Impact of breast density on computer-aided detection in full-field digital mammography

The goal of this study was to evaluate the performance of a computer-aided detection (CAD) system in full-field digital mammography (Senographe 2000D, General Electric, Buc, France) in finding out carcinomas depending on the parenchymal density. A total of 226 mediolateral oblique (MLO) and 186 craniocaudal (CC) mammographic views of histologically proven cancers were retrospectively evaluated with a digital CAD system (ImageChecker V2.3 R2 Technology, Los Altos, CA, USA). Malignant tumors were detected correctly by CAD in MLO view in 84.85% in breasts with parenchymal tissue density of the American College of Radiology (ACR) type 1, in 70.33% of the ACR type 2, in 68.12% of the ACR type 3, and in 69.70% of the ACR type 4. For the CC view, similar results were found according to the ACR types. Using the chi-square and McNemar tests, there was no statistical significance. However, a trend of better detection could be seen with decreasing ACR type. In conclusion, there seems to be a tendency for breast tissue density to affect the detection rate of breast cancer when using the CAD system.