MRI-based Habitat Imaging for Noninvasive Prediction of High-Grade Prostate Cancer

Purpose To evaluate the ability of habitat imaging to noninvasively assess high-grade prostate cancer (PCa). Materials and Methods This retrospective, multicenter study included patients with PCa undergoing MRI examination and subsequent radical prostatectomy (RP) between January 2018 and June 2024. Following the 2019 International Society of Urological Pathology (ISUP) guidelines, patients were categorized into low- to medium-grade (ISUP ≤ 3) and high-grade (ISUP ≥ 4) groups, using RP results as the reference. After integrating multimodal imaging data of each voxel, lesions were clustered into k habitat subregions. RP specimens were matched to these subregions, and each subregion's ISUP grade was evaluated to calculate the detection rate of high-grade lesions. Logistic regression identified high-grade PCa-related variables, forming the habitat imaging-clinical imaging (HICI) predictive model. The model's performance was validated using the area under the receiver operating characteristic curve (AUC). Results This study enrolled 359 male patients with PCa (median age, 68 years) divided into training (159 patients), internal test (69 patients), and external test (131 patients) sets. Habitat 1, which featured high cellular density, blood perfusion, and tissue structural complexity, showed a 92.6% (87 of 94) detection rate for high-grade PCa. Logistic regression identified the proportion of habitat 1 (odds ratio [OR], 3.18; P < .001), the prostate-specific antigen level (OR, 2.71; P = .004), and the Prostate Imaging Reporting and Data System score (OR, 1.69; P = .04) as independent risk factors. The HICI model (AUC, 0.87) outperformed the clinical imaging model (AUC, 0.81; P = .01). Conclusion The HICI model can noninvasively assess high-grade PCa. Keywords: MR-Diffusion Weighted Imaging, Prostate, MR-Imaging Supplemental material is available for this article. © RSNA, 2025.