earticle

영어

Prostate segmentation from MRI is a necessary first step and plays a key role in different stages of clinical decision making process. In this paper, we propose a MR T1 Image segmentation method of the prostate based on distance regularized level set evolution with a deformable shape prior. To smooth the prostate image to reduce the noise, we introduce a prostate image with Gaussian kernel and get an edge indicator. To avoid the leakage induced on account of prostate boundaries missing, or blending with surrounding tissues, we introduce priori shape information to construct an energy function with a distance regularization term and an external shape energy term containing the edge indicator and minimize it by solving the gradient flow which can be implemented with a finite difference scheme. To verify the MRI segmentation method of a prostate presented in this paper, we utilize the optimal value of parameters λ, μ, α and ε in the distance regularized level set evolution model and the deformable shape prior of prostate to segment a part of images from normal prostate, benign hyperplasia prostate and cancer prostate. The experiment results show that the MRI segmentation method of prostate presented in this paper is effective for different situation of different patients.