改进标记分水岭的乳腺X线图像肿块检测方法-《计算机技术与发展》

文章信息/Info

Title:: Mass Detection Method in Mammograms Based on Modified Maker-controlled Watershed Transform

Keywords:: mammogram; mass detection; marker-controlled watershed; shape feature; morphology

摘要:: 乳腺X线图像肿块大小不一，固定参数的传统标记分水岭算法无法实现乳腺X线图像肿块的有效检测。针对此问题，文中提出了一种结合形状特征和改进型标记分水岭的乳腺X线图像肿块检测方法。在计算前景标记时，结合标记的形状特征判定前景标记，通过对前景标记进行形态学膨胀并提取边缘以获得背景标记，利用改进型自适应参数标记分水岭算法实现肿块检测。实验结果表明，文中算法通过结合形状特征，自适应地选择合适的形态学参数，使得肿块检测准确率高于传统标记分水岭算法。

Abstract:: Due to different sizes and shapes of masses in mammograms,traditional watershed method with fixed parameters can't detect masses effectively. To solve this problem,propose an algorithm integrated with shape feature and modified watershed transform to detect masses in mammograms. The proposed method calculates inner markers with shape feature and via morphological dilation and edge detec-tion to locate external markers. Then apply modified watershed algorithm with adaptive parameters to detect masses. The result shows that the proposed approach has higher detection rate than the traditional watershed method.