Objectives: To assess liver fibrosis in patients with chronic liver disease using diffusion kurtosis imaging (DKI) in comparison with conventional diffusion-weighted imaging, with histology as reference standard.
Methods: This prospective study included 81 patients and DKI with b-values of 0, 200, 500, 1,000, 1,500, 2,000 s/mm2 were performed. Mean diffusivity (MD), mean kurtosis (MK) and apparent diffusion coefficient (ADC) maps were calculated. The diagnostic efficacy of MD, MK and ADC for predicting stage 2 fibrosis or greater, and stage 3 fibrosis or greater were compared.
Results: The MD (rho=-0.491, p<0.001), MK (rho=0.537, p<0.001) and ADC (rho=-0.496, p<0.001) correlated significantly with fibrosis stages, and ADC exhibited a strong negative correlation with MK (rho=-0.968; p<0.001) and a moderate association with MD (rho=0.601, p<0.001). Areas under the curves (AUCs) for predicting stage 2 fibrosis or greater were not significantly different (p>0.05) between MK (0.809) and ADC (0.797) as well as between MD (0.715) and ADC. AUCs were also similar for MD (0.710), MK (0.768) and ADC (0.747) for predicting stage 3 fibrosis or greater.
Conclusion: Although DKI is feasible for predicting liver fibrosis in patients with chronic liver disease, MD and MK offer similar diagnostic performance to ADC values.
Key points: • Diffusion kurtosis imaging is feasible for staging liver fibrosis. • Diffusion kurtosis and monoexponential model are highly correlated. • The kurtosis model offers no added value to the conventional, monoexponential model.
Keywords: Diagnostic imaging; Diffusion-weighted MRI; Liver disease; Liver fibrosis; Magnetic resonance imaging.