The value of computer aided detection system in breast cancer difficult to detect at screening mammography

Abstract

© 2019 Russian Electronic Journal of Radiology. All rights reserved. Purpose. To analyze the features of breast cancers (BC) missed at mammography and to assess the computer-aided detection system (CAD) MammCheck II value in their detection. Materials and methods. We retrospectively analyzed mammography results of 68 patients who were subsequently diagnosed the BC. Results. High parenchymal density was the reason of missing in 23 cases (34.8%); in 10 cases (43.5%) CAD marked invisible changes in the area of subsequently diagnosed BC, 0.8 years (mean) before the diagnosis. In remaining 45 cases (66.2%) the changes were visible, however were not identified or classified as suspicious. The first situation was represented by small low-intensity lesions (10 cases; 23.3%; visualized 2 years (mean) before the diagnosis; mean size: 7 mm; 8 were marked by CAD) or lesions that were very similar to patchy surrounding parenchyma (3 cases; 7.0%; 2.6 years; 11 mm, 3, correspondingly). The visible changes were represented by the lobular masses (7 cases; 15.7%; 2.2 years; 13 mm, 5, correspondingly); architectural distortions (2 cases; 4.7%; 2 years; 13 mm, 2, correspondingly); round lesions with smooth contour (4 cases; 9.3%; 2.2 years; 16 mm, 4, correspondingly); local asymmetric areas of parenchyma (17 cases; 27.9%; 1,5 years; 15 mm, 15, correspondingly); as well as special situations (2 cases; 4.7%; 1.5 years; 12 mm, 2, correspondingly). Discussion. CAD was able to mark 39 of 45 (86.7%) areas that corresponded to the early BC diagnosed later. Conclusion. In most cases the changes corresponded to the early BC were visible, however were not identified as suspicious due to non-specific appearance, small size and low density. CAD marked the significant part of these cases 2 year (mean) before the diagnosis.