一种基于Mumford-Shah模型的脑肿瘤水平集分割算法

提出了一种新的基于Mumford—Shah模型的脑肿瘤水平集分割方法．它能提供一客观的、可重复的脑肿瘤分割，且分割结果和专家人工分割结果很接近．它可以用来探测边界不一定由梯度来定义的对象，也能自动探测内部轮廓．通过对来自2个病人的共42个（含有或不含水肿）脑肿瘤MRI切片进行分割来评价该算法的效率，结果取得了令人满意的效果．用匹配的百分比（PM）和一致率（CR）来定量评价分割的质量，结果肿瘤分割的PM和CR分别为93．20%和0．92，水肿分割的PM和CR分别为97．33％和0．76，满足临床的需要．

Brain Tumor Segmentation Based on Mumford-Shah Model Level Set

A novel approach for automatic brain tumor segmentation based on Mumford-Shah model level set was proposed.This new approach can provide an objective and reproducible segmentation,and its results are close to the manual results by specialists.It can automatically detect objects whose boundary is not necessarily defined by gradient,as well as interior contours.A total of 42 MR images with brain tumor(with or without edema) of two patients were used to evaluate the efficiency of the segmentation method,and satisfactory results were achieved.Two quantitative measures for tumor segmentation quality estimation,namely,correspondence ratio(CR) and percent matching(PM),were performed.PM and CR is (93.20%) and 0.92 respectively for brain tumor segmentation,and is 97.33% and 0.76 respectively for edema segmentation,which meets clinical needs.