摘要
目的:将临床信息和多参数MRI(mpMRI)数据输入前列腺计算机辅助诊断(CAD)系统,研究其诊断效能。方法:选取本院前列腺MRI数据库中连续266例患者的病例资料,所有患者均以定性诊断为目的行前列腺MRI检查,临床资料和病理结果完整,且完成随访。由1位高年资泌尿系统影像诊断医师根据前列腺影像报告和数据系统(PI-RADS)进行阅片,以PI-RADS评分为mpMRI信息,输入CAD系统。选取患者资料中的年龄、T-PSA、F/T-PSA、直肠指诊、超声表现等作为临床信息,输入CAD系统。采用受试者工作特征曲线(ROC)分析,以最终诊断结果为金标准,比较mpMRI数据、临床信息、mpMRI+临床信息作为输入项时,CAD输出结果的诊断效能。结果:以临床信息为输入项,CAD诊断前列腺癌的ROC下面积(AUC)为0.888;以mpMRI信息作为输入项,CAD诊断前列腺的AUC为0.980;将mpMRI和临床信息整合作为输入项时,诊断效能最高,AUC为0.986,相应的诊断敏感度、特异度和符合率分别为93.65、96.15和94.12。结论:整合临床信息和mpMRI信息输入前列腺CAD系统,可得到最高的诊断效能。
Objective：To evaluate the diagnostic efficacy of computer-aided diagnosis （CAD） system in detection of prostate cancer （PCa）. Methods： 266 consecutive patients were recruited from institutional prostate MRI database. All the patients were suspected of PCa,who underwent mpMRI （T2 WI,DWI,MRS and DCE） examination with subsequent ultrasound guided biopsy within 3 months. PI-RADS scores given by uroradiologists were input as mpMRI information for CAD. Clinical data including age,T-PSA,F/T-PSA,digital rectal examination （DRE） and ultrasound findings were input as clinical information for CAD,the output of CAD was the probability of PCa. Taking the clinical outcome as reference standard, the clinical information,mpMRI information or integration of clinical and mpMRI information were input into the CAD respectively,and the corresponding diagnostic efficacy of CAD was studied using ROC analysis. Results： Based on clinical information,the area under the ROC （AUC） of the CAD system for diagnosis of PCa was 0. 888;based on mpMRI,the AUC was 0. 980;and based on clinical information combined with mpMRI, the diagnostic efficacy was the highest with AUC of 0. 986 ,and the corresponding diagnostic sensitivity, specificity and accuracy were 93.65,96.15 and 94.12. Conclusion： The input of CAD system in detection of PCa should consist of both the clinical information and the information of mpMRI.
出处
《放射学实践》
北大核心
2016年第12期1143-1145,共3页
Radiologic Practice
关键词
诊断
计算机辅助
磁共振成像
前列腺肿瘤
前列腺特异性抗原
Diagnosis, computer-assisted
Magnetic resonance imaging
Prostatic neoplasm
Prostate-specific antigen
作者简介
高歌（1988-），女，河南郑州人，博士研究生，主要从事MR新技术和影像信息学研究工作。
通讯作者：王霄英，E-mail：cjr．wangxiaoying@vip．163．com
