RESUMO
Detectability of breast cancer with magnetic resonance (MR) imaging versus xeromammography was quantitatively compared. MR images were obtained of breasts of 120 women who underwent xeromammography. T1 values were determined for masses larger than 2 cm. Cancer was histologically confirmed in 39 breasts and was considered excluded from 81 due to results of biopsy, cyst aspiration, or sonography or absence of change in xeromammographic findings over time. Images were blindly interpreted by three observers, and results were expressed as receiver operating characteristic curves. Detectability of breast cancer was substantially better with xeromammography than with MR imaging for all observers (P less than .03, 10(-6), and .001). On MR images, spiculation of a mass, distorted architecture, skin thickening, and nipple or skin retraction were specific but relatively insensitive indicators of cancer. Masses with smooth, distinct margins and signal intensity greater than that of fat on T2-weighted images were always benign. Other findings and T1 values were not diagnostically useful. The authors conclude that xeromammography is superior to MR imaging in detection of breast cancer.
