Suite à des erreurs XML, 2 publications n'ont pas été affichées.
Article de revue avec comité de lecture (5)
HAMROUNI CHTOUROU Sameh, ROUGON Nicolas, DEUX Jean-François
Quantitative assessment of myocardial perfusion in MRI. From registration to clinical measurements. International journal of computer assisted radiology and surgery (IJCARS), june 2011, vol. 6, n° Supplement 1, pp. 23-28
abstract
This paper presents a novel methodology for the non-rigid registration of cardiac perfusion MRI exams. The target medical application is the automated quantitative assessment of myocardial perfusion for clinical diagnosis and longitudinal study of ischemic pathologies. Specifically, an original variational method for the groupwise registration of p-MRI exams based on high-dimensional feature distribution matching using (normalized) mutual information, is developed. The hard issue of estimating information in high-dimensional spaces is solved using state-of-the-art kth-nearest neighbor (kNN) entropy estimators. Combined with mean-shift approximation, the latter allow to efficiently optimize (normalized) mutual information over finite- and infinite-dimensional motion spaces. This framework is applied to the groupwise alignment of cardiac p-MRI exams using local contrast enhancement curves as a feature set, and a B-spline model for cardio-thoracic motions. Preliminary experimental assessment suggests that the technique allows for accurately aligning up to 34 p-MRI images simultaneously, and for further reliably computing perfusion parameters whose joint analysis strongly correlates with expert-based visual diagnosis.
PETITJEAN Caroline, ROUGON Nicolas, CLUZEL Philippe
Assessment of myocardial function : a review of quantification methods and results using tagged MRI. Journal of cardiovascular magnetic resonance, april 2005, vol. 7, n° 2, pp. 501-516
abstract
Tagged MRI provides a noninvasive way to assess the regional function of the heart. Clinical use of myocardial strain measurements from tagged MRI requires identifying new normative values. As for cardiac motion estimation, a variety of methods for quantifying myocardial deformations have been proposed in the image analysis and medical literature, based on heart geometry and continuum mechanics. This article comparatively reviews existing quantification techniques, and synthesizes their results to establish confidence intervals for the standard deformation parameters.
ROUGON Nicolas, PETITJEAN Caroline, PRETEUX Francoise, CLUZEL Philippe, GRENIER Philippe
A non-rigid registration approach for quantifying myocardial contraction in tagged MRI using generalized information measures. Medical image analysis, august 2005, vol. 9, n° 4, pp. 353-375
URL: http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6W6Y-4G9Y5JT-1&_user=1052425&_rdoc=1&_fmt=&_orig=search&_sort=d&view=c&_acct=C000051058&_version=1&_urlVersion=0&_userid=1052425&md5=cc0eb0c03b851611b27e52fbb9460f62
abstract
We address the problem of quantitatively assessing myocardial function from tagged MRI sequences. We develop a two-step method comprising (i) a motion estimation step using a novel variational non rigid registration technique based on generalized information measures, and (ii) a measurement step, yielding local and segmental deformation parameters over the whole myocardium. Experiments on healthy and pathological data demonstrate that this method delivers, within a reasonable computation time and in a fully unsupervised way, reliable measurements for normal subjects and quantitative pathology-specific information. Beyond cardiac MRI, this work redefines the foundations of variational non rigid registration for information-theoretic similarity criteria with potential interest in multimodal medical imaging.
PETITJEAN Caroline, ROUGON Nicolas, CLUZEL Philippe, PRETEUX Francoise, GRENIER Philippe
Quantification of myocardial function using tagged MR and cine MR images. International journal of cardiovascular imaging, december 2004, vol. 20, n° 6, pp. 497-507
abstract
Magnetic Resonance Imaging (MRI) is recognized as a relevant modality for dynamically imaging the heart anatomy and function, and achieving satisfying qualitative diagnosis. Reliable methods for quantitatively analyzing cardiac motion from MR images remain, however, to be elaborated. This paper presents a novel approach for measuring myocardial deformations from tagged MRI sequences. Based on efficient pixel-based statistical non rigid registration, it allows for automatically extracting local/global deformation parameters at the pixel and myocardial segment scales. Its performances for assessing the myocardial function are illustrated both for the normal heart and for pathologies of the ischemic and dilated cardiomyopathic type
ROUGON Nicolas, PRETEUX Francoise
Directional adaptive deformable models for segmentation. Journal of electronic imaging, january 1998, vol. 7, n° 1, pp. 231-256
abstract
In this paper, we address the problem of adapting the functions controlling the material properties of 2D snakes, and show how introducing oriented smoothness constraints results in a novel class of active contour models for segmentation which extends standard isotropic inhomogeneous membrane/thin-plate stabilizers. These constraints, expressed as adaptive L2 matrix norms, are defined by two 2nd-order symmetric and positive definite tensors which are invariant with respect to rigid motions in the image plane. These tensors, equivalent to directional adaptive stretching and bending densities, are quadratic with respect to 1st- and 2nd-order derivatives of the image luminance, respectively. A representation theorem specifying their canonical form is established and a geometrical interpretation of their effects is developed. Within this framework, it is shown that, by achieving a directional control of regularization, such non-isotropic constraints consistently relate the differential properties (metric and curvature) of the deformable model with those of the underlying luminance surface, yielding a satisfying preservation of image contour characteristics. In particular, this model adapts to non-stationary curvature variations along image contours to be segmented, thus providing a consistent solution to curvature under-estimation problems encountered near high curvature contour points by classical snakes evolving with constant material parameters. Optimization of the model within continuous and discrete frameworks is discussed in details. Finally, accuracy and robustness of the model are established on synthetic images. Its efficacy is further demonstrated on 2D MRI sequences for which comparisons with segmentations obtained using classical snakes are provided.
