Objective: To assess the capability of the true fast imaging with steady state free precession (true FISP) sequence in the distinction between hemangiomas and malignant liver lesions.
Methods: Sixty-eight patients with 45 hemangiomas and 51 liver malignancies were included in this study. A 1.5-T magnetic resonance system and a phased-array body coil were used. In addition to true FISP, breath-hold and fat-suppressed, T2-weighted, half-Fourier single-shot turbo spin echo (HASTE) and both unenhanced and gadolinium (Gd)-enhanced T1-weighted sequences were acquired. Two radiologists evaluated the magnetic resonance images independently in a blinded fashion. Interobserver variations with true FISP and HASTE were determined. Lesion contrast-to-noise ratios were calculated from true FISP images.
Results: With true FISP, readers 1 and 2 made a correct distinction between hemangiomas and liver malignancies in 43 of 45 (96%) cases and 40 of 45 (89%) cases, respectively. The kappa value was 0.65. With HASTE, the success rates were 40 of 45 cases (89%) and 36 of 45 cases (80%), respectively, and the kappa value was 0.33. With a Gd-enhanced T1-weighted sequence, the correct classifications were 35 of 45 cases (78%) and 37 of 45 cases (82%), respectively. All hemangiomas appeared as bright and well-demarcated lesions on true FISP images. Malignant liver foci were heterogeneous with unsharp margins and nearly isointense relative to liver. The specificities of true FISP in lesion differentiation were 100% and 98% for readers 1 and 2, respectively. The mean contrast to noise ratio value of hemangiomas was 21.2 (standard deviation [SD] = 9.2), and that of malignant lesions was 4.9 (SD = 3.9). This difference was highly significant (P < 0.0001).
Conclusion: Noninvasive, rapid, and reliable differentiation between hemangiomas and malignant liver lesions is possible by using the true FISP sequence.