Marc-Emmanuel Bellemare

Multi-object segmentation framework using deformable models for medical imaging analysis

AbstractSegmenting structures of interest in medical images is an important step in different tasks such as visualization, quantitative analysis, simulation, and image-guided surgery, among several other clinical applications. Numerous segmentation methods have been developed in the past three decades for extraction of anatomical or functional structures on medical imaging. Deformable models, which includen the active contour models or snakes, are among the most popular methods for image segmentation combining several desirable features such as inherent connectivity and smoothness. Even though different approaches have been proposed and significant work has been dedicated to the improvement of such algorithms, there are still challenging research directions as the simultaneous extraction of multiple objects and the integration of individual techniques. This paper presents a novel open-source framework called deformable model array (DMA) for the segmentation of multiple and complex structures of interest in different imaging modalities. While most active contour algorithms can extract one region at a time, DMA allowsn integrating several deformable models to deal with multiple segmentation scenarios. Moreover, it is possible to consider any existing explicit deformable model formulation and even to incorporate new active contour methods, allowing to select a suitable combination in different conditions. The framework also introduces a control module that coordinates the cooperativen evolution of the snakes and is able to solve interaction issues toward the segmentation goal. Thus, DMA can implement complex object and multi-object segmentations in both 2D and 3D using the contextual information derived from the model interaction. These are important features for several medical image analysis tasks in which different but related objects need to be simultaneously extracted. Experimental results on both computed tomography and magnetic resonance imaging show that the proposed framework has a wide range of applications especially in the presence of adjacent structures of interest or under intra-structure inhomogeneities giving excellent quantitative results.

A Diffeomorphic Mapping Based Characterization of Temporal Sequences: Application to the Pelvic Organ Dynamics Assessment

In various imaging applications, shape variations are studied in order to define the transformations involved or to quantify a distance between each change performed. Regardless of the way the shapes may be extracted, with 2D imaging, shapes concern essentially curves or sets of points depending on the available data. Wether time is related to the shape variations or not, one can consider a set of shapes as the observation of the temporal evolution of an initial shape. In this context, we present a methodology aiming at quantifying the evolution of a set of contours without landmarks. Our characterization of temporal sequences is based on the large deformation diffeomorphic mapping paradigm and the shape representation based on currents, which allow both to propose a shape metric and a curve matching of the timed variations. Then, mechanics related features are extracted as they are physically meaningful and quite painless understandable.In this paper, the process is applied within the scope of a pelviperineology study. Available clinical diagnoses are combined with statistical analysis to show the soundness of the approach. Indeed, pelvic floor disorders are characterized by abnormal organ descents and deformations during abdominal strains. As they are soft-tissue organs, the pelvic organs have no fixed landmarks, in addition to wide shape differences. Routinely used, 2D sagittal mri sequences are segmented to provide the contour sets from which the characterization should highlight pelvic organ behaviors. We believe that a statistical analysis of these behaviors on several dynamic mri sequences could help to a better understanding of the pelvic floor pathophysiology. The methodology is applied on a dataset of 30 patients with different clinical diagnoses. Some promising results are presented, where the pathology detection capability of the deformation features is assessed, and the principal organ dynamics modes are computed, through an inter-patient analysis. Also, an organ parcellation is proposed thanks to the local deformation analysis, it identifies spatial references which are clinically relevant.

Geometric modeling of pelvic organs with thickness

Physiological changes in the spatial configuration of the internal organs in the abdomen can induce different disorders that need surgery. Following the complexity of the surgical procedure, mechanical simulations are necessary but the in vivo factor makes complicate the study of pelvic organs. In order to determine a realistic behavior of these organs, an accurate geometric model associated with a physical modeling is therefore required. Our approach is integrated in the partnership between a geometric and physical module. The Geometric Modeling seeks to build a continuous geometric model: from a dataset of 3D points provided by a Segmentation step, surfaces are created through a B-spline fitting process. An energy function is built to measure the bidirectional distance between surface and data. This energy is minimized with an alternate iterative Hoschek-like method. A thickness is added with an offset formulation, and the geometric model is finally exported in a hexahedral mesh. Afterward, the Physical Modeling tries to calculate the properties of the soft tissues to simulate the organs displacements. The physical parameters attached to the data are determined with a feedback loop between finite-elements deformations and ground-truth acquisition (dynamic MRI).

Automatic estimation of pelvic organ anatomical references

Pelvic floor diseases cover pathologies of which physiopathology is not well understood. 2D sagittal MRI sequences used in the clinical assessment allow to visualize the dynamic behavior of the main organs involved (bladder, uterus-vagina and rectum). Clinicians use anatomical landmarks and measurements related to the pelvic organs in their pathology assessment. Usually, those tasks are performed manually which results in being both tedious and subject to operator dependency. A methodology is proposed to attempt a quantitative and objective characterization of the organ behaviors under abdominal strain condition. This approach automatically assesses the organ movements, through the estimation of characteristic angles (anorectal angle, uterovaginal angle, bladder inclination), and the tracking of anatomically significant points (anorectal angle vertex, uterovaginal angle vertex, bladder neck). From a multi-subject analysis, pathological organs have been distinguished from healthy ones, which shows the relevance of the computed features. In addition, a stability analysis has shown the soundness of the approach.

Uterus segmentation in dynamic MRI using LBP texture descriptors

Pelvic floor disorders cover pathologies of which physiopathology is not well understood. However cases get prevalent with an ageing population. Within the context of a project aiming at modelization of the dynamics of pelvic organs, we have developed an efficient segmentation process. It aims at alleviating the radiologist with a tedious one by one image analysis. From a first contour delineating the uterus-vagina set, the organ border is tracked along a dynamic mri sequence. The process combines movement prediction, local intensity and texture analysis and active contour geometry control. Movement prediction allows a contour intitialization for next image in the sequence. Intensity analysis provides image-based local contour detection enhanced by local binary pattern (lbp) texture descriptors. Geometry control prohibits self intersections and smoothes the contour. Results show the efficiency of the method with images produced in clinical routine.

A diffeomorphic matching based characterization of the pelvic organ dynamics

The analysis of the behavior of the pelvic organs on dynamic mri sequences could help to a better understanding of pelvic floor pathophysiology. The main pelvic organs (bladder, uterus-vagina, rectum) are soft-tissue organs, they undergo deformations and displacements under an abdominal strain. Moreover, the inter-patient morphological variabilities of these organs are very important. In this paper, we present a methodology for the analysis of the pelvic organ dynamics based on a diffeormorphic matching method called large deformation diffeomorphic metric mapping. It allows to define a unique contour parametrization of the pelvic organs, and to estimate the organ deformations after matching the organ shape against its initial state (t=0). Some promising results are presented, where the pathology detection capability of the deformation features is analyzed through an inter-patient analysis. Also, an organ parcellation is proposed by performing a local deformation analysis.