|Home | About | Journals | Submit | Contact Us | Français|
To retrospectively evaluate the sensitivity of computer-aided detection (CAD) for full-field digital mammography (FFDM) in 360 breast cancers with regard to mammographic appearance and histopathological tumor type.
Three hundred and sixty breast cancers imaged with FFDM were evaluated retrospectively using CAD. A CAD mark was scored true positive (TP) if it correctly indicated a malignant lesion. All other CAD marks were considered false. Cancer cases were classified as microcalcifi cations (64), masses (196), or both (100). Histopathological findings were classified as IDC, ILC, DCIS or other. Sensitivity values for CAD according to mammographic appearance and, histopathological findings were analyzed using chi-squared tests.
A TP mark was observed on 319 out of the 360 cancers (89%). Calcifications were significantly (P <0.001) more likely to be marked than masses, 163/164 (99%) compared with 244/296 (82%). The probability of a mass TP mark was significantly related to histological type (P <0.001), with sensitivity being greatest for IDC at 76% (167/221) compared with 67% (38/57) for ILC, 31% (16/51) for DCIS and 74% (23/31) for other types. The probability of a calcification TP mark was also significantly related to histological type, with sensitivity being greatest for DCIS at 75% (38/51), compared with 43% (96/221) for IDC, 33% (19/57) for ILC and 32% (10/31) for other types.
CAD prompted the significant majority of radiological abnormalities related to cancers and was most accurate for detection of calcification and DCIS, less so for mass lesions and ILC.