OpenIrisDPI: An Open-Source Digital Dual Purkinje Image Eye Tracker for Visual Neuroscience

Video-based eye trackers are widely used in vision science, psychology, clinical assessment, and studies of human-computer interaction. Most of these eye trackers use the pupil center and corneal reflection (P-CR) to estimate the direction of gaze. However, P-CR eye trackers are often too imprecise to be used in applications with stringent requirements on eye tracking quality. To address this limitation, we present OpenIrisDPI, an opensource plugin for the OpenIris framework that implements dual Purkinje image (DPI) tracking. The system simultaneously tracks both P-CR and DPI signals, enabling a direct comparison between them. We leveraged this fact to demonstrate that the P-CR method overestimates the amount of eye movement during fixation compared to DPI. We further validated the accuracy of our DPI signals by simultaneously recording from neurons in the macaque lateral geniculate nucleus. These data revealed that reconstructing visual inputs using DPI produced tighter correlations between stimuli and neural responses than using P-CR. Through open-source implementation and detailed documentation, OpenIrisDPI makes high-precision eye tracking readily available to the research community.