基于马尔可夫多特征随机场模型的脑部mr图像分割分析-segmentation analysis of brain mr images based on markov multi-feature random field model.docx

AbstractIn recent years, segmentation of brain magnetic resonance (MR) images becomes an important research issue. However, due to various factors appearing on brain MR images such as intensity inhomogeneities and noise, segmentation remains the difficulty of the current research. In the existing segmentation methods, the Markov random field (MRF) model based algorithms giving reasonable segmentation results, have attracted more and more experts and scholars of the world to its in-depth study. In this paper, we focus on the MRF theory and its application to the brain MR image segmentation. The main results and contributions are as follows:We study the MRF model theory, define the concept of multiple feature random fields (MFRFs) and propose a novel MRF model based on the MFRFs named as Markov multiple feature random fields (MMFRF) model, which effectively integrated a variety of feature information in the objects and combined them into a unified decision model using the Bayesian framework. And then, we construct the corresponding estimate criteria by maximum a posteriori (MAP) and the optimization solving method using iterated conditional modes (ICM). Particularly, the traditional MRF model becomes the special case of our MMFRF model when the number of feature random fields is one (n=1). Furthermore, as a very general method, the developed MMFRF model presents a unified probabilistic framework for solving a series of image and vision analysis applications.We study on the application of brain MR image segmentation based on the proposed MMFRF model. We investigate the local MMFRF models defining on brain sub-volumes, and through which to construct a MMFRF-based compound model for the whole brain image domain (MMFRF-SVPA). And then, we present the theoretical concepts and practical computations of intensity feature random field, texture feature random field, shape feature random field, and label random field, respectively.We conduct the brain MR segmentation experiment