One-stop early noninvasive evaluation of renal allograft rejection and fibrosis: microstructural mapping via time-dependent diffusion MRI.

BACKGROUND: Functional MRI offers promise for noninvasive detection of renal allograft injury, but conventional models lack biological specificity. This study aimed to evaluate the clinical utility of time-dependent diffusion MRI (td-dMRI) for detecting rejection and fibrosis in renal allografts. METHODS: In this prospective observational study at a tertiary center in China, adult patients with renal allograft undergoing indication biopsies between April 2024 and June 2025 were enrolled. Microstructural maps were derived from td-dMRI, intravoxel incoherent motion, diffusion kurtosis imaging, stretched exponential model, fractional order calculus, and continuous-time random-walk model. To address overfitting risk from limited samples, final multivariate models were constructed using Firth-penalised multinomial logistic regression. Model performance was evaluated via confusion matrices, sensitivity, specificity, and overall accuracy. The incremental diagnostic value of MRI parameters was quantified by comparing the area under the curve of baseline clinical and final multivariable models. FINDINGS: A total of 100 patients (67 males, 33 females) were analysed. Among them, 41 patients had acute rejection and 13 had chronic rejection; 67 had mild fibrosis, 19 moderate, and 14 severe fibrosis. Td-dMRI-derived cortical cell diameter (diametercortex) significantly distinguished fibrosis grades, while cellularitymedulla identified early fibrotic changes. For rejection, diametercortex, cortical intracellular volume fraction (fincortex), and medullary perfusion fraction (fpmedulla) differentiated acute rejection, and cortical anomalous diffusion coefficient of continuous time random walk model (D-CTRWcortex), fincortex, fpmedulla, and diametercortex distinguished chronic rejection. Both final fibrosis and rejection models achieved high overall accuracy (≈88%) with substantial agreement, significantly outperforming baseline clinical model. INTERPRETATION: Td-dMRI provides a biologically specific, one-stop contrast-free imaging approach for transplant kidney surveillance. It enables noninvasive detection of cellular changes associated with rejection and fibrosis. Integrating td-dMRI with serum biomarkers may improve diagnostic accuracy and reduce unnecessary biopsies, promoting personalised immunosuppression and earlier intervention. Future research should validate its utility in multicenter and longitudinal settings. FUNDING: This study has received funding from the National Natural Science Foundation of China. (Nos. 82371942, U24A200313, and 32427802). The funders had no role in study design, data collection, data analysis, interpretation of results, or writing of the manuscript.