registration accuracy of the histology and in vivo imaging data (measured by overlap), II) samples are weighted by the local contour distance of the lumen and outer wall between histology and in vivo data, and III) 10% of each class is rejected by Gaussian outlier rejection. Classification was evaluated on the relative volumes (% of tissue type in the vessel wall) for calcified, fibrous and lipid-rich necrotic tissue, using linear discriminant (LDC) and support vector machine (SVM) classification. In addition, the combination of MRI and CTA data was compared to using only one imaging modality. Best results were obtained by LDC and outlier rejection: the volume error per vessel was 0.91.0% for calcification, 12.77.6% for fibrous and 12.18.1% for necrotic tissue, with Spearman rank correlation coefficients of 0.91 (calcification), 0.80 (fibrous) and 0.81 (necrotic). While segmentation using only MRI features yielded low accuracy for calcification, and segmentation using only CTA features yielded low accuracy for necrotic tissue, the combination of features from MRI and CTA gave good results for all studied components.

 
Jens Petersen, Mads Nielsen, Pechin Lo, Lars Haug Nordenmark, Jesper Holst Pedersen, Mathilde Marie Winkler Wille, Asger Dirksen, and Marleen de Bruijne. Optimal surface segmentation using flow lines to quantify airway abnormalities in chronic obstructive pulmonary disease. Medical Image Analysis, 18(3):531-541, 2014. [ link ]
This paper introduces a graph construction method for multi-dimensional and multi-surface segmentation problems. Such problems can be solved by searching for the optimal separating surfaces given the space of graph columns defined by an initial coarse surface. Conventional straight graph columns are not well suited for surfaces with high curvature, we therefore propose to derive columns from properly generated, non-intersecting flow lines. This guarantees solutions that do not self-intersect. The method is applied to segment human airway walls in computed tomography images in three-dimensions. Phantom measurements show that the inner and outer radii are estimated with sub-voxel accuracy. Two-dimensional manually annotated cross-sectional images were used to compare the results with those of another recently published graph based method. The proposed approach had an average overlap of 89.3±5.8%, and was on average within 0.096±0.097mm of the manually annotated surfaces, which is significantly better than what the previously published approach achieved. A medical expert visually evaluated 499 randomly extracted cross-sectional images from 499 scans and preferred the proposed approach in 68.5%, the alternative approach in 11.2%, and in 20.3% no method was favoured. Airway abnormality measurements obtained with the method on 490 scan pairs from a lung cancer screening trial correlate significantly with lung function and are reproducible; repeat scan R(2) of measures of the airway lumen diameter and wall area percentage in the airways from generation 0 (trachea) to 5 range from 0.96 to 0.73.

 
Ruwan Tennakoon, Alireza Bab-Hadiashar, Zhenwei Cao, and Marleen de Bruijne. Non-rigid registration of volumetric images using ranked order statistics. IEEE Transactions on Medical Imaging, 33:422 - 432, 2014. [ link ]
Non-rigid image registration techniques using intensity based similarity measures are widely used in medical imaging applications. Due to high computational complexities of these techniques, particularly for volumetric images, finding appropriate registration methods to both reduce the computation burden and increase the registration accuracy has become an intensive area of research. In this paper we propose a fast and accurate non-rigid registration method for intra-modality volumetric images. Our approach exploits the information provided by an order statistics based segmentation method, to find the important regions for registration and use an appropriate sampling scheme to target those areas and reduce the registration computation time. A unique advantage of the proposed method is its ability to identify the point of diminishing returns and stop the registration process. Our experiments on registration of endinhale to end-exhale lung CT scan pairs, with expert annotated landmarks, show that the new method is both faster and more accurate than the state of the art sampling based techniques, particularly for registration of images with large deformations.

 
N. Baka, B.L. Kaptein, J.E. Giphart, M. Staring, M. de Bruijne, B.P.F. Lelieveldt, and E. R. Valstar. Evaluation of automated statistical shape model based knee kinematics from biplane fluoroscopy. Journal of Biomechanics, 47(1):122-129, 2014. [ link ]
State-of-the-art fluoroscopic knee kinematic analysis methods require the patient-specific bone shapes segmented from CT or MRI. Substituting the patient-specific bone shapes with personalizable models, such as statistical shape models (SSM), could eliminate the CT/MRI acquisitions, and thereby decrease costs and radiation dose (when eliminating CT). SSM based kinematics, however, have not yet been evaluated on clinically relevant joint motion parameters. Therefore, in this work the applicability of SSM-s for computing knee kinematics from biplane fluoroscopic sequences was explored. Kinematic precision with an edge based automated bone tracking method using SSM-s was evaluated on 6 cadaver and 10 in-vivo fluoroscopic sequences. The SSMs of the femur and the tibia-fibula were created using 61 training datasets. Kinematic precision was determined for medial-lateral tibial shift, anterior-posterior tibial drawer, joint distraction-contraction, flexion, tibial rotation and adduction. The relationship between kinematic precision and bone shape accuracy was also investigated.The SSM based kinematics resulted in sub-millimeter (0.48-0.81 mm) and approximately one degree (0.69-0.99?) median precision on the cadaveric knees compared to bone-marker-based kinematics. The precision on the in-vivo datasets was comparable to the cadaveric sequences when evaluated with a semi-automatic reference method. These results are promising, though further work is necessary to reach the accuracy of CT-based kinematics. We also demonstrated that a better shape reconstruction accuracy does not automatically imply a better kinematic precision. This result suggests that the ability of accurately fitting the edges in the fluoroscopic sequences has a larger role in determining the kinematic precision than the overall 3D shape accuracy.

 
P. Ciet, P. Wielopolski, R. Manniesing, S. Lever, M. de Bruijne, G. Morana, P.C. Muzzio, M.H. Lequin, and H.A.W.M. Tiddens. Spirometer controlled cine-magnetic resonance imaging to diagnose tracheobronchomalacia in pediatric patients. European Respiratory Journal, 43(1):115-124, 2014. [ link ]
Tracheobronchomalacia (TBM) is defined as an excessive collapse of the intrathoracic trachea. Bronchoscopy is the gold standard to diagnose TBM, but bronchoscopy has major disadvantages, such as general anaesthesia. Cine-CT is a non-invasive alternative to diagnose TBM, but its use in children is restricted by ionizing radiation. Our aim was to evaluate the feasibility of spirometer-controlled cine-MRI as alternative to cine-CT in a retrospective study. 12 children (mean 12 years, range 7–17), suspected to have TBM, underwent cine-MRI. Static scans were acquired at end-inspiration and expiration covering the thorax using a 3D SPGR sequence. 3D-Dynamic-scans were performed covering only the central airways. TBM was defined as a decrease of the trachea or bronchi diameter greater than 50 in the static and dynamic scans. The success rate of the cine-MRI protocol was 92 with bronchoscopy or chest-CT in 7 subjects. TBM was diagnosed by cine-MRI in 7 out of 12 children (58 or CT. In 4 patients, cine-MRI demonstrated tracheal narrowing that was not present in the static scans. Spirometer-controlled cine–MRI is a promising technique to assess TBM in children and has the potential to replace bronchoscopy.

 
M. Loeve, G. P. Krestin, M. Rosenfeld, M. de Bruijne, S. Stick, and H.A.W.M. Tiddens. Chest computed tomography; a validated surrogate endpoint of cystic fibrosis lung disease? European Respiratory Journal, 42:844-857, 2013. [ link ]
Clinical trials for the treatment of cystic fibrosis (CF) lung disease are important to test and optimize new therapeutic interventions. To evaluate the effect of these interventions, sensitive and accurate outcome measures are needed. The most commonly used endpoints are spirometric variables such as the forced expiratory volume in one second (FEV1) and respiratory tract exacerbations (RTE). Unfortunately, these endpoints are relatively insensitive to monitor progression of CF lung disease, and thus require a large number of patients when used in clinical studies. In addition, these endpoints are not suitable to study CF lung disease in young children. Chest computed tomography (CT) holds great promise for use as a sensitive surrogate endpoint in CF. A large body of evidence has been produced to validate the use of chest CT as primary endpoint to study CF lung disease. However, before chest CT can be used in clinical trials, it has to be recognized as a validated surrogate endpoint by regulatory agencies. The aim of this review is to summarize what is currently known about the use of chest CT as surrogate endpoint in clinical trials in CF.

 
N. Baka, C. T. Metz, C. Schultz, L. Neefjes, R. J. van Geuns, B P F. Lelieveldt, W. J. Niessen, T. van Walsum, and M. de Bruijne. Statistical coronary motion models for 2D+t/3D registration of X-ray coronary angiography and CTA. Medical Image Analysis, 17(6):698-709, 2013. [ link ]
Accurate alignment of intra-operative X-ray coronary angiography (XA) and pre-operative cardiac CT angiography (CTA) may improve procedural success rates of minimally invasive coronary interventions for patients with chronic total occlusions. It was previously shown that incorporating patient specific coronary motion extracted from 4D CTA increases the robustness of the alignment. However, pre-operative CTA is often acquired with gating at end-diastole, in which case patient specific motion is not available. For such cases, we investigate the possibility of using population based coronary motion models to provide constraints for the 2D+t/3D registration. We propose a methodology for building statistical motion models of the coronary arteries from a training population of 4D CTA datasets. We compare the 2D+t/3D registration performance of the proposed statistical models with other motion estimates, including the patient specific motion extracted from 4D CTA, the mean motion of a population, the predicted motion based on the cardiac shape. The coronary motion models, constructed on a training set of 150 patients, had a generalization accuracy of 1mm root mean square point-to-point distance. Their 2D+t/3D registration accuracy on one cardiac cycle of 12 monoplane XA sequences was similar to, if not better than, the 4D CTA based motion, irrespective of which respiratory model and which feature based 2D/3D distance metric was used. The resulting model based coronary motion estimate showed good applicability for registration of a subsequent cardiac cycle.

 
A. Feragen, P. Lo, M. de Bruijne, M. Nielsen, and F. Lauze. Towards a theory of statistical tree-shape analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence, 35:2008 - 2021, 2013. [ link ]
To develop statistical methods for shapes with a tree-structure, we construct a shape space framework for tree-shapes and study metrics on the shape space. This shape space has singularities which correspond to topological transitions in the represented trees. We study two closely related metrics on the shape space, TED and QED. QED is a quotient euclidean distance arising naturally from the shape space formulation, while TED is the classical tree edit distance. Using Gromov's metric geometry, we gain new insight into the geometries defined by TED and QED. We show that the new metric QED has nice geometric properties that are needed for statistical analysis: Geodesics always exist and are generically locally unique. Following this, we can also show the existence and generic local uniqueness of average trees for QED. TED, while having some algorithmic advantages, does not share these advantages. Along with the theoretical framework we provide experimental proof-of-concept results on synthetic data trees as well as small airway trees from pulmonary CT scans. This way, we illustrate that our framework has promising theoretical and qualitative properties necessary to build a theory of statistical tree-shape analysis

 
A van Engelen, W.J. Niessen, S. Klein, H.C. Groen, K. van Gaalen, H.J.M. Verhagen, J.J. Wentzel, A van der Lugt, and M. de Bruijne. Automated segmentation of atherosclerotic histology based on pattern classification. Journal of Pathology Informatics, 4, 2013. [ link ]
Background: Histology sections provide accurate information on atherosclerotic plaque composition, and are used in various applications. To our knowledge, no automated systems for plaque component segmentation in histology sections currently exist. Materials and Methods: We perform pixel-wise classification of fibrous, lipid, and necrotic tissue in Elastica Von Gieson-stained histology sections, using features based on color channel intensity and local image texture and structure. We compare an approach where we train on independent data to an approach where we train on one or two sections per specimen in order to segment the remaining sections. We evaluate the results on segmentation accuracy in histology, and we use the obtained histology segmentations to train plaque component classification methods in ex vivo Magnetic resonance imaging (MRI) and in vivo MRI and computed tomography (CT). Results: In leave-one-specimen-out experiments on 176 histology slices of 13 plaques, a pixel-wise accuracy of 75.7 ± 6.8 This increased to 77.6 ± 6.5 of the specimen to be segmented were used for training. Rank correlations of relative component volumes with manually annotated volumes were high in this situation (P = 0.82-0.98). Using the obtained histology segmentations to train plaque component classification methods in ex vivo MRI and in vivo MRI and CT resulted in similar image segmentations for training on the automated histology segmentations as for training on a fully manual ground truth. The size of the lipid-rich necrotic core was significantly smaller when training on fully automated histology segmentations than when manually annotated histology sections were used. This difference was reduced and not statistically significant when one or two slices per section were manually annotated for histology segmentation. Conclusions: Good histology segmentations can be obtained by automated segmentation, which show good correlations with ground truth volumes. In addition, these can be used to develop segmentation methods in other imaging modalities. Accuracy increases when one or two sections of the same specimen are used for training, which requires a limited amount of user interaction in practice.

 
A. Bab-Hadiashar, R. Tennakoon, and M. de Bruijne. Quantification of smoothing requirement for 3D optic flow calculation of volumetric images. IEEE Transactions on Image Processing, 22:2128 - 2137, 2013. [ link ]
Complexities of dynamic volumetric imaging challenge the available computer vision techniques on a number of different fronts. This paper examines the relationship between the estimation accuracy and required amount of smoothness for a general solution from a robust statistics perspective. We show that a (surprisingly) small amount of local smoothing is required to satisfy both the necessary and sufficient conditions for accurate optic flow estimation. This notion is called “just enough” smoothing, and its proper implementation has a profound effect on the preservation of local information in processing 3D dynamic scans. To demonstrate the effect of “just enough” smoothing, a robust 3D optic flow method with quantized local smoothing is presented, and the effect of local smoothing on the accuracy of motion estimation in dynamic lung CT images is examined using both synthetic and real image sequences with ground truth.

 
P. Lo, B. van Ginneken, J.M. Reinhardt, Tarunashree Y., P.A. de Jong, B. Irving, C. Fetita, M. Ortner, R. Pinho, J. Sijbers, M. Feuerstein, A. Fabijanska, C. Bauer, R. Beichel, C. S. Mendoza, R. Wiemker, J. Lee, A. P. Reeves, S. Born, O. Weinheimer, E. M. van Rikxoort, J. Tschirren, K. Mori, B. Odry, D.P. Naidich, I. Hartmann, E.A. Hoffman, M. Prokop, J.H. Pedersen, and M. de Bruijne. Extraction of airways from CT (EXACT’09). IEEE Transactions on Medical Imaging, 31(11):2093-2107, 2012. [ link | pdf ]
This paper describes a framework for establishing a reference airway tree segmentation, which was used to quantitatively evaluate fifteen different airway tree extraction algorithms in a standardized manner. Because of the sheer difficulty involved in manually constructing a complete reference standard from scratch, we propose to construct the reference using results from all algorithms that are to be evaluated. We start by subdividing each segmented airway tree into its individual branch segments. Each branch segment is then visually scored by trained observers to determine whether or not it is a correctly segmented part of the airway tree. Finally, the reference airway trees are constructed by taking the union of all correctly extracted branch segments. Fifteen airway tree extraction algorithms from different research groups are evaluated on a diverse set of twenty chest computed tomography (CT) scans of subjects ranging from healthy volunteers to patients with severe pathologies, scanned at different sites, with different CT scanner brands, models, and scanning protocols. Three performance measures covering different aspects of segmentation quality were computed for all participating algorithms. Results from the evaluation showed that no single algorithm could extract more than an average of 74 reference standard, indicating substantial differences between the algorithms. A fusion scheme that obtained superior results is presented, demonstrating that there is complementary information provided by the different algorithms and there is still room for further improvements in airway segmentation algorithms.

 
S.B. Shaker, A. Dirksen, P. Lo, L.T. Skovgaard, M. de Bruijne, and J.H. Pedersen. Factors influencing decline in lung density in a Danish lung cancer screening cohort. European Respiratory Journal, 40(5):1142-1148, 2012. [ link ]
Lung cancer screening trials provide an opportunity to study the natural history of emphysema by using CT lung density as a surrogate parameter.In the Danish Lung Cancer Screening Trial, 2,052 participants were included. At screening rounds, smoking habits were recorded and spirometry was performed. CT lung density was measured as the volume-adjusted 15th percentile density (PD15). A mixed effects model was used with former smoking males with <30 pack-years and without airflow obstruction (AFO) at entry as a reference group. At study entry, 893 (44 had AFO. For the reference group, PD15 was 72.6 g·l-1 with an annual decline of -0.33 g·l-1. Female sex and current smoking increased PD15 at baseline, 17.3 g·l-1 (p<0.001) and 10 g/l (p<0.001), respectively; and both increased the annual decline in PD15 (female: -0.3 g/l; current smoking: -0.4 g/l). The presence and severity of AFO was a strong predictor of low PD15 at baseline (GOLD I: -1.4 g/l; GOLD II: -6.3 g/l; GOLD III: -17 g/l) and of increased annual decline in PD15 (GOLD I: -0.2 g/l; GOLD II: -0.5 g/l; GOLD III: -0.5 g/l).

 
Fan Liu, Fedde van der Lijn, Claudia Schurmann, Gu Zhu, M. Mallar Chakravarty, Pirro G. Hysi, Andreas Wollstein, Oscar Lao, Marleen de Bruijne, M. Arfan Ikram, Aad van der Lugt, Fernando Rivadeneira, Andre. Uitterlinden, Albert Hofman, Wiro J. Niessen, Georg Homuth, Greig de Zubicaray, Katie L. McMahon, Paul M. Thompson, Amro Daboul, Ralf Puls, Katrin Hegenscheid, Liisa Bevan, Zdenka Pausova, Sarah E. Medland, Grant W. Montgomery, Margaret J. Wright, Carol Wicking, Stefan Boehringer, Timothy D. Spector, Tomas Paus, Nicholas G. Martin, Reiner Biffar, and Manfred Kayser. A genome-wide association study identifies five loci influencing facial morphology in europeans. PLoS Genetics, 8(9):e1002932, 2012. [ link ]
Monozygotic twins look more alike than dizygotic twins or other siblings, and siblings in turn look more alike than unrelated individuals, indicating that human facial morphology has a strong genetic component. We quantitatively assessed human facial shape phenotypes based on statistical shape analyses of facial landmarks obtained from three-dimensional magnetic resonance images of the head. These phenotypes turned out to be highly promising for studying the genetic basis of human facial variation in that they showed high heritability in our twin data. A subsequent genome-wide association study (GWAS) identified five candidate genes affecting facial shape in Europeans: PRDM16, PAX3, TP63, C5orf50, and COL17A1. In addition, our data suggest that genetic variants associated with NSCL/P also influence normal facial shape variation. Overall, this study provides novel and confirmatory links between common DNA variants and normal variation in human facial morphology. Our results also suggest that the high heritability of facial phenotypes seems to be explained by a large number of DNA variants with relatively small individual effect size, a phenomenon well known for other complex human traits, such as adult body height.

 
N. Baka, M. de Bruijne, T. van Walsum, B.L. Kaptein, J.E. Giphart, M. Schaap, W.J. Niessen, and B.P.F. Lelieveldt. Statistical shape model based femur kinematics from bi-plane fluoroscopy. IEEE Transactions on Medical Imaging, 31(8), 2012. [ link ]
Studying joint kinematics is of interest to improve prosthesis design and to characterize post-operative motion. State of the art techniques register bones segmented from prior CT or MR scans with X-ray fluoroscopic sequences. Elimination of the prior 3D acquisition could potentially lower costs and radiationdose. Therefore, we propose to substitute the segmented bone surface with a statistical shape model based estimate. A dedicated dynamic reconstruction and tracking algorithm was developed estimating the shape based on all frames, and pose per frame. Thealgorithm minimizes the difference between the projected bone contour and image edges. To increase robustness, we employ a dynamic prior, image features, and prior knowledge about bone edge appearances. This enables tracking and reconstruction from a single initial pose per sequence. We evaluated our method on the distal femur using eight biplane fluoroscopic drop-landing sequences. The proposed dynamic prior and features increased the convergence rate of the reconstruction from 71 limit of 3 mm. The achieved root-mean-square point-to-surface accuracy at the converged frames was 1.48 +/- 0.41 mm. The resulting trackingprecision was 1-1.5 millimeter, with the largest errors occurring in the rotation around the femoral shaft (about 2.5°precision).

 
X. Wang, T. Heimann, P. Lo, M. Sumkauskaite, M. Puderbach, M. de Bruijne, H.-P. Meinzer, and I. Wegner. Statistical tracking of tree-like tubular structures with efficient branching detection in 3D medical image data. Physics in Medicine and Biology, 57(16):5325-42, 2012. [ link ]
The segmentation of tree-like tubular structures such as coronary arteries and airways is an essential step for many 3D medical imaging applications. Statistical tracking techniques for the extraction of elongated structures have received considerable attention in recent years due to their robustness against image noise and pathological changes. However, most tracking methods are limited to a specific application and do not support branching structures efficiently. In this work, we present a novel statistical tracking approach for the extraction of different types of tubular structures with ringlike cross-sections. Domain-specific knowledge is learned from training data sets and integrated into the tracking process by simple adaption of parameters. In addition, an efficient branching detection algorithm is presented. This approach was evaluated by extracting coronary arteries from 32 CTA data sets and distal airways from 20 CT scans. These data sets were provided by the organizers of the workshop '3D Segmentation in the Clinic: A Grand Challenge II-Coronary Artery Tracking (CAT08)' and 'Extraction of Airways from CT 2009 (EXACT'09)'. On average, 81.5 coronary arteries were achieved. For the extraction of airway trees, 51.3 and a 4.98 our approach is comparable to state-of-the-art methods.

 
M. Ganz, M de Bruijne, E. Dam, P. Pettersen, M. Karsdal, C. Christiansen, and M Nielsen. Distribution, size, and shape of abdominal aortic calcified deposits and their relationship to mortality in postmenopausal women. International Journal of Biomedical Imaging, 2012. [ link ]
Abdominal aortic calcifications (AACs) correlate strongly with coronary artery calcifications and can be predictors of cardiovascular mortality. We investigated whether size, shape, and distribution of AACs are related to mortality and how such prognostic markers perform compared to the state-of-the-art AC24 marker introduced by Kauppila. Methods. For 308 postmenopausal women, we quantified the number of AAC and the percentage of the abdominal aorta that the lesions occupied in terms of their area, simulated plaque area, thickness, wall coverage, and length. We analysed inter-/intraobserver reproducibility and predictive ability of mortality after 8-9 years via Cox regression leading to hazard ratios (HRs). Results. The coefficient of variation was below 25 were the number of calcifications (HR=2.4) and the simulated area percentage (HR=2.96) of a calcified plaque, and, unlike AC24 (HR=1.66), they allowed mortality prediction also after adjusting for traditional risk factors. In a combined Cox regression model, the strongest complementary predictors were the number of calcifications (HR=2.76) and the area percentage (HR=-3 .84). Conclusion. Morphometric markers of AAC quantified from radiographs may be a useful tool for screening and monitoring risk of CVD mortality.

