CAJ | 학술논문

针对当前医学图像特征提取仅从局部特征出发的问题,提出一种基于图割的脑部肿瘤提取方法.为克服图割仅适用于较少像素的图像和倾向于小割集的缺陷,对图像进行聚类以提高特征提取的准确度和速度,并构建聚类图的Gomory-Hu割树.按照割值的大小,依次去掉值较小的边提取出聚类图的子图,映射到原图像得到对应的子图像.将该方法应用到MRI脑部图像肿瘤提取中,实验结果表明,其能准确提取出MRI脑部图像中的肿瘤.
침대당전의학도상특정제취부종국부특정출발적문제,제출일충기우도할적뇌부종류제취방법.위극복도할부괄용우교소상소적도상화경향우소할집적결함,대도상진행취류이제고특정제취적준학도화속도,병구건취류도적Gomory-Hu할수.안조할치적대소,의차거도치교소적변제취출취류도적자도,영사도원도상득도대응적자도상.장해방법응용도MRI뇌부도상종류제취중,실험결과표명,기능준학제취출MRI뇌부도상중적종류.
Aiming at current feature extraction problems in medical image, a cerebral tumor extraction method based on graph-cuts is proposed. In order to overcome the defects that graph-cuts is only oriented to small image and easily lead to a small cut-set, a clustering is applied to image to improve the accuracy and speed of feature extraction and then build the Gomory-Hu cuts tree of image. Subgraph of clustering image can be got by removing the edges of Gomory-Hu cuts tree for the value of cuts, and the sub-image can be obtained after the subgraph is mapped to original image. This method is applied to tumor extraction of MRI cerebral image. Experimental result shows that the method can precisely extract tumor of cerebral image.