In this paper we introduce a new approach to knowledgebased segmentation. Our method consists of a novel representation to model shape variations as well as an efficient inference procedure to fit the model to new data. The considered shape model is similarity-invariant and refers to an incomplete graph that consists of intra and intercluster connections representing the inter-dependencies of control points. The clusters are determined according to the co-dependencies of the deformations of the control points within the training set. The connections between the components of a cluster represent the local structure while the connections between the clusters account for the global structure. The distributions of the normalized distances between the connected control points encode the prior model. During search, this model is used together with a discrete markov random field (MRF) based segmentation, where the unknown variables are the positions of the control points in the image domain. To encode the image support, a Voronoi decomposition of the domain is considered and regional based statistics are used. The resulting model is computationally efficient, can encode complex statistical models of shape variations and benefits from the image support of the entire spatial domain.

International audienceIn this paper, we propose a new segmentation algorithm that combines a graph-based shape model with image cues based on boosted features. The landmark-based shape model encodes prior constraints through the normalized Euclidean distances between pairs of control points, alleviating the need of a large database for the training. Moreover, the graph topology is deduced from the dataset using manifold learning and unsupervised clustering. In a graph-matching-like manner, we formulate the segmentation task as a labeling problem where we seek to match the model landmarks to image points that are extracted using the boosted classifiers. We also propose to overcome the limitation of missing correspondences by incorporating an additional label to account for outliers. Then, we repair the outlier positions to complete the segmentation. State-of-the-art discrete optimization techniques are used to provide our experimental results for the segmentation of the right lung in 2D chest radiographs, demonstrating the potentials of our method