Representing an image as a graph captures its spatial and contextual relationships effectively.Using Graph Neural Networks (GNNs) on graph-based images has considerably enhanced image segmentation.This paper investigates Multi-View GNNs for image segmentation, comparing Intermediate and Late Fusion methods.Experiments show that Intermediate Fusion achieves high accuracy on synthetic data by integrating relational features upfront.On a real dataset, Late Fusion methods, particularly RVCons, outperform Intermediate Fusion by dynamically aggregating multi-view predictions.Indeed, Late Fusion effectively mitigates issues arising from view-specific noise and variance.The results underscore the complementary strengths of both fusion strategies.

International audience

This paper presents a method for recovering and identifying image regions from an initial oversegmentation using qualitative knowledge of its content. Compared to recent works favoring spatial information and quantitative techniques, our approach focuses on simple a priori qualitative inclusion and photometric relationships such as "region A is included in region B", "the intensity of region A is lower than the one of region B" or "regions A and B depict similar intensities" (photometric uncertainty). The proposed method is based on a two steps' inexact graph matching approach. The first step searches for the best subgraph isomorphism candidate between expected regions and a subset of regions resulting from the initial oversegmentation. Then, remaining segmented regions are progressively merged with appropriate already matched regions, while preserving the coherence with a priori declared relationships. Strengths and weaknesses of the method are studied on various images (grayscale and color), with various intial oversegmentation algorithms (k-means, meanshift, quickshift). Results show the potential of the method to recover, in a reasonable runtime, expected regions, a priori described in a qualitative manner. For further evaluation and comparison purposes, a Python opensource package implementing the method is provided, together with the specifically built experimental database.

Phase-Contrast (PC) velocimetry MRI is a useful modality to explore cardiovascular pathologies, but requires the automatic segmentation of vessels. In this paper, we evaluate the ability of region-based active contours to efficiently segment carotid vessels, compared to traditionnal edge-based active contours considered in related works. Experiments are performed using both a sequence of a homemade phantom and a sequence of a patient carotid which has been manually analyzed by a radiologist expert. Results obtained on phantom sequence show that the edge-based approach leads to an underestimate of carotid lumen area and related flows of respectively 18.4 % and 3.6 %. This appears significantly larger than weak errors obtained using the region-based approach (respectively 2.3 % and 0.7 %). Benefits appear even better on the patient sequence. The edge-based approach leads to underestimates of 20.5 % for areas and 13.8 % for blood flows, compared to limited errors of 5.2 % and 4 % with our proposal. This improvement is confirmed by similarity indices used to compare segmentations.

This paper deals with the design aspect of a software for modeling the anatomical and pathological structures of patients from medical images, for diagnosis purposes. In terms of functionalities, it allows to combine image processing algorithms, and to visualize and manipulate 3D models and images. The proposed software uses appropriate design patterns, specific extensible and reusable components and a system managing their combination, thanks to a formal XML-based description of their interfaces. This architecture facilitates the dynamic integration of new functionalities, in particular in terms of image processing algorithms. We describe the structural and behavioral aspects of the proposed component-based architecture.

This paper concerns image segmentation and presents a method to automically determine optimal regions of interest (ROI) according to topological information. The use of ROI avoids the processing of irrelevant image points, therefore improving and accelerating segmentations. ROI determination is based on the optimal use of both the a priori knowledge about topological structure of an image and the contextual information. Contextual information concerns the nature of already segmented regions in the case of the hierarchical segmentation approach we consider. We describe this general purpose method and propose a formulation for the optimal determination of ROIs according to both informations. Then, we illustrate the use and the implementation of such a method in the particular case of medical image segmentation.

This paper deals with the improvement of the genericity of the well-known ITK medical image processing library. Such a library is the core of any medical system/software based on the medical image analysis (e.g. computer aided diagnosis, surgery planning...). This proposed improvement consists in, without algorithm rewriting, extending ITK iterators (leading to an ITK++framework) in order to constrain algorithms to user-specified image areas. We experimentally evaluate this work by considering the practical case of liver vessel segmentation from CT-scan images, where it is pertinent to constrain processing's to the liver area: this reduces the number of voxels to process. Experimental results clearly prove the interest of this work: for example, the anisotropic filtering of this area is performed in only 16 seconds with our proposed solution, while 52 seconds are required using the native limited ITK framework. Moreover, we also show that the code resulting from the proposed improvement remains easy to manage. A major advantage of the proposed solution is that the native ITK library is not modified because the improvement consists in some add-ons: this facilitates the further evaluation of the pertinence of the proposed design while preserving the native ITK framework.