 
C.T. Metz, N. Baka, H. Kirisli, , M. Schaap, S. Klein, L.A. Neefjes, N.R. Mollet, B. Lelieveldt, M. de Bruijne, W.J. Niessen, and T. van Walsum. Regression-based cardiac motion prediction from single-phase CTA. IEEE Transactions on Medical Imaging, 31(6), 2012. [ link ]
State of the art cardiac CT enables the acquisition of imaging data of the heart over the entire cardiac cycle at concurrent high spatial and temporal resolution. However, in clinical practice, acquisition is increasingly limited to 3D images. Estimating the shape of the cardiac structures throughout the entire cardiac cycle from a 3D image is therefore useful in applications such as the alignment of preoperative CTA to intra-operative X-ray images for improved guidance in coronary interventions. We hypothesize that the motion of the heart is partially explained by its shape and therefore investigate the use of three regression methods for motion estimation from single-phase shape information. Quantitative evaluation on 150 4D computed tomography angiography (CTA) images showed a small, but statistically significant, increase in the accuracy of the predicted shape sequences when using any of the regression methods, compared to shape-independent motion prediction by application of the mean motion. The best results were achieved using principal component regression resulting in point-to-point errors of 2.3 ± 0.5 mm, compared to values of 2.7 ± 0.6 mm for shape-independent motion estimation. Finally, we showed that this significant difference withstands small variations in important parameter settings of the landmarking procedure.

 
V. Gorbunova, J. Sporring, P. Lo, M. Loeve, H. Tiddens, M Nielsen, A. Dirksen, and M de Bruijne. Mass preserving image registration for lung CT. Medical Image Analysis, 16(4):786 - 795, 2012. [ link | pdf ]
This paper presents a mass preserving image registration algorithm for lung CT images. To account for the local change in lung tissue intensity during the breathing cycle, a tissue appearance model based on the principle of preservation of total lung mass is proposed. This model is incorporated into a standard image registration framework with a composition of a global affine and several free-form B-Spline transformations with increasing grid resolution. The proposed mass preserving registration method is compared to registration using the sum of squared intensity differences as a similarity function on four groups of data: 44 pairs of longitudinal inspiratory chest CT scans with small difference in lung volume; 44 pairs of longitudinal inspiratory chest CT scans with large difference in lung volume; 16 pairs of expiratory and inspiratory CT scans; and 5 pairs of images extracted at end exhale and end inhale phases of 4D-CT images. Registration errors, measured as the average distance between vessel tree centerlines in the matched images, are significantly lower for the proposed mass preserving image registration method in the second, third and fourth group, while there is no statistically significant difference between the two methods in the first group. Target registration error, assessed via a set of manually annotated landmarks in the last group, was significantly smaller for the proposed registration method.

 
M. Loeve, M. de Bruijne, I.C.J. Hartmann, M. van Straten, W.C.J. Hop, and H.A.W.M. Tiddens. Three-section expiratory CT: Insufficient for trapped air assessment in patients with cystic fibrosis? Radiology, 262(3):969-976, 2012. [ link ]
Purpose: To estimate the effect of the number of computed tomography (CT) sections on trapped air (TA) assessment in patients with cystic fibrosis (CF) by using an established scoring system and a new quantitative scoring system and to compare CT and pulmonary function test (PFT) estimates of TA in a cross-sectional and longitudinal study.Materials and Methods: In this institutional review board-approved pilot study, 20 subjects aged 6-20 years (12 female and eight male; median age, 12.6 years) contributed two expiratory CT studies (three-section baseline CT, volumetric follow-up CT) and two PFT studies over 2 years after parental informed consent was obtained. From follow-up CT studies, seven sets were composed: Set 1 was volumetric. Sets 2, 3, 4, and 5, had spacing of 2.4, 4.8, 9.6, and 20.4 mm, respectively, between sections. Sets 6 and 7 contained five and three sections, respectively. Longitudinal follow-up was performed with three sections. All images were deidentified and randomized, and TA was scored with the Brody II system and a new quantitative system. Statistical analysis included the Wilcoxon signed rank test, calculation of Spearman and intraclass correlation coefficients, and use of three-section and linear mixed models.Results: For the Brody II system, the intraclass correlation coefficient for set 1 versus those for sets 2 through 7 was 0.75 versus 0.87; however, mean scores from sets 6 and 7 were significantly lower than the mean score from set 1 (P = .01 and P < .001, respectively). For the quantitative system, the number of sections did not affect TA assessment (intraclass correlation coefficient range, 0.82-0.88; P > .13 for all). CT and PFT estimates were not correlated (rs = 20.19 to 0.09, P = .43-.93). No change in TA over time was found for CT or PFT (P > .16 for all).Conclusion: The number of sections affected Brody II estimates, suggesting that three-section protocols lead to underestimation of TA assessment in patients with CF when using the Brody II system; CT and PFT estimates of TA showed no correlation and no significant change over time.

 
M. Loeve, W.C.J. Hop, M. de Bruijne, P.Th.W. van Hal, P. Robinson, M.L. Aitken, J.D. Dodd, H.A.W.M. Tiddens, and on behalf of the Computed Tomography Cystic Fibrosis Survival study group. Chest computed tomography scores are predictive of survival in CF patients awaiting lung transplantation. American Journal of Respiratory and Critical Care Medicine, 2012. [ link ]
Rationale: Up to a third of cystic fibrosis (CF) patients awaiting lung transplantation (LTX) die while waiting. Inclusion of computed tomography (CT) scores may improve survival prediction models such as the lung allocation score (LAS). Objectives: This study investigated the association between CT and survival in CF patients screened for LTX. Methods: Clinical data and chest CTs of 411 CF patients screened for LTX between 1990 and 2005 were collected from 17 centers. CTs were scored with the Severe Advanced Lung Disease (SALD) 4-category scoring system, including the components "infection/inflammation" (INF), air trapping/hypoperfusion (AT), normal/hyperperfusion (NOR) and bulla/cysts (BUL). The volume of each component was computed using semi-automated software. Survival analysis included Kaplan-Meier curves, and Cox-regression models. Measurements and main results: 366 (186 males) out of 411 patients entered the waiting list (median age 23, range 5-58 years). Subsequently, 67/366(18 263/366(72 the census date. INF and LAS were significantly associated with waiting list mortality in univariate analyses. The multivariate Cox model including INF and LAS grouped in tertiles and comparing tertiles 2 and 3 to tertile 1, showed waiting list mortality hazard ratios of 1.62 (95 for INF and 1.42 (0.63-3.24, p=0.40), and 2.32 (1.17-4.60, p=0.016) for LAS, respectively. These results indicated that INF and LAS had significant, independent predictive value for survival. Conclusions: CT score INF correlates with survival, and adds to the predictive value of LAS.

 
K. Murphy, J.P.W. Pluim, E.M. van Rikxoort, P.A. de Jong, B. de Hoop, H.A. Gietema, O. Mets, M. de Bruijne, P. Lo, M. Prokop, and B. van Ginneken. Toward automatic regional analysis of pulmonary function using inspiration and expiration thoracic CT. Medical Physics, 39(3):1650-1662, 2012. [ link ]
Purpose: To analyze pulmonary function using a fully automatic technique which processes pairs of thoracic CT scans acquired at breath-hold inspiration and expiration, respectively. The following research objectives are identified to: (a) describe and systematically analyze the processing pipeline and its results; (b) verify that the quantitative, regional ventilation measurements acquired through CT are meaningful for pulmonary function analysis; (c) identify the most effective of the calculated measurements in predicting pulmonary function; and (d) demonstrate the potential of the system to deliver clinically important information not available through conventional spirometry. Methods: A pipeline of automatic segmentation and registration techniques is presented and demonstrated on a database of 216 subjects well distributed over the various stages of COPD (chronic obstructive pulmonary disorder). Lungs, fissures, airways, lobes, and vessels are automatically segmented in both scans and the expiration scan is registered with the inspiration scan using a fully automatic nonrigid registration algorithm. Segmentations and registrations are examined and scored by expert observers to analyze the accuracy of the automatic methods. Quantitative measures representing ventilation are computed at every image voxel and analyzed to provide information about pulmonary function, both globally and on a regional basis. These CT derived measurements are correlated with results from spirometry tests and used as features in a kNN classifier to assign COPD global initiative for obstructive lung disease (GOLD) stage. Results: The steps of anatomical segmentation (of lungs, lobes, and vessels) and registration in the workflow were shown to perform very well on an individual basis. All CT-derived measures were found to have good correlation with spirometry results, with several having correlation coefficients, r, in the range of 0.85–0.90. The best performing kNN classifier succeeded in classifying 67 into the correct COPD GOLD stage, with a further 29 a class neighboring the correct one. Conclusions: Pulmonary function information can be obtained from thoracic CT scans using the automatic pipeline described in this work. This preliminary demonstration of the system already highlights a number of points of clinical importance such as the fact that an inspiration scan alone is not optimal for predicting pulmonary function. It also permits measurement of ventilation on a per lobe basis which reveals, for example, that the condition of the lower lobes contributes most to the pulmonary function of the subject. It is expected that this type of regional analysis will be instrumental in advancing the understanding of multiple pulmonary diseases in the future.

Keywords: computerised tomography; diseases; image classification; image registration; image segmentation; lung; medical image processing; pneumodynamics
 
A. Crimi, M. Loog, M. de Bruijne, M. Nielsen, and M. Lillholm. Shape-based assessment of vertebral fracture risk in postmenopausal women using discriminative shape alignment. Academic Radiology, 19(4):446 - 454, 2012. [ link ]
Rationale and Objectives: Risk assessment of future osteoporotic vertebral fractures is currently based mainly on risk factors, such as bone mineral density, age, prior fragility fractures, and smoking. It can be argued that an osteoporotic vertebral fracture is not exclusively an abrupt event but the result of a decaying process. To evaluate fracture risk, a shape-based classifier, identifying possible small prefracture deformities, may be constructed. Materials and Methods: During a longitudinal case-control study, a large population of postmenopausal women, fracture free at baseline, were followed. The 22 women who sustained at least one lumbar fracture on follow-up represented the case group. The control group comprised 91 women who maintained skeletal integrity and matched the case group according to the standard osteoporosis risk factors. On radiographs, a radiologist and two technicians independently performed manual annotations of the vertebrae, and fracture prediction using shape features extracted from the baseline annotations was performed. This was implemented using posterior probabilities from a standard linear classifier. Results: The classifier tested on the study population quantified vertebral fracture risk, giving statistically significant results for the radiologist annotations (area under the curve, 0.71 ± 0.013; odds ratio, 4.9; 95 Conclusions: The shape-based classifier provided meaningful information for the prediction of vertebral fractures. The approach was tested on case and control groups matched for osteoporosis risk factors. Therefore, the method can be considered an additional biomarker, which combined with traditional risk factors can improve population selection (eg, in clinical trials), identifying patients with high fracture risk.

 
F. van der Lijn, M. de Bruijne, S. Klein, T. den Heijer, Y.Y. Hoogendam, A. van der Lugt, M.M.B. Breteler, and W.J. Niessen. Automated brain structure segmentation based on atlas registration and appearance models. IEEE Transactions on Medical Imaging, 31(2):276 - 286, 2012. [ link | pdf ]
Accurate automated brain structure segmentation methods facilitate the analysis of large-scale neuroimaging studies. This work describes a novel method for brain structure segmentation in magnetic resonance images that combines information about a structures location and appearance. The spatial model is implemented by registering multiple atlas images to the target image and creating a spatial probability map. The structures appearance is modeled by a classifier based on Gaussian scale-space features. These components are combined with a regularization term in a Bayesian framework that is globally optimized using graph cuts. The incorporation of the appearance model enables the method to segment structures with complex intensity distributions and increases its robustness against errors in the spatial model. The method is tested in crossvalidation experiments on two datasets acquired with different magnetic resonance sequences, in which the hippocampus and cerebellum were segmented by an expert. Furthermore, the method is compared to two other segmentation techniques that were applied to the same data. Results show that the atlas and appearance-based method produces accurate results with mean Dice similarity indices of 0.95 for the cerebellum, and 0.87 for the hippocampus. This was comparable to or better than the other methods, whereas the proposed technique is more widely applicable and robust.

 
F. van der Lijn, B.F.J. Verhaaren, M.A. Ikram, S. Klein, M. de Bruijne, H.A. Vrooman, M.W. Vernooij, A. Hammers, D. Rueckert, A. van der Lugt, M.M.B. Breteler, and W.J. Niessen. Automated measurement of local white matter lesion volume. NeuroImage, 59(4):3901-3908, 2012. [ link ]
It has been hypothesized that white matter lesions at different locations may have different etiology and clinical consequences. Several approaches for the quantification of local white matter lesion load have been proposed in the literature, most of which rely on a distinction between lesions in a periventricular region close to the ventricles and a subcortical zone further away. In this work we present a novel automated method for local white matter lesion volume quantification in magnetic resonance images. The method segments and measures the white matter lesion volume in 43 regions defined by orientation and distance to the ventricles, which allows a more spatially detailed study of lesion load. The potential of the method was demonstrated by analyzing the effect of blood pressure on the regional white matter lesion volume in 490 elderly subjects taken from a longitudinal population study. The method was also compared to two commonly used techniques to assess the periventricular and subcortical lesion load. The main finding was that high blood pressure was primarily associated with lesion load in the vascular watershed area that forms the border between the periventricular and subcortical regions. It explains the associations found for both the periventricular and subcortical load computed for the same data, and that were reported in the literature. But the proposed method can localize the region of association with greater precision than techniques that distinguish between periventricular and subcortical lesions only.

 
A. van Engelen, W.J. Niessen, S. Klein, H.C. Groen, H.J.M. Verhagen, J.J. Wentzel, A. van der Lugt, and M. de Bruijne. Multi-feature-based plaque characterization in ex vivo MRI trained by registration to 3D histology. Physics in Medicine and Biology, 57(1):241-256, 2012. [ link ]
We present a new method for automated characterization of atherosclerotic plaque composition in ex vivo MRI. It uses MRI intensities as well as four other types of features: smoothed, gradient magnitude and Laplacian images at several scales, and the distances to the lumen and outer vessel wall. The ground truth for fibrous, necrotic and calcified tissue was provided by histology and micro-CT in 12 carotid plaque specimens. Semi-automatic registration of a 3D stack of histological slices and micro-CT images to MRI allowed for 3D rotations and inplane deformations of histology. By basing voxelwise classification on different combinations of features, we evaluated their relative importance. To establish whether training by 3D registration yields different results than training by 2D registration, we determined plaque composition using (1) a 2D slice-based registration approach for three manually selected MRI and histology slices per specimen, and (2) an approach that uses only the three corresponding MRI slices from the 3D-registered volumes. Voxelwise classification accuracy was best when all features were used (73.3 ± 6.3%) and was significantly better than when only original intensities and distance features were used (Friedman, p < 0.05). Although 2D registration or selection of three slices from the 3D set slightly decreased accuracy, these differences were non-significant.

 
L. Sørensen, M. Nielsen, P. Lo, H. Ashraf, J.J.H. Pedersen, and M. de Bruijne. Texture-based analysis of COPD: a data-driven approach. IEEE Transactions on Medical Imaging, 31(1):70 - 78, 2012. [ link ]
This study presents a fully automatic, data-driven approach for texture-based quantitative analysis of chronic obstructive pulmonary disease (COPD) in pulmonary computed tomography (CT) images. The approach uses supervised learning where the class labels are, in contrast to previous work, based on measured lung function instead of on manually annotated regions of interest (ROIs). A quantitative measure of COPD is obtained by fusing COPD probabilities computed in ROIs within the lung fields where the individual ROI probabilities are computed using a k nearest neighbor (kNN) classifier. The distance between two ROIs in the kNN classifier is computed as the textural dissimilarity between the ROIs, where the ROI texture is described by histograms of filter responses from a multi-scale, rotation invariant Gaussian filter bank. The method was trained on 400 images from a lung cancer screening trial and subsequently applied to classify 200 independent images from the same screening trial. The texture-based measure was significantly better at discriminating between subjects with and without COPD than were the two most common quantitative measures of COPD in the literature, which are based on density. The proposed measure achieved an area under the receiver operating characteristic curve (AUC) of 0.713 whereas the best performing density measure achieved an AUC of 0.598. Further, the proposed measure is as reproducible as the density measures, and there were indications that it correlates better with lung function and is less influenced by inspiration level.

 
N. Baka, B.L. Kaptein, M. de Bruijne, T. van Walsum, J.E. Giphart, W.J. Niessen, and B.P.F. Lelieveldt. 2D-3D shape reconstruction of the distal femur from stereo X-ray imaging using statistical shape models. Medical Image Analysis, 15(6):840-850, 2011. [ link ]
Three-dimensional patient specific bone models are required in a range of medical applications, such as pre-operative surgery planning and improved guidance during surgery, modeling and simulation, and in vivo bone motion tracking. Shape reconstruction from a small number of X-ray images is desired as it lowers both the acquisition costs and the radiation dose compared to CT. We propose a method for pose estimation and shape reconstruction of 3D bone surfaces from two (or more) calibrated X-ray images using a statistical shape model (SSM). User interaction is limited to manual initialization of the mean shape. The proposed method combines a 3D distance based objective function with automatic edge selection on a Canny edge map. Landmark-edge correspondences are weighted based on the orientation difference of the projected silhouette and the corresponding image edge. The method was evaluated by rigid pose estimation of ground truth shapes as well as 3D shape estimation using a SSM of the whole femur, from stereo cadaver X-rays, in vivo biplane fluoroscopy image-pairs, and an in vivo biplane fluoroscopic sequence. Ground truth shapes for all experiments were available in the form of CT segmentations. Rigid registration of the ground truth shape to the biplane fluoroscopy achieved sub-millimeter accuracy (0.68mm) measured as root mean squared (RMS) point-to-surface (P2S) distance. The non-rigid reconstruction from the biplane fluoroscopy using the SSM also showed promising results (1.68mm RMS P2S). A feasibility study on one fluoroscopic time series illustrates the potential of the method for motion and shape estimation from fluoroscopic sequences with minimal user interaction.

 
K. Murphy, B. van Ginneken, J. Reinhardt, S. Kabus, K. Ding, X. Deng, K. Cao, K. Du, G. Christensen, V. Garcia, T. Vercauteren, N. Ayache, O. Commowick, G. Malandain, B. Glocker, N. Paragios, N. Navab, V. Gorbunova, J. Sporring, M. de Bruijne, X. Han, M. Heinrich, J. Schnabel, M. Jenkinson, C. Lorenz, M. Modat, J. McClelland, S. Ourselin, S. Muenzing, M. Viergever, D. De Nigris, D. Collins, T. Arbel, M. Peroni, R. Li, G. Sharp, A. Schmidt-Richberg, J. Ehrhardt, R. Werner, D. Smeets, D. Loeckx, G. Song, N. Tustison, B. Avants, J. Gee, M. Staring, S. Klein, B. Stoel, M. Urschler, M. Werlberger, J. Vandemeulebroucke, S. Rit, D. Sarrut, and J. Pluim. Evaluation of registration methods on thoracic CT: The EMPIRE10 challenge. IEEE Transactions on Medical Imaging, 30(11):1901-1920, 2011. [ link ]
EMPIRE10 (Evaluation of Methods for Pulmonary Image REgistration 2010) is a public platform for fair and meaningful comparison of registration algorithms which are applied to a database of intra-patient thoracic CT image pairs. Evaluation of non-rigid registration techniques is a non trivial task. This is compounded by the fact that researchers typically test only on their own data, which varies widely. For this reason, reliable assessment and comparison of different registration algorithms has been virtually impossible in the past. In this work we present the results of the launch phase of EMPIRE10, which comprised the comprehensive evaluation and comparison of 20 individual algorithms from leading academic and industrial research groups. All algorithms are applied to the same set of 30 thoracic CT pairs. Algorithm settings and parameters are chosen by researchers expert in the configuration of their own method and the evaluation is independent, using the same criteria for all participants. All results are published on the EMPIRE10 website (http://empire10.isi.uu.nl). The challenge remains ongoing and open to new participants. Full results from 24 algorithms have been published at the time of writing. This article details the organisation of the challenge, the data and evaluation methods and the outcome of the initial launch with 20 algorithms. The gain in knowledge and future work are discussed.

 
M. Schaap, T. van Walsum, L. Neefjes, C. Metz, E. Capuano, M. de Bruijne, and W.J. Niessen. Robust shape regression for supervised vessel segmentation and its application to coronary segmentation in CTA. IEEE Transactions on Medical Imaging, 30(11):1974-1986, 2011. [ link ]
This paper presents a vessel segmentation method which learns the geometry and appearance of vessels in medical images from annotated data and uses this knowledge to segment vessels in unseen images. Vessels are segmented in a coarse-to-fine fashion. First the vessel boundaries are estimated with multi-variate linear regression using image intensities sampled in a region of interest around an initialization curve. Subsequently the position of the vessel boundary is refined with a robust nonlinear regression technique using intensity profiles sampled across the boundary of the rough segmentation and using information about plausible cross-sectional vessel shapes. The method was evaluated by quantitatively comparing segmentation results to manual annotations of 229 coronary arteries. On average the difference between the automatically obtained segmentations and manual contours was smaller than the inter-observer variability, which is an indicator that the method outperforms manual annotation. The method was also evaluated by using it for centerline refinement on 24 publicly available datasets of the Rotterdam Coronary Artery Evaluation Framework. Centerlines are extracted with an existing method and refined with the proposed method. This combination is currently ranked 2nd out of 10 evaluated interactive centerline extraction methods. An additional qualitative expert evaluation in which 250 automatic segmentations were compared to manual segmentations showed that the automatically obtained contours were rated on average better than manual contours.

 
M. Lillholm, A. Ghosh, P. C. Pettersen, M. de Bruijne, E. B. Dam, M. A. Karsdal, C. Christiansen, H. K. Genant, and M. Nielsen. Vertebral fracture risk (vfr) score for fracture prediction in postmenopausal women. Osteoporosis International, 22(7):2119-2128, 2011. [ link ]
Early prognosis of osteoporosis risk is not only important to individual patients but is also a key factor when screening for osteoporosis drug trial populations. We present an osteoporosis fracture risk score based on vertebral heights. The score separated individuals who sustained fractures (by follow-up after 6.3 years) from healthy controls at baseline.This case-control study was designed to assess the ability of three novel fracture risk scoring methods to predict first incident lumbar vertebral fractures in postmenopausal women matched for classical risk factors such as BMD, BMI, and age.This was a case-control study of 126 postmenopausal women, 25 of whom sustained at least one incident lumbar fracture and 101 controls that maintained skeletal integrity over a 6.3-year period. Three methods for fracture risk assessment were developed and tested. They are based on anterior, middle, and posterior vertebral heights measured from vertebrae T12-L5 in lumbar radiographs at baseline. Each score's fracture prediction potential was investigated in two variants using (1) measurements from the single most deformed vertebra or (2) average measurements across vertebrae T12-L5. Emphasis was given to the vertebral fracture risk (VFR) score.All scoring methods demonstrated significant separation of cases from controls at baseline. Specifically, for the VFR score, cases and controls were significantly different (0.67?±?0.04 vs. 0.35?±?0.03, p?<?10?(-6)) with an AUC of 0.82. Dividing the VFR scores into tertiles, the fracture odds ratio for the highest versus lowest tertile was 35 (p?<?0.001). Sorting the combined case-control group according to VFR score resulted in 90% of cases in the top half.At baseline, the three scores separated cases from controls and, especially, the VFR score appears to be predictive of fractures. Control experiments, however also, indicate that VFR-based fracture prediction is operator/annotator dependent and high-quality annotations are needed for good fracture prediction.

