一种基于小波变换的乳腺X线图肿块分割方法

目的：乳腺癌的早期诊断和治疗是能够降低乳腺癌患者死亡率的有效途径。通过乳腺X线图像观察乳腺状况是目前乳腺癌普查的首选影像方法。随着图像处理技术的高速发展，计算机辅助检测技术在乳腺癌的检测方面起到越来越重要的作用。方法：本文首先利用图像处理领域的形态学处理、区域增长等相关知识，对乳腺X线图像进行预处理操作，去除图像中所包含的干扰信息。之后提出一种对图像的灰度直方图进行小波变换，并根据其小波变换的模极大值点确定图像分割阈值的方法对乳腺X线图像中的疑似肿块区域进行粗分割。在通过粗分割过程获得乳腺肿块的大致位置信息之后，再利用区域增长的方法获得肿块的边缘信息。结果：本文选取MIAS乳腺图像数据库中的65幅图像作为测试图像，保证每幅图像至少包含一个乳腺肿块。利用本文所提方法对这65幅图像进行实验，并将实验结果与该数据库中的专家标注信息作对比，实验结果为当采用db40的小波系数时的检出率为95．5％。结论：本文所述方法能够有效地分割出乳腺X线图中的肿块区域，并且有较高的检出率，具有进一步研究和应用的价值。

A Wavelet-based Segmentation Method of Breast Mass in Mammography

Objective： Early diagnosis and treatment of breast cancer is an efficient way of reducing the morality of breast can- cer. Observation of abnormalities through mammograms is the preferred method for breast cancer census. With the rapid devel- opment of image processing technology, computer aided detection （CAD） becomes more and more important in breast cancer detection. Methods： The proposed method firstly preprocesses the X-ray image using morphological processing, region grow- ing and other related image processing technology. After the preprocessing stage, all the interference information contained in the image are removed. Then we put forward a new method to segment the mass in a mammogram. The proposed method gets the gray-level histogram of the X-ray image at first, and then using wavelet transformation to find the Wavelet Transform Mod- ulus Maxima. According to this modulus maxima, the coarse segmentation threshold can be found and the suspected regions in the X-ray image can be detected. The rough location information of the breast mass can be got after the coarse segmentation stage. When combined with the region growing method, the edge information of the breast mass can be got. Results： Test im- ages are 65 M/AS mammography which contains at least one mass per image. By using the proposed method and comparing the results with the experts＇ marking information, we get a detection rate of 95.5% when we use db40 as the wavelet coeffi- cient. Conclusions： The proposed method can effectively segment the masses in mammograms. It＇s proved to have a high de- tection rate and further research and application value.