This study aims to validate and evaluate AI algorithms for detection and characterization of early GI neoplasia.
This is an investigator initiated; multi-centre study and will be conducted in two phases. Validation \& Optimization phase, where routinely collected and anonymized endoscopic images and videos from the departmental teaching and training library will be used to validate, and if needed fine-tune and improve, the accuracy of an AI algorithm. Testing phase, where the evaluation of AI algorithms performance will be conducted on a prospective basis. The ground truth (standard) will be expert assessment and histological diagnosis. This study has no direct impact on patient's clinical care, and collection of all endoscopic data will take place during standard endoscopy procedures done for purely clinical indications. No additional procedures, biopsies or interventions will be performed as a part of this study.
Study Type
OBSERVATIONAL
Enrollment
1,650
collection of endoscopic images and videos
Portsmouth Hospitals NHS University Trust
Portsmouth, Hampshire, United Kingdom
RECRUITINGneoplasia detection rate
neoplasia detection rate of the AI algorithms in different locations of the GI tract.
Time frame: during the procedure, up to 24 weeks, at 1 year, through study completion
Accuracy of AI diagnosis
Accuracy of real time optical diagnosis (characterization) of GI neoplasia by the AI algorithm.
Time frame: during the procedure, up to 24 weeks, at 1 year, through study completion
Endoscopist neoplasia detection rate
Detection rate of endoscopists at various levels of experience
Time frame: during the procedure, up to 24 weeks, at 1 year, through study completion
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.