 
A. Crimi, M. Lillholm, A. Ghosh, M. de Bruijne, E. Dam, M. Nielsen, and J. Sporring. Maximum a posteriori Bayes estimation of linear shape variation with application to vertebra and cartilage modeling. IEEE Transactions on Medical Imaging, 30(8):1514 - 1526, 2011. [ link ]
The estimation of covariance matrices is a crucial step in several statistical tasks. Especially when using few samples of a high dimensional representation of shapes, the standard Maximum Likelihood estimation (ML) of the covariance matrix can be far from the truth, is often rank deficient, and may lead to unreliable results. In this paper, we discuss regularization by prior knowledge using Maximum A Posteriori (MAP) estimates. We compare ML to MAP using a number of priors and to Tikhonov regularization. We evaluate the covariance estimates on both synthetic and real data, and we analyze the estimates influence on a missing-data reconstruction task, where high resolution vertebra and cartilage models are reconstructed from incomplete and lower dimensional representations. Our results demonstrate that our methods outperform the traditional ML method and Tikhonov regularization.

 
H. Ashraf, P. Lo, S.B. Shaker, M. de Bruijne, A. Dirksen, P. Tønnesen, M. Dahlbäck, and J.H. Pedersen. Short term effect of changes in smoking behaviour on emphysema quantification by computed tomography (CT). Thorax, 66(1):55-60, 2011. [ link ]
The effect of smoking cessation and smoking relapse on lung density was studied using low-dose CT.Spiral, multidetector, low-dose CT was performed on 726 current and former smokers (>20 pack-years) recruited from a cancer screening trial. Lung density was quantified by calculating the 15th percentile density (PD15), which was adjusted to predicted total lung capacity. Data were analysed by linear regression models.At baseline mean PD15 was 45 g/l in former smokers (n=178) and 55 g/l in current smokers (n=548), representing a difference of 10 g/l (p<0.001). After smoking cessation (n=77) PD15 decreased by 6.2 g/l (p<0.001) in the first year, and by a further 3.6 g/l (p<0.001) in the second year, after which no further change could be detected. Moreover, the first year after relapse to smoking (n=18) PD15 increased by 3.7 g/l (p=0.02).Current smoking status has a major influence on lung density assessed by CT, and the difference in lung density between current and former smokers observed in cross-sectional studies corresponds closely to the change in lung density seen in the years after smoking cessation. Current smoking status, and time since cessation or relapse, should be taken into account when assessing the severity of diseases such as emphysema by CT lung density.

 
M. Nielsen, M. Ganz, F. Lauze, P.C. Pettersen, M. de Bruijne, T.B. Clarkson, E.B. Dam, C. Christiansen, and M.A. Karsdal. Distribution, size, shape, growth potential and extent of abdominal aortic calcified deposits predict mortality in postmenopausal women. BMC Cardiovascular Disorders, 10:56, 2010. [ link ]
Aortic calcification is a major risk factor for death from cardiovascular disease. We investigated the relationship between mortality and the composite markers of number, size, morphology and distribution of calcified plaques in the lumbar aorta.308 postmenopausal women aged 48-76 were followed for 8.3 ± 0.3 years, with deaths related to cardiovascular disease, cancer, or other causes being recorded. From lumbar X-rays at baseline the number (NCD), size, morphology and distribution of aortic calcification lesions were scored and combined into one Morphological Atherosclerotic Calcification Distribution (MACD) index. The hazard ratio for mortality was calculated for the MACD and for three other commonly used predictors: the EU SCORE card, the Framingham Coronary Heart Disease Risk Score (Framingham score), and the gold standard Aortic Calcification Severity score (AC24) developed from the Framingham Heart Study cohorts.All four scoring systems showed increasing age, smoking, and raised triglyceride levels were the main predictors of mortality after adjustment for all other metabolic and physical parameters. The SCORE card and the Framingham score resulted in a mortality hazard ratio increase per standard deviation (HR/SD) of 1.8 (1.51-2.13) and 2.6 (1.87-3.71), respectively. Of the morphological x-ray based measures, NCD revealed a HR/SD >2 adjusted for SCORE/Framingham. The MACD index scoring the distribution, size, morphology and number of lesions revealed the best predictive power for identification of patients at risk of mortality, with a hazard ratio of 15.6 (p < 0.001) for the 10% at greatest risk of death.This study shows that it is not just the extent of aortic calcification that predicts risk of mortality, but also the distribution, shape and size of calcified lesions. The MACD index may provide a more sensitive predictor of mortality from aortic calcification than the commonly used AC24 and SCORE/Framingham point card systems.

 
P. Lo, J. Sporring, H. Ashraf, J.H. Pedersen, and M. de Bruijne. Vessel-guided airway tree segmentation: A voxel classification approach. Medical Image Analysis, 14(4):527-538, 2010. [ link | pdf ]
This paper presents a method for airway tree segmentation that uses a combination of a trained airway appearance model, vessel and airway orientation information, and region growing. We propose a voxel classification approach for the appearance model, which uses a classifier that is trained to differentiate between airway and non-airway voxels. This is in contrast to previous works that use either intensity alone or hand crafted models of airway appearance. We show that the appearance model can be trained with a set of easily acquired, incomplete, airway tree segmentations. A vessel orientation similarity measure is introduced, which indicates how similar the orientation of an airway candidate is to the orientation of the neighboring vessel. We use this vessel orientation similarity measure to overcome regions in the airway tree that have a low response from the appearance model. The proposed method is evaluated on 250 low dose computed tomography images from a lung cancer screening trial. Our experiments showed that applying the region growing algorithm on the airway appearance model produces more complete airway segmentations, leading to on average 20 trees, and 50 model with vessel orientation similarity, the improvement is even more significant (p<0.01) than only using the airway appearance model, with on average 7 correctly.

 
L. Sørensen, S. Shaker, and M. de Bruijne. Quantitative analysis of pulmonary emphysema using local binary patterns. IEEE Transactions on Medical Imaging, 29(2):559-569, 2010. [ link | pdf ]
<para> We aim at improving quantitative measures of emphysema in computed tomography (CT) images of the lungs. Current standard measures, such as the relative area of emphysema (RA), rely on a single intensity threshold on individual pixels, thus ignoring any interrelations between pixels. Texture analysis allows for a much richer representation that also takes the local structure around pixels into account. This paper presents a texture classification-based system for emphysema quantification in CT images. Measures of emphysema severity are obtained by fusing pixel posterior probabilities output by a classifier. Local binary patterns (LBP) are used as texture features, and joint LBP and intensity histograms are used for characterizing regions of interest (ROIs). Classification is then performed using a k nearest neighbor classifier with a histogram dissimilarity measure as distance. A 95.2 annotated ROIs, comprising the three classes: normal tissue, centrilobular emphysema, and paraseptal emphysema. The measured emphysema severity was in good agreement with a pulmonary function test (PFT) achieving correlation coefficients of up to |r| = 0.79 in 39 subjects. The results were compared to RA and to a Gaussian filter bank, and the texture-based measures correlated significantly better with PFT than did RA.

Keywords: Emphysema, local binary patterns (LBPs), quantitative computed tomography (CT), texture analysis, tissue classification
 
M. Loeve, M.H. Lequin, M. de Bruijne, I.J.C. Hartmann, K. Gerbrands, M. van Straten, W.C.J. Hop, and H.A.W.M. Tiddens. Cystic fibrosis: Are volumetric ultra-low-dose expiratory CT scans sufficient for monitoring related lung disease? Radiology, 253(1):223-229, 2009. [ link ]
Purpose: To assess whether chest computed tomography (CT) scores from ultra-low-dose end-expiratory scans alone could suffice for assessment of all cystic fibrosis (CF)-related structural lung abnormalities. Materials and Methods: In this institutional review board-approved study, 20 patients with CF aged 6-20 years (eight males, 12 females) underwent low-dose end-inspiratory CT and ultra-low-dose end-expiratory CT. Informed consent was obtained. Scans were randomized and scored by using the Brody-II CT scoring system to assess bronchiectasis, airway wall thickening, mucus plugging, and opacities. Scoring was performed by two observers who were blinded to patient identity and clinical information. Mean scores were used for all analyses. Statistical analysis included assessment of intra- and interobserver variability, calculation of intraclass correlation coefficients (ICCs), and Bland-Altman plots. Results: Median age was 12.6 years (range, 6.3-20.3 years), median forced expiratory volume in 1 second was 100% (range, 46%-127%) of the predicted value, and median forced vital capacity was 99% (range, 61%-123%) of the predicted value. Very good agreement was observed between end-inspiratory and end-expiratory CT scores for Brody-II total score (ICC = 0.96), bronchiectasis (ICC = 0.98), airway wall thickening (ICC = 0.94), mucus plugging (ICC = 0.96), and opacities (ICC = 0.90). Intra- and interobserver agreement were good to very good (ICC range, 0.70-0.98). Bland-Altman plots showed that differences in scores were independent of score magnitude. Conclusion: In this pilot study, CT scores from end-expiratory and end-inspiratory CT match closely, suggesting that ultra-low-dose end-expiratory CT alone may be sufficient for monitoring CF-related lung disease. This would help reduce radiation dose for a single investigation by up to 75%.

 
M. Nielsen, M. de Bruijne, C. Jørgensen, S. Olsen, K. Pedersen, and J. Sporring. Special issue on tribute workshop for Peter Johansen. Journal of Mathematical Imaging and Vision, pages 119-120, 2008. [ link ]
 
M. de Bruijne, M.T. Lund, L.B. Tankó, P.C. Pettersen, and M. Nielsen. Quantitative vertebral morphometry using neighbor-conditional shape models. Medical Image Analysis, 11(5):503-512, 2007. [ link | pdf ]
A novel method for vertebral fracture quantification from X-ray images is presented. Using pairwise conditional shape models trained on a set of healthy spines, the most likely normal vertebra shapes are estimated conditional on the shapes of all other vertebrae in the image. The difference between the true shape and the reconstructed normal shape is subsequently used as a measure of abnormality. In contrast with the current (semi-)quantitative grading strategies this method takes the full shape into account, it develops a patient-specific reference by combining population-based information on biological variation in vertebral shape and vertebra interrelations, and it provides a continuous measure of deformity. The method is demonstrated on 282 lateral spine radiographs with in total 93 fractures. Vertebral fracture detection is shown to be in good agreement with semi-quantitative scoring by experienced radiologists and is superior to the performance of shape models alone.

 
J.E. Iglesias and M. de Bruijne. Semi-automatic segmentation of vertebrae in lateral X-rays using a conditional shape model. Academic Radiology, 14(10):1156-1165, 2007. [ link | pdf ]
RATIONALE AND OBJECTIVES: Manual annotation of the full contour of the vertebrae in lateral x-rays for subsequent morphometry is time-consuming. The standard six-point morphometry is commonly used instead. It has been shown that the information from the complete contour improves the quality of the study. In this article, the six landmarks are given and the vertebrae are segmented taking advantage of that information. The result is a semiautomatic system in which the full contour is found with high precision, and that only requires a radiologist to mark six points per vertebra. MATERIALS AND METHODS: A shape model was built for both the six landmarks and the full contours of the vertebrae L1, L2, L3, and L4 of 142 patients. The distribution of the principal components of the full contour was then modeled as a Gaussian conditional distribution depending on the principal components of the six landmarks. The conditional mean was used as initialization for active shape model optimization, and the conditional variance was used to constrain the solution to plausible shapes. RESULTS: The achieved point-to-line error was 0.48 mm, and 95 were located within 1.36 mm of the annotated contour. The accuracy is superior to those of previously published studies, at the expense of requiring the six points to be marked. Fractures and osteophytes are well approximated by the model, although they are sometimes oversmoothed. CONCLUSIONS: The proposed method provides hence a richer description than the six points, and can be used as input for shape-based morphometry to detect and quantify vertebral deformation more accurately.

 
P.C. Pettersen, M. de Bruijne, J. Chen, Q. He, C. Christiansen, and L.B. Tankó. A computer-based measure of irregularity in vertebral alignment is a BMD-independent predictor of fracture risk in postmenopausal women. Osteoporosis International, 18(11), 2007. [ link ]
Prevalent fracture and BMD are core elements of fracture prediction. In this control study case, we demonstrate that a simple computer-based estimation of local irregularities in the alignment of the lumbar vertebrae independently contributes to the fracture risk, thus supplementing current diagnostic tools. INTRODUCTION: We tested the hypothesis that degree of lordosis and/or irregularity in the alignment of lumbar vertebrae could be contributors to the risk of fragility fractures. METHODS: This was a case-control analysis including 144 elderly women; 108 maintaining skeletal integrity, whereas 36 sustaining a lumbar vertebral fracture during a 7.5-year observation period. The two groups of women were carefully matched for age, BMI, spine BMD and numerous classic risk factors. Lateral X-rays of the lumbar spine were digitized and the four corner points of endplates on each vertebra from Th22 to L5 were annotated. The degree of lordosis and irregularity of vertebral alignment was assessed by image analysis software. RESULTS: Degree of lordosis was not predictive for fractures. In contrast, irregularity was significantly higher in those who later sustained a fracture (1.6 x 10(-2)vs. 2.0 x 10(-3) cm(-1), p < 0.001), and further increased upon a sustained fracture (2.8 x 10(-2) cm(-1), p < 0.001), but was unchanged in controls (1.6 x 10(-2) cm(-1)). The predictive value of irregularity was independent of classic risk factors of fractures, including BMD (p < 0.01). CONCLUSION: Our results suggest that the herein introduced simple measure of irregularities in vertebral alignment could provide useful supplement to the currently used diagnostic tools of fracture prediction in elderly women.

 
M. de Bruijne, B. van Ginneken, M.A. Viergever, and W.J. Niessen. Interactive segmentation of abdominal aortic aneurysms in CTA images. Medical Image Analysis, 8(2):127-138, 2004. [ link | pdf ]
A model-based approach to interactive segmentation of abdominal aortic aneurysms from CTA data is presented. After manual delineation of the aneurysm sac in the first slice, the method automatically detects the contour in subsequent slices, using the result from the previous slice as a reference. If an obtained contour is not sufficiently accurate, the user can intervene and provide an additional manual reference contour. The method is inspired by the active shape model (ASM) segmentation scheme (Cootes et al., 1995), in which a statistical shape model, derived from corresponding landmark points in manually labeled training images, is fitted to the image in an iterative manner. In our method, a shape model of the contours in two adjacent image slices is progressively fitted to the entire volume. The contour obtained in one slice thus constrains the possible shapes in the next slice. The optimal fit is determined on the basis of multiresolution gray level models constructed from gray value patches sampled around each landmark. We propose to use the similarity of adjacent image slices for this gray level model, and compare these to single-slice features that are more generally used with ASM. The performance of various image features is evaluated in leave-one-out experiments on 23 data sets. Features that use the similarity of adjacent image slices outperform measures based on single-slice features in all cases. The average number of slices in our datasets is 51, while on average eight manual initializations are required, which decreases operator segmentation time by a factor of 6.

 
M. de Bruijne, W.J. Niessen, J.B.A. Maintz, and M.A. Viergever. Localization and segmentation of aortic endografts using marker detection. IEEE Transactions on Medical Imaging, 22(4):473-482, 2003. [ link | pdf ]
A method for localization and segmentation of bifurcated aortic endografts in computed tomographic angiography (CTA) images is presented. The graft position is determined by detecting radiopaque markers sewn on the outside of the graft. The user indicates the first and the last marker, whereupon the remaining markers are automatically detected. This is achieved by first detecting marker-like structures through second-order scaled derivative analysis, which is combined with prior knowledge of graft shape and marker configuration. The identified marker centers approximate the graft sides and, derived from these, the central axis. The graft boundary is determined by maximizing the local gradient in the radial direction along a deformable contour passing through both sides. Three segmentation methods were tested. The first performs graft contour detection in the initial CT-slices, the second in slices that were reformatted to be orthogonal to the approximated graft axis, and the third uses the segmentation from the second method to find a more reliable approximation of the axis and subsequently performs contour detection. The methods have been applied to ten CTA images and the results were compared to manual marker indication by one observer and region growing aided segmentation by three observers. Out of a total of 266 markers, 262 were detected. Adequate approximations of the graft sides were obtained in all cases. The best segmentation results were obtained using a second iteration orthogonal to the axis determined from the first segmentation, yielding an average relative volume of overlap with the expert segmentations of 92 difference in volume measured by the automated method and by the experts equals the difference among the experts: 3.5%.

 
F.A. Karelse, M. de Bruijne, C.J. Barth, M.N.A. Beurskens, G.M.D. Hogeweij, and N.J. Lopes Cardozo. Measurements of the current density profile with tangential Thomson scattering in RTP. Plasma Physics and Controlled Fusion, 43(4):443-468, 2001. [ link ]
The electron drift velocity ( v d,e ) profile is measured with tangential Thomson scattering (TTS). From this, the current density ( j ) profile is determined, extending over 60% of the plasma diameter ( 2 a ) witha spatial resolution of 12% of a . The statistical error of ˜10% is in agreement with the expected accuracy based on simulation studies. Two analysis methods are compared and found to agree. Results of TTS in electron cyclotron heated discharges with peaked and hollow j profiles are in good agreement with neo-classical resistivity.

Papers in conference proceedings

 
G. van Tulder and M. de Bruijne. Learning features for tissue classification with the classification restricted boltzmann machine. In MICCAI 2014 Workshop on Medical Computer Vision: Algorithms for Big Data, 2014. [ link ]
Performance of automated tissue classification in medical imaging depends on the choice of descriptive features. In this paper, we show how restricted Boltzmann machines (RBMs) can be used to learn features that are especially suited for texture-based tissue classification. We introduce the convolutional classification RBM, a combination of the existing convolutional RBM and classification RBM, and use it for discriminative feature learning. We evaluate the classification accuracy of convolutional and non-convolutional classification RBMs on two lung CT problems. We find that RBM-learned features outperforms conventional RBM-based feature learning, which is unsupervised and uses only a generative learning objective, as well as often-used filter banks. We show that a mixture of generative and discriminative learning can produce filters that give a higher classification accuracy.

 
Veronika Cheplygina, Lauge Sørensen, David M.J. Tax, Jesper Holst Pedersen, Marco Loog, and Marleen de Bruijne. Classification of COPD with multiple instance learning. In 22nd International Conference on Pattern Recognition, pages 1508-1513, 2014. [ link ]
 
Aasa Feragen, Niklas Kasenburg, Jens Petersen, Marleen de Bruijne, and Karsten Borgwardt. Scalable kernels for graphs with continuous attributes. In C.J.C. Burges, L. Bottou, Z. Ghahramani, and K.Q. Weinberger, editors, Neural Information Processing Systems (NIPS), pages 216-224, 2013. [ link ]
 
A Arias, D Carvalho, J Petersen, A van Dijk, A van der Lugt, W Niessen, S Klein, and M de Bruijne. Carotid artery lumen segmentation on 3D free-hand ultrasound images using surface graph cuts. In Medical Image Computing & Computer-Assisted Intervention, Lecture Notes in Computer Science. Springer, 2013. [ link ]
 
F. Ciompi, C. Gatta, and M. de Bruijne. Iterated stacked classifiers for lung segmentation in computed tomography. In R. Beichel, M. de Bruijne, S. Kabus, A. Kiraly, T. Kitasaka, J.M. Kuhnigk, J.R. McClelland, K. Mori, E.M. van Rikxoort, and S. Rit, editors, Proc. of Fifth International Workshop on Pulmonary Image Analysis. CreateSpace, 2013. ISBN-13: 978-1492186977. [ link ]
 
A.G. van Opbroek, M.A. Ikram, M.W. Vernooij, and M. de Bruijne. A transfer-learning approach to image segmentation across scanners by maximizing distribution similarity. In Guorong Wu, Daoqiang Zhang, Dinggang Shen, Pingkun Yan, Kenji Suzuki, and Fei Wang, editors, Machine Learning in Medical Imaging, volume 8184 of Lecture Notes in Computer Science. Springer, 2013. [ link ]
 
J. Petersen, M. Modat, M. Cardoso, A. Dirksen, S. Ourselin, and M. de Bruijne. Quantitative airway analysis in longitudinal studies using groupwise registration and 4D optimal surfaces. In Medical Image Computing & Computer-Assisted Intervention, Lecture Notes in Computer Science. Springer, 2013. [ link ]
 
A.G. van Opbroek, F. van der Lijn, and M. de Bruijne. Automated brain-tissue segmentation by multi-feature SVM classification. In MICCAI Grand Challenge on MR Brain Image Segmentation (MRBRains13), 2013. [ link ]
 
A. Feragen, M. Owen, J. Petersen, M.M.W. Wille, L.H. Thomsen, A. Dirksen, and M. de Bruijne. Tree-space statistics and approximations for large-scale analysis of anatomical trees. In W.M. Wells, S. Joshi, and K.M. Pohl, editors, Information Processing in Medical Imaging, Lecture Notes in Computer Science. Springer, 2013. [ link ]
 
A. Feragen, J. Petersen, D. Grimm, A. Dirksen, J.H. Pedersen, K. Borgwardt, and M. de Bruijne. Geometric tree kernels: Classification of COPD from airway tree geometry. In W.M. Wells, S. Joshi, and K.M. Pohl, editors, Information Processing in Medical Imaging, Lecture Notes in Computer Science. Springer, 2013. [ link ]
 
Ruwan Bandara Tennakoon, Alireza Bab-Hadiashar, Marleen de Bruijne, and Zhenwei Cao. Efficient nonrigid registration using ranked order statistics. In IEEE International Symposium on Biomedical Imaging: Macro to Nano (ISBI'13), 2013. [ link ]
Non-rigid image registration techniques are widely used in medical imaging applications. Due to high computational complexities of these techniques, finding appropriate registration method to both reduce the computation burden and increase the registration accuracy has become an intense area of research. In this paper we propose a fast and accurate nonrigid registration method for intra-modality volumetric images. Our approach exploits the information provided by an order statistics based segmentation method, to find the important regions for registration and use an appropriate sampling scheme to target those areas and reduce the registration computation time. A unique advantage of the proposed method is its ability to identify the point of diminishing returns and stop the registration process. Our experiments on registration of real lung CT images, with expert annotated landmarks, show that the new method is both faster and more accurate than the state of the art techniques, particularly for registration of images with large deformations.

