Novel approach improves gaze direction detection in closed eyelid conditions, suggesting benefits for clinical applications.
Accurate assessment of pupil dynamics and gaze direction is vital in both clinical and research settings. However, conventional pupillometry methods typically require open eye conditions, which limit their applicability in patients under sedation, anesthesia, or during sleep. We previously introduced a novel, touchless short-wave infrared (SWIR) imaging method capable of separately monitoring pupil size or gaze direction through closed eyelids. Yet, clinical bedside applications demand a more robust approach that performs pupillometry while gaze direction changes dynamically, when both eyes are closed, and yields improved signal-to noise ratio (SNR). Here, we expand upon the initial proof-of-concept by incorporating polarizing filters into the SWIR imaging setup to enhance contrast and signal quality. Multiple polarizer configurations, including crossed and parallel setups, were evaluated in both open and closed-eye conditions to assess their effects on SNR during pupillary light reflex (PLR) imaging. In addition, we implemented improved image-processing algorithms to enable accurate pupil localization and gaze direction estimation under naturally closed eyelid conditions. We find that our improved setup employing crossed polarizer configuration significantly improved PLR SNR compared to a setup without polarizers, due to improved pupil localization and reduced surface glare. These findings suggest that the integration of polarizers into SWIR-based pupillometry systems enables more accurate and reliable measurements in clinically relevant scenarios where eyes remain closed. This advancement opens opportunities for broader applications in sleep medicine, anesthesiology, and neurocritical care.
No takes yet. Share an insight, caveat, or question.
Tepper et al. (2025) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: