TMIS: a new image‐based software application for the measurement of tear meniscus height

Abstract

Purpose To present a new automated image recognition software for the measurement of tear meniscus height (TMH) and investigate its correlation and efficacy compared with an open‐source software (NIH ImageJ) and manual evaluation. Methods A total of 520 slit lamp photographs, among which 276 were in ×16 magnification and 244 were ×40 magnified, captured from 138 eyes of 69 healthy subjects were assessed for TMH by the new automated Tear Meniscus Identification Software (TMIS), ImageJ and human graders. Images processing of TMIS included filtration, recognition and measurement of slit lamp photographs under certain algorithm, which output two measurement patterns, TMISMax and TMISMean. TMH measured by ImageJ software, considered as the reference value, was conducted by a masked observer while four masked ophthalmologists performed the manual evaluation. Results In both magnifications, TMH measured by TMISMean showed similar values with ImageJ while manual evaluation demonstrated underestimated results, and a strong correlation was detected between TMIS and ImageJ. In ×16 magnified photographs, manually obtained TMH revealed a higher correlation with ImageJ, whereas a notably stronger correlation of TMIS with ImageJ was observed in ×40 photographs. Correspondingly, the accuracy for both TMISMax and TMISMean appeared to be lower than most doctors in ×16 slit lamp images, in contrast to a better precision of TMISMean in ×40 ones. Conclusion The new software displayed high accuracy and efficacy in ×40 magnification and TMISMean pattern, suggesting the possibility of this automated TMH measurement platform to be a valid tool in dry eye screening and follow‐up practice.

Artificial Intelligence in Dry Eye Disease

“…Consequently, the tear meniscus height can be used as a quantitative indicator for DED caused by aqueous deficiency. When connected component labelling was applied to slit-lamp images, the Pearson's correlation between the predicted meniscus heights and an established software methodology (ImageJ [76]) was high, ranging between 0.626 and 0.847 [48].…”

“…It is difficult to compare the studies due to different applications and evaluation metrics. Despite promising results, most of the studies [60,59,55,52,34,35,50,36,49,61,48] did not evaluate their systems on external data. The systems should be tested on independent data before they can be considered for clinical application.…”

“…The systems should be tested on independent data before they can be considered for clinical application. Moreover, some studies were small [61,36] or pilots [48,29], and the suggested models should be tested on a larger number of subjects.…”

Artificial Intelligence in Dry Eye Disease

“…Interferometry is a useful tool that gives a snapshot of the status of the tear film lipid layer, which can be used to aid diagnosis of DED. Machine learning systems have been applied to interferometry and slit-lamp images for lipid layer classification based on morphological properties [60,59,55,54,52,34,35], estimation of the lipid layer thickness [50,36], diagnosis of DED [49,47], determination of ocular redness [61] and estimation of tear meniscus height [48,29].…”

“…The systems should be tested on independent data before they can be considered for clinical application. Moreover, some studies were small [61,36] or pilots [48,29],…”