PNET-PRISM: a multicenter-validated radiomics nomogram for noninvasive grading of pancreatic neuroendocrine tumors - PubMed
5 hours ago
- #Radiomics
- #Pancreatic Neuroendocrine Tumors
- #Artificial Intelligence
- PNET-PRISM is a radiomics nomogram for noninvasive grading of pancreatic neuroendocrine tumors (PNETs).
- The study included 407 patients across training, validation, and external test cohorts.
- PNET-PRISM integrates CT radiomics features from intratumoral, peritumoral, habitat, and deep learning domains.
- The model achieved high AUCs (0.92, 0.89, 0.87) in training, validation, and external test sets.
- PNET-PRISM showed 100% sensitivity in external validation and improved grading in cases where EUS-FNA failed.
- The M-DLR Score was associated with progression-free survival (HR = 2.050).
- The tool addresses limitations of EUS-FNA and aids in treatment selection when biopsy is inconclusive.