基于SVM的乳腺癌X光照片计算机辅助诊断模型

SVM-based Computer-aided Diagnosis Model in Digitized Mammogram

乳腺癌是现代女性最常见的恶性肿瘤之一。支持向量机SVM是一种基于统计学习理论的机器学习算法,它能在训练样本很少的情况下达到良好的分类效果。本文提出一个基于支持向量机的超声乳腺肿瘤图像计算机辅助诊断系统,它由图像预处理、ROI特征提取和SVM分类器异常诊断三个模块构成。通过实验证明,在处理相同的样本数据集时,基于SVM算法的计算机辅助诊断系统相对于BP神经网络,有更高的诊断灵敏度。统计学习理论的发展将更加完善SVM,具有高分类性能的分类器将使计算机辅助诊断的能力进一步提高。

Breast cancer is one of the most popular malignant diseases of the modem women. Mammography has been widely used in screening of breast cancer and many computer aided diagnose（CAD） technologies are developed to help radiologists to improve the diagnostic performance. SVM is a machine learning method based on statistics. It＇ s a very efficient binary classification algorithm designed for solving the two - class pattern recognition problems. The proposed scheme is evaluated by a data set of 200 clinical mammograms from DDSM. Experimental results demonstrate that the proposed SVM - based CAD system offers a very satisfactory performance for MCs detection in digitizing mammograms. Compare with previous ANN classifier, it provides higher classification accuracy and computational speed.