Video/Image Library of Endoscopy Procedures for the Development of AI-empowered Endoscopy Quality Reporting and Educational Modules

Creation of a Video/Image Library of Annotated Full-length Endoscopy Procedures for the Development of Artificial Intelligence-empowered Endoscopy Quality Reporting and Educational Modules

The goal of this observational study is to establish a video/image library dataset of complete endoscopy or partial colonoscopy procedures for patients with rectal cancer or inflammatory bowel disease (IBD). With this video/image library, the aims are:

• to develop and validate novel AI-empowered solutions to automatically detect and report endoscopy quality metrics
• to develop automated endoscopy reporting solutions, auditing, and educational tools for residents and fellows to enhance their endoscopy skills.

The hypothesis is that a heterogeneous video/image library will provide:

• comprehensive and robust source material to develop AI models
• real-time quality feedback at the end of an endoscopy procedure.

The video/image library will be constructed by prospectively recruiting patients referred for endoscopy procedures for the upper or lower gastrointestinal tract diseases. Patients who have had a screening, surveillance, or diagnostic upper or lower endoscopy recorded as part of a research project at the CHUM will all be invited to participate in this banking.

A qualified and adequately trained study team representative will invite potential study patients to participate during and outpatient clinic visit, or at their arrival in the endoscopy unit on the day of the endoscopy.

Board-certified gastroenterologists, or trainees under supervison, will perform all upper or lower endoscopies at the CHUM. Patients will be prepared for endoscopies according to the most recent guidelines and current standards of care.

The complete or partial endoscopy will be video recorded and the data files will be saved without loss of image quality using Medicapture USB 300 (or equivalent) devices and stored locally in an on-site storage cloud. All videos will be recorded with timestamps burned into the recording. A research assistant will document all relevant informations and annotations during the procedure.

All esophagogastroduodenoscopies (EGDs) will be performed using the high-definition endoscopes. All findings (e.g., lesions, polyps, inflammations), diagnostics, and therapeutic procedures will be documented during EGDs.

The colonoscopies will be performed using high-definition colonoscopes equipped with image-enhanced techniques, or CAD systems, to provide real-time support to endoscopists to detect polyps of all sizes. If required, the endoscopists will be instructed to use white-light and narrow-band imaging to perform optical diagnosis (OD) for all 1-10mm polyps. An OD will first be performed without CADx, then CADx will be used to make a CAD-based OD.

After detecting a polyp and performing ODs, the endoscopists will remove all polyps using standard techniques. Endoscopists will annotate the withdrawal time before initiating a resection, and again when the resection is complete. All biopsy specimens and main polyp specimens will be sent to the institutional pathology laboratories (in separate jars) for histopathology assessments.

Inclusion Criteria:

• ≥ 18 y.o.
• indication of undergoing a screening, surveillance, diagnostic, or therapeutic upper (EGD) or lower (colonoscopy) endoscopy

Exclusion Criteria:

• Coagulopathy defined as an elevated INR ≥ 2.5
• Platelet count ≤ 50,000/mm3
• Emergency endoscopy
• Poor general health defined as the American Society of Anesthesiologists physical status class >3