BackgroundMagnetic resonance imaging (MRI) and fluorodeoxyglucose positron emission tomography/computed tomography (FDG PET/CT) show promise for molecular cancer imaging. We evaluated 3T MRI, FDG PET/CT, and ultrasound images for asymptomatic women with an abnormal screening mammogram. MethodsThe Institutional Review Board of Kaohsiung Veterans General Hospital (Kaohsiung, Taiwan) approved the study. Patients provided written informed consent. A total of 11,865 screening mammograms of 118,65 women were performed at our facility between January 2011 and December 2012. Fifty-three asymptomatic women (mean age, 53.3 years) whose screening mammograms had a Breast Imaging Reporting and Data System (BI-RADS) category of 4 or 5 were ultimately enrolled in this study. Breast 3T MRI, FDG PET/CT, and breast ultrasound were performed before biopsy. All imaging modalities were compared by lesion-by-lesion analyses. ResultsFifty-nine breast lesions (28 malignant and 31 benign lesions) from 53 women were analyzed. The sensitivity, specificity, and accuracy for 28 breast cancers were 96%, 77%, and 86%, respectively, for breast 3T MRI; 50%, 100%, and 76%, respectively, for FDG PET,CT; and 61%, 87%, and 74%, respectively, for breast ultrasound. One 0.8-cm invasive breast cancer was missed by the screening mammogram, but detected by breast 3T MRI and FDG PET/CT. The sensitivity for detecting breast cancer was significantly higher with MRI than with PET/CT or ultrasound (for all, p < 0.01). The specificity for detecting breast cancer was significantly higher for PET/CT than for breast MRI (p = 0.02). The sensitivity exhibited by 3T breast MRI and FDG PET/CT for 16 noninvasive breast cancers was 94% and 25%, respectively. ConclusionOn screening mammograms, breast 3T MRI showed higher sensitivity but less specificity than FDG PET/CT for detecting asymptomatic breast cancers.