Mean difference

Infrared

Ophthalmology

Economics

Youden's J statistic

Receiver operating characteristic

Physics

Mathematics

Optics

Internal medicine

Operations management

Thermography

Confidence interval

Area under the curve

Medicine

Metric (unit)

Nuclear medicine

Contact Lens and Anterior Eye

Purpose To evaluate the efficacy of infrared (IR) ocular thermography in screening for dry eye disease (DED). Methods IR ocular thermography was performed on 62 dry eye and 63 age- and sex-matched control subjects. Marking of ocular surface and temperature acquisition was done using a novel ‘diamond’ demarcation method. 30 static- and 30 dynamic-metrics were studied and receiver operating characteristic curves were plotted. Efficacy of the temperature metrics in detecting DED were evaluated singly and in combination in terms of their area under the curve (AUC), Youden’s index and discrimination power (DP). Results Absolute temperature of the extreme nasal conjunctiva 5 s and 10 s after eye opening were best detectors for DED. With threshold value for the first metric set at 34.7 °C, sensitivity and specificity was 87.1% (95% CI: 76.2–94.3%) and 50.8% (95% CI: 37.9–63.6%) respectively. With threshold value for the second metric set at 34.5 °C, sensitivity and specificity was 77.6% (95% CI: 64.7–87.5%) and 61.9% (95% CI: 48.8–73.9%) respectively. The two metrics had moderate accuracy and limited performances with AUC of 72% (95% CI: 63–81%) and 73% (95% CI: 64–82%); Youden index of about 0.4 and DP of 1.07 and 1.05 respectively. None of the dynamic metrics was good detector for DED. Combining metrics was not able to increase the AUC. Conclusions This work suggests some utility for the application of IR ocular thermography for evaluation of dry eye patients.

Screening for dry eye disease using infrared ocular thermography