Detection of reactions to sound via gaze and global eye motion analysis using camera streaming

This work is focused on the field of automatic hearing assessment for patients presenting cognitive decline or severe communication difficulties. Audiometry is a test of behavior requiring intense interaction between patient and audiologist, so it is extremely difficult to properly assess patients with severe communication disorders. However, patients with cognitive decline often make some eye gestures in reaction to auditory stimuli. These reactions are interpreted by expert audiologists. On the other hand, a manual assessment of the patient creates problems such as subjectivity or low reproducibility, to name but two. Bearing this in mind, this paper introduces a novel methodology to analyze video recordings acquired during audiometric evaluations and characterize movements of the eye so they can be interpreted as a positive gestural reaction to sound. Motion analysis in the eye region helps the human expert to establish the existence of a reaction to sound, thus increasing the reproducibility and objectivity of the test.