Quantitative analysis of the diffusion-weighted steady-state free precession signal in vertebral bone marrow lesions.

OBJECTIVES : Diffusion-weighted steady-state free precession (DW-SSFP) sequences have shown great potential for the differential diagnosis of benign osteoporotic and malignant neoplastic vertebral compression fractures, which appear hypo- to isointense or hyperintense in DW-SSFP magnetic resonance imaging, respectively. In contrast to other diffusion weighting sequences, the DW-SSFP signal depends not only on the apparent diffusion coefficient (ADC), but also on the tissue relaxation times and sequence parameters. The purpose of the present study was to provide a detailed analysis of the DW-SSFP signal in benign and malignant vertebral lesions (VLs) and in vertebral bone marrow (VBM) to understand the observed signal alterations and their dependence on tissue and sequence parameters. MATERIALS AND METHODS : Magnetic resonance imaging was performed in 40 patients with benign (n = 20) or malignant (n = 20) VLs to determine the fat fraction and tissue parameters (ADC, T1, T2, T2*) for both the water and fat signal. With these values, the DW-SSFP signal was simulated and compared with the measured signals for different diffusion gradients by determining the signal intensity ratio between the SSFP signals of the lesions and of normal-appearing VBM for both malignant and benign VLs. RESULTS : The simulated DW-SSFP contrast agreed well with the measured contrast and provided a very good differentiation between benign osteoporotic and malignant VLs. ADCs were significantly different in both lesion types (malignant 1.36 vs. osteoporotic 1.77 × 10 mm/s); however, the observed contrast differences were caused predominantly by an opposed-phase readout in combination with significantly different T2* values (malignant 22 vs. osteoporotic 14 ms) and fat fractions (malignant 3.9% vs. osteoporotic 12%) in the lesions as well as significantly different fat fractions in normal-appearing VBM (malignant 42% vs. osteoporotic 52%) of both patient groups. CONCLUSIONS : Although the ADCs of the evaluated malignant and benign VLs showed highly significant differences, the influence of diffusion on the DW-SSFP signal contrast is relatively low compared with other tissue parameters due to the very complex signal mechanism of the SSFP sequence. Thus, the observed DW-SSFP signal contrast of different VLs (hypo-/isointense vs. hyperintense signal) is rather fat- and T2*-weighted than diffusion-weighted. The intermediate diffusion weighting of the applied SSFP sequence, however, helps to shift the different contrasts into a signal range that is easily visually accessible.