【摘要】 提出了一种基于支持向量机(SVM)算法的钙化点检测方法。通过对乳腺图像进行预处理并提取可能含有微钙化点的感兴趣区域(ROI),对样本ROI进行小波变换确定优化参数,利用SVM检测微钙化点。试验中研究了SVM参数的选取对分类效果的影响,并利用ROC评估准则对SVM的检测效果进行评估。结果表明,SVM在微钙化点检测中是有效的,解决了目前微钙化点检测中普遍存在的假阳性率高、效率低的问题。更多还原

【Abstract】 An improved Support Vector Machine (SVM) for the detection of calcifications in digital mammograms is proposed,which consists of preprocessing and extracting the Region of Interest (ROI) as the possible microcalcification,obtaining characteristic vectors by wavelet transform,detecting microcalcification in all ROIs with SVM,examining the parameters of the impact of SVM,and using the assessment criteria of ROC to assess the test results of SVM.Experimental results show that SVM is effective in the detection of microcalcifications and can solve the problem of high false positive rate and low efficiency in microcalcification detection.更多还原