Communication dans une conférence à comité de lecture (17)
HAMROUNI CHTOUROU Sameh, ROUGON Nicolas, PRETEUX Francoise
Groupwise registration of cardiac perfusion MRI sequences using normalized mutual information in high dimension. SPIE Medical Imaging 2011 : Image Processing, Belligham, VA : SPIE, 14-16 february 2011, Lake Buena Vista, United States, 2011, vol. 7962, pp. 796208:1-796208:12, ISBN 978-0-8194-8504-5
abstract
In perfusion MRI (p-MRI) exams, short-axis (SA) image sequences are captured at multiple slice levels along the long-axis of the heart during the transit of a vascular contrast agent (Gd-DTPA) through the cardiac chambers and muscle. Compensating cardio-thoracic motions is a requirement for enabling computer-aided quantitative assessment of myocardial ischaemia from contrast-enhanced p-MRI sequences. The classical paradigm consists of registering each sequence frame on a reference image using some intensity-based matching criterion. In this paper, we introduce a novel unsupervised method for the spatio-temporal groupwise registration of cardiac p-MRI exams based on normalized mutual information (NMI) between high-dimensional feature distributions. Here, local contrast enhancement curves are used as a dense set of spatio-temporal features, and statistically matched through variational optimization to a target feature distribution derived from a registered reference template. The hard issue of probability density estimation in high-dimensional state spaces is bypassed by using consistent geometric entropy estimators, allowing NMI to be computed directly from feature samples. Specifically, a computationally efficient kth-nearest neighbor (kNN) estimation framework is retained, leading to closed-form expressions for the gradient flow of NMI over finite- and infinite-dimensional motion spaces. This approach is applied to the groupwise alignment of cardiac p-MRI exams using a free-form Deformation (FFD) model for cardio-thoracic motions. Experiments on simulated and natural datasets suggest its accuracy and robustness for registering p-MRI exams comprising more than 30 frames.
HAMROUNI CHTOUROU Sameh, ROUGON Nicolas, PRETEUX Francoise
Multi-feature statistical nonrigid registration using high-dimensional generalized information measures. MICCAI '11 : 14th International Conference on Medical Image Computing and Computer Assisted Intervention, Heidelberg;Dordrecht;London [etc.] : Springer, 18-22 september 2011, Toronto, Canada, 2011, vol. 6891/2011, pp. 524-531, ISBN 978-3-642-23622-8
abstract
Nonrigid image registration methods based on the optimization of information-theoretic measures provide versatile solutions for robustly aligning mono-modal data with nonlinear variations and multi-modal data in radiology. Whereas mutual information and its variations arise as a first choice, generalized information measures offer relevant alternatives in specific clinical contexts. Their usual application setting is the alignement of image pairs by statistically matching scalar random variables (generally, greylevel distributions), handled via their probability densities. In this paper, we address the issue of estimating and optimizing generalized information measures over high-dimensional state spaces to derive multi-feature statistical nonrigid registration models. Specifically, we introduce novel consistent and asymptotically unbiaised k nearest neighbors estimators of alpha-informations, and study their variational optimization over finite and infinite dimensional smooth transform spaces. The resulting theoretical framework provides a well-posed and computationally efficient alternative to entropic graph techniques. Its performances are assessed on two cardiological applications: measuring myocardial deformations in tagged MRI, and compensating cardio-thoracic motions in perfusion MRI.
