Evaluating the clinical value of MRI multi-model diffusion-weighted imaging on liver fibrosis in chronic hepatitis B patients

Purpose To explore the value of various diffusion parameters obtained from monoexponential, biexponential, and stretched exponential in assessing liver fibrosis in chronic hepatitis B (CHB). Methods DWI and intravoxel incoherent motion (IVIM) MRI were performed prospectively on liver for 146 patients with CHB and 21 healthy volunteers. ADC values were obtained from monoexponential model imaging. Diffusion coefficient (D), pseudodiffusion coefficient (D*), and perfusion fraction (f) obtained by biexponential model imaging, and stretched exponential model to obtain diffusion distribution coefficient (DDC) and diffusion heterogeneity index (alpha). Blood draw were performed on patients to obtain AST, ALT, and PLT, and then APRI and FIB-4 index were determined based on the serological diagnostic models. The fibrosis stage was staged (S0-S4) according to the pathology of liver puncture. Independent samplettest was used to compare the parameter values between liver fibrosis group and control group. One-way ANOVA was used to compare the parameters of different liver fibrosis grades. Bonferroni test was used for correcting multiple comparisons. Spearman correlation was used to analyze the correlation between each parameter and liver fibrosis grades. ROC was used to predict the diagnostic power of each parameter for liver fibrosis stages >= S2 and >= S3. Results ADC,D,D*,f, and DDC values were significantly different between normal control group and hepatic fibrosis group (P < 0.05). There were significant differences in ADC,D*,f, and DDC value among liver fibrosis groups (P < 0.05). D* and DDC values were moderately negatively correlated with the grades of liver fibrosis (r = - 0.483,P < 0.001;r = - 0.622,P < 0.001). ADC andfvalues were slightly negatively correlated with the grades of liver fibrosis (r = - 0.295,P < 0.001;r = - 0.312,P < 0.001). DDC values have the highest diagnostic efficiency in liver fibrosis stages >= S2 and >= S3. The areas under ROC curve (AUC) were 0.813 and 0.832 for >= S2 and >= S3, respectively, the sensitivity is 83.72% and 73.53%, and the specificity of 83.33% and 66.04%, which were better than APRI and FIB-4. Conclusion D* obtained from biexponential and DDC obtained from stretched exponential DWI have better value in evaluating the degree of liver fibrosis in CHB.