 
Hakim C. Achterberg, Dirk H. J. Poot, Fedde van der Lijn, Meike W. Vernooij, M Arfan Ikram, Wiro J. Niessen, and Marleen de Bruijne. Robust local appearance features for MRi brain structure segmentation across scanning protocols. In S. Ourselin and D.R. Haynor, editors, Medical Imaging: Image Processing, volume 8669 of Proceedings of SPIE. SPIE Press, 2013. [ link ]
 
Sepp de Raedt, Inger Mechlenburg, Maiken Stilling, Lone Rømer, Kjeld Søballe, and Marleen de Bruijne. Automated measurement of diagnostic angles for hip dysplasia. In C.L. Novak and S. Aylward, editors, Medical Imaging: Computer-Aided Diagnosis, volume 8670 of Proceedings of SPIE. SPIE Press, 2013. [ link ]
 
A. Arias, J. Petersen, A. van Engelen, H. Tang, M. Selwaness, J.C.M. Witteman, A. van der Lugt, W.J. Niessen, and M. de Bruijne. Carotid artery wall segmentation by coupled surface graph cuts. In B. Menze, G. Langs, L. Lu, A. Montillo, Z. Tu, and A. Criminisi, editors, Medical Computer Vision 2012. Recognition Techniques and Applications in Medical Imaging, volume 7766 of Lecture Notes in Computer Science, pages 38-47. Springer, 2013. [ link ]
 
A. Feragen, J. Petersen, M. Owen, P. Lo, L.H. Thomsen, M.M.W. Wille, A. Dirksen, and M. de Bruijne. A hierarchical scheme for geodesic anatomical labeling of airway trees. In N. Ayache, H. Delingette, P. Golland, and K. Mori, editors, Medical Image Computing & Computer-Assisted Intervention, volume 7510 of Lecture Notes in Computer Science. Springer, 2012. [ link ]
We present a fast and robust supervised algorithm for labeling anatomical airway trees, based on geodesic distances in a geometric tree-space. Possible branch label configurations for a given tree are evaluated based on distances to a training set of labeled trees. In tree-space , the tree topology and geometry change continuously, giving a natural way to automatically handle anatomical differences and noise. The algorithm is made efficient using a hierarchical approach , in which labels are assigned from the top down. We only use features of the airway centerline tree, which are relatively unaffected by pathology. A thorough leave-one-patient-out evaluation of the algorithm is made on 40 segmented airway trees from 20 subjects labeled by 2 medical experts. We evaluate accuracy, reproducibility and robustness in patients with Chronic Obstructive Pulmonary Disease (COPD). Performance is statistically similar to the inter- and intra-expert agreement, and we found no significant correlation between COPD stage and labeling accuracy.

 
A.G. van Opbroek, M.A. Ikram, M.W. Vernooij, and M. de Bruijne. Supervised image segmentation across scanner protocols: A transfer learning approach. In Fei Wang, Dinggang Shen, Pingkun Yan, and Kenji Suzuki, editors, Machine Learning in Medical Imaging, volume 7588 of Lecture Notes in Computer Science, pages 160-167. Springer, 2012. [ link ]
Supervised classification techniques are among the most powerful methods used for automatic segmentation of medical images. A disadvantage of these methods is that they require a representative training set and thus encounter problems when the training data is acquired e.g. with a different scanner protocol than the target segmentation data. We therefore propose a framework for supervised biomedical image segmentation across different scanner protocols, by means of transfer learning.We establish a transfer learning algorithm for classification, which can exploit a large amount of labeled samples from different sources in addition to a small amount of samples from the target source. The algorithm iteratively re-weights the contribution of training samples from these different sources based on classification by a weighted SVM classifier. We evaluate this technique by performing tissue classification on MRI brain data from four substantially different scanning protocols. For a small number of labeled samples from a single image obtained with the same protocol, the proposed transfer learning method outperforms classification on all available training data as well as classification based on the labeled target samples only. The classification errors for these cases can be reduced with up to 40 percent compared to traditional classification techniques.

 
A van Engelen, W.J. Niessen, S. Klein, H.C. Groen, K. van Gaalen, H.J.M. Verhagen, J.J. Wentzel, A van der Lugt, and M. de Bruijne. Automated segmentation of atherosclerotic histology based on pattern classification. In MICCAI workshop on Histopathology Image Analysis: Image computing in digital pathology, 2012.
 
N. Baka, C.T. Metz, C. Schultz, L. Neefjes, R.J. van Geuns, B.P.F. Lelieveldt, W.J. Niessen, M. de Bruijne, and T. van Walsum. 3D+t/2D+t CTA-XA registration using population-based motion estimates. In MICCAI-Workshop on Computer Assisted Stenting Proceedings, MICCAI-STENT'12, 2012.
 
A. van Engelen, W.J. Niessen, S. Klein, H. Verhagen, H.C. Groen, J.J. Wentzel, A. van der Lugt, and M. de Bruijne. Supervised in-vivo plaque characterization incorporating class label uncertainty. In IEEE International Symposium on Biomedical Imaging: Macro to Nano (ISBI'12), pages 246-249, 2012. [ link ]
We segment atherosclerotic plaque components in in-vivo MRI and CT data using supervised voxelwise classification. The most reliable ground truth can be obtained from histology sections, however, it is not straightforward to use this for classifier training as the registration with in-vivo data often shows misalignments. Therefore, for training we incorporate uncertainty in the ground truth via ”soft” labels that indicate a probability for each class. Soft labels are created by Gaussian blurring of the original ”hard” segmentations, and weighted by the registration accuracy. Classification is evaluated on the relative volumes for fibrous, lipid-rich necrotic and calcified tissue. Using conventional hard labels, the differences between the ground truth and classification result per subject are -0.4±3.6 for calcification, +7.6±14.9 tissue. Using the new approach accuracy is improved: for calcification -0.6±1.6%, fibrous +3.6±16.8% and necrotic tissue -2.9±16.1%.

 
B. Ghafary, F. van der Lijn, M. Poels, H. Vrooman, M.A. Ikram, W.J. Niessen, A. van der Lugt, M. Vernooij, and M. de Bruijne. A computer aided detection system for cerebral microbleeds in brain MRI. In IEEE International Symposium on Biomedical Imaging: Macro to Nano (ISBI'12), pages 138-141, 2012. [ link ]
Advances in MR technology have improved the potential for visualization of small lesions in brain images. This has resulted in the opportunity to detect cerebral microbleeds (CMBs), small hemorrhages in the brain that are known to be associated with risk of ischemic stroke and intracerebral bleeding. In this paper, we propose a computer aided detection (CAD) system for the detection of CMBs to speed up visual analysis. Our method consists of three steps: (i) skull-stripping (ii) initial candidate selection and (iii) reduction of false-positives (FPs) using a two layer classification. Geometrical, intensity-based and local image descriptor features were used in the classification steps. The training and test set consist of 156 subjects (448 CMBs) and 81 subjects (183 CMBs), respectively. The sensitivity for CMB detection was 91% with, on average, 4.1 false-positives per subject.

 
C. Chen, L. Sørensen, F. Lauze, C. Igel, M. Loog, A. Feragen, M. de Bruijne, and M. Nielsen. Towards exaggerated emphysema stereotypes. In B. van Ginneken and C.L. Novak, editors, Medical Imaging: Computer-Aided Diagnosis, volume 8315 of Proceedings of SPIE. SPIE Press, 2012. [ link ]
Classification is widely used in the context of medical image analysis and in order to illustrate the mechanism of a classifier, we introduce the notion of an exaggerated image stereotype based on training data and trained classifier. The stereotype of some image class of interest should emphasize/exaggerate the characteristic patterns in an image class and visualize the information the employed classifier relies on. This is useful for gaining insight into the classification and serves for comparison with the biological models of disease. In this work, we build exaggerated image stereotypes by optimizing an objective function which consists of a discriminative term based on the classification accuracy, and a generative term based on the class distributions. A gradient descent method based on iterated conditional modes (ICM) is employed for optimization. We use this idea with Fisher's linear discriminant rule and assume a multivariate normal distribution for samples within a class. The proposed framework is applied to computed tomography (CT) images of lung tissue with emphysema. The synthesized stereotypes illustrate the exaggerated patterns of lung tissue with emphysema, which is underpinned by three different quantitative evaluation methods.

 
N. Baka, C. Metz, M. Schaap, B. Lelieveldt, W.J. Niessen, and M. de Bruijne. Comparison of shape regression methods under landmark position uncertainty. In G. Fichtinger, A. Martel, and T. Peters, editors, Medical Image Computing & Computer-Assisted Intervention, volume 6893 of Lecture Notes in Computer Science. Springer, 2011. [ link ]
Despite the growing interest in regression based shape estimation, no study has yet systematically compared different regression methods for shape estimation. We aimed to fill this gap by comparing linear regression methods with a special focus on shapes with landmark position uncertainties. We investigate two scenarios: In the first, the uncertainty of the landmark positions was similar in the training and test dataset, whereas in the second the uncertainty of the training and test data were different. Both scenarios were tested on simulated data and on statistical models of the left ventricle estimating the end-systolic shape from end-diastole with landmark uncertainties derived from the segmentation process, and of the femur estimating the 3D shape from one projection with landmark uncertainties derived from the imaging setup. Results show that in the first scenario linear regression methods tend to perform similar. In the second scenario including estimates of the test shape landmark uncertainty in the regression improved results.

 
F. Ciompi, A. Palaioroutas, M. Loeve, O. Pujol, P. Radeva, H. Tiddens, and M. de Bruijne. Lung tissue classification in severe advanced cystic fibrosis from CT scans. In R. Beichel, M. de Bruijne, B. van Ginneken, S. Kabus, A. Kiraly, J.M. Kuhnigk, J.R. McClelland, K. Mori, E.M. van Rikxoort, and S. Rit, editors, Proc. of Fourth International Workshop on Pulmonary Image Analysis, pages 57-68. CreateSpace, 2011. ISBN-13: 978-1466200166. [ link ]
A framework for lung tissue classification in Computed Tomography (CT) scans is presented. The method combines supervised and unsupervised learning techniques, with the aim of classifying four tissue types in lung: (i) inflammation, (ii) air-trapped / hypoperfused,(iii) normal / hyperperfused and (iv) bulla / cyst. The framework has been tested on a large heterogeneous dataset, collected over the last 20 years from 17 sites worldwide. The overall accuracy of the proposed methodology is 72.6 and 89.6 be used for the assessment of Severe Advanced Lung Disease (SALD) in patients affected by Cystic Fibrosis.

 
P. Lo, E.M. van Rikxoort, J. Goldin, F. Abtin, M. de Bruijne, and M. Brown. A bottom-up approach for labeling of human airway trees. In R. Beichel, M. de Bruijne, B. van Ginneken, S. Kabus, A. Kiraly, J.M. Kuhnigk, J.R. McClelland, K. Mori, E.M. van Rikxoort, and S. Rit, editors, Proc. of Fourth International Workshop on Pulmonary Image Analysis, pages 23-34. CreateSpace, 2011. ISBN-13: 978-1466200166. [ link ]
In this paper, an airway labeling algorithm that allows for gaps between the labeled branches is introduced. A bottom-up approach for arriving to an optimal set of branches and their associated labels is used in the proposed method. A K nearest neighbor based appearance model is used to differentiate the different anatomical branches. The proposed method was applied on 33 computed tomography scans of different subjects, where an average of 24 anatomical branches were correctly detected out of a total of 29 anatomical branches. Additionally the proposed method was also evaluated on trees with simulated errors, such as missing branches and having falsely detected branches, where we showed that such errors have little or no effect on the proposed method.

 
J. Petersen, V. Gorbunova, M. Nielsen, A. Dirksen, P. Lo, and M de Bruijne. Longitudinal analysis of airways using registration. In R. Beichel, M. de Bruijne, B. van Ginneken, S. Kabus, A. Kiraly, J.M. Kuhnigk, J.R. McClelland, K. Mori, E.M. van Rikxoort, and S. Rit, editors, Proc. of Fourth International Workshop on Pulmonary Image Analysis, pages 11-22. CreateSpace, 2011. ISBN-13: 978-1466200166. [ link ]
Longitudinal investigations of airway abnormalities associated with Chronic Obstructive Pulmonary Disease (COPD) has been very limited so far, partly due to the difficulties in obtaining reproducible measures. We propose to improve on this by limiting measurements to corresponding branches found using image registration. The results obtained from scans of 237 subjects show increased intrasubject correlation when measurements are conducted in branches found in each scan compared to similar measurements not limited to corresponding branches. This indicates the method could be useful for longitudinal analysis. Yearly changes in CT measures showed that airways increase in size and decrease in density with time. Changes were in general not found to be significantly correlated with changes in lung function and neither were there any significant differences between COPD GOLD stages.

 
A. Feragen, P. Lo, V. Gorbunova, M. Nielsen, A. Dirksen, J. Reinhardt, F. Lauze, and M. de Bruijne. An airway tree-shape model for geodesic airway branch labeling. In X. Pennec, S. Joshi, and M. Nielsen, editors, Mathematical Foundations of Computational Anatomy, 2011. [ link ]
We present a mathematical airway tree-shape framework where airway trees are compared using geodesic distances. The framework consists of a rigorously defined shape space for treelike shapes, endowed with a metric such that the shape space is a geodesic metric space. This means that the distance between two tree-shapes can be realized as the length of the geodesic, or shortest deformation, connecting the two shapes. By computing geodesics between airway trees, as well as the corresponding airway deformation, we generate airway branch correspondences. Correspondences between an unlabeled airway tree and a set of labeled airway trees are combined with a voting scheme to perform automatic branch labeling of segmented airways from the challenging EXACT'09 test set. In spite of the varying quality of the data, we obtain robust labeling results.

 
D.H.J. Poot, M. de Bruijne, M. Vernooij, M.A. Ikram, and W.J. Niessen. Improved tissue segmentation by including an MR acquisition model. In Tianming Liu, Dinggang Shen, Luis Ibanez, and Xiaodong Tao, editors, Multimodal Brain Image Analysis, volume 7012 of Lecture Notes in Computer Science, pages 152-159. Springer Berlin / Heidelberg, 2011. [ link ]
This paper presents a new MR tissue segmentation method. In contrast to most previous methods the image formation model includes the point spread function of the image acquisition. This allows optimal combination of images acquired with di erent contrast weighting, resolutions, and orientations. The proposed method computes the regularized maximum likelihood partial volume segmentation from the images. The quality the resulting segmentation is studied with a simulation experiment and by testing the reproducibility of the segmentation on repeated brain MRI scans. Our results demonstrate improved segmentation quality, especially at tissue edges.

 
J. Petersen, M. Nielsen, P. Lo, Z. Saghir, A. Dirksen, and M de Bruijne. Optimal graph based segmentation using flow lines with application to airway wall segmentation. In G. Székely and H.K. Hahn, editors, Information Processing in Medical Imaging, volume 6801 of Lecture Notes in Computer Science, pages 49-60. Springer, 2011. [ link | pdf ]
This paper introduces a novel optimal graph construction method that is applicable to multi-dimensional, multi-surface segmentation problems. Such problems are often solved by refining an initial coarse surface within the space given by graph columns. Conventional columns are not well suited for surfaces with high curvature or complex shapes but the proposed columns, based on properly generated flow lines, which are non-intersecting, guarantee solutions that do not self-intersect and are better able to handle such surfaces. The method is applied to segment human airway walls in computed tomography images. Comparison with manual annotations on 649 cross-sectional images from 15 different subjects shows significantly smaller contour distances and larger area of overlap than are obtained with recently published graph based methods. Airway abnormality measurements obtained with the method on 480 scan pairs from a lung cancer screening trial are reproducible and correlate significantly with lung function.

 
L. Sørensen, P. Lo, J. Petersen, A. Dirksen, and M de Bruijne. Dissimilarity-based classification of anatomical tree structures. In G. Székely and H.K. Hahn, editors, Information Processing in Medical Imaging, volume 6801 of Lecture Notes in Computer Science. Springer, 2011. [ link ]
A novel method for classification of abnormality in anatomical tree structures is presented. A tree is classified based on direct comparisons with other trees in a dissimilarity-based classification scheme. The pair-wise dissimilarity measure between two trees is based on a linear assignment between the branch feature vectors representing those trees. Hereby, localized information in the branches is collectively used in classification and variations in feature values across the tree are taken into account. An approximate anatomical correspondence between matched branches can be achieved by including anatomical features in the branch feature vectors. The proposed approach is applied to classify airway trees in computed tomography images of subjects with and without chronic obstructive pulmonary disease (COPD). Using the wall area percentage (WA as anatomical features to characterize each branch, an area under the receiver operating characteristic curve of 0.912 is achieved. This is significantly better than computing the average WA%.

 
A. van Engelen, M. de Bruijne, S. Klein, H. Verhagen, H.C. Groen, J.J. Wentzel, A. van der Lugt, and W.J. Niessen. Plaque characterization in ex vivo MRI evaluated by dense 3D correspondence with histology. In R.M. Summers and B. van Ginneken, editors, Medical Imaging: Computer-Aided Diagnosis, volume 7963 of Proceedings of SPIE. SPIE Press, 2011. [ link ]
Automatic quantification of carotid artery plaque composition is important in the development of methods that distinguish vulnerable from stable plaques. MRI has shown to be capable of imaging different components noninvasively. We present a new plaque classification method which uses 3D registration of histology data with ex vivo MRI data, using non-rigid registration, both for training and evaluation. This is more objective than previously presented methods, as it eliminates selection bias that is introduced when 2D MRI slices are manually matched to histological slices before evaluation. Histological slices of human atherosclerotic plaques were manually segmented into necrotic core, fibrous tissue and calcification. Classification of these three components was voxelwise evaluated. As features the intensity, gradient magnitude and Laplacian in four MRI sequences after different degrees of Gaussian smoothing, and the distances to the lumen and the outer vessel wall, were used. Performance of linear and quadratic discriminant classifiers for different combinations of features was evaluated. Best accuracy (72.5 ± 7.7 when all features were used. Although this was only a minor improvement to the accuracy of a classifier that only included the intensities and distance features (71.6 ± 7.9 significant (paired t-test, p<0.05). Good sensitivity and specificity for calcification was reached (83 differentiation between fibrous (sensitivity 85 and necrotic tissue (sensitivity 49%, specificity 89%) was more difficult.

 
A. Feragen, F. Lauze, P. Lo, M. de Bruijne, and M Nielsen. Geometries on spaces of treelike shapes. In R. Kimmel, R. Klette, and A. Sugimoto, editors, Asian Conference on Computer Vision (ACCV), volume 6493 of Lecture Notes in Computer Science, pages 160-173. Springer-Verlag, 2010. [ link ]
In order to develop statistical methods for shapes with a tree-structure, we construct a shape space framework for treelike shapes and study metrics on the shape space. The shape space has singularities, which correspond to topological transitions in the represented trees. We study two closely related metrics, TED and QED. The QED is a quotient euclidean distance arising from the new shape space formulation, while TED is essentially the classical tree edit distance. Using Gromov's metric geometry we gain new insight into the geometries defined by TED and QED. In particular, we show that the new metric QED has nice geometric properties which facilitate statistical analysis, such as existence and local uniqueness of geodesics and averages. TED, on the other hand, has algorithmic advantages, while it does not share the geometric strongpoints of QED. We provide a theoretical framework as well as computational results such as matching of airway trees from pulmonary CT scans and geodesics between synthetic data trees illustrating the dynamic and geometric properties of the QED metric.

 
H.C. Achterberg, F. van der Lijn, T. den Heijer, A. van der Lugt, M.M.B. Breteler, W.J. Niessen, and M. de Bruijne. Prediction of dementia by hippocampal shape analysis. In Fei Wang, Pingkun Yan, Kenji Suzuki, and Dinggang Shen, editors, Machine Learning in Medical Imaging, volume 6357 of Lecture Notes in Computer Science, pages 42-49. Springer, 2010. [ link ]
This work investigates the possibility of predicting future onset of dementia in subjects who are cognitively normal, using hippocampal shape and volume information extracted from MRI scans. A group of 47 subjects who were non-demented normal at the time of the MRI acquisition, but were diagnosed with dementia during a 9 year follow-up period, was selected from a large population based cohort study. 47 Age and gender matched subjects who stayed cognitively intact were selected from the same cohort study as a control group. The hippocampi were automatically segmented and all segmentations were inspected and, if necessary, manually corrected by a trained observer. From this data a statistical model of hippocampal shape was constructed, using an entropy-based particle system. This shape model provided the input for a Support Vector Machine classifier to predict dementia. Cross validation experiments showed that shape information can predict future onset of dementia in this dataset with an accuracy of 70 By incorporating both shape and volume information into the classifier, the accuracy increased to 74%.

 
M.J. Gangeh, L. Sørensen, S.B. Shaker, M.S. Kamel, and M. de Bruijne. Multiple classifier systems in texton-based approach for the classification of CT images of lung. In B. Menze, G. Langs, Z. Tu, and A. Criminisi, editors, Medical Computer Vision: Recognition Techniques and Applications in Medical Imaging, volume 6533 of Lecture Notes in Computer Science, pages 153-163. Springer, 2010. [ link ]
In this paper, we propose using texton signatures based on raw pixel representation along with a parallel multiple classifier system for the classification of emphysema in computed tomography images of the lung. The multiple classifier system is composed of support vector machines on the texton signatures as base classifiers and combines their decisions using product rule. The proposed approach is tested on 168 annotated regions of interest consisting of normal tissue, centrilobular emphysema, and paraseptal emphysema. Texton-based approach in texture classification mainly has two parameters, i.e., texton size and k value in k-means. Our results show that while aggregation of single decisions by SVMs over various k values using multiple classifier systems helps to improve the results compared to single SVMs, combining over different texton sizes is not beneficial. The performance of the proposed system, with an accuracy of 95 to a recently proposed approach based on local binary patterns, which performs almost the best among other approaches in the literature.

 
M.J. Gangeh, L. Sørensen, S. Shaker, M.S. Kamel, M. de Bruijne, and M. Loog. A texton-based approach for the classification of lung parenchyma in CT images. In T. Jiang, N. Navab, J.P.W. Pluim, and M.A. Viergever, editors, Medical Image Computing & Computer-Assisted Intervention, volume 6361 of Lecture Notes in Computer Science, pages 595-602. Springer, 2010. [ link ]
In this paper, a texton-based classification system based on raw pixel representation along with a support vector machine with radial basis function kernel is proposed for the classification of emphysema in computed tomography images of the lung. The proposed approach is tested on 168 annotated regions of interest consisting of normal tissue, centrilobular emphysema, and paraseptal emphysema. The results show the superiority of the proposed approach to common techniques in the literature including moments of the histogram of filter responses based on Gaussian derivatives. The performance of the proposed system, with an accuracy of 96.43%, also slightly improves over a recently proposed approach based on local binary patterns.

