This single-center retrospective observational study investigates the preoperative risk-stratification value of integrated ¹⁸F-PSMA-1007 PET/MR combined with diffusion-relaxation correlation spectrum imaging (DR-CSI) in patients with histologically confirmed prostate cancer who subsequently underwent radical prostatectomy.
Eligible patients underwent both preoperative ¹⁸F-PSMA-1007 PET/CT and integrated PET/MR, including conventional magnetic resonance imaging and DR-CSI. Quantitative PET parameters, conventional MRI parameters, PI-RADS assessment, and DR-CSI-derived microstructural parameters are evaluated together with available preoperative clinical information. Postoperative whole-gland pathological findings serve as the reference standard.
The study compares the predictive performance of different preoperative information strategies, including clinical variables, PET/CT imaging, PET/MR imaging, and PET/MR supplemented with DR-CSI. Particular attention is given to whether DR-CSI provides complementary microstructural information beyond conventional PET/MR and whether the integrated imaging model can approach the predictive performance of a clinical model for high postoperative ISUP grade.
Model performance is assessed using internally validated prediction methods with repeated nested cross-validation and out-of-fold predictions. Discrimination, prediction error, calibration, and decision-curve characteristics are evaluated. Because this is an exploratory single-center study without external validation, comparisons between imaging strategies are interpreted as internal exploratory evidence rather than definitive proof of clinical superiority or equivalence.
