Gastroenterology and Artificial Intelligence: 3rd Annual Artificial Intelligence Summit (On-Demand) | September 2021-Opportunity to Redefine Quality Metrics in Endoscopy Using AI

Opportunity to Redefine Quality Metrics in Endoscopy Using AI

Pdf Summary

The article discusses the opportunity to redefine quality metrics in endoscopy using artificial intelligence (AI). The author emphasizes the need to quantify quality in healthcare and identify practices that lead to higher quality care and efficiency. The current challenges in measuring quality metrics in endoscopy are feasibility and data burden. The potential unintended consequences are also highlighted.<br /><br />The article mentions recent developments in AI in gastrointestinal (GI) endoscopy. One study discusses an AI-based automatic quality control system for colonoscopy that uses deep convolutional neural network (DCNN) models to perform tasks such as timing the withdrawal phase, supervising withdrawal stability, evaluating bowel preparation, and detecting colorectal polyps in real-time.<br /><br />Another study introduces a machine learning-based approach called Colonoscopy Coverage Deficiency via Depth algorithm (C2D2) to improve colonoscopy coverage. C2D2 computes a depth image from a single RGB image and identifies which areas of the colon were covered and which remained outside the field of view.<br /><br />The article also mentions AI systems developed for assessing bowel preparation and quantifying bowel cleanliness. Additionally, a case study illustrates the use of computer vision in esophagogastroduodenoscopy (EGD) to obtain photodocumentation during the procedure.<br /><br />The article concludes by stating that AI has the potential to standardize quality metrics, remove subjectivity in measurement, improve outcomes, and augment the intelligence of clinicians. It emphasizes the importance of imagination and being inspired in this field.

Keywords

redefine quality metrics

endoscopy

artificial intelligence

quantify quality

healthcare

efficiency

measuring quality metrics

feasibility

data burden

unintended consequences