Abstract
In this work we use loopy part models to segment ensembles of organs in medical images. Each organ’s shape is represented as a cyclic graph, while shape consistency is enforced through inter-shape connections. Our contributions are two-fold: fifirstly, we use an effifi- cient decomposition-coordination algorithm to solve the resulting optimization problems: we decompose the model’s graph into a set of open, chain-structured, graphs each of which is effificiently optimized using Dynamic Programming with Generalized Distance Transforms. We use the Alternating Direction Method of Multipliers (ADMM) to fifix the potential inconsistencies of the individual solutions and show that ADMM yields substantially faster convergence than plain Dual Decomposition-based methods. Secondly, we employ structured prediction to encompass loss functions that better reflflect the performance criteria used in medical image segmentation. By using the mean contour distance (MCD) as a structured loss during training, we obtain clear test-time performance gains. We demonstrate the merits of exact and effificient inference with rich, structured models in a large X-Ray image segmentation benchmark, where we obtain systematic improvements over the current state-of-the-art