基于DICOM格式的肝脏肿瘤CT图像分割
黄荔丽, 王博亮, 黄晓阳
(厦门大学 计算机科学系,福建 厦门 361005)
摘 要:为减少数据损失与处理时间,直接从DICOM格式的CT胸腹部图像中分割出肝脏肿瘤。为有效地分割出肿瘤,先设置恰当的窗宽窗位。利用ITK读取图像,用均值曲率流滤波法去除图像上的噪声。对目前几种在图像分割领域应用较多的阈值分割及区域生长分割算法进行了探讨,结合获得的活体CT图像进行实验研究,得到较为满意和有效的结果。实验表明:肝脏肿瘤这一目标区域的面积较小,区域生长分割算法中的“置信连接阈值法”能从胸腹腔CT图像中很好地分割出肝脏肿瘤。
关键词:CT图像; 阈值分割; 区域生长
中图分类号: 文献标识码: 文章编号:1673-629X(2008)01-0048-04
Segmentation of Liver Tumor in CT Image Based on DICOM Format
HUANG Li-li, WANG Bo-liang, HUANG Xiao-yang
(Computer Science Department ,Xiamen University, Xiamen 361005,China)
Abstract: For reducing data loss and proceeding time, liver tumor is directly segmented from thorax an d abdominal part in CT image based on DICOM format. Proper window width and window center are set first for efficient segmentation. ITK software is used to read the image, and curvature flow smoothing filter is used to denoise the image. Several histogram and region growing segmentation algorithms requently applied in image segmentation field are discussed, and also experimented combined with vivo CT image. Finally satisfied and effective result is obtained, which shows that the area of liver tumor in the image is small, so the target can be availably segmented by “confidence connected threshold method” in region growing
Key words: CT image; histogram segmentation; region growing |
|