基于支持向量机的乳腺超声计算机辅助诊断临床分析

摘要

Objective To investigate the value of based on support vector machine (SVM) breast ultrasonography technology of Computer-Assisted diagnosis (CAD) for differential diagnosis of benign and malignant breast masses.Methods Total of 143 patients who had 151 breast masses were collected in Fujian Maternity and Children Health Hospital or The Fist Affiliated Hospital of Fujian Medical University from June 2014 to December 2015.Based on pathological results as the gold standard,the diagnostic efficiency of CAD and ultrasonography were compared.Results The sensitivity,specificity,accuracy,positive predictive value and negative predictive value of ultrasonography were 80.1％,71.0％,76.8％,80.0％ and 72.1％,respectively.And the sensitivity,specificity,accuracy,positive predictive value and negative predictive value of CAD were 96.6％,90.3％,94.0％,93.5％ and 94.9％,respectively.The specificity,accuracy,positive predictive value and negative predictive of CAD technology were significantly higher than those of ultrasonography (P ＜ 0.05).The area under the receiver operator characteristic curve of ultrasonography and CAD were 0.759 and 0.935 respectively,with significant difference (P ＜ 0.05).Conclusions CAD has a higher specificity and accuracy rate than ultrasonography in the diagnosis of breast masses.It could help to differentiate benign from malignant breast masses.%目的 探讨基于支持向量机(SVM)的乳腺超声计算机辅助诊断(CAD)技术在诊断乳腺良恶性肿物中的应用价值.方法 收集2014年6月至2015年12月在福建省妇幼保健院或福建医科大学附属第一医院就诊的143例患者,共151个乳腺肿物,以病理结果为金标准,分析比较CAD与常规超声诊断乳腺肿物的效能.结果 常规超声检查的灵敏度80.1％、特异度71.0％、准确度76.8％、阳性预测值80.0％、阴性预测值72.1％;CAD的灵敏度96.6％、特异度90.3％、准确度94.0％、阳性预测值93.5％、阴性预测值94.9％.CAD的特异度、准确度、阳性预测值和阴性预测值均高于超声检查(P＜0.05).超声检查和CAD的受试者工作特征曲线(ROC)下面积分别为0.759、0.935,差异有统计学意义(P＜0.05).结论 CAD与超声检查相比可提高诊断乳腺肿物的特异度和准确度,有助于良恶性肿物的鉴别.