Projective invariant of surface ratio: application to pupil measurement through simulations and proof-of-concept recordings
Pierre M. Daye, Pierre Pouget · Methods · 2026
Despite the usefulness of pupil size assessment as a proxy for subjects' state of mind in clinical and neuroscientific studies, its measure remains confined to non-ecological, controlled experiments where the subject fixates a stationary target. Here, we propose a novel method to assess pupil metrics based on a projective invariant, allowing the measurement to be insensitive to eye orientation, and to the video eye-tracker experimental setup. We used computer simulations to demonstrate the improvements achieved by the novel method in various scenarios. Finally, the new method was applied during a proof-of-concept free viewing experiment with a single subject that underwent a chemically induced miosis on one of his eyes. Limbus and pupil edges were manually labeled using a custom-made code developed to avoid perceptual bias of the edges' location. We demonstrated that the method tracks the pupil size independently of the eye orientation, negating the need to post-process the data as all solutions previously proposed. Our work could be used to continuously evaluate the pupil size, while relaxing the fixation constraint, paving the way to drastically increase the number of applications using pupil measurements in more ecological conditions.