基于形态学灰度重构的乳腺微钙化点提取

On Microcalcifications Detection in Mammograms Based on Morphological Grayscale Reconstruction

针对乳腺X线摄片,提出了一种计算机辅助检测乳腺微钙化点的方法:首先利用灰度形态重构完成穹顶提取,除去乳腺背景结构同时又能很好地保留微钙化点;其次运用拉普拉斯算子在空间域内对图像进行锐化以增强微钙化点;然后在频率域内利用巴特沃思高通滤波器对图像进行高通滤波;最后利用二维最大熵阈值分割对图像二值化,从而检测出微钙化点并在图像上标识出来。计算机辅助诊断的方法不仅避免了医生因大量阅片引起疲劳而造成误诊,起到了协助医生诊断的作用,而且提高了诊断的效率和准确率。

A novel approach to computer aided microcalcifications detection in mammograms was presented. Firstly dome extraction was accomplished by using morphological grayscale reconstruction, which could not only remove the breast structures but also remain microcalcifications commendably. Secondly microcalcifications were enhanced by utilizing Laplacian operator in spatial domain. Thirdly Butterworth highpass filters was adopted to located microcalcifications incipiently. Lastly two-dimensional entropic thresholding was applied to segmentation image, thus microcalcifications were detected and were marked in mammograms. The proposed method could avoid misdiaguosis because of doctors＇ fatigue causing by reading large amount of films. It played an important role of assisting doctors to diagnose and it also improved the efficiency and accuracy of diagnosis.