关键词: 变形模型, 梯度向量流场, 图像分割, 胼胝体

Abstract: A novel method is introduced for medical image segmentation based on the gradient vector flow.Experiments demonstrate that this method can increase the capture range of the deformable contour.After several iterations，the method can move the deformable contour into the boundary，which is close to the real boundary.In addition，the method is robust and can be applied to noisy images，which is applicable for medical images.This method is applied to segment corpus callosum from the brain in MRI images and it is effective and speedy to achieve a good result of great diagnostic value.

Key words: deformable models, gradient vector flow, image segmentation, corpus callosum