 
V. Gorbunova, S.S.A.M. Jacobs, A. Bab-Hadiashar, P. Lo, M. Nielsen, A. Dirksen, and M. de Bruijne. Early detection of emphysema progression. In T. Jiang, N. Navab, J.P.W. Pluim, and M.A. Viergever, editors, Medical Image Computing & Computer-Assisted Intervention, volume 6362 of Lecture Notes in Computer Science, pages 193-200. Springer, 2010. [ link ]
Emphysema is one of the most widespread diseases in subjects with smoking history. The gold standard method for estimating the severity of emphysema is a lung function test, such as forced expiratory volume in first second (FEV1). However, several clinical studies showed that chest CT scans offer more sensitive estimates of emphysema progression. The standard CT densitometric score of emphysema is the relative area of voxels below a threshold (RA). The RA score is a global measurement and reflects the overall emphysema progression. In this work, we propose a framework for estimation of local emphysema progression from longitudinal chest CT scans. First, images are registered to a common system of coordinates and then local image dissimilarities are computed in corresponding anatomical locations. Finally, the obtained dissimilarity representation is converted into a single emphysema progression score. We applied the proposed algorithm on 27 patients with severe emphysema with CT scans acquired five time points, at baseline, after 3, after 12, after 21 and after 24 or 30 months. The results showed consistent emphysema progression with time and the overall progression score correlates significantly with the increase in RA score.

 
C.T. Metz, N. Baka, H.A. Kirisli, M. Schaap, T. van Walsum, S. Klein, L. Neefjes, N.R.A. Mollet, B.P.F. Lelieveldt, M. de Bruijne, and W.J. Niessen. Conditional shape models for cardiac motion estimation. In T. Jiang, N. Navab, J.P.W. Pluim, and M.A. Viergever, editors, Medical Image Computing & Computer-Assisted Intervention, volume 6361 of Lecture Notes in Computer Science, pages 452-459. Springer, 2010. [ link ]
We propose a conditional statistical shape model to predict patient specific cardiac motion from a 3D end-diastolic CTA scan. The model is built from 4D CTA sequences by combining atlas based seg- mentation and 4D registration. Cardiac motion estimation is, for exam- ple, relevant in the dynamic alignment of pre-operative CTA data with intra-operative X-ray imaging. Due to a trend towards prospective elec- trocardiogram (ECG) gating techniques, 4D imaging data, from which this motion information could be extracted, is not commonly available. The prediction of motion from shape information is therefore relevant for this purpose. Evaluation of the accuracy of the predicted motion was per- formed using imaging data of 50 patients, showing an average accuracy of 1.1 mm.

 
L. Sørensen, M. Loog, P. Lo, H. Ashraf, A. Dirksen, R.P.W. Duin, and M. de Bruijne. Image dissimilarity-based quantification of lung disease from CT. In T. Jiang, N. Navab, J.P.W. Pluim, and M.A. Viergever, editors, Medical Image Computing & Computer-Assisted Intervention, volume 6361 of Lecture Notes in Computer Science, pages 37-44. Springer, 2010. [ link ]
In this paper, we propose to classify medical images using dissimilarities computed between collections of regions of interest. The images are mapped into a dissimilarity space using an image dissimilarity measure, and a standard vector space-based classifier is applied in this space. The classification output of this approach can be used in computer aided-diagnosis problems where the goal is to detect the presence of abnormal regions or to quantify the extent or severity of abnormalities in these regions. The proposed approach is applied to quantify chronic obstructive pulmonary disease in computed tomography (CT) images, achieving an area under the receiver operating characteristic curve of 0.817. This is significantly better compared to combining individual region classifications into an overall image classification, and compared to common computerized quantitative measures in pulmonary CT.

 
V. Gorbunova, J. Sporring, P. Lo, A. Dirksen, and M. de Bruijne. Mass preserving image registration: Results of EMPIRE 2010 challenge. In B. van Ginneken, K. Murphy, T. Heimann, V. Pekar, and X. Deng, editors, Medical Image Analysis For The Clinic - A Grand Challenge, pages 155-164, 2010.
 
L. Sørensen, M. Loog, D.M.J. Tax, W. Lee, M. de Bruijne, and R.P.W. Duin. Dissimilarity-based multiple instance learning. In Structural, Syntactic, and Statistical Pattern Recognition (SSPR & SPR), 2010. [ link ]
 
N. Baka, M. de Bruijne, J.H.C. Reiber, W.J. Niessen, and B.P.F. Lelieveldt. Confidence of model based shape reconstruction from sparse data. In IEEE International Symposium on Biomedical Imaging: Macro to Nano (ISBI'10), 2010. [ link ]
Statistical shape models (SSM) are commonly applied for plausible interpolation of missing data in medical imaging. However, when fitting a shape model to sparse information, many solutions may fit the available data. In this paper we derive a constrained SSM to fit noisy sparse input landmarks and assign a confidence value to the resulting reconstructed shape. An evaluation study is performed to compare three methods used for sparse SSM fitting w.r.t. specificity, generalization ability, and correctness of estimated confidence limits with an increasing amount of input information. We find that the proposed constrained shape model outperforms the other models, is robust against the selection and amount of sparse information, and indicates the shape confidence well.

 
V. Gorbunova, S. Durrleman, P. Lo, X. Pennec, and M. de Bruijne. Lung CT registration combining intensity, curves and surfaces. In IEEE International Symposium on Biomedical Imaging: Macro to Nano (ISBI'10), 2010. [ link ]
In this paper we propose a registration method that combines intensity information with geometrical information in the form of curves and surfaces derived from lung CT images. Vessel tree centerlines and lung surfaces were extracted from segmented structures. First, a current-based registration was applied to align the pulmonary vessel tree and the lung surfaces. Subsequently, the resulting deformation field was used to constrain an intensity-based registration method. We applied the combined registration on a set of image pairs, extracted at the end exhale and the end inhale phases of 4DCT scans. The proposed combined registration was compared to intensity-based registration, using a set of manually selected landmarks. The proposed registration decreases the mean and the standard deviation of the target registration errors for all 5 cases to on average 1.47 ± 1.05 mm, compared to the intensity-based registration without constraint 1.74 ± 1.31 mm.

 
S. Klein, M. Loog, F. van der Lijn, T. den Heijer, A. Hammers, M. de Bruijne, A. van der Lugt, R. Duin, M. Breteler, and W. Niessen. Early diagnosis of dementia based on intersubject whole-brain dissimilarities. In IEEE International Symposium on Biomedical Imaging: Macro to Nano (ISBI'10), 2010. [ link ]
 
P. Lo, B. van Ginneken, and M. de Bruijne. Vessel tree extraction using locally optimal paths. In IEEE International Symposium on Biomedical Imaging: Macro to Nano (ISBI'10), 2010. [ link ]
This paper proposes a method to extract vessel trees by continually extending detected branches with locally optimal paths. Our approach uses a cost function from a multiscale vessel enhancement filter. Optimal paths are selected based on rules that take into account the geometric characteristics of the vessel tree. Experiments were performed on 10 low dose chest CT scans for which the pulmonary vessel trees were extracted. The proposed method is shown to extract a better connected vessel tree and extract more of the small peripheral vessels in comparison to applying a threshold on the output of the vessel enhancement filter.

 
M. Ganz, M. de Bruijne, and M. Nielsen. MACD: an imaging marker for cardiovascular disease. In N. Karssemeijer and R.M. Summers, editors, Medical Imaging: Computer-Aided Diagnosis, volume 7624 of Proceedings of SPIE. SPIE Press, 2010. [ link ]
Despite general acceptance that a healthy lifestyle and the treatment of risk factors can prevent the development of cardiovascular diseases (CVD), CVD are the most common cause of death in Europe and the United States. It has been shown that abdominal aortic calcifications (AAC) correlate strongly with coronary artery calcifications. Hence an early detection of aortic calcified plaques helps to predict the risk of related coronary diseases. Also since two thirds of the adverse events have no prior symptoms, possibilities to screen for risk in low cost imaging are important. To this end the Morphological Atherosclerotic Calcification Distribution (MACD) index was developed. In the following several potential severity scores relating to the geometrical outline of the calcified deposits in the lumbar aortic region are introduced. Their individual as well as their combined predictive power is examined and a combined marker, MACD, is constructed. This is done using a Cox regression analysis, also known as survival analysis. Furthermore we show how a Cox regression yields MACD to be the most efficient marker. We also demonstrate that MACD has a larger individual predictive power than any of the other individual imaging markers described. Finally we present that the MACD index predicts cardiovascular death with a hazard ratio of approximately four.

 
J. Petersen, P. Lo, M. Nielsen, G. Edula, H. Ashraf, A. Dirksen, and M. de Bruijne. Quantitative analysis of airway abnormalities in CT. In N. Karssemeijer and R.M. Summers, editors, Medical Imaging: Computer-Aided Diagnosis, volume 7624 of Proceedings of SPIE. SPIE Press, 2010. [ link ]
A coupled surface graph cut algorithm for airway wall segmentation from Computed Tomography (CT) images is presented. Using cost functions that highlight both inner and outer wall borders, the method combines the search for both borders into one graph cut. The proposed method is evaluated on 173 manually segmented images extracted from 15 different subjects and shown to give accurate results, with 37 than the Full Width at Half Maximum (FWHM) algorithm and 62 than a similar graph cut method without coupled surfaces. Common measures of airway wall thickness such as the Interior Area (IA) and Wall Area percentage (WA on a total of 723 CT scans from a lung cancer screening study. These measures were significantly different for participants with Chronic Obstructive Pulmonary Disease (COPD) compared to asymptomatic participants. Furthermore, reproducibility was good as confirmed by repeat scans and the measures correlated well with the outcomes of pulmonary function tests, demonstrating the use of the algorithm as a COPD diagnostic tool. Additionally, a new measure of airway wall thickness is proposed, Normalized Wall Intensity Sum (NWIS). NWIS is shown to correlate better with lung function test values and to be more reproducible than previous measures IA, WA perimeter of 10 mm (PI10).

 
V. Gorbunova, S. Durrleman, P. Lo, X. Pennec, and M. de Bruijne. Curve- and surface-based registration of lung CT images via currents. In M. Brown, M. de Bruijne, B. van Ginneken, A. Kiraly, J.M. Kuhnigk, C. Lorenz, J.R. McClelland, K. Mori, A.P. Reeves, and J. Reinhardt, editors, Proc. of Second International Workshop on Pulmonary Image Analysis, 2009. ISBN-13: 978-1448680894. [ link ]
Feature-based registration methods offer a robust alternative to intensity-based methods when intensities change because of pathology, image artifacts or differences in acquisition. For registration of lung CT images, we propose to use distinctive anatomical structures, such as the pulmonary vessel tree centerlines and lung surfaces, to establish correspondences between pairs of scans. In this respect, we develop and evaluate a curve- and surface-based registration method using currents. This method does not require point correspondence between structures. We conducted experiments on five pairs of images, where each pair consists of image volumes extracted at the end inhale and end exhale phases of a 4D-CT scan. To evaluate the registration, we used a set of 300 anatomical landmarks marked on each image pair. Using both vessel centerlines curves and lung surfaces yields better alignment (median error of 1.85 mm) than using only curves (2.37 mm) or surfaces (3.53 mm). The combined method achieves overall registration accuracy comparable to that of intensity-based registration, whereas the errors are made in different locations. This suggests that low dimensional geometrical features capture sufficient information to drive a reliable registration, while results can still be improved by combining intensity and feature based registration approaches into one framework.

 
P. Lo, B. van Ginneken, J.M. Reinhardt, and M. de Bruijne. Extraction of airways from CT (EXACT’09). In M. Brown, M. de Bruijne, B. van Ginneken, A. Kiraly, J.M. Kuhnigk, C. Lorenz, J.R. McClelland, K. Mori, A.P. Reeves, and J. Reinhardt, editors, Proc. of Second International Workshop on Pulmonary Image Analysis, pages 175-189, 2009. ISBN-13: 978-1448680894. [ link ]
This paper describes a framework for evaluating airway extraction algorithms in a standardized manner and establishing reference segmentations that can be used for future algorithm development. Because of the sheer difficulty of constructing a complete reference standard manually, we propose to construct a reference using results from the algorithms being compared, by splitting each airway tree segmentation result into individual branch segments that are subsequently visually inspected by trained observers. Using the so constructed reference, a total of seven performance measures covering different aspects of segmentation quality are computed. We evaluated 15 airway tree extraction algorithms from different research groups on a diverse set of 20 chest CT scans from subjects ranging from healthy volunteers to patients with severe lung disease, who were scanned at different sites, with several different CT scanner models, and using a variety of scanning protocols and reconstruction parameters.

 
P. Lo, J. Sporring, and M. de Bruijne. Multiscale vessel-guided airway tree segmentation. In M. Brown, M. de Bruijne, B. van Ginneken, A. Kiraly, J.M. Kuhnigk, C. Lorenz, J.R. McClelland, K. Mori, A.P. Reeves, and J. Reinhardt, editors, Proc. of Second International Workshop on Pulmonary Image Analysis, 2009. ISBN-13: 978-1448680894. [ link ]
This paper presents a method for airway tree segmentation that uses a combination of a trained airway appearance model, vessel and airway orientation information, and region growing. The method uses a voxel classification based appearance model, which involves the use of a classifier that is trained to differentiate between airway and non-airway voxels. Vessel and airway orientation information are used in the form of a vessel orientation similarity measure, which indicates how similar the orientation of the an airway candidate is to the orientation of the neighboring vessel. The method is evaluated within EXACT’09 on a diverse set of CT scans. Results show a favorable combination of a relatively large portion of the tree detected correctly with very few false positives.

 
P. Lo, J. Sporring, J.H. Pedersen, and M. de Bruijne. Airway tree extraction with locally optimal paths. In D. Hawkes, D. Rueckert, and G. Yang, editors, Medical Image Computing & Computer-Assisted Intervention, volume 5762 of Lecture Notes in Computer Science, pages 51-58. Springer, 2009. [ link | pdf ]
This paper proposes a method to extract the airway tree from CT images by continually extending the tree with locally optimal paths. This is in contrast to commonly used region growing based approaches that only search the space of the immediate neighbors. The result is a much more robust method for tree extraction that can overcome local occlusions. The cost function for obtaining the optimal paths takes into account of an airway probability map as well as measures of airway shape and orientation derived from multi-scale Hessian eigen analysis on the airway probability. Significant improvements were achieved compared to a region growing based method, with up to 36% longer trees at a slight increase of false positive rate.

 
K. Murphy, B. van Ginneken, E.M van Rikxoort, B.J. de Hoop, M. Prokop, P. Lo, M. de Bruijne, and J.P.W. Pluim. Obstructive pulmonary function: Patient classification using 3D registration of inspiration and expiration CT images. In M. Brown, M. de Bruijne, B. van Ginneken, A. Kiraly, J.M. Kuhnigk, C. Lorenz, J.R. McClelland, K. Mori, A.P. Reeves, and J. Reinhardt, editors, Proc. of Second International Workshop on Pulmonary Image Analysis, 2009. ISBN-13: 978-1448680894. [ link ]
 
L. Sørensen, P. Lo, H. Ashraf, J. Sporring, M. Nielsen, and M. de Bruijne. Learning COPD sensitive filters in pulmonary CT. In D. Hawkes, D. Rueckert, and G. Yang, editors, Medical Image Computing & Computer-Assisted Intervention, Lecture Notes in Computer Science, pages 699-706. Springer, 2009. [ link | pdf ]
The standard approaches to analyzing emphysema in computed tomography (CT) images are visual inspection and the relative area of voxels below a threshold (RA). The former approach is subjective and impractical in a large data set and the latter relies on a single threshold and independent voxel information, ignoring any spatial correlation in intensities. In recent years, supervised learning on texture features has been investigated as an alternative to these approaches, showing good results. However, supervised learning requires labeled samples, and these samples are often obtained via subjective and time consuming visual scoring done by human experts. In this work, we investigate the possibility of applying supervised learning using texture measures on random CT samples where the labels are based on external, non-CT measures. We are not targeting emphysema directly, instead we focus on learning textural differences that discriminate subjects with chronic obstructive pulmonary disease (COPD) from healthy smokers, and it is expected that emphysema plays a major part in this. The proposed texture based approach achieves an 69% classification accuracy which is significantly better than RA's 55% accuracy.

 
A. Crimi, J. Sporring, M. de Bruijne, M. Lillholm, and M. Nielsen. Prior knowledge regularization in statistical medical image tasks. In MICCAI Workshop on Probabilistic Models for Medical Image Analysis, 2009.
 
M. Loog and M. de Bruijne. Discriminative shape alignment. In Jerry L. Prince, Dzung L. Pham, and Kyle J. Myers, editors, Information Processing in Medical Imaging, volume 5636 of Lecture Notes in Computer Science. Springer, 2009. [ link ]
The alignment of shape data to a common mean before its subsequent processing is an ubiquitous step within the area shape analysis. Current approaches to shape analysis or, as more specifically considered in this work, shape classification perform the alignment in a fully unsupervised way, not taking into account that eventually the shapes are to be assigned to two or more different classes. This work introduces a discriminative variation to well-known Procrustes alignment and demonstrates its benefit over this classical method in shape classification tasks. The focus is on two-dimensional shapes from a two-class recognition problem.

 
F. van der Lijn, M. de Bruijne, Y.Y. Hoogendam, S. Klein, R. Hameeteman, M. Breteler, and W. Niessen. Cerebellum segmentation in MRI using atlas registration and local multi-scale image descriptors. In IEEE International Symposium on Biomedical Imaging: Macro to Nano (ISBI'09), pages 221-224, 2009. [ link ]
We propose a novel cerebellum segmentation method for MRI, based on a combination of statistical models of the structure's expected location in the brain and its local appearance. The appearance model is obtained from a k-nearest-neighbor classifier, which uses a set of multi-scale local image descriptors as features. The spatial model is constructed by registering multiple manually annotated datasets to the unlabeled target image. The two components are then combined in a Bayesian framework. The method is quantitatively validated in a leave-one-out experiment using 18 MR images of elderly subjects. The experiment showed that the method produces accurate segmentations. The mean Dice similarity index compared to the manual reference was 0.953 for left and right, and the mean surface distance was 0.49 mm for left and 0.50 mm for right. The combined atlas- and appearance-based method was found to be more accurate than a method based on atlas-registration alone.

 
S. Sommer, A. Tatu, C. Chen, D.R. Jørgensen, M. de Bruijne, M. Loog, M. Nielsen, and F. Lauze. Bicycle chain shape models. In D. Metaxas and C. Kambhamettu, editors, IEEE Workshop on Mathematical Methods in Biomedical Image Analysis, 2009. [ link ]
In this paper we introduce landmark-based preshapes which allow mixing of anatomical landmarks and pseudo-landmarks, constraining consecutive pseudo-landmarks to satisfy planar equidistance relations. This defines naturally a structure of Riemannian manifold on these preshapes, with a natural action of the group of planar rotations. Orbits define the shapes. We develop a Geodesic Generalized Procrustes Analysis procedure for a sample set on such a preshape spaces and use it to compute Principal Geodesic Analysis. We demonstrate it on an elementary synthetic example as well on a dataset of manually annotated vertebra shapes from X-ray. We re-landmark them consistently and show that PGA captures the variability of the dataset better than its linear counterpart, PCA.

 
V. Gorbunova, P. Lo, M. Loeve, H.A. Tiddens, J. Sporring, M. Nielsen, and M. de Bruijne. Mass preserving registration for lung CT. In J.P.W. Pluim and B.M. Dawant, editors, Medical Imaging: Image Processing, volume 7259 of Proceedings of SPIE. SPIE Press, 2009. [ link ]
In this paper, we evaluate a novel image registration method on a set of expiratory-inspiratory pairs of computed tomography (CT) lung scans. A free-form multi resolution image registration technique is used to match two scans of the same subject. To account for the differences in the lung intensities due to differences in inspiration level, we propose to adjust the intensity of lung tissue according to the local expansion or compression. An image registration method without intensity adjustment is compared to the proposed method. Both approaches are evaluated on a set of 10 pairs of expiration and inspiration CT scans of children with cystic fibrosis lung disease. The proposed method with mass preserving adjustment results in significantly better alignment of the vessel trees. Analysis of local volume change for regions with trapped air compared to normally ventilated regions revealed larger differences between these regions in the case of mass preserving image registration, indicating that mass preserving registration is better at capturing localized differences in lung deformation.

 
L. Sørensen and M. de Bruijne. Dissimilarity representations in lung parenchyma classification. In N. Karssemeijer and M.L. Giger, editors, Medical Imaging: Computer-Aided Diagnosis, volume 7260 of Proceedings of SPIE. SPIE Press, 2009. [ link | pdf ]
A good problem representation is important for a pattern recognition system to be successful. The traditional approach to statistical pattern recognition is feature representation. More specifically, objects are represented by a number of features in a feature vector space, and classifiers are built in this representation. This is also the general trend in lung parenchyma classification in computed tomography (CT) images, where the features often are measures on feature histograms. Instead, we propose to build normal density based classifiers in dissimilarity representations for lung parenchyma classification. This allows for the classifiers to work on dissimilarities between objects, which might be a more natural way of representing lung parenchyma. In this context, dissimilarity is defined between CT regions of interest (ROI)s. ROIs are represented by their CT attenuation histogram and ROI dissimilarity is defined as a histogram dissimilarity measure between the attenuation histograms. In this setting, the full histograms are utilized according to the chosen histogram dissimilarity measure. We apply this idea to classification of different emphysema patterns as well as normal, healthy tissue. Two dissimilarity representation approaches as well as different histogram dissimilarity measures are considered. The approaches are evaluated on a set of 168 CT ROIs using normal density based classifiers all showing good performance. Compared to using histogram dissimilarity directly as distance in a k nearest neighbor classifier, which achieves a classification accuracy of 92.9 is significantly better with a classification accuracy of 97.0 = 0.046).

 
V. Gorbunova, P. Lo, H. Ashraf, A. Dirksen, M. Nielsen, and M. de Bruijne. Weight preserving image registration for monitoring disease progression in lung CT. In D. Metaxas, L. Axel, G. Fichtinger, and G. Székely, editors, Medical Image Computing & Computer-Assisted Intervention, volume 5242 of Lecture Notes in Computer Science, pages 863-870. Springer, 2008. [ link | pdf ]
We present a new image registration based method for monitoring regional disease progression in longitudinal image studies of lung disease. A free-form image registration technique is used to match a baseline 3D CT lung scan onto a following scan. Areas with lower intensity in the following scan compared with intensities in the deformed baseline image indicate local loss of lung tissue that is associated with progression of emphysema. To account for differences in lung intensity owing to differences in the inspiration level in the two scans rather than disease progression, we propose to adjust the density of lung tissue with respect to local expansion or compression such that the total weight of the lungs is preserved during deformation. Our method provides a good estimation of regional destruction of lung tissue for subjects with a significant difference in inspiration level between CT scans and may result in a more sensitive measure of disease progression than standard quantitative CT measures.

