Regional appearance modeling based on the clustering of intensity profiles

► Regional clustering with no requirement for accurate pointwise registration. ► Propose multimodal assignment: each vertex may have several appearance modes. ► First use of spectral clustering on intensity profiles with selection of number of classes. ► New boosted clustering method with localization criterion to optimize segmentation. ► Comparison with PCA-based appearance prior on a database of 35 liver CT data. Model-based image segmentation is a popular approach for the segmentation of anatomical structures from medical images because it includes prior knowledge about the shape and appearance of structures of interest. This paper focuses on the formulation of a novel appearance prior that can cope with large variability between subjects, for instance due to the presence of pathologies. Instead of relying on Principal Component Analysis such as in Statistical Appearance Models, our approach relies on a multimodal intensity profile atlas from which a point may be assigned to several profile modes consisting of a mean profile and its covariance matrix. These profile modes are first estimated without any intra-subject registration through a boosted EM classification based on spectral clustering. Then, they are projected on a reference mesh whose role is to store the appearance information in a common geometric representation. We show that this prior leads to better performance than the classical monomodal Principal Component Analysis approach while relying on fewer profile modes.