Purpose: Development and validation of a non-invasive imaging system to capture spontaneous blinking and characterise blink dynamics using a custom image analysis algorithm. Methods: A pilot study investigated the influence of illumination type on blink dynamics. Spontaneous blinking was recorded in eight subjects using a high- speed camera under two illumination conditions: white light and infrared. Custom semi-automated software assessed palpebral aperture height (PAH), blink rate, blink speed, blink completeness and blink duration. The main clinical study compared two different image analysis techniques. Spontaneous blinking was recorded in 20 subjects using a high-speed infrared camera. Blink speed and duration were determined using two techniques: manual tracking and semi-automated tracking. Agreement between the two techniques was assessed using Bland-Altman analysis. Coefficients of repeatability (COR) were calculated for the semi-automated technique.
Results: There were significant differences between white light and infrared illumination for PAH (p < 0.0001), blink rate (p = 0.04), closing speed (p = 0.009), opening speed (p < 0.0001) and blink duration (p = 0.0003). The mean differences (95% limits of agreement) between the two techniques were 0.6 mm/s (-15.9 to 17.0) closing speed, 1.5 mm/s (-6.8 to 9.8) opening speed, 2.4 ms (-6.8 to 11.5) closed phase duration and 5.0 ms (-19.4 to 29.3) total blink duration. COR values were 10.1 mm/s closing speed, 3.7 mm/s opening speed, 6.7 ms closed phase duration and 11.2 ms total blink duration. Conclusions: This study has shown that spontaneous blinking can be characterised using a non-invasive imaging system. The semi-automated analysis pro- vides a rapid characterisation of blink dynamics, allowing its application in large-scale trials in a number of clinical areas.