Mobile Smartphone-Based Digital Pupillometry Curves in the Diagnosis of Traumatic Brain Injury
Lynn B. McGrath, Jessica Eaton, Isaac Joshua Abecassis, Anthony J. Maxin, Cory Kelly, Randall M. Chesnut, Michael R. Levitt
Abstract
Objective The pupillary light reflex (PLR) and the pupillary diameter over time (the PLR curve) is an important biomarker of neurological disease, especially in the diagnosis of traumatic brain injury (TBI). We investigated whether PLR curves generated by a novel smartphone pupillometer application could be easily and accurately interpreted to aid in the diagnosis of TBI. Methods A total of 120 PLR curves from 42 healthy subjects and six patients with TBI were generated by PupilScreen. Eleven clinician raters, including one group of physicians and one group of neurocritical care nurses, classified 48 randomly selected normal and abnormal PLR curves without prior training or instruction. Rater accuracy, sensitivity, specificity, and interrater reliability were calculated. Results Clinician raters demonstrated 93% accuracy, 94% sensitivity, 92% specificity, 92% positive predictive value, and 93% negative predictive value in identifying normal and abnormal PLR curves. There was high within-group reliability ( k = 0.85) and high interrater reliability ( K = 0.75). Conclusion The PupilScreen smartphone application-based pupillometer produced PLR curves for clinical provider interpretation that led to accurate classification of normal and abnormal PLR data. Interrater reliability was greater than previous studies of manual pupillometry. This technology may be a good alternative to the use of subjective manual penlight pupillometry or digital pupillometry.