Time-dependent diffusion MRI for assessing tumor microstructure and prognostic risk factors in cervical cancer.

PURPOSE: To evaluate the utility of td-dMRI for noninvasively characterizing tumor microstructure and its potential value in prediction of prognostic risk factors for cervical cancer. MATERIALS AND METHODS: In this prospective study, 117 women with suspected cervical cancer underwent td-dMRI on a 3T scanner between January 2024 and February 2025. Microstructural parameters including intracellular volume fraction (fin), cell diameter (d), intracellular diffusivity (Din), extracellular diffusivity (Dex), intracellular water exchange rate (kin), intracellular water exchange time (τ) were derived using a two-compartment model, and multiple ADCs were obtained. Statistical analysis included interreader agreement, ROC, and logistic regression. Histologic validation was performed on H&E-stained slides. RESULTS: After FDR correction, ADC50Hz was significantly lower in SCC than in non-SCC and showed moderate performance for differentiating pathological type (AUC, 0.760). Parameter τ remained the only FDR-significant parameter for histological grade and showed the highest diagnostic performance (AUC, 0.821). For LVSI, cellularity, d, Din, and ADC25Hz remained significant after FDR correction, with AUCs ranging from 0.711 to 0.759. An exploratory combined LVSI model incorporating cellularity, d, Din, and ADC25Hz yielded an apparent AUC of 0.762, a leave-one-out cross-validated AUC of 0.663, and a bootstrap-corrected AUC of 0.700. Parameter d showed a moderate positive correlation with histopathology-derived nuclear diameter (r = 0.595, p < 0.001). CONCLUSION: Td-dMRI-derived parameters may provide noninvasive imaging biomarkers reflecting cervical cancer microstructure and show preliminary value for assessing selected prognostic risk factors, particularly histological grade and LVSI. These findings require further validation in larger, multicenter cohorts.