In this retrospective multicenter study of 488 patients with suspected prostate cancer, investigators developed and externally validated a multimodal radiomics model integrating ^68Ga-PSMA PET/CT and multiparametric MRI to distinguish clinically significant prostate cancer (Gleason score ≥ 3 + 4) from nonclinically significant disease. Using six machine learning classifiers trained on combined diffusion-weighted imaging, T2-weighted imaging, PET, and CT features, the full multimodal approach achieved the strongest performance, with area under the curve values up to 0.91 in the internal test cohort and 0.82–0.89 across two external validation cohorts. Multimodal feature integration consistently outperformed single-modality models and improved discrimination and patient reclassification. Automated prostate segmentation using an nnU-Net model produced comparable results to expert-defined contours, with most radiomic features demonstrating good reproducibility. The authors conclude that multimodal radiomics may support noninvasive prostate cancer risk stratification, although the model is intended as a decision-support tool rather than a replacement for biopsy.