HAMROUNI CHTOUROU Sameh, ROUGON Nicolas
Recalage non rigide statistique multi-attributs par optimisation d'informations généralisées en grande dimension. GRETSI '11 : XXIIIe Colloque Traitement du Signal et des Images, 05-08 septembre 2011, Bordeaux, France, 2011
abstract
Adressant la problématique du recalage non rigide statistique multi-attributs, nous présentons dans cet article un cadre théorique pour l'estimation et l'optimisation variationnelle de mesures d'informations généralisées en grande dimension. Nous introduisons de nouveaux estimateurs géométriques de type k plus proches voisins, consistants et asymptotiquement non biaisés, de la classe des alpha-informations (informations de Havrda-Charvát et de Renyi), et dérivons les expressions analytiques de leur gradients sur des espaces de transformations spatiales régulières de dimension finie et infinie. Le cadre résultant fournit une alternative efficace aux techniques de graphes entropiques. Ses performances sont évaluées dans deux contextes cliniques en IRM cardiaque: l'estimation des déformations myocardiques en IRM de marquage, et la compensation de mouvements cardio-respiratoires en IRM de perfusion.
We address the problem of multi-feature statistical nonrigid image registration, and focus our interest on the issue of estimating and optimizing generalized information measures over high-dimensional state-spaces. To this end, we introduce novel consistent and asymptotically unbiased k-nearest neighbors estimators of alpha-informations (comprising Havrda-Charvát and Renyi informations), and study their variational optimization over finite and infinite dimensional smooth transform spaces. The resulting theoretical framework provides an efficient alternative to entropic graph techniques. Its performances are assessed on two cardiological applications: the measurement of myocardial deformations in tagged MRI, and the compensation of cardio-thoracic motions in perfusion MRI.
We address the problem of multi-feature statistical nonrigid image registration, and focus our interest on the issue of estimating and optimizing generalized information measures over high-dimensional state-spaces. To this end, we introduce novel consistent and asymptotically unbiased k-nearest neighbors estimators of alpha-informations (comprising Havrda-Charvát and Renyi informations), and study their variational optimization over finite and infinite dimensional smooth transform spaces. The resulting theoretical framework provides an efficient alternative to entropic graph techniques. Its performances are assessed on two cardiological applications: the measurement of myocardial deformations in tagged MRI, and the compensation of cardio-thoracic motions in perfusion MRI.
HAMROUNI CHTOUROU Sameh, ROUGON Nicolas, PRETEUX Francoise
Multi-feature information-theoretic image registration : application to groupwise registration of perfusion MRI exams. ISBI '11 : IEEE International Symposium on Biomedical Imaging : From Nano to Macro, IEEE, 30 march - 02 april 2011, Chicago, United States, 2011, pp. 574-577, ISBN 978-1-4244-4127-3
abstract
Investigating multi-feature information-theoretic image registration, we introduce consistent and asymptotically unbiased kth-nearest neighbor (kNN) estimators of mutual information (MI), normalized MI and exclusive information applicable to high-dimensional random variables, and derive under closed-form their gradient flows over finite- and infinite-dimensional transform spaces. Using these results, we devise a novel unsupervised method for the groupwise registration of cardiac perfusion MRI exams. Here, local time-intensity curves are used as a dense set of spatio-temporal features, and statistically matched through variational optimization. Experiments on simulated and real datasets suggest the accuracy of the model for the affine registration of exams with up to 34 frames.
ROUGON Nicolas, DISCHER Antoine, PRETEUX Francoise
Info-snakes : segmentation statistique orientée région par contours actifs informationnels. GRETSI '07 : 21ème Colloque sur le Traitement du Signal et des Images, 11-14 Septembre, Troyes, France, 2007, pp. 753-756
URL: http://documents.irevues.inist.fr/bitstream/2042/17615/1/GRETSI_2007_753.pdf
abstract
Les approches variationnelles hybrides pour la segmentation d'image, fondées sur des fonctionnelles alliant termes référencés contour et termes référencés région, permettent d'intégrer simultanément des contraintes de forme et de texture sur les objets d'intérêt d'une scène. Ces fonctionnelles peuvent être minimisées efficacement par des techniques d'évolution de courbes, conduisant à une adaptation itérative des frontières de segmentation par compétition de régions. Dans ce contexte, nous introduisons une nouvelle classe de contours actifs statistiques orientés région, dénommés info-snakes, définissant un cadre général pour la segmentation d'image par critères informationnels. Le problème de segmentation est ici formulé comme la maximisation d'une mesure d'information entre la distribution de luminance observée et une distribution d'étiquettes associée à un gabarit régional de scène multi-objets, sous contraintes de régularité des frontières des régions. Nous concentrons notre étude sur la classe de mesures d'information généralisées d'Ali-Silvey, dont nous dérivons les flots de gradient sur des espaces non paramétriques de courbes régulières. Les équations d'évolution des frontières du gabarit s'interprêtent comme des modèles de compétition statistique de régions, promouvant les régions statistiquement homogènes au sens de la métrique d'information choisie. Ces équations généralisent les modèles de contours actifs à information mutuelle maximale proposés dans la littérature. Une implantation efficace par ensembles de niveau multiphasés est proposée. Enfin, les performances de cette approche sont illustrées pour la segmentation de structures cardiaques en IRM de perfusion.
