Introduction Optical imaging is increasingly advocated for characterisation of polyps during colonoscopy. Accuracy of these techniques during surveillance colonoscopy in colonic inflammatory bowel disease (IBD-C) is unclear, with variable results reported. We performed a systematic review and meta-analysis of the diagnostic accuracy of optical imaging techniques including dye-based chromoendoscopy (DCE), virtual chromoendoscopy (VCE), magnification endoscopy and confocal laser endomicroscopy (CLE).

Method We searched Medline and Embase for relevant papers. Full articles or abstracts were eligible when characterisation performance of DCE, VCE (narrow-band imaging [NBI], i-scan, Fujinon intelligent chromoendoscopy [FICE]), magnification endoscopy and CLE had been compared with histopathology, as the reference standard. Enough information had to be required to obtain 2 × 2 contingency table. Pooled analysis was done using random-effects model. Sub-group analysis was performed at real-time (RT) Kudo pit pattern based and real-time (RT) CLE for characterisation of visible lesions. Heterogeneity was assessed using X² and I² statistics.

Results Our search stratergy identified 172 studies of which 20 met the inclusion criteria. Pooled results are outlined in the table.

Conclusion Real-time CLE and magnification endoscopy had best performance characteristics. However there is a lot of herterogeneity in the results. Most CLE and magnification studies were single centre, single expert user which could explain the results. CLE studies were also affected by attrition bias with some studies reporting non-interpretable images in a significant proportion.

Disclosure of Interest None Declared