乳腺X线图像结构扭曲检测的研究

针对乳腺X线图像结构扭曲 （Architectural distortion，AD）检测假阳性率偏高的问题，提出了一种新的乳腺X线图像结构扭曲 检测方法相似度收敛指数（Similarity convergence index，SCI）方法.首先利用马氏距离比计算出毛刺的相似度，然后通过计算相似度加权的收敛指数增强放射状毛 刺，最后提取出收敛指数的局部最大值作为候选点，并对这些候选点进行分类，检测出结构扭曲. 该方法在Mini-MIAS （Mammographic Image Analysis Society）乳腺图像和北京大学人民医院乳腺中心乳腺图像上进行验证，实验结果表明，本文提出的方法有效降低了假阳 性率，同时适用于脂肪型乳腺X线图像和致密型乳腺X线图像.

Detection of Architectural Distortion in Mammograms

For the problem of the high false positive rate in the detection of architectural distortion (AD) in mammograms, a novel method called similarity convergence index (SCI) is proposed. Firstly the spiculation similarity based on Mahalanobis distance is presented and used to compute the SCI to enhance radiating spiculations. Then local maximum values of SCI are extracted as AD candidates, and lastly these candidates are classified into AD and normal tissues. The proposed method is evaluated on mammograms from the Mini-MIAS (Mammographic Image Analysis Society) and mammograms from the Breast Disease Center of Peking University People's Hospital. The results show that the proposed detection method can reduce the false positive rate significantly, and can be applied to both fat and compact mammograms.