Influence of computer-aided detection on performance of screening mammography.

摘要: BACKGROUND Computer-aided detection identifies suspicious findings on mammograms to assist radiologists. Since the Food and Drug Administration approved technology in 1998, it has been disseminated into practice, but its effect accuracy of interpretation is unclear. METHODS We determined association between use computer-aided at mammography facilities performance screening from 1998 through 2002 43 three states. had complete data for 222,135 women (a total 429,345 mammograms), including 2351 who received a diagnosis breast cancer within 1 year after screening. calculated specificity, sensitivity, positive predictive value with without detection, as well rates biopsy breast-cancer overall accuracy, measured area under receiver-operating-characteristic (ROC) curve. RESULTS Seven (16%) implemented during study period. Diagnostic specificity decreased 90.2% before implementation 87.2% (P<0.001), 4.1% 3.2% (P=0.01), rate increased by 19.7% (P<0.001). The increase sensitivity 80.4% 84.0% was not significant (P=0.32). change cancer-detection (including invasive cancers ductal carcinomas situ) (4.15 cases per 1000 4.20 implementation, P=0.90). Analyses all showed that associated significantly lower than nonuse (area ROC curve, 0.871 vs. 0.919; P=0.005). CONCLUSIONS reduced mammograms. clearly improved cancer.