Hybrid variational image segmentation techniques, involving energy functionals which combine contour- and region-based terms, have been actively investigated due to their ability to jointly integrate shape and texture cues about scene objects. Minimizing these functionals can be efficiently achieved using curve evolution techniques, yielding region competition models along the deforming segmentation boundaries. Within this framework, this paper presents a generic region-based statistical active contour approach to segmentation, refered to as info-snakes, which provides a generalization of mutual information-based active contour models. Here, the segmentation problem is expressed as the maximization of an information-theoretic similarity measure between the image luminance distribution, and the label distribution of a regional template defining a multi-object geometric prior model, subject to regularization constraints on region boundaries. We focus our attention on the Ali-Silvey class of generalized information measures, and derive the corresponding gradient flows over nonparametric smooth curve spaces. As expected, the evolution equations for the template boundaries interpret as a statistical region competition model, promoting statistically consistent regions relative to the chosen information metrics. These equations generalize the mutual information-based active contour models proposed in the literature. An efficient implementation using a multiphase level set technique is finally provided. Experiments on a cardiac perfusion MRI dataset are presented, demonstrating the relevance of info-snakes for implementing computer-assisted diagnosis tools in cardiology.
Hybrid variational image segmentation techniques, involving energy functionals which combine contour- and region-based terms, have been actively investigated due to their ability to jointly integrate shape and texture cues about scene objects. Minimizing these functionals can be efficiently achieved using curve evolution techniques, yielding region competition models along the deforming segmentation boundaries. Within this framework, this paper presents a generic region-based statistical active contour approach to segmentation, refered to as info-snakes, which provides a generalization of mutual information-based active contour models. Here, the segmentation problem is expressed as the maximization of an information-theoretic similarity measure between the image luminance distribution, and the label distribution of a regional template defining a multi-object geometric prior model, subject to regularization constraints on region boundaries. We focus our attention on the Ali-Silvey class of generalized information measures, and derive the corresponding gradient flows over nonparametric smooth curve spaces. As expected, the evolution equations for the template boundaries interpret as a statistical region competition model, promoting statistically consistent regions relative to the chosen information metrics. These equations generalize the mutual information-based active contour models proposed in the literature. An efficient implementation using a multiphase level set technique is finally provided. Experiments on a cardiac perfusion MRI dataset are presented, demonstrating the relevance of info-snakes for implementing computer-assisted diagnosis tools in cardiology.
ROUGON Nicolas, DISCHER Antoine, PRETEUX Francoise
Region-based statistical segmentation using informational active contours. Mathematics of data/image pattern recognition, compression, and encryption with applications IX, August 15-16, San Diego, California, USA, Bellingham, WA : SPIE, 2006, pp. 63150I:1-63150I:12
abstract
Hybrid variational image segmentation techniques, involving energy functionals which combine contour- and region-based terms, have been actively investigated due to their ability to jointly integrate shape and texture cues about scene objects. Minimizing these functionals can be efficiently achieved using curve evolution techniques, yielding region competition models along the deforming segmentation boundaries. Within this framework, this paper presents a novel region-based statistical active contour approach to segmentation, refered to as info-snakes. Here, the segmentation problem is expressed as the maximization of an information-theoretic similarity measure between the image luminance distribution, and the label distribution of a regional template defining a multi-object geometric prior model, subject to regularization constraints on region boundaries. The probability densities associated with luminance distributions within each template region are estimated using a nonparametric Parzen technique, which avoids resorting to prior assumptions on image statistics or to a training phase. We shall focus our attention on the Ali-Silvey class of information measures, and derive the corresponding gradient flows over nonparametric smooth curve spaces. As expected, the evolution equations for the template boundaries interpret as a statistical region competition model, promoting statistically consistent regions relative to the chosen information metrics. An efficient implementation using a multiphase level set technique is finally provided. Experiments on a cardiac perfusion MRI dataset are presented, demonstrating the relevance of info-snakes for implementing computer-assisted diagnosis tools in cardiology.
