Eye-tracking for efficient database labelling: Applications to automatic analysis of colonoscopy video

Abstract

In this paper we present our preliminary results in the automatic analysis of colonoscopy video using eye-tracking. We propose that eye-tracking can be successfully applied to solve different problems in computer assisted colonoscopy, such as database labelling, expertise assessment and abnormality detection. We provide results in these three areas, including a machine learning-based system for colon cancer detection using data generated with eye-tracking.