AI approach predicts demographic and clinical characteristics from meibography images

Spread the love

November 04, 2022

1 minute read



Lin M, et al. AI helps to better understand the morphology and function of the Meibomian glands. Presented to: Academy; Oct. 26-29, 2022; San Diego.

Lin reports receiving a grant from the NIH for future work in this area. The project was supported by the Robert Smith Research Fund.

We have not been able to process your request. Please try again later. If you continue to have this problem, please contact

SAN DIEGO – A deep learning model was able to predict patient demographics, clinical and subjective outcomes, and ocular surface health using meibography images, according to a study presented at the 2022 Academy .

Meng C. Lin, DO, PhD, FAAO, a professor at the Herbert Wertheim School of Optometry & Vision Science, director of the clinical research center and co-head of the dry eye clinic at the University of California, Berkeley, shared these study results at a conference of virtual press sponsored by the academy.

doctor on computer

AI can uncover more relationships between meibomian gland morphology and clinical signs. Source: Adobe Stock

“Changes in meibomian gland morphology are one of the main causes of meibomian gland dysfunction,” Lin said. “Current meibography assessment methods lack standards and are time-consuming, and little evidence is available to show downstream effects.”

Data-driven prediction models take three sources, she said: the Meibography image, gland-level features, and metadata such as demographics.

The researchers collected 689 infrared meibography images from 363 patients (170 contact lens wearers, 193 non-wearers).

According to the study, Lin and his colleagues trained a deep learning model to anonymize meibography images and learn the morphological characteristics of the meibomian gland, then predict demographics, clinical outcomes related to gland function , tear film stability, ocular surface health, and subjective outcomes related to discomfort and dryness.

Age and ethnicity were predicted with an average accuracy of 75.6% for Asian patients and 85.8% for non-Asian patients, according to the study. In addition, morphological characteristics of the meibomian glands ranged from 65% to 96% accuracy as highly weighted predictors for eyelid notch and vascularity, quality of meibomian gland expression, extent of corneal staining and mean comfort rating on the visual analog scale (VAS).

“Meibography images can predict an expert clinical diagnosis of MGD with 75% to 85% accuracy using the number of visible glands as the most weighted predictor,” Lin said. “Lower glandular density, fewer visible glands and more glandular atrophy were all main features used to predict clinical signs. However, they cannot predict dry eye symptoms with a high level of confidence.”

Lin concluded, “Machine learning is able to uncover more relationships between meibomian gland morphology and clinical signs than current methods. It can detect the age and ethnicity of patients based on algorithms.

“It is possible that meibography images could soon be used as a biometric fingerprint to identify an individual patient,” she added.

Leave a Comment

Your email address will not be published.