Segmenting the Left Ventricle in 3D Using a Coupled ASM and a Learned Non-Rigid Spatial Model

Abstract

This paper presents a new approach to higher dimensional segmentation. We present an extended Active Shape Model (ASM) formulation for the segmentation of multi-contour anatomical structures. We employ coupling and weighting schemes to improve the robustness of ASM segmentation. 3D segmentation is achieved through propagation of a 2D ASM using a learned non-rigid spatial model. This approach does not suffer from the training and aligning difficulties faced by direct 3D model-based methods. Experimental results are encouraging at this early stage, and future directions of research are provided.