Abstract
A novel medical image segmentation approach, termed as threshold-based level set approach, is presented, which combines the threshold segmentation method with the fast marching method. In this approach, the original image is firstly preprocessed by anisotropic filter in order to enhance the image edge and filter noise, and then the threshold segmentation method is used to control the diffusion coefficient of curve within the threshold set, and finally the image is segmented by using improved fast marching method (FMM). The speed function of FMM is defined based on the statistical similarity degree of the regions. The experimental results show that this approach can improve the segmentation effect and raise the segmentation speed.