二维主成分分析在乳腺钼靶X线片钙化点感兴趣区域提取中的应用

Extracting calcification ROIs in mammograms using 2DPCA

目的针对乳腺钼靶X线影像,将基于二维主成分分析(Two-Dimensional Principal ComponentAnalysis,2DPCA)的方法提取的图像特征用于乳腺感兴趣区域的自动提取,实现计算机辅助检测乳腺X线影像中微钙化点的前期预处理阶段。方法对乳腺图像进行预处理,通过改进的2DPCA方法提取乳腺图像特征,利用边缘检测算法对乳腺图像进行边缘特征提取,最后利用神经网络分类器提取乳腺感兴趣区域。结果实验结果表明该方法可以得到95%的阳性检出率。结论综合运用二维主成分分析方法、边缘特征提取方法和神经网络进行乳腺感兴趣区域提取,准确率更高。

Objective In order to preprocess mammograms for diagnosing the early cases of breast cancer and realize the computer-aided detection of micro-calcifications in mammograms,this paper presented a method based on two-dimensional principal component analysis（2DPCA） to extract the region of interests（ROI） automatically.Methods First we preprocessed the mammograms,and then extracted mammography features by 2DPCA method and edge-detection algorithm.Finally,ROI was extracted by neural network classifier.Results The results showed that we obtained better positive detection ratio with this method.Conclusion Our method could obtain better extraction effect by integrating 2DPCA algorithm,edge-detection algorithm and neural network.