MRI habitat and peritumoral radiomics nomogram predicts high-grade prostate cancer preoperatively

A combined MRI radiomics nomogram using intratumoral habitats and peritumoral features predicted high-grade prostate cancer with AUC 0.914 (training), 0.903 (internal), 0.797 (external) in a dual-center study of 274 patients.

• The combined nomogram achieved AUCs of 0.914 (training), 0.903 (internal validation), and 0.797 (external test).
• The 5-mm peritumoral shell yielded the best discriminative performance among peritumoral regions.
• SHAP analysis identified habitat shape sphericity as the most influential feature; the model performed similarly in peripheral and transition zones (AUCs 0.796 and 0.753).

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.