摘要目的 探讨超声S-Detect分类技术在乳腺包块良恶性鉴别诊断中的应用价值.方法 选取我院2016年1-12月间经手术或病理穿刺活检证实的47例乳腺包块患者(共61个病灶).所有病灶分别进行二维超声成像BI-RADS分类(由3类不同年资乳腺专科超声医师进行判别)以及计算机S-Detect分类,分别计算人为BI-RADS分类及S-Detect分类对乳腺包块良恶性诊断的敏感性、特异性、准确性、阳性预测值及阴性预测值.绘制各组的ROC曲线,比较ROC曲线下面积.结果 61个乳腺病灶中,病理证实良性病灶36个,恶性病灶25个.BI-RADS分类诊断的敏感性、特异性及准确性分别为:工作2年医师,69.4％ 、72.0％ 、70.5％;工作5年医师,64.0％ 、92.0％ 、75.4％;工作7年医师,69.4％ 、92.0％ 、78.7％.计算机S-Detect分类诊断敏感性、特异性及准确性分别为80.6％ 、96.0％ 、86.9％.S-Detect分类诊断特异性、准确性及阳性预测值均高于工作2年医师BI-RADS分类,差异均具有统计学意义(P<0.05).各组ROC曲线下面积分别为:工作2年医师,0.729;工作5年医师,0.786;工作7年医师,0.801;S-Detect分类,0.917.结论 与人工BI-RADS分类诊断相比,S-Detect分类在乳腺包块良恶性诊断中具有优势,尤其对于低年资医师,S-Detect分类有助于提高其诊断准确率.

abstractsObjective To investigate the value of S-Detect classification in differential diagnosis of breast mass . Methods The data of forty-seven patients with breast mass lesions ( n=61) from our hospital during January to December in 2016 were retrospectively analyzed . Both the man-made BI-RADS classification ( identified by three different specialist physicians with 2 ,5 and 7 years of experience , respectively) and computer S-Detect classification were performed . The sensitivity ,specificity ,accuracy , positive predictive value and negative predictive value of the man-made BI-RADS classification and S-Detect classification of the benign or malignant diagnosis of breast lumps were calculated . The ROC curve was further plotted ,and the area under the curve ( AUC) of each group was compared ,respectively . Results Sixty-one breast mass lesions were confirmed 36 benign lesions and 25 malignant lesions by pathological biopsy . The sensitivity ,specificity and accuracy of man-made BI-RADS classification were as follows:2-year experience physicians 69 .4％ ,72 .0％ and 70 .5％ ;5-year experience physicians:64 .0％ ,92 .0％ and 75 .4％ ;7-year experience physicians:69 .4％ , 92 .0％ and 78 .7％ . The diagnostic sensitivity , specificity , and accuracy of S-Detect classification were 80 .6％ ,96 .0％ and 86 .9％ . The specificity ,accuracy and positive predictive value of S-Detect classification were significantly higher than those of 2-year experience physicians by BI-RADS classification ( P <0 .05) . The area under the ROC curve of each group was 0 .729 ,0 .786 and 0 .801 for 2 , 5 and 7-year experience physicians , respectively , and 0 .917 for S-Detect classification . Conclusions Compared with the man-made BI-RADS classification ,S-Detect classification has advantages in diagnosis of the benign or malignant of breast mass and is helpful to improve the accuracy of diagnosis , especially for junior physicians .