During upper digestive tract endoscopy, vision (and thus detection, diagnosis, etc.) is quite disturbed by saliva, bubbles, and other viscous liquids. Studies show that the images obtained can vary in quality and disable a correct diagnosis.

A protocol is established, which is based on an analysis of the images by an operator, using an existing scoring method to compare several clinical options (e.g. using special liquid that limits bubble formation).

The aim of the project is to provide the user with a quantitative score directly extracted from the raw images, possibly in real time, to assist the human expert.

The project will involve image processing and analysis, deep learning and embedded development.

Contacts:

Olivier DEBEIR odebeir@ulb.ac.be, Pierre EISENDRATH pierre_eisendrath@stpierre-bru.be