RESUMO
PURPOSE: To evaluate the diagnostic accuracy of quantitative, three-dimensional (3D) magnetic resonance (MR) spectroscopic imaging at 3 T for the differentiation of benign and malignant breast lesions, on the basis of choline (Cho) signal-to-noise ratio (SNR) threshold levels, in a clinically feasible measurement time. MATERIALS AND METHODS: Institutional review board approval and written informed consent were obtained from all subjects. Fifty female patients (mean age, 50 years; age range, 25-82 years) with mammographic or ultrasonographic (US) abnormalities were successfully examined in the prone position with a 3-T MR system by using a dedicated breast coil. Lesions were verified by either histopathologic examination or follow-up of at least 24 months. For 3D MR spectroscopic imaging, a point-resolved spectroscopic sequence (repetition time msec/echo time msec, 750/145; field of view, 12 × 12 × 12 cm(3); matrix size, 12 × 12 × 12, interpolated to 16 × 16 × 16; acquisition time, 11 minutes 17 seconds) was used. The maximum Cho SNR was assessed in all lesions and correlated with the histopathologic results. RESULTS: Thirty-two malignant and 12 benign lesions were confirmed in 43 patients with histopathologic examination. Seven patients without biopsy underwent imaging follow-up. In 31 of 32 (97%) malignant and 10 of 19 (53%) benign lesions, Cho was detected. The median Cho SNR in malignant lesions was 5.7, compared with 2.0 in benign lesions. With a Cho SNR threshold level of 2.6, 3D MR spectroscopic imaging provided a sensitivity of 97% and a specificity of 84% for the differentiation of benign and malignant breast lesions. CONCLUSION: At 3T, 3D MR spectroscopic imaging yields high diagnostic sensitivity and specificity for discrimination of benign and malignant breast lesions within reasonable measurement times. This technique allows the study of heterogeneous and multicentric breast tumors and simplifies acquisition planning.