 
P. Lo, J. Sporring, H. Ashraf, J.H. Pedersen, and M. de Bruijne. Vessel-guided airway segmentation based on voxel classification. In M. Brown, M. de Bruijne, B. van Ginneken, A. Kiraly, J.M. Kuhnigk, C. Lorenz, K. Mori, and J. Reinhardt, editors, Proc. of First International Workshop on Pulmonary Image Analysis, 2008. ISBN-13: 978-1435759527. [ link ]
This paper presents a method for improving airway tree segmentation using vessel orientation information. We use the fact that an airway branch is always accompanied by an artery, with both structures having similar orientations. This work is based on a voxel classification airway segmentation method proposed previously. The probability of a voxel belonging to the airway, from the voxel classification method, is augmented with an orientation similarity measure as a criterion for region growing. The orientation similarity measure of a voxel indicates how similar is the orientation of the surroundings of a voxel is to that of a neighboring vessel. The proposed method is tested on 20 CT images from different subjects selected randomly from a lung cancer screening study. The proposed method improves the results significantly (p = 0.0125) in terms of longer airway branches extracted, as compared to only using probability from the voxel classification method.

 
L. Sørensen, S. Shaker, and M. de Bruijne. Classification of lung texture in CT using local binary patterns. In D. Metaxas, L. Axel, G. Fichtinger, and G. Székely, editors, Medical Image Computing & Computer-Assisted Intervention, volume 5241 of Lecture Notes in Computer Science, pages 934-941. Springer, 2008. [ link | pdf ]
In this paper we propose to use local binary patterns (LBP) as features in a classification framework for classifying different texture patterns in lung computed tomography. Image intensity is included by means of the joint LBP and intensity histogram, and classification is performed using the k nearest neighbor classifier with histogram similarity as distance measure. The proposed method is evaluated on a set of 168 regions of interest comprising normal tissue and different emphysema patterns, and compared to a filter bank based on Gaussian derivatives. The joint LBP and intensity histogram, achieving a classification accuracy of 95.2%, shows superior performance to using the common approach of taking moments of the filter response histograms as features, and slightly better performance than using the full filter response histograms instead. Classification results are better than some of those previously reported in the literature.

 
L. Sørensen, S. Shaker, and M. de Bruijne. Texture based emphysema quantification in lung CT. In M. Brown, M. de Bruijne, B. van Ginneken, A. Kiraly, J.M. Kuhnigk, C. Lorenz, K. Mori, and J. Reinhardt, editors, Proc. of First International Workshop on Pulmonary Image Analysis, 2008. ISBN-13: 978-1435759527. [ link ]
In this paper we propose to use texture based pixel classification in lung computed tomography (CT) for measuring emphysema. Two quantitative parameters for emphysema, based on the pixel classification, are suggested; relative class area and mean class posterior. The approach is evaluated on a group of 39 patients, of whom 20 have been diagnosed with chronic obstructive pulmonary disease, using two different feature groups, local binary patterns and a filter bank based on Gaussian derivatives. The pixel classification based quantitative parameters correlate well with lung function (r = 0.80, p < 10-5 for the parameter with the highest correlation) and correlate significantly better than the most commonly used CT based emphysema quantification method, namely relative area of low attenuation.

 
M. de Bruijne and P.C. Pettersen. Supervised shape analysis for risk assessment in osteoporosis. In IEEE International Symposium on Biomedical Imaging: Macro to Nano (ISBI'08), pages 1581-1584, 2008. [ link | pdf ]
Early diagnosis and treatment of patients at high risk of developing fragility fractures is crucial in the management of osteoporosis. In this paper we propose to estimate the risk of future vertebral fractures using a training set of longitudinal data to learn the shape characteristics of vertebrae and spines that will sustain a fracture in the near future. A discriminant classifier is trained to discriminate between subjects developing one or more vertebral fractures in the course of 5 years and subjects maintaining a healthy spine. This approach is compared to a one-class system where the classifier is trained only on the subjects staying healthy. In a case-control study with 218 subjects, all unfractured at baseline and matched for main vertebral fracture risk factors such as spine BMD and age, we were able to predict future fractures with a sensitivity of 76% and a specificity of 72%.

 
P. Lo and M. de Bruijne. Voxel classification-based airway tree segmentation. In J.M. Reinhardt and J.P. Pluim, editors, Medical Imaging: Image Processing, volume 6914 of Proceedings of SPIE. SPIE Press, 2008. [ link | pdf ]
This paper presents a voxel classification based method for segmenting the human airway tree in volumetric computed tomography (CT) images. In contrast to standard methods that use only voxel intensities, our method uses a more complex appearance model based on a set of local image appearance features and Kth nearest neighbor (KNN) classification. The optimal set of features for classification is selected automatically from a large set of features describing the local image structure at several scales. The use of multiple features enables the appearance model to differentiate between airway tree voxels and other voxels of similar intensities in the lung, thus making the segmentation robust to pathologies such as emphysema. The classifier is trained on imperfect segmentations that can easily be obtained using region growing with a manual threshold selection. Experiments show that the proposed method results in a more robust segmentation that can grow into the smaller airway branches without leaking into emphysematous areas, and is able to segment many branches that are not present in the training set.

 
E.M. van Rikxoort, M. de Bruijne, and B. van Ginneken. Integrating local voxel classification and global shape models for medical image segmentation. In J.M. Reinhardt and J.P. Pluim, editors, Medical Imaging: Image Processing, volume 6914 of Proceedings of SPIE. SPIE Press, 2008. [ link ]
Segmentation of anatomical structures is a prerequisite for many medical image analysis tasks. We propose a method that integrates local voxel classification and global shape models. The method starts by computing a local feature vector for every voxel and mapping this, via a classifier trained from example segmentations, to a probability that the voxel belongs to the structure to be segmented. Next, this probabilistic output is entered into a global shape model. This shape model is constructed by mapping aligned blurred versions of reference segmentations of the training data into a vector space and applying principal component analysis (PCA). The mapping onto a vector space that is applied guarantees valid results from the PCA. An advantage of using such a shape model is that there is no need to define corresponding landmarks on all training scans, which is a hard task on 3D data. Segmentation of unseen test data is performed by a least squares fit of the results of the voxel classification, after alignment and blurring, into the PCA space. The result of this procedure is for each voxel a probability that it belongs to the structure to be segmented conditioned on both local and global information. We demonstrate the effectiveness of the method on segmentation of lungs containing pathologic abnormalities in 3D CT data.

 
L. Sørensen, J. Ostergaard, N. Jørgensen, P. Johansen, and M. de Bruijne. Multiobject tracking of human spermatozoa. In J.M. Reinhardt and J.P. Pluim, editors, Medical Imaging: Image Processing, volume 6914 of Proceedings of SPIE. SPIE Press, 2008. [ link ]
We propose a system for tracking of human spermatozoa in phase-contrast microscopy image sequences. One of the main aims of a computer-aided sperm analysis (CASA) system is to automatically assess sperm quality based on spermatozoa motility variables. In our case, the problem of assessing sperm quality is cast as a multi-object tracking problem, where the objects being tracked are the spermatozoa. The system combines a particle filter and Kalman filters for robust motion estimation of the spermatozoa tracks. Further, the combinatorial aspect of assigning observations to labels in the particle filter is formulated as a linear assignment problem solved using the Hungarian algorithm on a rectangular cost matrix, making the algorithm capable of handling missing or spurious observations. The costs are calculated using hidden Markov models that express the plausibility of an observation being the next position in the track history of the particle labels. Observations are extracted using a scale-space blob detector utilizing the fact that the spermatozoa appear as bright blobs in a phase-contrast microscope. The output of the system is the complete motion track of each of the spermatozoa. Based on these tracks, different CASA motility variables can be computed, for example curvilinear velocity or straight-line velocity. The performance of the system is tested on three different phase-contrast image sequences of varying complexity, both by visual inspection of the estimated spermatozoa tracks and by measuring the mean squared error (MSE) between the estimated spermatozoa tracks and manually annotated tracks, showing good agreement.

 
L.A. Conrad-Hansen, M. de Bruijne, F.B. Lauze, L.B. Tankó, P.C. Pettersen, Q. He, J. Chen, C. Christiansen, and M. Nielsen. Quantifying calcification in the lumbar aorta on X-ray images. In N. Ayache, S. Ourselin, and A. Maeder, editors, Medical Image Computing & Computer-Assisted Intervention, volume 4792 of Lecture Notes in Computer Science, pages 352-359. Springer, 2007. [ link | pdf ]
In this paper we propose to use inpainting to estimate the severity of atherosclerotic plaques from X-ray projections. Inpainting allows to “remove” the plaque and estimate what the background image for an uncalcified aorta would have looked like. A measure of plaque severity can then be derived by subtracting the inpainting from the orig- inal image. In contrast to the current standard of categorical calcification scoring from X-rays, our method estimates both the size and the density of calcified areas and provides a continuous severity score, thus allowing for measurement of more subtle differences. We discuss a class of smooth inpainting methods, compare their ability to reconstruct the original images, and compare the inpainting based calcification score to the conventional categorical score in a longitudinal study on 49 patients addressing correlations of the calcification scores with hypertension, a known cardiovascular risk factor.

 
J.E. Iglesias, M. de Bruijne, M. Loog, F.B. Lauze, and M. Nielsen. A family of principal component analyses for dealing with outliers. In N. Ayache, S. Ourselin, and A. Maeder, editors, Medical Image Computing & Computer-Assisted Intervention, volume 4792 of Lecture Notes in Computer Science, pages 178-185. Springer, 2007. [ link | pdf ]
Principal Component Analysis (PCA) has been widely used for dimensionality reduction in shape and appearance modeling. There have been several attempts of making PCA robust against outliers. However, there are cases in which a small subset of samples may appear as outliers and still correspond to plausible data. The example of shapes corresponding to fractures when building a vertebra shape model is addressed in this study. In this case, the modeling of "outliers" is important, and it might be desirable not only not to disregard them, but even to enhance their importance. A variation on PCA that deals naturally with the importance of outliers is presented in this paper. The technique is utilized for building a shape model of a vertebra, aiming at segmenting the spine out of lateral X-ray images. The results show that the algorithm can implement both an outlier-enhancing and a robust PCA. The former improves the segmentation performance in fractured vertebrae, while the latter does so in the unfractured ones.

 
M. de Bruijne, P.C. Pettersen, L.B. Tankó, and M. Nielsen. Vertebral fracture classification. In J.P. Pluim and J.M. Reinhardt, editors, Medical Imaging: Image Processing, volume 6512 of Proceedings of SPIE. SPIE Press, 2007. [ link | pdf ]
A novel method for classification and quantification of vertebral fractures from X-ray images is presented. Using pairwise conditional shape models trained on a set of healthy spines, the most likely unfractured shape is estimated for each of the vertebrae in the image. The difference between the true shape and the reconstructed normal shape is an indicator for the shape abnormality. A statistical classification scheme with the two shapes as features is applied to detect, classify, and grade various types of deformities. In contrast with the current (semi-)quantitative grading strategies this method takes the full shape into account, it uses a patient-specific reference by combining population-based information on biological variation in vertebra shape and vertebra interrelations, and it provides a continuous measure of deformity. Good agreement with manual classification and grading is demonstrated on 204 lateral spine radiographs with in total 89 fractures.

 
L.A. Conrad-Hansen, M. de Bruijne, M. Nielsen, and M. Petrou. Semi-automatic detection of calcification using non-linear stretching on X-ray images. In M.L. Giger and N. Karssemeijer, editors, Medical Imaging: Computer-Aided Diagnosis, volume 6514 of Proceedings of SPIE. SPIE Press, 2007. [ link ]
A central problem in the development of a mass-screening tool for atherosclerotic plaque is automatic calcification detection. The mass-screening aspect implies that the detection process should be fast and reliable. In this paper we present a first step in this direction by introducing a semi-automatic calcification classification tool based on non-linear stretching, an image enhancement method that focusses on local image statistics. The calcified areas are approximated by a coarse brush, which in our case is mimicked by taking the ground truth, provided by radiologists, and dilating it with circular structuring elements of varying sizes. Thresholds are then examined for the different structuring elements, that yield optimal results on the enhanced image. The results in this preliminary study which contains 19 images of varying calcification degree, fully annotated by medical experts, show a significant increase in accuracy when the methodology is validated on a region of interest containing the areas of a simulated coarse brush.

 
F.B. Lauze and M. de Bruijne. Toward automated detection and segmentation of aortic calcifications from radiographs. In J.P. Pluim and J.M. Reinhardt, editors, Medical Imaging: Image Processing, volume 6512 of Proceedings of SPIE. SPIE Press, 2007. [ link | pdf ]
This paper aims at automatically measuring the extent of calcified plaques in the lumbar aorta from standard radiographs. Calcifications in the abdominal aorta are an important predictor for future cardiovascular morbidity and mortality. Accurate and reproducible measurement of the amount of calcified deposit in the aorta is therefore of great value in disease diagnosis and prognosis, treatment planning, and the study of drug effects. We propose a two-step approach in which first the calcifications are detected by an iterative statistical pixel classification scheme combined with aorta shape model optimization. Subsequently, the detected calcified pixels are used as the initialization for an inpainting based segmentation. We present results on synthetic images from the inpainting based segmentation as well as results on several X-ray images based on the two-steps approach.

 
M. de Bruijne, M.T. Lund, L.B. Tankó, P.C. Pettersen, and M. Nielsen. Quantitative vertebral morphometry using neighbor-conditional shape models. In R. Larsen, M. Nielsen, and J. Sporring, editors, Medical Image Computing & Computer-Assisted Intervention, volume 4190 of Lecture Notes in Computer Science, pages 1-8. Springer, 2006. [ link | pdf ]
A novel method for vertebral fracture quantification from X-ray images is presented. Using pairwise conditional shape models trained on a set of healthy spines, the most likely normal vertebra shapes are estimated conditional on all other vertebrae in the image. The differences between the true shape and the reconstructed normal shape is subsequently used as a measure of abnormality. In contrast with the current (semi-)quantitative grading strategies this method takes the full shape into account, it uses a patient-specific reference by combining population-based information on biological variation in vertebra shape and vertebra interrelations, and it provides a continuous measure of deformity. The method is demonstrated on 212 lateral spine radiographs with in total 78 fractures. The distance between prediction and true shape is 1.0 mm for unfractured vertebrae and 3.7 mm for fractures, which makes it possible to diagnose and assess the severity of a fracture.

 
M. de Bruijne. Shape particle guided tissue classification. In P. Golland and D. Rueckert, editors, IEEE Workshop on Mathematical Methods in Biomedical Image Analysis, 2006. [ link | pdf ]
In many cases, the accuracy of statistical pixel classification can be improved by applying a spatially varying prior that can be derived from a shape model. We propose to represent the prior knowledge on the spatial distribution of tissue classes by a distribution of shape particles, each representing one possible distribution of tissue classes. Classification and shape can then be optimized jointly by alternating a particle filtering step, in which the shape particle distribution is evolved under the influence of the current classification, with an update of the classification estimate using the shape distribution. Since a large number of shape hypotheses is used this method does not easily get trapped in local maxima. By applying shape models that are conditional on other, more easily discernible, objects in the image one can perform shape guided classification even if the shapes themselves are hardly visible. The method is demonstrated on the task of detecting aortic calcifications in X-ray images, in which calcifications can only be present inside the aorta — mainly on the aortic wall — but the aorta itself is not visible.

 
L.A. Conrad-Hansen, M. de Bruijne, F.B. Lauze, L.B. Tankó, and M. Nielsen. A pixelwise inpainting based refinement scheme for quantizing calcification in the lumbar aorta on 2D lateral X-ray images. In J.M. Reinhardt and J.P. Pluim, editors, Medical Imaging: Image Processing, volume 6144 of Proceedings of SPIE, pages 474-484. SPIE Press, 2006. [ link ]
In this paper we seek to improve the standard method of assessing the degree of calcification in the lumbar aorta visualized on lateral 2-D X-rays. The semiquantitative method does not take density of calcification within the individual plaques into account and is unable to measure subtle changes in the severity of calcification over time. Both of these parameters would be desirable to assess, since they are the keys to assessing important information on the impact of risk factors and candidate drugs aiming at the prevention of atherosclerosis. As a further step for solving this task, we propose a pixelwise inpainting-based refinement scheme that seeks to optimize the individual plaque shape by maximizing the signal-to-noise ratio. Contrary to previous work the algorithm developped for this study uses a sorted candidate list, which omits possible bias introduced by the choice of starting pixel. The signal-to-noise optimization scheme will be discussed in different settings using TV as well as Harmonic inpainting and comparing these with a simple averaging process.

 
M. de Bruijne. A pattern classification approach to aorta calcium scoring in radiographs. In Y. Liu, T. Jiang, and C. Zhang, editors, Proceedings of the ICCV Workshop on Computer Vision for Biomedical Image Applications: Current Techniques and Future Trends (CVBIA), volume 3765 of Lecture Notes in Computer Science, pages 170-177. Springer, 2005. [ link | pdf ]
A method for automated detection of calcifications in the abdominal aorta from standard X-ray images is presented. Pixel classification based on local image structure is combined with a spatially varying prior that is derived from a statistical model of the combined shape variation in aorta and spine. Leave-one-out experiments were performed on 87 standard lateral lumbar spine X-rays, resulting in on average 93.7% of the pixels within the aorta being correctly classified.

 
L.A. Conrad-Hansen, M. de Bruijne, F.B. Lauze, L.B. Tankó, and M. Nielsen. Quantizing calcification in the lumbar aorta on 2-D lateral X-ray images. In Y. Liu, T. Jiang, and C. Zhang, editors, Proceedings of the ICCV Workshop on Computer Vision for Biomedical Image Applications: Current Techniques and Future Trends (CVBIA), volume 3765 of Lecture Notes in Computer Science, pages 409-418. Springer, 2005. [ link ]
In this paper we seek to improve the standard method of assessing the degree of calcification in the lumbar aorta visualized on lateral 2-D X-rays. The semiquantitative method does not take density of calcification within the individual plaques into account and is unable to measure subtle changes in the severity of calcification over time. Both of these parameters would be desirable to access, since they are the keys to assessing important information on the impact of risk factors and candidate drugs aiming at the prevention of atherosclerosis. Herein we propose to estimate the background of a calcification using inpainting, a technique used in image restoration as well as postprocessing of film, and measure the plaque density as the difference between the background estimation and the original image. Furthermore, we compare total variation inpainting with harmonic inpainting and discuss the potential implications of inpainting for characterizing aortic calcification.

 
M. de Bruijne and M. Nielsen. Multi-object segmentation using shape particles. In G.E. Christensen and M. Sonka, editors, Information Processing in Medical Imaging, volume 3565 of Lecture Notes in Computer Science, pages 762-773. Springer, 2005. [ link | pdf ]
Deformable template models, in which a shape model and its corresponding appearance model are deformed to optimally fit an object in the image, have proven successful in many medical image segmentation tasks. In some applications, the number of objects in an image is not known a priori. In that case not only the most clearly visible object must be extracted, but the full collection of objects present in the image. We propose a stochastic optimization algorithm that optimizes a distribution of shape particles so that the overall distribution explains as much of the image as possible. Possible spatial interrelationships between objects are modelled and used to steer the evolution of the particle set by generating new shape hypotheses that are consistent with the shapes currently observed. The method is evaluated on rib segmentation in chest X-rays.

 
M.T. Lund, M. de Bruijne, L.B. Tankó, and M. Nielsen. Shape regression for vertebra fracture quantification. In M.J. Fitzpatrick and J.M. Reinhardt, editors, Medical Imaging: Image Processing, volume 5747 of Proceedings of SPIE, pages 723-731. SPIE Press, 2005. [ link ]
Accurate and reliable identification and quantification of vertebral fractures constitute a challenge both in clinical trials and in diagnosis of osteoporosis. Various efforts have been made to develop reliable, objective, and reproducible methods for assessing vertebral fractures, but at present there is no consensus concerning a universally accepted diagnostic definition of vertebral fractures. In this project we want to investigate whether or not it is possible to accurately reconstruct the shape of a normal vertebra, using a neighbouring vertebra as prior information. The reconstructed shape can then be used to develop a novel vertebra fracture measure, by comparing the segmented vertebra shape with its reconstructed normal shape. The vertebrae in lateral x-rays of the lumbar spine were manually annotated by a medical expert. With this dataset we built a shape model, with equidistant point distribution between the four corner points. Based on the shape model, a multiple linear regression model of a normal vertebra shape was developed for each dataset using leave-one-out cross-validation. The reconstructed shape was calculated for each dataset using these regression models. The average prediction error for the annotated shape was on average 3%.

 
M. de Bruijne and M. Nielsen. Shape particle filtering for image segmentation. In C. Barillot, D.R. Haynor, and P. Hellier, editors, Medical Image Computing & Computer-Assisted Intervention, volume 3216 of Lecture Notes in Computer Science, pages 186-175. Springer, 2004. [ link | pdf ]
Deformable template models are valuable tools in medical image segmentation. Current methods elegantly incorporate global shape and appearance, but can not cope with localized appearance variations and rely on an assumption of Gaussian gray value distribution. Furthermore, initialization near the optimal solution is required. We propose a maximum likelihood shape inference that is based on pixel classification, so that local and non-linear intensity variations are dealt with naturally, while a global shape model ensures a consistent segmentation. Optimization by stochastic sampling removes the need for accurate initialization. The method is demonstrated on three different medical image segmentation problems: vertebra segmentation in spine radiographs, lung field segmentation in thorax X rays, and delineation of the myocardium of the left ventricle in MRI slices. Accurate results were obtained in all tasks.

 
M. de Bruijne and M. Nielsen. Image segmentation by shape particle filtering. In J. Kittler, M. Petrou, and M. Nixon, editors, International Conference on Pattern Recognition, pages III:722-725. IEEE Computer Society Press, 2004. [ link | pdf ]
Statistical appearance models are valuable tools in medical image segmentation. Current methods elegantly incorporate global shape and appearance, but can not cope with local appearance variations and rely on an assumption of Gaussian gray value distribution. Furthermore, initialization near the optimal solution is required. We propose a shape inference method that is based on pixel classification, so that local and non-linear intensity variations are dealt with naturally, while a global shape model ensures a consistent segmentation. Optimization by stochastic sampling removes the need for accurate initialization. The method is demonstrated on vertebra segmentation in spine radiographs. Segmentation errors are below 2 mm in 88 out of 91 cases, with an average error of 1.4 mm.

 
M. de Bruijne, B. van Ginneken, W. Bartels, M.J. van der Laan, J.D. Blankensteijn, W.J. Niessen, and M.A. Viergever. Automated segmentation of abdominal aortic aneurysms in multi-spectral MR images. In T. Peters and R. Ellis, editors, Medical Image Computing & Computer-Assisted Intervention, volume 2879 of Lecture Notes in Computer Science, pages 538-545. Springer, 2003. [ link | pdf ]
An automated method for segmenting the outer boundary of abdominal aortic aneurysms in MR images is presented. The method is based on the well known Active Shape Models (ASM), which fit a global landmarkbased shape model on the basis of local boundary appearance models. The original three-dimensional ASM scheme is modified to deal with multi-spectral image information and inconsistent boundary appearance in a principled way, with only a limited amount of training data. In addition, a framework for user interaction is proposed. If required, the obtained segmentation can be corrected in an interactive manner by indicating points on the desired boundary. The methods are evaluated in leave-one-out experiments on 21 datasets. A segmentation scheme combining gray level information from two or three MR sequences produces significantly better results than a single-scan model. Average volume errors with respect to the manual segmentation are 4.0 In the cases in which the obtained error is large, results can easily be improved using the interactive scheme.

