多尺度区域增长的肿瘤区域分割方法  

A Measurement of Tumor Segmentation Based upon Multi-Tolerance Region Growing Procedure

作　　者：李丙春[1,2] 周明全[1,2] 耿国华[1,2] 孙蕾[1,2] 

机构地区：[1]喀什师范学院网络中心,新疆喀什844000 [2]西北大学计算机科学系,西安710069

出　　处：《计算机工程与应用》2006年第30期218-219,232,共3页Computer Engineering and Applications

基　　金：国家自然科学基金资助项目(编号:60372072)

摘　　要：数字化的医学图像在早期临床诊断中具有重要的意义。肿瘤形状分析是区分良性肿瘤和恶性肿瘤的重要方法之一。对乳腺肿瘤检测的最终结果很大程度上依赖于对肿瘤区域的分割。为了对肿瘤区域进行特征的提取和识别,首先需要把待处理的肿瘤区域从背景中分离出来。论文使用多尺度区域增长的图像分割技术实现肿瘤分割及肿瘤边缘的准确定位,该方法具有精度高、鲁棒性强等优点。实验结果表明,效果非常理想。Digital medical image plays an important role in the early clinical diagnosis .Analysis of tumor shape is one of the important method for distinguishing malignant tumor from benign tumor.The fiual detection result of breast tumor relies on segmentation of tumor region to a great extent.The tumor needs to be segmented from background for extracting shape features based upon the contour of region.A muhi-tolerance region growing method is proposed for the detection of tumor region and extraction of their contours.The experiment shows that the proposed method is very effective with high precision and robustness.

关 键 词：区域增长 图像分割 肿瘤区域 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]