OBJECTIVES:This study aimed to develop and evaluate an optimized DWI-based virtual MR elastography (vMRE) technique for diagnosing focal liver lesions (FLLs) and monitoring therapeutic responses in malignant tumors, comparing its performance with traditional MRE.
MATERIALS AND METHODS:This retrospective analysis of 142 patients with FLLs using six b-value DWI (150-1500 s/mm2) to calculate eight apparent diffusion coefficient (ADC) combinations. Linear regression analysis was performed to evaluate the correlation between ADC and traditional MRE, and a vMRE formula was constructed. Diagnostic performance of vMRE and MRE was compared for FLLs. Prognostic capabilities of both techniques were assessed in malignant tumors using paired t-tests across three response groups at two time points.
RESULTS:Significant negative correlations were found between ADC values and MRE stiffness (all P < 0.001), and strongest at b = 200/1500 s/mm2 (rho = - 0.785). vMRE and MRE effectively differentiated benign/malignant lesions (AUCs: 0.957 vs. 0.944, P = 0.440), with vMRE showing higher specificity (95.1 % vs. 85.4 %) and MRE better sensitivity (89.2 % vs. 86.3 %). Longitudinal analysis revealed MRE detected earlier stiffness reductions in partial responders (-31.0 %, P = 0.01), while vMRE showed a non-significant decrease (-12.5 %, P > 0.05). Both modalities identified significant stiffness escalation in progressors (vMRE + 40.1 %, MRE + 43.7 %, both P < 0.01) but not in stable disease (all P > 0.05).
CONCLUSIONS:Optimized DWI-based vMRE effectively differentiates FLLs, offering high specificity for diagnosis and comparable prognostic capabilities to MRE in monitoring malignant tumor responses.