Purpose: To evaluate a digital photography method of pupil size estimation over a broad range of illumination conditions and to compare this method with common clinical techniques.
Setting: College of Optometry, Ohio State University, Columbus, Ohio, USA.
Methods: Two examiners measured the pupil diameter in 45 right eyes at 3 illumination levels: <0.63 lux (dark), 5 lux (dim), and 1000 lux (bright). Estimation by infrared video recording, the reference standard, was compared with measurements by digital photography, ruler, semicircular templates, and the Colvard pupillometer. Masked graders measured pupil size from infrared video recordings and digital photographs.
Results: The repeatability of the measurement method determined by the mean intraclass correlation coefficients was highest for video recording across conditions (0.86-0.97), followed by digital photography (0.76-0.94), Colvard pupillometry (0.63-0.82), ruler (0.71-0.85), and templates (0.70-0.83). An analysis of variance showed a significant difference in pupil size by method (P<.001). All methods except digital photography estimated smaller pupil sizes under dark and dim illumination than infrared video measurements (all P<.01). Under bright illumination, the ruler measurements were significantly smaller (-0.15 mm) and the Colvard pupillometer measurements were greater (+0.30 mm) than the reference (P<.01). The 95% limits of agreement (LoA) between examiners were smallest for video measurements at all light levels. The remaining measures ranked from best to worst by 95% LoA were digital photography, Colvard pupillometry, ruler, and templates.
Conclusions: Estimation of pupil size by digital photography was more repeatable and accurate than estimates by common clinical techniques over a wide range of illumination. Although not as quick as other methods, digital photography is relatively inexpensive, permits lasting documentation, and allows independent grading suitable for clinical research purposes.