Study of the CADDIE AI Device to Improve Diagnosis of Small Colon Polyps During Colonoscopy in Adults

What this Trial Is About

Researchers are evaluating whether the CADDIE device, which uses artificial intelligence AI to assist in identifying and characterizing small colorectal polyps during colonoscopy, improves the accuracy of endoscopists. The study compares colonoscopies performed with the CADDIE AI system against those done with routine practice without AI assistance. This observational study uses random allocation to place participants into either the AI-assisted or control group to assess the devices impact on optical diagnosis accuracy of polyps 5mm or smaller. The CADDIE system consists of a computer connected to the endoscopy equipment and displays real-time video with AI-generated overlays highlighting polyp features. Endoscopists can activate the AI characterisation by pressing a foot pedal during the procedure. Both groups have colonoscopies recorded, but AI assistance is only active for the intervention group. Polyps under 10mm are measured, imaged in different lighting modes, optically diagnosed, and resected or biopsied if hyperplastic and located in the rectosigmoid area. Each polyp is labeled and sent for histopathology, which serves as the reference standard. Participants undergo colonoscopy with data entered into an electronic system to randomize group assignment. Endoscopists document polyp details, complete diagnosis forms, and record procedure timing and quality markers. The study also evaluates the devices safety through adverse event monitoring and assesses acceptability via questionnaires for staff and participants. The primary outcome is the accuracy of optical polyp diagnosis over one year, with secondary measures including diagnostic confidence, surveillance interval concordance, and device integration into clinical workflow.

Early DiAgnosis Real-Time Healthcare System for CANcer Trial