肝脏MR图像自适应分割方法  

Adaptive segmentation of liver MR images

作　　者：郭庆燕[1,2] 杨贤栋[1,2] 叶少珍[1,3] 

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]福建省气象信息中心,福建福州350001 [3]福州大学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）》2012年第3期327-333,共7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：提出一种基于自适应边缘检测和自适应灰度阈值的肝脏MR图像序列分割方法.方法通过构造频率图像自动选取序列中含有最大肝脏区域的图像切片,以此图像切片为起点,向上、向下对序列图像依次进行分割.在分割的过程中,将相邻图像的分割结果作为当前分割图片的掩膜图像,以表示当前图像中肝脏区域的大致轮廓,据此轮廓,统计当前图像中肝脏区域灰度值变化范围,进行自适应阈值分割,最后利用集合关系及质心位置合并阈值分割后的肝脏区域.以此方法对福建省肿瘤医院提供的30个腹部MR图像序列进行分割实验,结果证明该方法对肝脏MR图像分割的有效性.Paper proposes a new method based on adaptive edge detection and adaptive grayseale thresholding. This method automatically selects the slice in which liver has a large cross section by constructing frequency image, then the slice is regarded as the starting slice, from which we segment the liver forward and backward. In the process of segmentation, we use the segmented liver in previous slice as an approximate contour for current slice, estimating the gray value range of the current image and segment the liver with adaptive thresholding. Finally the segmented liver regions are merged by u- sing the set of relation and the location of eentroid. The research experiment is done by using the 30 abdominal MR image sequences provided by Fujian Tumor Hospital. The result of experiment shows the effectiveness of this method.

关 键 词：肝脏 MR图像 分割 自适应边缘检测 自适应灰度阈值 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]