Segmentation of the Right Ventricle in MR images using dual active shape model in the Bookstein coordinates
Hossam El-Rewaidy, Ahmed Shamel Fahmy · 2015
Segmentation of the Right Ventricle (RV) from cardiac MRI images is necessary for evaluating a number of cardiopulmonary and cardiovascular disorders. Active Shape Models (ASM) have been proposed to capture the variability among the different RV shapes and used to segment the RV. Nevertheless, the method is challenged by the complexity and the large variability among the RV shapes. In this work, we propose two modifications of the basic ASM method that improve the performance significantly. First, the RV contour is divided into two simpler parts: free-wall and septal, each is modeled using a separate ASM model. Secondly, the ASM is evolved within the Bookstein coordinate space which fixes the RV insertion points as landmarks to nonlinearly align the different contours. Results on real MRI images show that the proposed Dual-Bookstein ASM framework outperforms standard ASM algorithm.