5.0T multi-parametric diffusion MRI histogram analysis for prostate cancer pathology grading and adverse pathological factors.

Objective: To evaluate 5.0-T ultra-high-field multi-parametric diffusion MRI (mp-dMRI) combined with histogram analysis for assessing the pathological grade, stage, and adverse pathological factors in prostate cancer (PCa). Methods: This retrospective study included 138 patients with histopathologically confirmed PCa. All patients underwent preoperative 5.0-T MRI, including diffusion-weighted imaging (DWI), diffusion kurtosis imaging (DKI), and intravoxel incoherent motion (IVIM) sequences. Histogram parameters were extracted from parametric maps of the apparent diffusion coefficient (ADC), mean diffusivity (MD), mean kurtosis (MK), true diffusion coefficient (D), perfusion fraction (f), etc. using 3D Slicer software. Between-group comparisons were performed using the Mann-Whitney U test or the independent-samples t-test, as appropriate. Spearman's rank correlation analysis was used to evaluate the relationships between histogram parameters and both the ISUP grade group (ISUP GG) and pathological T stage (pT). Logistic regression was used for variable selection and model construction. Diagnostic performance was assessed using receiver operating characteristic (ROC) curve analysis. Results: Among the 138 patients, most histogram parameters of ADC, MD, MK, D, and f differed significantly between clinically significant PCa (csPCa) and clinically insignificant PCa (ciPCa). For distinguishing csPCa from ciPCa, D_90th percentile achieved the highest AUC among single parameters (0.756). Univariate and multivariate logistic regression analyses identified ADC_mean, MD_Minimum, D_90th percentile, and f_Maximum as independent predictors across the DWI, DKI, and IVIM sequences. The combined model comprising these four parameters achieved an AUC of 0.809. After internal validation using bootstrap resampling (1,000 repetitions), the resulting AUC was 0.808, indicating good model robustness. Further correlation analysis showed that multiple histogram parameters of ADC, MD, MK, D, and f also demonstrated mild to moderate correlations with ISUP GG and pT stage. D_RMS and MD_skewness exhibited the strongest correlations with ISUP GG and pT stage, respectively ( Conclusion: Quantitative histogram parameters derived from 5.0-T ultra-high-field mp-dMRI show promise for the non-invasive assessment of tumor aggressiveness and adverse prognostic factors in PCa.