A Hybrid Model for Liver Shape Segmentation with Customized Fast Marching and Improved GMM-EM

摘要

This paper describes an approach to segment liver shape from abdominal CT sequences, required by the analysis of liver diseases. A rough segmentation is first conducted via a customized Fast Marching method to obtain an approximate 3D liver region for subsequent procedure. Then, an improvement of GMM-EM algorithm is made to extract the accurate liver region. Experimental results, evaluated on non-tumor series and tumor series of 10 cases, show that the proposed method performs better than several other typical segmentation models in running time and precision.