Machine learning approach for ambient-light-corrected parameters and the Pupil Reactivity (PuRe) score in...

Area of Science:

• Ophthalmology and Neuroscience
• Biomedical Engineering

Background:

• The pupillary light reflex (PLR) is a critical indicator of neurological function.
• PLR measurements are highly sensitive to ambient illumination, complicating interpretation.
• Existing methods lack robust correction for environmental light variations.

Purpose of the Study:

• To develop and validate a method for correcting PLR parameters for ambient light.
• To create a reliable clinical score for pupil reactivity.
• To enhance the diagnostic utility of pupillometry in diverse settings.

Main Methods:

• Utilized a smartphone-based pupillometer for 345 measurements across varied light conditions.
• Applied nonlinear models to identify correction functions for PLR parameters.
• Developed a machine learning model combining corrected parameters into a Pupil Reactivity (PuRe) score.

Main Results:

• Ambient light significantly affected pupil size, constriction amplitude, and velocity.
• The developed correction method stabilized PLR parameters across four orders of light magnitude.
• The PuRe score achieved 100% accuracy in discriminating reactive from unreactive pupils.

Conclusions:

• A novel method effectively mitigates ambient light's confounding effect on PLR.
• The PuRe score provides a robust, clinically applicable measure of pupil reactivity.
• Openly available formulae promote wider adoption and research in pupillometry.