一种提取乳腺癌DCE-MRI感兴趣区域的分割方法

A method to extract the region of interest in dynamic contrast- enhanced MR imaging of breast cancer

针对医院临床诊断中希望能够克服灰度不均匀并且减少非病灶增强区域的干扰,从而更精确地进行DCE-MRI医学病灶分割的情况,在水平集方法的几何活动轮廓模型基础上,提出结合了局部灰度聚类和尺度停止函数的方法,以克服灰度不均匀和非病灶噪声干扰.对福建省肿瘤医院等医院的临床乳腺癌DEC-MRI数据进行实验,结果表明改进的水平集方法具有较好的分割效果.

These problems need to be solved in clinic so that the focus of DCE - MRIs can be segmented more accurately. The paper is based on level set segmentation, combining with local clustering criterion and a scale change function, in order to overcome the interferences. The experiment DCE - MRI data come from Fujian Provincial Tumor Hospital. The result shows that the segmentation is improved.