DISCHER Antoine, ROUGON Nicolas, PRETEUX Francoise
An unsupervised approach for measuring myocardial perfusion in MR image sequences. Mathematical Methods in Pattern and Image Analysis, August 3-4, San Diego, California, USA, Bellingham, WA : SPIE, 2005, pp. 59160C:1-59160C:12
abstract
Quantitatively assessing myocardial perfusion is a key issue for the diagnosis, therapeutic planning and patient follow-up of cardio-vascular diseases. To this end, perfusion MRI (p-MRI) has emerged as a valuable clinical investigation tool thanks to its ability of dynamically imaging the first pass of a contrast bolus in the framework of stress/rest exams. However, reliable techniques for automatically computing regional first pass curves from 2D short-axis cardiac p-MRI sequences remain to be elaborated. We address this problem and develop an unsupervised four-step approach comprising: (i) a coarse spatio-temporal segmentation step, allowing to automatically detect a region of interest for the heart over the whole sequence, and to select a reference frame with maximal myocardium contrast; (ii) a model-based variational segmentation step of the reference frame, yielding a bi-ventricular partition of the heart into left ventricle, right ventricle and myocardium components; (iii) a respiratory/cardiac motion artifacts compensation step using a novel region-driven intensity-based non rigid registration technique, allowing to elastically propagate the reference bi-ventricular segmentation over the whole sequence; (iv) a measurement step, %based on a standard pharmaco-kinetic model for the contrast agent delivering first-pass curves over each region of a segmental model of the myocardium. The performance of this approach is assessed over a database of 15 normal and pathological subjects, and compared with perfusion measurements delivered by a MRI manufacturer software package based on manual delineations by a medical expert.
ROUGON Nicolas, DISCHER Antoine, PRETEUX Francoise
Recalage variationnel non rigide statistique référencé région. GRETSI '05 : 20ème Colloque sur le Traitement du Signal et des Images, 6-9 Septembre, Louvain-La-Neuve, Belgium, 2005, pp. 1228-1231
URL: http://documents.irevues.inist.fr/bitstream/2042/14091/1/A544_85924.pdf
abstract
Les mesures de similarité statistiques classiquement utilisées en recalage non rigide iconique exploitent des statistiques de luminance globales n'intégrant aucune information géométrique. Cette absence de prise en compte du contexte spatial peut conduire dans certains cas à un appariement imprécis voire incohérent de structures géométriques homologues, rendant ces critères mal adaptés à la segmentation par propagation d'atlas. Nous proposons de résoudre cette limitation en injectant dans la mesure de similarité un a priori spatial sous forme d'un modèle étiqueté de scène associé à l'image cible. Ce dernier induit un critère régionalisé fondé sur des densités de probabilité régionales, estimées via des noyaux de Parzen spécifiques sur chaque région du modèle. Nous formalisons ainsi les notions de rapport de corrélation et de f-information régionalisés, dont nous calculons les flots de gradient sur des espaces de transformations non paramétriques et paramétriques. L'application au modèle de scène de la transformation optimale inverse induit de facto une segmentation de l'image source. Nous illustrons les performances de cette approche pour la compensation de mouvements respiratoires et cardiaques complexes lors du transit d'un produit de contraste en IRM de perfusion.
Classical statistical similarity measures used in intensity-based non rigid registration rely on global luminance statistics and do not incorporate geometrical information. Ignoring spatial correlations may lead to inaccurate or geometrically inconsistent alignment of homologous geometric structures, making these criteria unreliable for atlas-based segmentation purposes. This paper addresses this limitation and presents a region-driven approach to statistical non rigid registration based on regional non-parametric estimates of luminance distributions. The latter are derived from a regional segmentation of the target image which is used as a fixed object/scene template and induces regionalized statistical similarity measures. We provide their expressions in the case of f-informations and correlation ratio, and derive the corresponding gradient flows over parametric and non-parametric transforms spaces. The relevance of this approach is illustrated for the joint respiratory/cardiac motion artifacts compensation and heart segmentation in short-axis cardiac perfusion MR sequences. using a bi-ventricular heart template.
Classical statistical similarity measures used in intensity-based non rigid registration rely on global luminance statistics and do not incorporate geometrical information. Ignoring spatial correlations may lead to inaccurate or geometrically inconsistent alignment of homologous geometric structures, making these criteria unreliable for atlas-based segmentation purposes. This paper addresses this limitation and presents a region-driven approach to statistical non rigid registration based on regional non-parametric estimates of luminance distributions. The latter are derived from a regional segmentation of the target image which is used as a fixed object/scene template and induces regionalized statistical similarity measures. We provide their expressions in the case of f-informations and correlation ratio, and derive the corresponding gradient flows over parametric and non-parametric transforms spaces. The relevance of this approach is illustrated for the joint respiratory/cardiac motion artifacts compensation and heart segmentation in short-axis cardiac perfusion MR sequences. using a bi-ventricular heart template.
