Characterization of musculoskeletal lesions on 3-T proton MR spectroscopy

OBJECTIVE. The purpose of our study was to determine the feasibility and value of proton MR spectroscopy at 3 T for characterizing musculoskeletal tumors. SUBJECTS AND METHODS. At 3 T, 18 patients with musculoskeletal lesions (four histologically proven to be malignant, 14 proven benign histologically or at clinical follow-up) underwent 23 MR spectroscopy studies, 20 with a single-voxel technique and three with a multivoxel technique. Seventeen patients were imaged with a surface coil and six with a body coil. Choline signal (3.2 ppm) was measured in each voxel and expressed relative to background noise as signal-to-noise ratio (SNR). Choline SNRs of malignant tumors and benign lesions were compared. RESULTS. Diagnostic spectra were obtained in 20 of 23 lesions. For malignant lesions (osteosarcoma with two MR spectroscopy sites, metastasis, grade 1 sarcoma), choline SNRs were 5.2 and 4.2 (performed with body coil) and 4.8 and 18.7 (performed with surface coil), respectively. For benign lesions (neurofibroma, two stress reactions, bone cyst, hemangioma, lipoma, Baker cyst), choline SNR was 6.3 (with surface coil), 5.5 (with surface coil), and not detected for five cases. Seven postoperative patients with myocutaneous flaps showed either the typical spectrum of muscle or negligible choline. Only a water peak existed in a bone cyst and a significant lipid peak in a lipoma. Choline SNRs were different for malignant and benign lesions (11.7 vs 2.3, p = 0.04, as performed with a surface coil). CONCLUSION. At 3 T, both single-voxel and multivoxel MR spectroscopy are feasible. Proton MR spectroscopy is a potential noninvasive tool for characterizing lesion composition and malignant activity.