Introduction The advent of image enhanced endosopic modalities have paved the way for better optical diagnosis of colorectal polyps. Fujifilm recently introduced a new technology – Blue Light Imaging (BLI) that utilises powerful light emitting diode (LED) technology and short wavelength absorption of haemoglobin with no post processor digital reconstruction to enhance mucosal surface and vessel patterns. This is a novel concept. The aim of this study was to investigate the diagnostic ability of BLI in the optical diagnosis of diminutive colorectal polyps.

Method Images from a library of 32 polyps (size 1–9 mm, combination of 19 adenomas and 13 non adenomas) was used. 64 images with equal proportions of high definition white light (HDWL) and BLI were shown to a panel of 6 expert endoscopists with experience in using image enhanced endoscopy. The endoscopists were blinded to the proportion of adenomas in the images viewed. They independently classified each of the 64 images as adenoma or hyperplastic using the established Narrow band imaging International Colorectal Endoscopic (NICE) classification as no standard BLI classification exists. A level of confidence was assigned to each prediction. The sensitivity, specificity, negative (NPV) and positive predictive values (PPV) for adenoma detection on HDWL and BLI were calculated alongside the confidence levels.

Results A total of 384 observations were made with an adenoma prevalence of 59.4%.

The table below shows the sensitivity, specificity, NPV and PPV for adenoma detection on HDWL and BLI.

The diagnostic accuracy using BLI was significantly higher (109/119, 91.6%) for high confidence predictions compared to 75.6% (59/78) for predictions made with low confidence (p=0.002).

Conclusion We have demonstrated the feasibility of using novel BLI technology in the optical diagnosis of small colorectal polyps. BLI improves the accuracy and sensitivity of endoscopic diagnosis with accuracy rates reaching 92% when high confidence predictions were made. These gains can be maximised by developing a dedicated BLI specific classification system and training module to enable its validation and subsequent application in clinical practice.

Disclosure of Interest S. Subramaniam: None Declared, K Kandiah: None Declared, P Aepli: None Declared, G Wurm Johansson: None Declared, E Schoon: None Declared, M Stefanovic: None Declared, P Bhandari Conflict with: Educational grants from Fujifilm, Olympus and Pentax