Segmentation of Knee Joints Based on Improved Multiphase Chan-Vese Model

In this paper, we propose a modified multiphase Chan-Vese model which is applied to the medical image segmentation of knee joints. Firstly, according to the anatomical structures and intensity distributions of knee joints in horizontal section, we present a new energy functional to keep the evolving curves just as the same relative location as the initial ones. This measure corrects the mistakes during segmentation and ensures the veracity of segmentation results. Then another interior energy term is introduced to keep the evolving curves close to the signed distance functions. In this way, re-initialization which is time- consuming and necessary in traditional Chan-Vese model, is never required and the convergence of segmentations becomes rapidly and smoothly. The implements and results of segmenting knee joints in MRI and CT validate the effectiveness and efficiency of our approach, compared to the original multiphase Chan-Vese model.