ROUGON Nicolas, DISCHER Antoine, PRETEUX Francoise
Region-driven statistical non-rigid registration : application to model-based segmentation and tracking of the heart in perfusion MRI. Mathematical Methods in Pattern and Image Analysis, August 3-4, San Diego, California, USA, Bellingham, WA : SPIE, 2005, pp. 59160E:1-59160E:12
abstract
Intensity-based Non Rigid Registration (NRR) techniques using statistical similarity measures have been widely used to address mono- and multimodal image alignment problems in a robust and segmentation-free way. In these approaches, registration is achieved by minimizing the discrepancy between luminance distributions. Classical similarity criteria, including mutual information, f-information and correlation ratio, rely on global luminance statistics over the whole image domain and do not incorporate spatial information. This may lead to inaccurate or geometrically inconsistent (though visually satisfying) alignment of homologous image structures, making these criteria unreliable for atlas-based segmentation purposes. This paper addresses these limitations and presents a region-driven approach to statistical NRR based on regional non-parametric estimates of luminance distributions. The latter are derived from a regional segmentation of the target image which is used as a fixed object/scene template and induces regionalized statistical similarity measures. We provide the expressions of these criteria in the case of generalized information measures and correlation ratio, and derive the corresponding gradient flows over parametric and non-parametric transforms spaces. This approach is then applied to the joint non rigid segmentation and registration of short-axis cardiac perfusion MR sequences using a bi-ventricular heart template. In this framework, region-driven NRR allows for compensating for respiratory/cardiac motion artifacts, and fitting a segmental heart model used for quantitatively assessing regional myocardial perfusion. Experiments have been performed on a 15 pathological subjects database, demonstrating the relevance of region-driven NRR over global NRR in terms of computational performance and registration accuracy with respect to an expert reference.
ROUGON Nicolas, PETITJEAN Caroline, PRETEUX Francoise
Building and using a statistical 3D motion atlas for analyzing myocardial contraction in MRI. SPIE Medical Imaging 2004 : Image Processing, February 16, San Diego, California, USA, Bellingham, WA : SPIE, 2004, pp. 253-264
abstract
We address the issue of modeling and quantifying myocardial contraction from 4D MR sequences, and present an unsupervised approach for building and using a statistical 3D motion atlas for the normal heart. This approach relies on a state-of-the-art variational non rigid registration (NRR) technique using generalized information measures, which allows for robust intra-subject motion estimation and inter-subject anatomical alignment. The atlas is built from a collection of jointly acquired tagged and cine MR exams in short- and long-axis views. Subject-specific non parametric motion estimates are first obtained by incremental NRR of tagged images onto the end-diastolic (ED) frame. Individual motion data are then transformed into the coordinate system of a reference subject using subject-to-reference mappings derived by NRR of cine ED images. Finally, principal component analysis of aligned motion data is performed for each cardiac phase, yielding a mean model and a set of eigenfields encoding kinematic variability. The latter define an organ-dedicated hierarchical motion basis which enables parametric motion measurement from arbitrary tagged MR exams. To this end, the atlas is transformed into subject coordinates by reference-to-subject NRR of ED cine frames. Atlas-based motion estimation is then achieved by parametric NRR of tagged images onto the ED frame, yielding a compact description of myocardial contraction during diastole.
PETITJEAN Caroline, ROUGON Nicolas, PRETEUX Francoise, CLUZEL Philippe, GRENIER Philippe
Measuring myocardial deformations in tagged MR sequences using informational non-rigid registration. FIMH 2003 : 2nd International Workshop on Functional imaging and modeling of the heart, Berlin; Heidelberg : Springer-Verlag, 05-06 june 2003, Lyon, France, 2003, pp. 162-172, ISBN 978-3-540-40262-6
abstract
We address the problem of quantifying myocardial deformations from tagged MRI sequences. We develop a two-step method comprising (i) a motion estimation step using a novel variational non rigid registration technique based on generalized information measures, and (ii) a measurement step, yielding local and segmental deformation parameters over the whole myocardium. Experiments performed on healthy and pathological data originating from various imaging devices demonstrate that this method delivers, within a reasonable computation time and in a robust and fully unsupervised way, reliable measurements for normal subjects and provide quantitative pathology-specific information
PETITJEAN Caroline, ROUGON Nicolas, PRETEUX Francoise, CLUZEL Philippe, GRENIER Philippe
Measuring myocardial deformations from MR data using information-theoretic nonrigid registration. CARS 2003 : 17th International Congress and Exhibition on Computer Assisted Radiology and Surgery, Amsterdam : Elsevier, 25-28 june 2003, Londres, United Kingdom, 2003, pp. 1159-1164
abstract
