Noninvasive Estimation of Post-TIPS Portosystemic Pressure Gradient Using 4D Flow MRI Relative Pressure Mapping and CT-Based CFD Driven by MRI-Derived Inflow: A Retrospective Validation Study.

BACKGROUND: Post-transjugular intrahepatic portosystemic shunt (TIPS) portosystemic pressure gradient (PPG) is invasive and a key hemodynamic marker for post-TIPS management. Longitudinal monitoring motivates development of reliable noninvasive alternatives. OBJECTIVE: To validate the performance of non-invasively assessing post-TIPS PPG using four-dimensional flow MRI (MRI_PPG), CT (CT_PPG) (driven by MRI-derived inflow), and concordance with invasively measured PPG. MATERIALS AND METHODS: This retrospective study included 31 post-TIPS patients who underwent PPG measurement and 4D flow MRI within 7 days of the invasive measurement, a subset (n=18) underwent abdominal CT. MRI_PPG was computed from 4D flow MRI velocity fields using relative pressure mapping. For CT-Based CFD Driven by MRI-Derived Inflow, portal venous phase CT was segmented to create patient-specific geometry, and 4D flow MRI-derived inflow at the main portal vein was used to define three inlet protocols: (i) time-averaged peak velocity (CT_PPG1), (ii) time-resolved flow waveform (CT_PPG2), and (iii) time-averaged mean velocity (CT_PPG3). Correlation and agreement with PPG were assessed, and classification performance for clinically relevant PPG ranges was evaluated. RESULTS: In the full cohort (n=31), MRI_PPG (11.23±4.74mmHg) correlated strongly with PPG (11.84±4.72mmHg) (r=0.94, P<0.01) and showed excellent agreement (ICC=0.94), with Bland-Altman bias of 0.61mmHg and limits of agreement (LoA) from -2.70 to 3.92mmHg. In the CT subset (n=18), MRI_PPG remained highly correlated (r=0.91, P<0.01) and showed good agreement (ICC=0.90; bias 0.80mmHg; LoA -3.04 to 4.63mmHg). Among CT‑based estimates, CT_PPG1 demonstrated the best performance (12.47±5.31mmHg; r=0.79, P<0.01; ICC=0.77; bias -0.42mmHg; LoA -6.84 to 6.00mmHg). Three‑range classification accuracy in the CT subset was 77.78% for MRI_PPG, compared with 66.67% for CT_PPG1, 50% for CT_PPG2, and 61.11% for CT_PPG3. CONCLUSION: Both MRI_PPG and CT-PPG CFD estimates are feasible noninvasive approaches for post-TIPS hemodynamic assessment. 4D flow MRI-derived relative pressure mapping demonstrated superior performance, supporting its potential value for post-TIPS follow-up.