Development And Testing Of A Novel Dry Eye Diagnostic Device

Dry eye disease (DED) has a high prevalence in the population and its incidence is ever rising with increased screen time. DED also has a socio-economic impact. Current tests for DED are neither standardised nor universally accepted and there is no single test which is diagnostic of DED. There are quite a few devices which perform a plethora of tests for DED. On the basis of an online survey among Indian ophthalmologists we felt the need to develop a device that would address the lacunae in the current DED tests. We developed a device that does live lipid layer thickness over the central 5 mm of the cornea, quantitative Meibography, detailed blink analysis and Meniscometry

The device was conceptualised in a clinical setting spread across 5 ophthalmology clinics in India. Four of the top most important tests based on the online survey were designed to be incorporated on a single hardware platform driven by software. Validation was done in these clinics for intra and inter-observer variability as well as in head-to-head trials with two predicate devices.

A small footprint, slit-lamp integrated, USB connected device with cloud integration was achieved. Four tests were developed - lipid layer thickness (LLT) using a unique fiber optic cone that permits imaging of central 5mm of the cornea using low intensity white light. Measures LLT from the area most important for vision and displayes lipid distribution patterns in real-time as a colour coded thickness map. Blink Analysis: Calculates complete blinks and partial blinks, Inter-blink interval and lid contact time. Meibography: IR image acquisition, automatic segmentation of meibomian glands (MG) and calculating MG area as a percentage. Meniscometry: meniscus radius of curvature at the point of interest on the meniscus.

Intra and inter-observer variability was was measured by three observers making two consecutive observations on the same patient at the same sitting. 40 eyes were tested resulting in 240 observations in total. Average LLT and meniscus RoC were recorded. These values were tested for variability using ANOVA at p values <0.05. Both tests showed acceptable variability. Subsequently LLT values were compared with two predicate devices - Lipiview II (28 eyes) and Idra (29 eyes). Bland Altman plotting in both cases showed all data points to be within the +/- 2 SD lines. The values were also compared using ANOVA at p values <0.05. There was no significant difference between our device and the predicate devices.

We have successfully developed a DED testing device which started with a clinical question of what clinicians want to test to diagnose and monitor DED. A small form factor, slit lamp integrable, USB connected device was then tested successfully in the clinical setting and had acceptable intra and inter-observer variability. It also fared well against two predicate devices. The device provides some unique features like LLT measurements in real time of the central cornea and displaying it as a clinician friendly colour coded map, automated MG segmentation and area calculation, and advanced blink parameters analysis like lid contact time and inter-blilnk interval statistics.