MRI of the thorax during whole-body MRI: evaluation of different MR sequences and comparison to thoracic multidetector computed tomography (MDCT)

PURPOSE: To evaluate the accuracy of four MR sequences used as part of a whole-body MRI protocol to detect pulmonary lesions in cancer patients. MATERIALS AND METHODS: A total of 31 oncology patients were imaged in a 1.5T MR scanner (Magnetom Avanto; Siemens Medical Solutions, Germany) for whole-body staging. MR chest imaging included: axial and coronal T2-weighted (T2w)-short-tau inversion-recovery (STIR), axial T2w turbo spin-echo (TSE), and contrast-enhanced (CE) three-dimensional (3D) volumetric interpolated breathhold examination (VIBE). Multidetector computed tomography (MDCT) of the thorax served as the reference standard. The MDCT and MR images were evaluated independently by two radiologists. Comparative analysis was performed per lesion, per lobe, and per patient. Sensitivity, specificity, and predictive values were determined. RESULTS: Compared to MDCT that detected 268 pulmonary lesions ranging from 2 to 75 mm in diameter, the MR sensitivities were 91.1%, 92.5%, 90.8%, and 87.3% for the coronal STIR, the axial STIR, the axial T2w-TSE, and the axial CE 3D-VIBE, respectively. Undetected pulmonary lesions were either calcified or smaller than 10 mm in the axial diameter. With coronal STIR, six false-positive findings were detected; with axial STIR, 14 were detected; with axial T2w-TSE, 10 were detected; and with 3D-VIBE, seven were detected. CONCLUSION: Pulmonary MRI is feasible as part of a whole-body MRI protocol. In our study, STIR images achieved high accuracy compared to chest MDCT for pulmonary lesions of 3 mm in size or larger.