Related Experiment Video
Updated: Jul 17, 2026
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
Diagnostic performance of artificial intelligence based on PET/CT for lymph node involvement in prostate cancer: A
Dongxuan Song1, Zhiyu Zhao2, Kuangqi Chen3
1Clinical medical college of Tianjin Medical University, Tianjin 300270, China.
Purpose:
This meta-analysis aims to evaluate the diagnostic performance of PET/CT-based artificial intelligence (AI) models for detecting lymph node involvement (LNI) in prostate cancer and compare it with physician readers.
Methods:
We systematically searched PubMed, Embase, and Web of Science through March 4, 2025, for studies evaluating PET/CT-based AI (machine learning or deep learning) for LNI detection. Eligible studies used pathology or expert physician reads as reference standards. AI model studies were assessed with PROBAST + AI, and extractable physician-reader diagnostic accuracy datasets were additionally assessed with QUADAS-2. A bivariate random-effects model pooled sensitivity, specificity, diagnostic odds ratio (DOR), and AUC using patient-based datasets with extractable 2 x 2 data.
Results:
Of 884 records identified, 11 studies involving 1,143 patients were included. AI demonstrated a sensitivity of 0.71 (95% CI: 0.61-0.79), specificity of 0.89 (95% CI: 0.76-0.95), and AUC of 0.79 (95% CI: 0.75-0.82). Physician readers showed sensitivity of 0.67 (95% CI: 0.57-0.76), specificity of 0.88 (95% CI: 0.70-0.95), and AUC of 0.71 (95% CI: 0.67-0.75). AI achieved a higher AUC (Z = 2.95, P = 0.003), but only two studies directly supported AI-versus-physician comparisons.
Conclusions:
PET/CT-based AI models may show modestly improved discrimination for detecting LNI in prostate cancer, but current evidence does not establish clinical superiority over physician readers. The evidence is limited by small study numbers, mostly retrospective designs, substantial heterogeneity, limited external validation, and low certainty. Larger prospective, multi-center, externally validated studies are required before routine clinical implementation.
Registration Number:
PROSPERO (CRD420251128237).