Magnetic Resonance Imaging (MRI) is recognized as a relevant modality for dynamically imaging the heart anatomy and function, and achieving satisfying qualitative diagnosis. Reliable methods for quantitatively analyzing cardiac MR images remain, however, to be elaborated. This paper presents a novel approach for measuring myocardial and wall motion from MRI sequences. Based on efficient voxel-based statistical non rigid registration, it allows for automatically extracting local/global deformation parameters at the voxel, myocardial segment and whole myocardium scales. Its performances for assessing the myocardial contractile function are illustrated both for the normal heart and for dilated cardiomyopathies (DCM)
PETITJEAN Caroline, ROUGON Nicolas, PRETEUX Francoise, CLUZEL Philippe, GRENIER Philippe
A non rigid registration approach for measuring myocardial contraction in tagged MRI using exclusive f-information. ICISP 2003 : International Conference on Image and Signal Processing, IAPR, 25-27 june 2003, Agadir, Morocco, 2003, pp. 145-152
abstract
We address the problem of quantifying myocardial deformations from tagged MRI sequences. We develop a two-step approach comprising (i) a motion estimation step via a variational non rigid registration method using a novel class of generalized information measures, called exclusive f-information, and (ii) a measurement step, yielding local and segmental deformation parameters over the whole myocardium. Experiments performed on healthy and pathological data demonstrate that this approach delivers, within a reasonable computation time and in a robust and fully unsupervised way, reliable measurements for normal subjects and provides quantitative pathology-specific information
ROUGON Nicolas, PETITJEAN Caroline, PRETEUX Francoise
Variational non rigid image registration using exclusive F-information. ICIP '03 : IEEE International Conference on Image Processing, September 14-17, Barcelona, Spain, Piscataway, NJ : IEEE, 2003, pp. 703-706, ISBN 0-7803-7750-8
abstract
We address the problem of robust non rigid image registration and focus our interest on similarity criteria based on generalized information measures within the Ali-Silvey class. We introduce exclusive f-information as a novel class of similarity functionals, and develop a generic variational framework for their optimization over non-parametric and parametric transform spaces. This approach is applied to blind non-rigid alignement of face images with arbitrary lighting conditions and complex deformations.
ROUGON Nicolas, PRETEUX Francoise
Contours actifs géodésiques généralisés. RFIA '98 : 11ème Congrès AFCET : Reconnaissances des Formes et Intelligence Artificielle, 20-22 Janvier, Clermont-Ferrand, France, 1998, pp. 287-296
abstract
Fondés sur la théorie des flots géométriques et les méthodes d'évolution d'interfaces par ensembles de niveau, les contours actifs géodésiques apportent une réponse cohérente aux problèmes d'initialisation, d'optimisation et d'adaptativité topologique des contours actifs classiques. Dans ces modèles, la segmentation est formulée comme un calcul de surfaces minimales relativement à une métrique riemannienne isotrope fonction du gradient de luminance. Cet article présente une généralisation anisotrope de ces approches sous forme de flots réactifs-diffusifs intrinsèques, dans lesquels la géométrie de la variété déformable est contrôlée par un tenseur d'interaction dépendant du tenseur de structure de l'image. En exploitant une information géométrique tant positionnelle que directionnelle issue de l'image, ce nouveau modèle permet la détection de structures texturées ou faiblement contrastées. Nous énonçons un théorème établissant son caractère bien-posé, présentons une caractérisation spectrale des tenseurs d'interaction admissibles, et dérivons un schéma numérique préservant les propriétés théoriques du modèle continu dans un cadre discret. Son implantation algorithmique efficace par des techniques de bande étroite alliées à l'utilisation de files d'attente est envisagée. Sa pertinence est illustrée sur des images médicales, pour lesquelles des comparaisons avec les modèles isotropes sont fournies.
ROUGON Nicolas, PRETEUX Francoise
Représentations multi-échelles génériques par diffusion anisotrope contrôlée. RFIA '96 : 10ème Congrès AFCET : Reconnaissances des Formes et Intelligence Artificielle, Janvier, Rennes, France, 1996, pp. 289-298
abstract
Le concept de diffusion anisotrope a démontré sa pertinence tant comme technique non-linéaire de prétraitement permettant de conjuguer filtrage et rehaussement de contraste, que comme fondement théorique de représentations multi-échelles déterministes cohérentes. Dans cet article, nous présentons une étude générale des modèles de diffusion anisotrope via une caractérisation géométrique exploitant la notion de représentations locales d'image invariantes par action d'un groupe de transformations différentielles. Dans un premier temps, nous rappelons comment la géométrie locale (forme et échelle) d'une surface de luminance peut être intrinsèquement spécifiée via deux familles de courbes isométriquement invariantes, constituées par ses lignes de niveau et ses lignes de courant. Dans ce contexte, un processus diffusif induit un flot de déformation sur le réseau formé par ces courbes. Nous dérivons l'expression analytique générale de ce flot, et par une étude de l'Hamiltonien associé, proposons une représentation générique des fonctions de conduction garantissant un respect stable et optimal des propriétés géométriques de l'image. Par ailleurs, en reliant la notion d'échelle aux variations directionnelles de la densité de lignes de niveau, nous montrons comment une dynamique diffusive peut être adaptativement contrôlée au moyen d'une densité d'amortissement dont nous donnons la forme générique. L'ensemble de ces résultats est finalement généralisé en dimension arbitraire. Leur pertinence est illustrée comparativement dans un contexte de segmentation multi-échelles en imagerie médicale.
