477 Development and Validation of a Second-generation of PupilScreen, a Smartphone-based Pupillometer, in Healthy Population. (1st March 2022)

• Record Type: Journal Article
• Authors: Lim, Do; Maxin, Anthony; Gulek, Bernice G.; Buzytsky, Chungeun (Chloe) Lee; Ilya; Coppess, Steve; Alfieri, Robin; Bly, Randall; Law, Anthony; Levitt, Michael R.; Kelly, Cory; McGrath, Lynn
• Is Part Of: Neurosurgery. Volume 68(2022)Supplement 1
• Page Start: 117
• Page End: 118
• Publication Date: 2022-03-01
• Subjects: Nervous system -- Surgery -- Periodicals; 617.48005
• DOI: 10.1227/NEU.0000000000001880_477
• Languages: English
• ISSNs: 0148-396X

Abstract

Abstract : INTRODUCTION: Although there are continuous efforts towards promoting pupillary light reflex (PLR) as a biomarker, its use has been limited due to the expense and limited mobility of pupillometry devices. With demands for non-invasive methods, the expanded use and improved accuracy of PLR are vital to advancing neurological care in myriad settings. METHODS: PLR was captured with both a commercial pupillometer and the PupilScreen app for n=21 subjects, utilizing a standard iPhone camera in a standard clinical environment. The captured data was processed via our machine learning model utilizing a fully convolutional neural network to generate the PLR parameters. Then the parameters were compared to standard pupillometry for both eyes: initial pupil size(px), latency(s), the percentage change in size, average and peak constriction velocity (px/s), average and peak dilatation velocity (px/s), and construction amplitude (px) were analyzed and compared in consideration of age, gender, ethnicity, and the eye color. RESULTS: The second generation of PupilScreen continues to meet the gold standard accuracy of manual pupillometry, despite no longer requiring control of external conditions. Mean parameters were as followed for pupillometry and PupilScreen respectively: mean latency 0.22s(L)/0.23(R) vs via 0.17s(L/R) (p<0.05), mean percent diameter change 36%(L/R) vs 39% (L/R) (p=0.07), and mean constriction velocity 3.1mm/s(L)/3.0mm/s(R) vs 3.0mm/s(L)/2.9mm/s(R) (p<0.05). Demographics subjects: Mean age: 35 years; 62% female; Race/Ethnicity: White 76%, Black/African-American 10%, Asian 14%. Eye color was 38% brown, 29% blue, 33% mixed. CONCLUSION: These results validate the accuracy that the second generation of PupilScreen, can provide estimates of PLR that meet or exceed the accuracy of manual pupillometry and current clinical gold standard digital infrared pupillometry, with the advantage of its being able to be used anywhere there's a smartphone.