 
M. de Bruijne, B. van Ginneken, M.A. Viergever, and W.J. Niessen. Adapting active shape models for 3D segmentation of tubular structures in medical images. In C. Taylor and A. Noble, editors, Information Processing in Medical Imaging, volume 2732 of Lecture Notes in Computer Science, pages 136-147. Springer, 2003. [ link | pdf ]
Active Shape Models (ASM) have proven to be an effective approach for image segmentation. In some applications, however, the linear model of gray level appearance around a contour that is used in ASM is not sufficient for accurate boundary localization. Furthermore, the statistical shape model may be too restricted if the training set is limited. This paper describes modifications to both the shape and the appearance model of the original ASM formulation. Shape model flexibility is increased, for tubular objects, by modeling the axis deformation independent of the cross-sectional deformation, and by adding supplementary cylindrical deformation modes. Furthermore, a novel appearance modeling scheme that effectively deals with a highly varying background is developed. In contrast with the conventional ASM approach, the new appearance model is trained on both boundary and non-boundary points, and the probability that a given point belongs to the boundary is estimated non-parametrically. The methods are evaluated on the complex task of segmenting thrombus in abdominal aortic aneurysms (AAA). Shape approximation errors were successfully reduced using the two shape model extensions. Segmentation using the new appearance model significantly outperformed the original ASM scheme; average volume errors are 5.1% and 45% respectively.

 
M. de Bruijne, B. van Ginneken, W.J. Niessen, M. Loog, and M.A. Viergever. Model-based segmentation of abdominal aortic aneurysms in CTA images. In M. Sonka and M.J. Fitzpatrick, editors, Medical Imaging: Image Processing, volume 5032 of Proceedings of SPIE, pages 1560-1571. SPIE Press, 2003. [ link | pdf ]
Segmentation of thrombus in abdominal aortic aneurysms is complicated by regions of low boundary contrast and by the presence of many neighboring structures in close proximity to the aneurysm wall. We present an automated method that is similar to the well known Active Shape Models (ASM), combining a three-dimensional shape model with a one-dimensional boundary appearance model. Our contribution is twofold: we developed a non-parametric appearance modeling scheme that effectively deals with a highly varying background, and we propose a way of generalizing models of curvilinear structures from small training sets. In contrast with the conventional ASM approach, the new appearance model trains on both true and false examples of boundary profiles. The probability that a given image profile belongs to the boundary is obtained using k nearest neighbor (kNN) probability density estimation. The performance of this scheme is compared to that of original ASMs, which minimize the Mahalanobis distance to the average true profile in the training set. The generalizability of the shape model is improved by modeling the objects axis deformation independent of its cross-sectional deformation. A leave-one-out experiment was performed on 23 datasets. Segmentation using the kNN appearance model significantly outperformed the original ASM scheme; average volume errors were 5.9% and 46% respectively.

 
B. van Ginneken, M. de Bruijne, M. Loog, and M.A. Viergever. Interactive shape models. In M. Sonka and M.J. Fitzpatrick, editors, Medical Imaging: Image Processing, volume 5032 of Proceedings of SPIE, pages 1206-1216. SPIE Press, 2003. [ link ]
Supervised segmentation methods in which a model of the shape of an object and its gray-level appearance is used to segment new images have become popular techniques in medical image segmentation. However, the results of these methods are not always accurate enough. We show how to extend one of these segmentation methods, active shape models (ASM) so that user interaction can be incorporated. In this interactive shape model (iASM), a user drags points to their correct position thus guiding the segmentation process. Experiments for three medical segmentation tasks are presented: segmenting lung fields in chest radiographs, hand outlines in hand radiographs and thrombus in abdominal aorta aneurysms from CTAdata. By only fixing a small number of points, the part of sufficiently accurate segmentations can be increased from 20-70 can be used in many clinical applications.

 
M. de Bruijne, B. van Ginneken, W.J. Niessen, J.B.A. Maintz, and M.A. Viergever. Active shape model based segmentation of abdominal aortic aneurysms in CTA images. In M. Sonka and M.J. Fitzpatrick, editors, Medical Imaging: Image Processing, volume 4684 of Proceedings of SPIE, pages 463-474. SPIE Press, 2002. [ link ]
An automated method for the segmentation of thrombus in abdominal aortic aneurysms from CTA data is presented. The method is based on Active Shape Model (ASM) fitting in sequential slices, using the contour obtained in one slice as the initialisation in the adjacent slice. The optimal fit is defined by maximum correlation of grey value profiles around the contour in successive slices, in contrast to the original ASM scheme as proposed by Cootes and Taylor, where the correlation with profiles from training data is maximised. An extension to the proposed approach prevents the inclusion of low-intensity tissue and allows the model to refine to nearby edges. The applied shape models contain either one or two image slices, the latter explicitly restricting the shape change from slice to slice. To evaluate the proposed methods a leave-one-out experiment was performed, using six datasets containing 274 slices to segment. Both adapted ASM schemes yield significantly better results than the original scheme (p<0.0001). The extended slice correlation fit of a one-slice model showed best overall performance. Using one manually delineated image slice as a reference, on average a number of 29 slices could be automatically segmented with an accuracy within the bounds of manual inter-observer variability.

 
M. de Bruijne, W.J. Niessen, J.B.A. Maintz, and M.A. Viergever. Semi-automatic aortic endograft localisation for post-operative evaluation of endovascular aneurysm treatment. In M. Sonka and K.M. Hanson, editors, Medical Imaging: Image Processing, volume 4322 of Proceedings of SPIE, pages 395-406. SPIE Press, 2001. [ link ]
A semi-automatic method for localisation and segmentation of bifurcated aortic endografts in CTA images is presented. The graft position is established through detection of radiopaque markers sewn on the outside of the graft. The user indicates the first and the last marker, whereupon the rest of the markers are detected automatically by second order scaled derivative analysis combined with prior knowledge of graft shape and marker configuration. The marker centres obtained approximate the graft sides and central axis. The graft boundary is determined, either in the original CT slices or in reformatted slices orthogonal to the local graft axis, by maximizing the local gradient in the radial direction along a deformable contour passing through both sides. The method has been applied to ten CTA images. In all cases, an adequate segmentation is obtained. Compared to manual segmentations an average similarity (i.e. relative volume of overlap) of 0.93 +/- 0.02 for the graft body and 0.84 +/- 0.05 for the limbs is found.

 
F.A. Karelse, C.J. Barth, M.N.A. Beurskens, M. de Bruijne, G.M.D. Hogeweij, N.J. Lopes Cardozo, H.J. van der Meiden, F.J. Pijper, and the RTP-team. First current density measurements with tangential Thomson scattering at RTP. In Proceedings of the 8th International Symposium on Laser-aided Plasma Diagnostics, pages 253-258, 1997.
 
F.A. Karelse, M. de Bruijne, M.N.A. Beurskens, C.J. Barth, G.M.D. Hogeweij, N.J. Lopes Cardozo, H.J. van der Meiden, F.J. Pijper, and the RTP-team. Current density profiles measured with multi-position tangential Thomson scattering in the RTP tokamak. In M. Schittenhelm, R. Bartiromo, and F. Wagner, editors, Proceedings of the 24th EPS Conference on Controlled Fusion and Plasma Physics, volume 21A Part II of Europhysics Conference Abstracts, pages 621-624, 1997.

Published abstracts

 
W. Kuo, M. de Bruijne, H. Ozturk, B. Li, A. Perez Rovira, and H. Tiddens. Sensitive and objective method to diagnose bronchiectasis on inspiratory and expiratory chest computed tomography in children with cystic fibrosis. In Annual North American Cystic Fibrosis Conference, volume 49, pages 366-367, 2014.
 
Sepp de Raedt, Marleen de Bruijne, Inger Mechlenburg, Maiken Stilling, Lone Rømer, and Kjeld Søballe. Validation of intraoperative reported angle measurements in periactabular osteotomy. In Dansk Ortopædisk Selskab (DOS) Kongressen, 2014.
 
A van Engelen, A.C. van Dijk, M.T.B. Truijman, R. van 't Klooster, A van der Lugt, W.J. Niessen, M.E. Kooi, and M. de Bruijne. Automatic segmentation of atherosclerotic plaque components in different carotid MRI protocols. In European Congress of Radiology, 2014.
 
A van Engelen, T Wannarong, G Parraga, W.J. Niessen, A Fenster, J.D. Spence, and M. de Bruijne. Atherosclerotic plaque texture from 3d carotid ultrasound improves prediction of vascular events. In European Congress of Radiology, 2014.
 
Tim Rosenow, Wieying Kuo, Marleen de Bruijne, Conor Murray, Harm Tiddens, and Stephen Stick. A new gold standard for assessing ct in early cf lung disease? In European Respiratory Society (ERS), 24th International Congress, volume 44, 2014. [ link ]
Background: Chest computed tomography (CT) is an important tool to assess early cystic fibrosis (CF) structural lung disease, particularly early in life where no validated outcome measures exist. Current CF-CT scoring systems show poor agreement between observers in infants <3 yrs. CT scores must be reliable if they are to be validated as outcome measures.Aim: To determine the repeatability of a quantitative scoring system for chest CT scans early in life.Methods: Twenty-five volumetric CT scans from the AREST CF cohort obtained at age 1 or 3 yrs were scored by two observers using PRAGMA (Perth-Rotterdam Annotated Grid Morphometry Analysis). Briefly, a square grid was overlaid on 10 equidistant slices. Each cell was annotated according to whether it contained bronchiectasis, other abnormalities or healthy lung only, expressed as a proportion of the total lung for bronchiectasis (%Bx) or any lung abnormality (%Disease). This analysis was then repeated two months later by one observer. Repeatability was assessed using intraclass correlation coefficients (ICC).Results: %Bx ranged from 0-3.9 (median 0.23) and %Disease ranged from 0-6.9 (median 1.04). ICCs for inter-observer reliability were 0.77 (good agreement) for %Bx and 0.65 (good agreement) for %Disease. For intra-observer reliability, ICCs were 0.79 (good agreement) for %Bx and 0.71 (good agreement) for %Disease.Conclusion: Inter- and intra-observer agreement were high for PRAGMA, suggesting that this new method is more reliable in young children than the present gold standard. The biological relevance of this new scoring system will be further assessed in the AREST CF cohort by comparing outcomes with other markers of early CF lung disease.

 
T Rosenow, H Tiddens, M de Bruijne, and S Stick. PRAGMA: A new method of quantifying structural lung disease in young children with cystic fibrosis. In Thoracic Society of Australia & New Zealand and the Australian & New Zealand Society of Respiratory Science 2014 Annual Scientific Meetings, volume 19, page 48, 2014.
 
T Rosenow, H Tiddens, M Oudraad, M de Bruijne, and S Stick. PRAGMA: Further support for use as a quantitative CT outcome measure. In 28th Annual North American Cystic Fibrosis Conference, volume 49, page 295, 2014.
Rationale: Chest computed tomography (CT) is currently the only validated outcome measure to report lung disease in children under 6 years old with CF. The Perth-Rotterdam Annotated Grid Morphometric Analysis method (PRAGMA) is a quantitative measure of CF-related structural lung disease that has been shown to be a promising outcome for this age group, with good intra-observer reliability. Reliability between observers and ease of training new observers is essential for it to become a standard outcome for use in clinical trials. Method: A brief summary of the PRAGMA method is as follows. A square grid is overlaid on ten equidistant axial slices of the inspiratory scan. Each grid cell is annotated according to the presence of a hierarchy of the following abnormalities: bronchiectasis, mucus plugging, other airway abnormality, atelectasis, healthy tissue. Expiratory scans were processed with a Minimum Intensity Projection (MinIP) filter, and scored with the same grid method for the presence/absence of trapped air. Outcomes reported were: proportion of bronchiectasis ( disease ( A comprehensive training document including instructions and reference images was constructed to train new PRAGMA observers. Thirty de-identified volumetric inspiratory and expiratory CT scans were randomly selected from patients aged 1, 3 or 5 years in the AREST CF programme to act as the training set, to be scored with PRAGMA in four batches of five and a final batch of ten scans, with detailed feedback on performance provided to the trainee in between batches. After completing the training, an additional thirty scans were de-identified and provided in a random order to both an experienced observer and a newly trained observer. The results were compared using the intraclass correlation coefficient (ICC) to determine inter-observer reliability. Results: One expiratory scan was excluded due to technical problems. Median (range) outcomes from the experienced observer were as follows: ICCs between observers for each outcome were as follows: 0.85, results of >0.8 were considered excellent. Conclusions: New observers familiar with CF-CT scoring can easily be trained in PRAGMA with a simple training programme. After training, an excellent interobserver reliability can be achieved that exceeds the currently used scoring systems in this young age group. PRAGMA is therefore a reliable quantitative outcome measure for assessing structural disease in CF, further supporting its use in clinical trials.

 
S.C. Wens, P. Ciet, A. Perez-Rovira, K. Logie, E. Salamon, P. Wielopolski, M. Bruijne, M.E. Kruijshaar, H.W. Tiddens, N.A.M. van der Beek, P.A. van Doorn, and A.T. van der Ploeg. Cine-mri as a new tool to evaluate diaphragmatic dysfunction in pompe disease. volume 24, pages 839 -, 2014. 19th International Congress of The World Muscle Society. [ link ]
Severe pulmonary dysfunction is a serious threat to patients with Pompe disease, a treatable metabolic neuromuscular disorder caused by lysosomal acid alpha-glucosidase deficiency. This pulmonary dysfunction - which is particularly severe in the supine position - is mainly caused by diaphragmatic weakness. Standard pulmonary function tests provide only indirect information about diaphragmatic function, and they do not supply information about chest mechanics in detail. We therefore used cine-MRI to examine the dynamic performance of respiratory muscles, and compared these data with the results of simultaneously performed pulmonary function testing. Ten adult Pompe patients and six healthy volunteers participated. We performed two static scans at end-inspiration and end-expiration to evaluate lung anatomy and lung volumes. Three dynamic 3D acquisitions were performed to investigate overall respiratory dynamics. Using manual segmentation of the acquired images, three length ratios were calculated. Diaphragmatic displacement manifests itself by motion in cranio-caudal direction, while movement in antero-posterior and left-right directions reflects chest wall displacement. Pompe patients have a significantly reduced cranio-caudal length ratio compared to healthy volunteers (p < 0.001), indicating impaired diaphragmatic displacement. This ratio correlated strongly with forced vital capacity in supine position (r = 0.88), and severity of 'postural drop' (FVCsitting-FVCsupine; r = 0.89). The difference in antero-posterior length ratio was less pronounced (p = 0.04), while there was no difference in left-right length ratio (p = 0.1). Cine-MRI is a promising technique to assess chest mechanics and to visualize the severely impaired diaphragmatic function in Pompe patients. It may allow us to detect respiratory weakness at an earlier stage. Early diagnosis of diaphragmatic weakness may prove important in deciding when to start enzyme treatment.

 
Pierluigi Ciet, Stephan Wens, Adria Perez-Rovira, Karla Logie, Piotr Wielopolski, Marleen de Bruijne, Ans van der Ploeg, Pieter van Doorn, and Harm Tiddens. Respiratory muscle movement in pompe disease using cine magnetic resonance imaging. In Proceedings of the 99th Annual Meeting of the Radiological Society of North America (RSNA), 2013.
 
Sepp de Raedt, Marleen de Bruijne, Inger Mechlenburg, Maiken Stilling, Lone Rømer, and Kjeld Søballe. A novel program for manual measurement of acetabular angles with improved intra- and inter-rater reliability. In Dansk Ortopædisk Selskab (DOS) Kongressen, 2013.
 
P. Ciet, K. Gonzalez Graniel, S. Stick, M. de Bruijne, H. Tiddens, and M. van Straten. Chest-CT protocol standardization for multicentre trial in cystic fibrosis (CF) infants. In ECR - Annual meeting of the European Society of Radiology, 2013.
Purpose: To standardise the scan protocol for CT scanners of participating centres in a multi-centre study (clinicaltrials.gov NCT01270074) for the prevention of radiologically defined bronchiectasis in CF infants by ensuring the maximum image quality at the minimum radiation dose. Methods and Materials: Three different sized phantoms (QRM, Germany) were used to assess scanners’ performance of automatic exposure control (AEC). CTDI and DLP were recorded. The phantoms contained various inserts to assess slice-sensitivity-profile, in-plane spatial resolution, noise and the Hounsfield Unit (HU) scale. Scans were made for several dose levels and reconstruction kernels. Images were analysed with custom-made software (MatLab, USA) to obtain the standard deviation of the noise, point-spread-function (PSF) and slice thickness. Results: Eight different scanners with 64 slices or more from 4 manufacturers (GE, Philips, Siemens and Toshiba) were assessed. Despite differences in AEC’s performance, we obtained approximately the same dose level at each center by recommending site and age-specific AEC reference levels. A constant image quality was only possible by matching the different reconstruction kernels measured PSFs at full-width-at-half-maximum. In fact, large part of the differences between scanners was related to reconstruction kernels. The relatively high noise images corresponded with reconstructions using a kernel with edge enhancement such as the Siemens B70 kernel or the GE lung kernel that are routinely used in chest imaging. Conclusion: Objective measurements on CT images allowed for matching of scan protocols among CT scanners of different manufacturers. Use of routine protocols might introduce a bias in the (automated) image analysis.

 
Jens Petersen, Aasa Feragen, Megan Owen, Mathilde M.W. Wille, Laura H. Thomsen, Asger Dirksen, and Marleen de Bruijne. Automatic system for segmentation, longitudinal matching, anatomical labeling and measurements of airways from CT images. In Novel technology that shapes Radiology: EIBIR presents IMAGINE, 2013.
 
Jens Petersen, Aasa Feragen, Laura H. Thomsen, Mathilde M.W. Wille, Asger Dirksen, and Marleen de Bruijne. Manual airway labeling has limited reproducibility. In ECR - Annual meeting of the European Society of Radiology, 2013.
Purpose: Quantitative airway assessment is often performed in specific branches to enable comparison of measurements between patients and over time. Little is known on the accuracy in locating these branches. We determined inter- and intra-observer agreement of manual labeling of segmental bronchi from low-dose chest CT scans. Methods and Materials: We selected 40 participants of the Danish Lung Cancer Screening Trial, 10 of each category: asymptomatic, mild, moderate, and severe COPD. Each subject contributed 2 CT scans with an average interval of 4 years. The airways were segmented automatically using in-house developed software. Three trained observers placed labels L1-L10 and R1-R10 in each of the images, using 3D visualization and reformatted cross-sectional views. Inter-expert agreement for each segmental bronchus for a pair of experts was defined as the percentage of images in which both experts assigned that label to the same branch. Automatic deformable image registration was used to determine corresponding branches in the two scans of the same subject. Intra-expert agreement for a bronchus was then defined as the percentage of image pairs in which the expert assigned the label to the same branch in both scans. Results: Average inter-expert agreement was 73.9 Agreement was lowest in the lower left lobe (55.0 largest in R6 and L6 (95.0 was 75.4 Conclusion: We found considerable disagreement in expert labeling, possibly reflecting large anatomical heterogeneity and changes with inspiration. Consistent airway measurement cannot be guaranteed based on manual localization.

 
J. Petersen, M.M.W. Wille, L.H. Thomsen, A. Feragen, A. Dirksen, and M. de Bruijne. The effect of inspiration on airway dimensions measured in CT images from the danish lung cancer screening trial. In ECR - Annual meeting of the European Society of Radiology, 2013.
Purpose: Airway dimensions measured from CT are increasingly being used to investigate diseases such as Chronic Obstructive Pulmonary Disease (COPD). In this study we investigate the effect of differences in inspiration level on such measurements in voluntary inspiration breathhold scans. Methods and Materials: We selected from the Danish Lung Cancer Screening Trial 978 subjects without COPD who were scanned annually for 5 years with low-dose multi-slice CT. Using in-house developed software, the lungs and airways were automatically segmented and corresponding airway branches were found in all scans of the same subject using image registration. Mixed effect models were used to predict the relative change in lumen diameter (LD) and wall thickness (WT) in airways of generation 0 (trachea) to 6 based on relative changes in the segmented total lung volume (TLV). Results: On average, 1.0, 2.0, 3.9, 7.6, 15.0, 25.0 and 27.3 airways per subject were included from generation 0 to 6, respectively. Relative changes in LD were positively related to changes in TLV and coefficients increased with generation: 0.20 (+-0.02), 0.19 (+-0.02), 0.21 (+-0.01), 0.25 (+-0.01), 0.29 (+-0.01), 0.34 (+-0.01), 0.37 (+-0.01). Relative changes in WT were inversely related to changes in TLV and generation: -0.01 (+-0.02), 0.01 (+-0.01), -0.02 (+-0.01), -0.03 (+- 0.01), -0.05 (+-0.01), -0.09 (+- 0.00), -0.08 (+-0.00). Conclusion: Subjects who inspire deeper prior to scanning tend to have larger LD and smaller WT. This effect is more pronounced in higher generation airways. Thus, adjustment for inspiration level is needed to accurately assess airway dimensions.

 
T Rosenow, H Tiddens, M de Bruijne, L Berry, and S Stick. Quantitation of chest CT abnormalities in early life CF: back to the basics. In 27th Annual North American Cystic Fibrosis Conference, 2013.
 
L.A. Tepper, M. de Bruijne, D. Caudri, A. Perez-Rovira, and H.A.W.M. Tiddens. The history of bronchiectasis on CT in cystic fibrosis. In 27th Annual North American Cystic Fibrosis Conference, 2013.
 
Mathilde M.W. Wille, Jens Petersen, Asger Dirksen, Jesper Pedersen, and Marleen de Bruijne. Airway distensibility in chronic obstructive pulmonary disease - evaluation by CT airway segmentation and lung density measurement based on The Danish Lung Cancer Screening Trial. In American Thoracic Society International Conference (ATS), 2013. [ link ]
 
N. Baka, M. de Bruijne, T. van Walsum, B.L. Kaptein, J.E. Giphart, M. Schaap, W.J. Niessen, and B.P.F. Lelieveldt. Fluoroscopic assessment of femoral kinematics using a statistical shape model. In EORS 2012. European Orthopedics Research Society, 2012.
 