ROUGON Nicolas, PRETEUX Francoise
Régularisation directionnelle optimale et modèles déformables adaptatifs. RFIA '94 : 9ème Congrès AFCET : Reconnaissances des Formes et Intelligence Artificielle, Janvier, Paris, France, 1994, pp. 51-62
abstract
Dans cet article, nous montrons comment les problèmes liés à l'adaptativité des propriétés matérielles et dynamiques des modèles bidimensionnels de type snake peuvent être abordés dans un cadre théorique unifié, centré sur le concept de représentations locales d'image, invariantes par action d'un groupe de transformations. Ainsi, nous montrons comment l'introduction de contraintes régularisantes orientées, dérivées d'une caractérisation locale d'une image en terme d'isophote, permet d'élaborer une nouvelle classe de contours actifs pour la segmentation généralisant les stabilisateurs de type membrane/plaque-mince hétérogène isotrope. Ces contraintes, exprimées sous la forme de normes L2 tensorielles et adaptatives, sont spécifiées par des tenseurs du 2ème ordre, symétriques, définis positifs et invariants par transformations rigides dans le plan image, interprétables comme des densités directionnelles et adaptatives de tension et de flexion. Nous établissons un théorème de structure déterminant leur forme canonique, et élaborons une caractérisation géométrique de leurs effets. Dans ce contexte, nous démontrons que l'emploi de contraintes anisotropes permet, via un contrôle local directionnel de la régularisation, de relier de façon cohérente les propriétés différentielles (métrique et courbure) du modèle déformable à celles de la surface d'intensité, autorisant par là-même le respect des propriétés différentielles des contours photométriques. Par ailleurs, en reliant la notion de viscosité à la rugosité de la surface photométrique dans le cadre d'une analyse multi-échelle des propriétés directionnelles du potentiel externe, nous proposons une forme générique invariante des densités d'amortissement adaptatives pour des dynamiques lagrangiennes. La pertinence de cette approche est illustrée sur des images médicales 2D et 3D issues de modalités diverses, pour lesquelles des comparaisons avec les segmentations et les vitesses de convergence obtenues avec des modèles non-adaptatifs sont fournies.
Chapitre dans un livre (1)
ROUGON Nicolas, BROSSARD-PAILLEUX Marie-Ange, PRETEUX Francoise
Robust parametric estimation over optimal support of fluid flow structure in multispectral image sequences. Mathematical modeling, estimation, and imaging : 31 July-1 August 2000, San Diego, USA, Bellingham, WA : SPIE, 2000, (Proceedings of SPIE, the International Society for Optical Engineering, 4121), pp. 39-50, ISBN 0-8194-3766-2
abstract
This article presents a methodology for analyzing the Lagrangian structure of fluid flows generated by the evolution of cloud systems in meteorological multispectral image sequences. The correlation between the orientation of cloud texture and the underlying motion field Lagrangian component allows to adopt a static strategy. Following a scale-space approach, we therefore first construct a non-local robust estimator for the locally dominant orientation field in an image. This estimator, which is derived from the image structure tensor, is relevant in both mono- and multispectral contexts. In a second step, the Lagrangian component of the flow is estimated over some bounded image region by robustly fitting a hierarchical vector parametric model to the dominant orientation field. Here, a recurrent problem deals with adaptating the geometry of the model support to obtain unbiased estimates. To tackle this classic issue, we introduce a novel variational, semi-parametric approach which allows the joint optimization of model parameters and support. This approach is generic and, in particular, can be readily applied to motion estimation, yielding robust measurement of the Eulerian structure of the flow. Finally, a structural characterization of the resulting vector field is derived by means of classic differential geometry techniques. This methodology is applied to the analysis of temperated latitude depressions in Meteosat images.
![[del.icio.us]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/delicious.png)
![[Digg]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/digg.png)
![[Facebook]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/facebook.png)
![[Google]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/google.png)
![[Jamespot]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/jamespot.png)
![[LinkedIn]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/linkedin.png)
![[MySpace]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/myspace.png)
![[Reddit]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/reddit.png)
![[Technorati]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/technorati.png)
![[Twitter]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/twitter.png)
![[Email]](https://artemis.telecom-sudparis.eu/wp-content/plugins/bookmarkify/email.png)
Modifié le octobre 1, 2012
