A novel 3D segmentation approach for segmenting the prostate from dynamic contrast enhanced MRI using current appearance and learned shape prior

Abstract

Prostate segmentation is an essential step in developing any non-invasive Computer-Assisted Diagnostic (CAD) system for the early diagnosis of prostate cancer using Magnetic Resonance Images (MRI). In this paper, we propose, a novel framework for 3D segmentation of the prostate region from Dynamic Contrast Enhancement Magnetic Resonance Images (DCE-MRI). The framework is based on a maximum aposteriori (MAP) estimate of a new log-likelihood function that consists of three descriptors: (i) 1st-order visual appearance descriptors of the DCE-MRI, (ii) a 3D spatially invariant 2nd-order homogeneity descriptor, and (iii) a 3D prostate shape descriptor. The shape prior is learned from the co-aligned 3D segmented prostate DCE-MRI data. The visual appearances of the object and its background are described with marginal gray-level distributions obtained by separating their mixture over prostate volume. The spatial interactions between the prostate voxels are modeled by a 3D 2nd-order rotation-variant Markov-Gibbs random field (MGRF) of object/background labels with analytically estimated potentials. Experiments with real in vivo prostate DCE-MRI confirm the robustness and accuracy of the proposed approach.