Purpose: To qualitatively and quantitatively evaluate the diagnostic efficacy of the breath-hold magnetic resonance (MR) imaging sequences in widespread clinical use for detection and characterization of focal hepatic lesions.
Materials and methods: Fifty patients with 143 lesions [57 hepatocellular carcinomas (HCC), 10 borderline lesions, 18 metastatic tumors, 21 hemangiomas, and 37 cysts] underwent single-shot fast spin echo (SSFSE), fast spin echo (FSE), and gadolinium-enhanced dynamic fast spoiled gradient-recalled acquisition in steady state (FSPGR) breath-hold MR imaging of the liver. Alternative free receiver operating characteristic (AFROC) analysis was performed to independently and prospectively assess each sequence.
Results: For solid lesions, dynamic FSPGR allowed the most sensitive lesion detection and produced the highest lesion conspicuity and lesion-liver contrast-to-noise ratio (CNR). For non-solid benign lesions, SSFSE and FSE produced better results than dynamic FSPGR. SSFSE allowed the most sensitive detection and produced the best lesion conspicuity and lesion-liver CNR.
Conclusion: SSFSE and dynamic FSPGR comprise the optimal imaging protocol for breath-hold MR assessment of focal hepatic lesions. This combination of sequences allows acquisition of critical diagnostic information about both inherent T2 and T1 lesion contrast and lesion vascularity.