Evaluating liver fibrosis in chronic hepatitis B patients using multi-compartment restriction spectrum imaging

To optimize multi-compartment restriction spectrum imaging (RSI) models and validate their diagnostic performance for non-invasive staging of liver fibrosis in chronic hepatitis B (CHB), compared to conventional apparent diffusion coefficient (ADC).

This prospective study enrolled 80 CHB patients and 19 healthy controls (METAVIR F0-F4). All underwent 3.0 T MRI with multi-shell diffusion-weighted imaging (b = 0-2000 s/mm2). Four RSI models (2-5 compartments) were fitted; the optimal model was selected using Bayesian Information Criterion (ΔBIC). Conventional ADC and RSI compartmental fractions were compared across fibrosis stages. Diagnostic performance was assessed via ROC analysis.

The 4-compartment RSI model demonstrated the lowest ΔBIC. Compartment C4 (fast pseudo-diffusion fraction) decreased significantly with fibrosis (P = .004), showing a stronger inverse correlation with fibrosis stage than ADC (r = -0.493 vs -0.324, P < .001). For distinguishing F0-F2 vs F3-F4, C4 achieved: Higher AUC than ADC (0.774 [95% CI: 0.679-0.852] vs 0.654 [0.552-0.747]; P = .033); Superior specificity (75.0% vs 44.6%). For F0-F1 vs F2-F4, C4's AUC was 0.814 vs ADC's 0.686 (P = .020).

The 4-compartment RSI model is a promising non-invasive tool for liver fibrosis staging. By resolving distinct diffusion and perfusion compartments, RSI overcomes key limitations of conventional DWI, with the C4 fraction (thought to reflect perfusion decline) serving as a sensitive biomarker for fibrosis progression and outperforming ADC in staging accuracy.

(1) First application of RSI to liver fibrosis staging. (2) C4 fraction is a sensitive biomarker for perfusion loss in fibrosis. (3) RSI enables voxel-wise quantification of fibrosis, offering a potential advantage over ROI-averaged diffusion techniques for detailed spatial analysis.