Sepp de Raedt, Marleen de Bruijne, Inger Mechlenburg, Maiken Stilling, Lone Rømer, and Kjeld Søballe. Fully automated measurement of radiological angles in hip dysplasia using CT images. In Dansk Ortopædisk Selskab (DOS) Kongressen, 2012.
Background: Developmental hip dysplasia is a debilitating condition that iscommonly diagnosed by manual measurements performed on x-rays or CTslices. However, manual measurements are subjective and time consuming.Purpose / Aim of Study: To develop a fully automated method to measure theangles used in the diagnosis of hip dysplasia using CT images.Materials and Methods: Using the automatically segmented hip, the centerpoint of the femoral head was determined by fitting a sphere to the surface ofthe femoral head. The local axis was determined and corrected for the tilt ofthe pelvis. Starting from the center point, the necessary reference points wereidentified by automatically rotating a line until the edge of the acetabulum wasreached in the respective directions. Using the reference points, the CE-angle(CE), acetabular-anteversion (AcAV) angle, posterior sector (PS) angle andanterior sector (AS) angle were calculated. The method was validated againstmanual measurements as performed in daily practice on 52 hips. We reportmean, average difference with 95 concordancecorrelation coefficient (CCC) between the two methods.Findings / Results: The mean CE angle was 21.2° (Avg. diff. 0.0 ± 5.4 , CCC:0.93 ).The mean AcAV angle was 19.8° (Avg. diff. -0.9 ± 3.6 , CCC: 0.95 ).The mean PS angle was 86 .5° (Avg. diff. 1.8 ± 6.2 , CCC: 0.87 ). The meanAS angle was 49.0° (Avg. diff. 0.8 ± 5.6 , CCC: 0.95 ). Three patients (6 hips)were excluded from analysis due to outlying measurements (>2SD) due toosteophytes.Conclusions: The new automated method achieves acceptable accuracy asvalidated against manual measurements. In the future, the method will beimproved to correct for osteophytes and can be further developed to pre-operatively determine the optimal rotation of the acetabulum beforeperiacetabular osteotomy.

 
Z. Saghir, C. Jacobs, B. van Ginneken, M. de Bruijne, A. Dirksen, and J. H. Pedersen. Probability of malignancy based on automatic segmentation and software measurements of nodules in the Danish lung cancer screening trial (DLCST). In European Respiratory Society Annual Meeting, 2012. [ link ]
 
Z. Saghir, C. Jacobs, B. van Ginneken, M. de Bruijne, A. Dirksen, and J. Holst Pedersen. Performance of segmentation software on large longitudinal database of pulmonary nodules in the Danish lung cancer screening trial (DLCST). In European Respiratory Society Annual Meeting, 2012. [ link ]
 
Laura H Thomsen, Saher B Shaker, Haseem Ashraf, Thomas Rasmussen, Pechin Lo, Marleen de Bruijne, Asger Dirksen, and Jesper H. Pedersen. The relationship between NT proBNP and CT lung density in long term smokers. In European Respiratory Society Annual Meeting, 2012. [ link ]
 
P. Ciet, K. Gonzalez Graniel, S. Stick, M. de Bruijne, H. Tiddens, and M. van Straten. Standardization of a chest-CT protocol for multi-center trial in cystic fibrosis. In 20th Anniversary Meeting of the European Society of Thoracic Imaging (ESTI), 2012.
 
Wiro J. Niessen, Henri A. Vrooman, Renske de Boer, Fedde van der Lijn, Hakim C. Achterberg, Marcel Koek, Stefan Klein, Aad van der Lugt, Marleen de Bruijne, M. Arfan Ikram, and Meike W. Vernooij. Quantitative imaging biomarkers in neurologic disease: Population study perspective. In IEEE International Symposium on Biomedical Imaging: Macro to Nano (ISBI'12), page 911, 2012. [ link ]
The capacity of recognizing the first signs of disease has enormous socio-economic benefits. Population studies have the potential to see disease develop before your eyes, and when including advanced imaging techniques in these studies, literally so. Population imaging studies, especially when complemented with other biomedical and genetic data, provide unique databases that can be exploited with advanced analysis and search techniques for discovering methods for early detection and prediction of disease. This new way of medical research will have considerable impact in the practice of medicine at large. In this presentation we will focus on the development of quantitative imaging biomarkers in neurology using imaging data acquired in a population setting. Currently, effective treatment strategies are lacking in e.g. dementia and stroke. In order to develop such strategies, improved understanding of the early, preclinical stages, of disease, is essential. Quantitative imaging biomarkers for neurologic disease are developed within the context of the Rotterdam Study, a prospective population based study of the causes and determinants of chronic diseases in the elderly that was initiated in 1995. MR brain imaging was performed during this study in random subsets in 1995 and 1999, and since 2005, MR brain imaging is part of the core protocol of the Rotterdam Study. The large scale acquisition of MR brain imaging within the Rotterdam Study allows us to study whether morphologic brain pathology is already present years before clinical onset of neurologic disease, and whether MRI based measurements may be used for prognosis. More information on the Rotterdam Scan Study can be found in [1]. Within the context of the Rotterdam Scan Study, a standardized and validated image analysis workflow is being developed to enable the objective, accurate, and reproducible extraction of relevant parameters describing brain anatomy, possible brain pathologies, and brain connectivity from- multispectral MRI data. Image processing in the Rotterdam Scan Study has four main goals: First, owing to the sheer size and complexity of the imaging database being generated, automation of the tedious task of manual analysis is required. Second, qualitative image assessment should be replaced by objective quantitative analyses as much as possible. Third, we aim to limit or avoid altogether inter- and intraobserver variability. Fourth, image processing allows the extraction of relevant image-derived parameters that would not be feasible manually or cannot be assessed visually. This presentation will provide an overview of different quantitative imaging biomarkers that have been developed, or are currently developed as part of the Rotterdam Scan studies. These include brain tissue quantification (grey matter, white matter, also quantified per lobe), quantification of cerebrospinal fluid, volume and shape of neurostructures such as the hippocampus, ventricles and cerebellum, brain connectivity based on diffusion tensor MRI, and vascular brain pathologies such as white matter lesions and microbleeds.

 
M. Loeve, W.C.J. Hop, M. de Bruijne, P. Th.W. van Hal, P. Robinson, M.L. Aitken, J.D. Dodd, and H.A.W.M. Tiddens. CT scores are predictive of survival in CF patients awaiting lung transplantation. volume 31, pages S244 -, 2012. Abstract Issue: International Society for Heart and Lung Transplantation 32nd Annual Meeting and Scientific Sessions. [ link ]
 
H.C. Achterberg, M. de Bruijne, F. van der Lijn, T. den Heijer, M.W. Vernooij, M.A. Ikram, and W.J. Niessen. Hippocampal shape predicts development of dementia in the general population. In Proceedings of the 97th Annual Meeting of the Radiological Society of North America (RSNA), 2011.
 
M. Ganz, M Nielsen, M de Bruijne, E. Dam, M. Karsdal, and C. Christiansen. Distribution, size, and shape of abdominal aortic calcified deposits and their relationship to mortality in postmenopausal women. In Proceedings of the 97th Annual Meeting of the Radiological Society of North America (RSNA), 2011.
 
S.B. Shaker, A. Dirksen, P. Lo, L.T. Skovgaard, M. de Bruijne, and J.H. Pedersen. Decline in lung density is accelerated in active smokers. In European Respiratory Society, 2011. [ link ]
 
M. Loeve, P.van Hal, P. Robinson, M. Aitkin, J. Dodd, M. de Bruijne, W. Hop, and H. Tiddens on behalf of the CT CFS study group. CT scores correlate with survival in CF patients awaiting lung transplantation. In 34th European Cystic Fibrosis Conference, 2011.
 
P. Ciet, P. Wielopolski, S. Lever, R. Manniesing, M. de Bruijne, M.H. Lequin, and H.A.W.M. Tiddens. MRI tracheomalacia (TM) assessment in pediatric patients: feasibility study. In European Congress of Radiology (ECR), 2011.
Purpose: TM is an excessive narrowing of the intrathoracic part of the trachea. TM is a common congenital pediatric anomaly, but it’s often not recognized due to its unspecific clinical presentation. The aims of our study are: 1) to develop cine-MRI sequences to visualize central airways in static and dynamic conditions in patients that were able to follow specific breathing manoeuvres;2) to develop post-processing tools for image analysis. Methods and Materials: To date 10 subjects (7 males; 2 adults) were enrolled in the pilot study: mean age 15, (range 6 to 30yrs). Volunteers were trained to perform spirometry controlled breathing maneuvers (peak flow and coughing) using a MRI compatible spirometer. “Static” 13-second breath-hold scans covering the entire thoracic region were acquired at end-inspiration and end-expiration using a 3D GRE with TR/TE=1.2/0.5 ms, alpha = 2, sagittal isotropic volume (2.8) x 3mm3 voxels. “Dynamic” scans were performed with the same parameters but covering only the central thorax (1/3 volume), temporal resolution was 500 ms per volume using the TRICKS. In-house developed software for segmentation and analysis was used. Results: All subjects managed to follow the required breathing maneuvers. Images of central airways during static and dynamic conditions were acquired and could be analyzed. Three out of the 8 children had a TM just above the carina during forced expiration, confirmed by bronchoscopy. Conclusion: This pilot study shows that Dynamic-MRI is feasible in pediatric population and allows avoiding radiation exposure and bronchoscopy for the evaluation of central airway dimensions.

 
H.A. Vrooman, M.M.S. Jasperse, M. Koek, F. van der Lijn, H.C. Achterberg, M. de Bruijne, A. van der Lugt, and W.J. Niessen. A computer-aided diagnosis system for the early and differential diagnosis of neurodegenerative disease. In ECR Conference Proceedings 2011, 2011.
 
H.C. Achterberg, F. van der Lijn, T. den Heijer, A van der Lugt, M.M.B. Breteler, W.J. Niessen, and M. de Bruijne. Prediction of dementia by hippocampal shape analysis. In 3rd Dutch conference on Biomedical Engineering (BME), 2011.
 
A van Engelen, M. de Bruijne, S. Klein, H.C. Groen, J.J. Wentzel, A van der Lugt, and W.J. Niessen. Plaque characterization in ex vivo MRI evaluated by dense 3D correspondence with histology. In 3rd Dutch conference on Biomedical Engineering (BME), 2011.
 
P. Ciet, P.A. Wielopolski, S. Lever, M. de Bruijne, M.H. Lequin, and H.A. Tiddens. Tracheomalacia assessment in pediatric patients with MRI: Preliminary report. In Proceedings of the 96th Annual Meeting of the Radiological Society of North America (RSNA), 2010.
 
C. Mol, B. van Ginneken, M. de Bruijne, P. de Jong, M. Oudkerk, A. Dirksen, and P. Zanen. Correction of quantitative emphysema measures with density calibration based on measurements in the trachea. In Proceedings of the 96th Annual Meeting of the Radiological Society of North America (RSNA), 2010.
 
S.B. Shaker, P. Lo, L. Wigstrom, M. de Bruijne, and A. Dirksen. Optimal CT-technique for the detection of emphysema progression. In European Respiratory Society, 2010.
 
M. Nielsen, A. Ghosh, P. C. Pettersen, M. de Bruijne, M. A. Karsdal, H. K. Genant, and C. Christiansen. Computer-based BMD-independent vertebral deformity score (VDS) for fracture prediction in post-menopausal women. In 36th European Symposium on Calcified Tissues, volume 44, page S411, 2009. [ link ]
 
H. Ashraf, G. Edula, M. de Bruijne, M. Nielsen, A. Dirksen, M. Dahlbäck, and J.H. Pedersen. A longitudinal follow-up of COPD disease progression in a large cohort of smokers screened for early lung cancer: CBQ study. In American Thoracic Society International Conference, 2009. [ link ]
 
H. Ashraf, P. Lo, S.B. Shaker, M. de Bruijne, A. Dirksen, P. Tønnesen, M. Dahlbäck, and J.H. Pedersen. Change in smoking habits affects lung density by CT. In American Thoracic Society International Conference, 2009. [ link ]
 
G. Edula, H. Ashraf, P. Lo, M. de Bruijne, M. Dahlbäck, and J.H. Pedersen. Natural progression of lung density changes in COPD measured by MSCT: Correlation to lung function. In American Thoracic Society International Conference, 2009. [ link ]
 
M. Loeve, M.H. Lequin, M. de Bruijne, I.J. Hartmann, W.C. Hop, and H.A. TiddensNorth American Cystic Fibrosis Conference. Air trapping in cystic fibrosis (CF) evaluated using ultra low dose computed tomography (CT): implications for sample size in clinical studies. In 22nd Annual North American Cystic Fibrosis Conference, 2008.
 
M. Loeve, M.H. Lequin, M. de Bruijne, I.J. Hartmann, W.C. Hop, and H.A. Tiddens. Measuring trapped air by computed tomography (CT) in CF: In search of the optimal protocol. In European Respiratory Society Annual Congress, 2008. [ link ]
 
M. Loeve, M.H. Lequin, M. de Bruijne, I.J. Hartmann, W.C. Hop, and H.A. Tiddens. Monitoring of air trapping in a cohort of CF patients measured by computed tomography (CT) and pulmonary function tests. In 22nd Annual North American Cystic Fibrosis Conference, 2008.
 
M. Loeve, M.H. Lequin, M. de Bruijne, I.J. Hartmann, W.C. Hop, and H.A. Tiddens. Expiratory computed tomography (CT) protocols made during voluntary breath hold may be sufficient for monitoring cystic fibrosis (CF) lung disease. In 22nd Annual North American Cystic Fibrosis Conference, 2008.
 
M. de Bruijne, P.C. Pettersen, A. Ghosh, M. G. Sorensen, and M. Nielsen. Spine shape predicts vertebral fractures in postmenopausal women. In American Society for Bone and Mineral Research (ASBMR) 30th Annual Meeting, 2008. [ link ]
 
A. Ghosh, M. de Bruijne, P. C. Pettersen, M. A. Karsdal, D. J. Leeming, and C. Christiansen M. Nielsen. A computer-based measure using normalized heights predicts vertebral fracture independent of BMD in post-menopausal women. In American Society for Bone and Mineral Research (ASBMR) 30th Annual Meeting, 2008. [ link ]
 
M. Loeve, M.H. Lequin, M. de Bruijne, I.J. Hartmann, W.C. Hop, and H.A. Tiddens. Comparing inspiratory and expiratory computed tomography protocols for monitoring mild CF lung disease. In 45th Annual Meeting of the European Society of Paediatric Radiology, 2008. [ link ]
 
M. Loeve, M.H. Lequin, M. de Bruijne, I.J. Hartmann, W.C. Hop, and H.A. Tiddens. Can the distance from lung top to diaphragm on CT scans be used to estimate the lung inflation level for in- and expiratory chest CT? In 45th Annual Meeting of the European Society of Paediatric Radiology, 2008. [ link ]
 
M. Loeve, M.H. Lequin, M. de Bruijne, I.J. Hartmann, W.C. Hop, and H.A. Tiddens. Volumetric ultra low dose expiratory computed tomography (CT) protocols for the monitoring of mild cystic fibrosis (CF) lung disease may be sufficient. In 31st European Cystic Fibrosis Conference, 2008.
 
C. Christiansen, M.A. Karsdal, F. Lauze, E.B. Dam, M. Ganz, M. de Bruijne, M.H. Soerensen, N. Barascuk, and M. Nielsen. Automated quantification of the morphological atherosclerotic calcification distribution on X-rays is a strong predictor of mortality in postmenopausal women. In Arterioscler. Thromb. Vasc. Biol. 2008;28;e32-e149, 2008.
 
P.C. Pettersen, M. de Bruijne, J. Chen, Q. He, L.B. Tankó, and C. Christiansen. Computer based measure of kyphosis predicts fractures in the thoracic spine of postmenopausal women. In American Society for Bone and Mineral Research (ASBMR) 29th Annual Meeting, 2007. [ link ]
 
P.C. Pettersen, M. de Bruijne, J. Chen, Q. He, C. Christiansen, and L.B. Tankó. A computer-based measure of irregularity in the alignment of vertebrae in the lumbar spine is a BMD-independent predictor of fracture risk in postmenopausal women. In 34th European Symposium on Calcified Tissues (ECTS), 2007.
 
M. de Bruijne. Quantitative image analysis using statistical models. In International Workshop on Image Analysis in the Life Sciences: Theory and Applications, 2007. Invited talk.
 
M. de Bruijne, M.T. Lund, L.B. Tankó, P.C. Pettersen, and M. Nielsen. Quantitative vertebral morphometry using neighbor-conditional shape models. In E. Dam, S. Majumdar, and J.C. Buckland-Wright, editors, Joint Disease Workshop - Quantitative Automated Musculoskeletal Analysis, MICCAI 2006 Workshop Proceedings, page 41, 2006. [ link ]
 
M. de Bruijne, P.C. Pettersen, M.T. Lund, P. Alexandersen, L.B. Tankó, M. Nielsen, and C. Christiansen. A novel computer-based quantitative measurement of vertebral deformity in postmenopausal women: Implications for diagnosis and monitoring. In American Society for Bone and Mineral Research (ASBMR) 28th Annual Meeting, 2006. [ link ]
 
M.T. Lund, L.B. Tankó, M. de Bruijne, M. Nielsen, and C. Christiansen. Initial experience with a new computerized method for clinical assessment of vertebral deformations. In American Society for Bone and Mineral Research (ASBMR) 27th Annual Meeting, 2005. [ link ]

Patents

 
M. de Bruijne, L. Sørensen, and M. Nielsen. Classification of medical diagnostic images. Patent Application no US 13/103656, filed 9 May 2011, US Patent 8811724 issued August 2014, 2011.
 
M. de Bruijne, L. Sørensen, and M. Nielsen. Classification of medical diagnostic images. Patent Application No 61/333,513, filed 11 May 2010, US Patent 8811724 issued August 2014, 2010.
 
M. de Bruijne, P.C. Pettersen, M. Nielsen, A. Ghosh, M.A. Karsdal, and C. Christiansen. Vertebral fracture prediction. Patent Application No WO/2009/124995, US 12/936790, PCT/EP2009/054294, filed 9 April 2009, 2009.
 
M. Nielsen, F.B. Lauze, M. de Bruijne, E.B. Dam, M.A. Karsdal, and C. Christiansen. A method of deriving a quantitative measure of the instability of calcific deposits of a blood vessel. Patent Application No WO/2009/101128, PCT/EP2009/051626, filed 12 February 2009, 2009.
 
M. de Bruijne and J.E. Iglesias. Semi-automatic contour detection. Patent Application No. WO/2008/141996, PCT/EP2008/055956, filed 15 May 2008, 2008.
 
M. de Bruijne, P.C. Pettersen, and M. Nielsen. Vertebral fracture prediction. Patent Application No GB 0806509.6, filed 10 April 2008, 2008.
 
M. de Bruijne, P.C. Pettersen, C. Christiansen, and L.B. Tankó. Vertebral fracture prediction. Patent Application No WO/2008/116918, US/12/450421, PCT/EP2008/053664, filed 27 March 2008, 2008.
 
M. Nielsen, F.B. Lauze, M. de Bruijne, E.B. Dam, M.A. Karsdal, and C. Christiansen. A method of deriving a quantitative measure of the instability of calcific deposits of a blood vessel. Patent Application No US 12/069,894, filed 13 February 2008, 2008.
 
J.E. Iglesias and M. de Bruijne. Semi-automatic contour detection. Patent Application No. GB 0718369.2 filed 20 September 2007, 2007.
 
J.E. Iglesias and M. de Bruijne. Semi-automatic contour detection. Patent Application No. GB 0710540.6 filed 1 June 2007, 2007.
 
J.E. Iglesias and M. de Bruijne. Semi-automatic contour detection. Patent Application No. GB 0709599.5 filed 18 May 2007, 2007.
 
P.C. Pettersen, M. de Bruijne, and L.B. Tankó. Vertebral fracture prediction. Patent Application No GB 0705881.1, filed 27th March 2007, 2007.
 
M. de Bruijne, M. Nielsen, C. Christiansen, L.A. Conrad-Hansen, and F.B. Lauze. A method of deriving a quantitative measure of a degree of calcification of an aorta. Patent Application No WO/2006/128729, PCT/EP2006/005317, filed 2nd June 2006, 2006.
 
M. de Bruijne, M. Nielsen, C. Christiansen, L.A. Conrad-Hansen, and F.B. Lauze. A method of deriving a quantitative measure of a degree of calcification of an aorta. U.S. Patent Application No 11/921,372, filed 2 June 2006; US Patent 7844090 issued 30 November 2010, 2006.
 
M. de Bruijne, M.T. Lund, M. Nielsen, and P.C. Pettersen. Vertebral fracture quantification. Patent Application No WO/2006/087190, PCT/EP2006/001407, US 11/844,166 filed 16th February 2006, US patent US8126240 issued 28 February 2012, 2006.
 
M. de Bruijne, M. Nielsen, C. Christiansen, L.A. Conrad-Hansen, and F.B. Lauze. A method of deriving a quantitative measure of a degree of calcification of an aorta. Patent Application No. US 11/247,809 filed 10th October 2005, U.S. Patent No. 7561727 issued 14 July 2009, 2005.
 
L.A. Conrad-Hansen, M. de Bruijne, F.B. Lauze, C. Christiansen, and M. Nielsen. A method of deriving a quantitative measure of a degree of calcification of an aorta. Patent Application No. US 11/144,488 filed 2nd June 2005; U.S. Patent No. 7463758 issued 9th December 2008, 2005.
 
M. de Bruijne, M.T. Lund, M. Nielsen, and P.C. Pettersen. Vertebral fracture quantification. Patent Application No GB 0503236.2 filed 16th February 2005, 2005.

Theses

 
M. de Bruijne. Model-based segmentation of vascular images. PhD thesis, Utrecht University, 2003. ISBN 90-393-3396-3.
Contact me by email (marleen@diku.dk) if you'd like to receive a paper copy.

 
M. de Bruijne. Tangential Thomson scattering at the RTP tokamak. Master's thesis, Utrecht University, 1997. Tech. Rep. IR 97/003 FOM Institute of Plasma Physics, Nieuwegein.

Edited Books

 
R. Beichel, M. de Bruijne, S. Kabus, A. Kiraly, T. Kitasaka, J.M. Kuhnigk, J.R. McClelland, K. Mori, E.M. van Rikxoort, and S. Rit, editors. Fifth International Workshop on Pulmonary Image Analysis. CreateSpace, 2013. ISBN-13: 978-1492186977. [ link ]
 
R. Beichel, M. de Bruijne, B. van Ginneken, S. Kabus, A. Kiraly, J.M. Kuhnigk, J.R. McClelland, K. Mori, E.M. van Rikxoort, and S. Rit, editors. Fourth International Workshop on Pulmonary Image Analysis. CreateSpace, 2011. ISBN-13: 978-1466200166. [ link ]
 
M. Brown, M. de Bruijne, K. Ding, B. van Ginneken, A. Kiraly, J.M. Kuhnigk, J.R. McClelland, K. Mori, and J. Reinhardt, editors. Third International Workshop on Pulmonary Image Analysis, 2010. ISBN-13: 978-1453776001. [ link ]
 
M. Brown, M. de Bruijne, B. van Ginneken, A. Kiraly, J.M. Kuhnigk, C. Lorenz, J.R. McClelland, K. Mori, A.P. Reeves, and J. Reinhardt, editors. Second International Workshop on Pulmonary Image Analysis, 2009. ISBN-13: 978-1448680894. [ link ]
 
M. Brown, M. de Bruijne, B. van Ginneken, A. Kiraly, J.M. Kuhnigk, C. Lorenz, K. Mori, and J. Reinhardt, editors. First International Workshop on Pulmonary Image Analysis, 2008. ISBN-13: 978-1435759527. [ link ]