Real-time decision-making support and documentation in endoscopy (EndoAssist)

Colonoscopy is a frequently performed procedure, with about 500,000 interventions per year in Germany. As the examinations are dynamic diagnostic and therapeutic procedures, decisions have to be made in real time. Afterward, there is a need to summarize the results efficiently and comprehensively in a report that adheres to current nomenclature guidelines.

This project addresses these problems using AI-assisted endoscopic video analysis. Common but important pathologies such as polyps are recognized and labeled in the live video. While the physician is taking images for documentation, the software processes the images. The information collected is combined with free speech text input during the colonoscopy to generate a report, which adheres to current standards for colonoscopy reports. In collaboration with E&L medical systems GmbH, the aim is to directly integrate these systems into the clinical workflow in order to improve patient care.

Research for this project is sponsored by Bayern Innovativ GmbH, Nuremberg, grant number LSM-2007-0008.

The project has been developed in cooperation with Amar Hekalo and Prof. Dr. Frank Puppe, Head of Chair for Artificial Intelligence and Knowledge Systems, Julius Maximilians University of Wurzburg, as well as E&L medical systems GmbH.