MSRM和Snake算法相结合的MRI肿瘤提取

MRI tumor extraction via MSRM and Snake algorithms

提出一种将最大相似度区域合并(MSRM)算法与Snake算法相结合的新的分割方法,进而提取MR成像中的肿瘤区域。首先使用简单线性迭代聚类(SLIC)算法对图像进行超像素分割,其次使用MSRM算法得到图像肿瘤区域大致边界,最后将该边界作为Snake模型的初始轮廓线,完成对图像的肿瘤提取。实验结果表明,相比于MSRM算法,文中方法能更准确地提取MRI肿瘤区域,且不需要手动选取Snake模型的初始轮廓线。

A new segmentation method which combines maximum similarity region merging （MSRM） algorithm with Snake algorithm is proposed, and it can be used to extract tumor regions from MR images. Firstly, do the image super-pixel segmentation by using simple linear iterative clustering （SLIC） algorithm, and next get the approximate region boundary of the tumor by using MSRM algorithm, finally let the boundary be the initial contour of Snake model to complete the tumor extraction of the image. Compared with the MSRM algorithm, the new method which is proposed in this paper can more accurately extract MRI tumor regions showed from experiment results, and the initial contour of the Snake model does not need to be manually selected.