图论在脑肿瘤分割及提取中的应用研究  

Application of graph theory in brain tumor segmentation and extraction

作　　者：李鹏[1] 吴水才[1] 高宏建[1] 盛磊[1] 

机构地区：[1]北京工业大学生命科学与生物工程学院,北京100124

出　　处：《北京生物医学工程》2013年第3期243-247,共5页Beijing Biomedical Engineering

基　　金：北京市自然科学基金(3112005)资助

摘　　要：目的基于Matlab和VC++混合编程,实现了图论在脑肿瘤分割及提取中的应用,为之后脑肿瘤三维重建提供准确的分割结果。方法在Matlab和VC++开发平台下,首先读取含脑肿瘤的MRI图像,经过一定的预处理后,调用C++编写的图论分割函数,实现MRI图像的全局分割,然后通过肿瘤区域的颜色信息进行区域二值化和轮廓提取等后处理,很好地完成了脑肿瘤的分割提取。结果通过与专家手动分割的脑肿瘤区域进行比较以及对算法各模块运行时间的监测,显示脑肿瘤分割准确度高,且算法运行稳定。结论基于图论的分割算法能够反映图像全局特性,且运行稳定,是一种值得推广的脑肿瘤分割方法。Objective Based on Matlab and VC ＋ ＋ mixed programming, this paper realizes the application of graph theory in the brain tumor segmentation and extraction, providing accurate segmentation results for subsequent brain tumor three-dimensional reconstruction. Methods On Matlab and VC ＋ ＋ development platform, the MR images with brain tumors are read firstly, after certain preprocessing, the graph theory segmentation functions written in C ＋＋ are called to realize the global segmentation of MR images. Then some postprocessing including region binarization and contour extraction according to color information of tumor regions are done to complete the brain tumor segmentation and extraction. Results Compared with the manual segmentation of brain tumor region by expert, and with the monitoring on the running time of each module in the algorithm,the results are highly accurate in brain tumor segmentation and the segmentation algorithm runs stably. Conclusions The image segmentation algorithm based on graph theory reflects the global image properties, runs stably, and is worthy of popularization in brain tumor segmentation.

关 键 词：图论 脑肿瘤 图像分割 

分 类 号：R318.04[医药卫生—生物医学工程]