Near-Infrared Indocyanine Green (NIR-ICG) Fluorescence Imaging for Intraoperative Detection of Endometriosis: A Systematic Review and Meta-Analysis

Objective To evaluate the diagnostic performance, protocol determinants, and limitations of NIR-ICG fluorescence imaging for intraoperative detection of endometriosis compared with standard white-light laparoscopy. Methods Systematic literature searches of PubMed, Dimensions, ScienceDirect, DOAJ, and Google Scholar (through 31 July 2025) identified prospective and retrospective studies assessing NIR-ICG for endometriosis lesion detection versus standard white-light (WL) laparoscopy with histopathological confirmation. Risk of bias was assessed using QUADAS-2 and ROBINS-I; certainty of evidence by GRADE; reporting followed PRISMA 2020 guidelines. The review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO; CRD420251128472) on 18 August 2025. Primary outcomes were per-lesion sensitivity and specificity. Secondary outcomes included incremental detection of occult lesions and adverse events. Results Seven studies (230 patients) met inclusion criteria. Sensitivity ranged from 14.7% to 87.8% and specificity from 56.5% to 99.3% across the five studies with calculable diagnostic accuracy. Protocol-optimized studies (ICG dose ≥0.25 mg/kg, imaging delay ≥5 minutes post-injection) achieved pooled sensitivity of 86.3% (95% CI 83.0–89.1%; I² = 10%) versus 22.3% (95% CI 17.1–28.2%; I² = 93%) for non-optimized protocols. The confirmed incremental detection rate for optimized protocols was 14.7% (95% CI 11.8–18.0%), corresponding to a number-needed-to-image of approximately 7. No ICG-related adverse events were reported across all 230 patients. Certainty of evidence was rated moderate for optimized protocols and low overall, owing to heterogeneity and the absence of patient-level outcomes. Conclusion NIR-ICG fluorescence imaging enhances intraoperative endometriosis detection when applied with optimized protocols (adequate weight-based dosing, appropriate imaging delay, and stable visualization platforms), but adds minimal diagnostic value under non-optimized conditions. The marked difference in diagnostic performance between protocol-optimized and non-optimized studies suggests that protocol standardization is the central determinant of diagnostic success. Protocol standardization and large controlled trials evaluating patient-centered outcomes are required before routine clinical adoption.