Supporting dry eye diagnosis with a new method for non-invasive tear film quality assessment

TL;DR

This study evaluates a novel non-invasive method using high-speed videokeratoscopy and fractal analysis to assess tear film dynamics, significantly improving dry eye diagnosis accuracy over previous techniques.

Contribution

Introduces a new fractal dimension-based analysis method for tear film assessment, enhancing diagnostic sensitivity and specificity for dry eye detection.

Findings

Best parameter was BFI non-invasive break up time with 0.85 AUC

Achieved 100% sensitivity and 84% specificity at 10s cutoff

Method improves tear film homeostasis assessment over previous techniques

Abstract

Purpose: To evaluate the efficacy of a recently proposed method for characterizing tear film dynamics using non-invasive high speed videokeratoscopy in assessing the loss of homeostasis of tear film. Methods: Thirty subjects, from a retrospective study, of which 11 were classified as dry eye and 19 as normal, were included. High speed videokeratoscopy measurements were performed using E300 videokeratoscope (Medmont Pty., Ltd, Melbourne, Australia). Raw data was analyzed using a recently proposed method to estimate the dynamics of the tear film, based on a fractal dimension approach. This method provides three time-varying indicators related to the regularity of the reflected rings: Tear Film Surface Quality (TFSQ) indicator, Breaks Feature Indicator (BFI) and Distortions Feature Indicator (DFI). From each indicator five parameters were extracted and analyzed, including non-invasive break up time, mean value of the indicator in the stability phase, mean value of the indicator in the whole inter blink interval, mean value of the indicator in the levelling phase and the general trend of the time series. Receiver Operating Characteristic were used to determine the sensitivity and specificity of each parameter in dry eye detection. Results: The best discrimination performance between dry eye and normal subjects was achieved with the BFI non-invasive break up time parameter, with an area under the curve of 0.85. For a cut off value of 10 s the sensitivity was 100% and the specificity 84%. Conclusions: The analyzed method improves the assessment of tear film homeostasis in comparison to previous high speed videokeratoscopy methods showing higher potential in assisting dry eye diagnosis.