Optimizing Waveform Descriptors and Pupillometry of the Pupillary Light Reflex Using Data Acquired from Video-Based Eye Trackers

Previous models of the pupillary light reflex (PLR) provided a compelling narrative to the complex dynamics of pupil oscillations. However, simple descriptors, such as the time constant obtained from curve fits of a single exponential decay, can provide easily interpretable insights into the dynamics of the PLR. To optimize these time constant estimates and improve the overall quality of the pupillometric data, the ideal eye-tracking system should implement pupillary curve-fitted algorithms that are robust against occlusion artifacts and provide the most accurate centroid approximation.