一种提取MRI图像感兴趣区域的分割方法  

The method of extracting region of interest for MRI images

作　　者：郭庆燕[1] 杨贤栋[1] 叶少珍[1,2] 

机构地区：[1]福州大学数学与计算机科学学院,福建福州350108 [2]福建省医疗器械与医药技术重点实验室,福建福州350002

出　　处：《福州大学学报（自然科学版）》2010年第4期509-514,共6页Journal of Fuzhou University(Natural Science Edition)

基　　金：福建省自然科学基金资助项目(2009J01282);福建省科技平台建设资助项目(2008J1005)

摘　　要：以国际标准脑肿瘤MRI图像库为背景进行分割实验,提出一种结合模糊C均值聚类、区域生长和数学形态学的FCM_Region分割方法对MRI脑肿瘤感兴趣区域进行提取.先利用模糊C均值聚类算法对原图进行聚类粗分割,对分割的结果采用形态学双结构算子和区域生长法去除颅骨等非脑组织来获取脑部组织,并平滑图像,最后采用比对法获得肿瘤感兴趣区域.实验结果证明了该方法对MRI脑肿瘤图像分割的有效性.This paper proposed a new segmentation method FCM_Region which is combined the fuzzy C -means clustering algorithm, regional growing algorithm, and mathematical morphology. At first, the MIR image is roughly segmented by adopting the fuzzy C - means clustering algorithm, and then using the designed morphology operators and regional growing algorithm to further remove the skull and other non -, and smooth the partitioned image, finally, compared the gray levels between images, the tumor is partitioned from the locked area. This paper is based on brain tumor MRI image library of international standards which is developed by Harvard Medical College. The result of experiment shows the effectiveness of this method.

关 键 词：MRI图像 模糊C均值聚类 区域生长 图像分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]