Lesion detection and segmentation in uterine cervix images using an ARC-level mRF

This study develops a procedure for automatic extraction and segmentation of a class-specific object (or region) by learning class-specific boundaries. We present and evaluate the method with a specific focus on the detection of lesion regions in uterine cervix images. The watershed map of the input image is modeled using MRF in which watershed regions correspond to binary random variables indicating whether the region is part of the lesion tissue or not. The local pairwise factors on the arcs of the watershed map indicate whether the arc is part of the object boundary. The factors are based on supervised learning of a visual word distribution. Final lesion region segmentation is obtained using a loopy belief propagation applied to the watershed arc-level MRF. Experimental results on real data show state-of-the-art segmentation resultsin this very challenging task. If needed, the results can be interactively even improved.