基于双尺度Sech模板的浸润性乳腺癌检测方法研究

乳腺肿块的准确检测是目前计算机辅助诊断系统的关键。为了准确检测X线乳腺图像中的可疑肿块区域,提出一种基于双尺度Sech模板的早期乳腺癌检测方法。首先根据乳腺X线图像的特征对图像进行预处理,包括乳腺区域提取和内外侧斜位图像胸肌提取;然后对提取的乳腺图像进行基于两个不同尺寸的Sech模板的模板匹配,将模板匹配后得到的相似度图像进行融合和阈值截断;最后进行可疑乳腺肿块区域的定位。结果表明,基于双尺度的Sech模板的检测具有更好的检测效果,改善了基于单尺度模板检测准确率低的问题,并降低了假阳率。

Infiltrative Breast Cancer Detection Based on Double-size Sech Template Matching

Breast mass detection accurately is the key to Computer Aided Diagnosis System at present. To accurately detect breast masses in the mammograms, proposed a method to detect early infiltrative masses from mammographic images based on double-size Sech template matching. First is image preprocessing, including the breast region abstraction from the mammogram and the pectoral muscle detection in the MLO; then template matching based on double-size Sech template, after the images fusion and threshold processing, is the mass-like regions positioning. The test results show the effectiveness of the proposed algorithm, it improved the accuracy rate of mass detection based on single scale, and reduced the positive rate.