基于区域增长法的新疆哈萨克族食管癌X射线医学图像的分割研究  

Xinjiang Kazaks' Esophageal Cancer Medical Image Segmentation Based on Region Growing Method

作　　者：伊力扎提.阿力甫 木拉提.哈米提 严传波[1] 阿布都艾尼.库吐鲁克 姚娟[1] 杨芳[1] 

机构地区：[1]新疆医科大学,830011

出　　处：《实用癌症杂志》2016年第3期443-445,共3页The Practical Journal of Cancer

基　　金：国家自然科学基金项目(编号:81460281)

摘　　要：目的为了提高分类和检索准确率从哈萨克族食管癌X射线医学图像中提取感兴趣区域。方法通过40幅新疆哈萨克族食管癌X射线医学图像中得出新疆哈萨克族食管癌医学图像直方图特征并利用此特征分别采用区域增长法和全阈值法分割100新疆哈萨克族食管癌医学图像,并利用面积大小差异和平均表面距离评价分割结果。结果分割后的图像与手工分割的图像进行比较评价得出区域增长法与手工分割图像面积平均相差4.1606%,平均表面距离相差1个像素点而全阈值法分割后图像面积平均相差13.056%,平均表面距离相差3个像素点。结论区域增长法比较适合分割新疆哈萨克族食管癌X射线医学图像。此研究不仅能提高新疆哈萨克族食管癌辅助诊断系统的诊断准确率并对以后的食管癌医学图像分割研究奠定了基础。Objective To improve the accuracy of the classification and retrieval Xiujiang Kazak esophageal cancer medical image segmentation. Methods Segment 100 Xinjiang Kazak esophageal cancer medical images using regional growth method and full image threshold method which based on the features taken by other 40 Xinjiang Kazak esophageal cancer medical images. Evaluate the segmentation accuracy of the 2 methods by segment area and average surface distance. Results The average area gap between region growing method and experts was 4. 1606%. The average gray gap between region growing method and experts was 1 pixel. The average area gap between all threshold method and experts was 13. 056%. The average gray gap between all threshold method and experts was 3 pixels. Conclusion Regional growing method is suitable for segmentation of Xinjiang Kazak esophageal cancer medical image. This study can improve the diagnostic accuracy of Kazak esophageal cancer assistant diagnosis system and later laid the foundation for esophageal cancer medical image segmentation research.

关 键 词：新疆哈萨克族食管癌医学图像 区域增长法 分割评价 

分 类 号：R735[医药卫生—肿瘤]