Purpose: To compare the value of different MRI sequences of the lung for the detection of pulmonary metastases.
Materials and methods: A total of 28 patients with 225 pulmonary metastases confirmed at multidetector-row computed tomography (MDCT) underwent MRI of the lung, including breathhold T2-weighted single-shot turbo spin-echo (half-Fourier single-shot turbo spin-echo [HASTE] and inversion recovery [IR]-HASTE) and conventional turbo spin-echo (TSE and short-tau inversion recovery [STIR]) sequences, a respiratory- and pulse-triggered black-blood STIR sequence (triggered STIR), and breathhold pre- and postcontrast volumetric interpolated 3D gradient-echo (VIBE) sequences. MR images were reviewed by three independent observers and results were correlated with MDCT, which served as standard of reference. Lesion-to-lung contrast-to-noise ratios (CNRs) and image artifacts were also assessed.
Results: CNRs were highest on TSE images (P < 0.001). Mean sensitivities for lesion detection with triggered STIR, TSE, and STIR were 72.0%, 69.0%, and 63.4%, respectively. With HASTE, IR-HASTE, and pre- and postcontrast VIBE, significantly lower sensitivities were obtained (P < 0.05), although artifacts due to physiological motion were less distinct with these sequences compared to TSE and STIR (P < 0.05).
Conclusion: Conventional TSE sequences are more sensitive in depicting pulmonary metastases than single-shot TSE or 3D gradient-echo sequences. Respiratory and pulse triggering can improve lesion detection, but increases acquisition time substantially.