Validation of Intravoxel Incoherent Motion MRI Using Perfused Explanted Human Livers

Evaluate the feasibility of using perfused explanted human livers for validating intravoxel incoherent motion (IVIM).

Eight (n = 8) explanted livers from deceased donors were obtained. The portal vein and hepatic artery of each explanted liver were connected to a perfusion system. IVIM data were acquired at four total volumetric flow rates (0, 0.6, 0.9, and 1.2 L/min). For each IVIM dataset, diffusion coefficient (D), relaxation-corrected perfusion signal fraction (F_c), blood velocity SD (V_b), and/or pseudo-diffusion coefficient (D*) were estimated. Linear mixed-effects modeling was performed to determine if the effect of applied flow, flow rate, and fibrosis stage on the estimated IVIM parameters was significant (F-tests), while correcting for temporal effect(s). Liver fibrosis stages were obtained from clinical histology.

D was independent of applied flow and total volumetric flow rate (p ≥ 0.30). F_c was approximately zero when no flow was applied and a positive, non-negligible value that was independent of flow rate with applied flow (p < 0.001 and ≥ 0.58 for applied flow and flow rate effects, respectively). V_b and D* were dependent on applied flow and flow rate (p ≤ 0.04 with flow rate effect size ≥ 0.69 mm/s per L/min and 25.7 × 10^-3 mm²/s per L/min for V_b and D*, respectively). Significantly lower (p ≤ 0.01) F_c, V_b, and D* estimates were observed for livers with moderate-to-advanced fibrosis (stages F2-4) compared to no-to-mild fibrosis (stages F0-1) (effect sizes = -0.74%, -0.60 mm/s, and -19.8 × 10^-3 mm²/s, respectively).

Perfused explanted human livers from deceased donors may serve as biologically accurate systems for validation of quantitative IVIM techniques.