Fenton JJ et al. (2007) Influence of computer-aided detection on performance of screening mammography. N Engl J Med 356: 1399–1409

Computer-aided detection—designed to assist in the interpretation of mammograms—was approved by the FDA in 1998. Since its approval, an increasing number of mammography facilities in the US have adopted this technology; however, few studies have examined the effect of computer-aided detection on the accuracy of mammogram interpretation. To address this issue, Fenton et al. examined the performance of 43 screening facilities in the US during 1998–2002. Complete data were available for 222,135 women (429,345 mammograms).

During the study period, seven centers implemented computer-aided detection. In these centers, diagnostic specificity fell from 90.2% before implementation to 87.2% after implementation (P <0.001), and the positive predictive value decreased from 4.1% to 3.2% (P = 0.01). In addition, while the rate of biopsy increased significantly in those centers that implemented computer-aided detection (P <0.001), the slight increases in sensitivity and cancer-detection rate were not significant (P = 0.32 and P = 0.90, respectively). When data from all 43 centers were analyzed, the accuracy of mammogram interpretation was found to be significantly lower in those centers using computer-aided detection than in those centers that had not implemented this technology (P = 0.005). The authors conclude that the increased rates of biopsy and recall associated with computer-aided detection might not translate into an improved detection of breast cancer. They suggest that larger studies are required to determine whether the clinical benefits of this technology outweigh the potential distress caused to women recalled unnecessarily.