Dry eye disease is a common ocular condition that can significantly impair quality of life, and dysfunction of the meibomian glands is a major contributing factor. Advances in infrared meibography allow clinicians to visualize these glands, but interpretation remains highly subjective and varies between observers. While artificial intelligence (AI) offers powerful tools for automating image analysis, its clinical reliability depends on high-quality training datasets. This challenge is particularly acute in pediatric populations, where imaging is more difficult and normative data are scarce. Based on these challenges, there is a clear need to establish a rigorously curated, quantitative reference dataset to support standardized and AI-driven analysis of meibomian gland development in children and adolescents.
Researchers from institutions including The University of Melbourne and Fujian Medical University reported (DOI: 10.1186/s40662-025-00460-2) on November 6, 2025, in the journal Eye and Vision, the creation of the Children and Adolescents Meibomian Gland (CAMG) dataset. The study presents a large open-access collection of high-quality infrared images of upper eyelid meibomian glands, analyzed using AI. By combining multi-stage expert quality control with deep-learning segmentation, the team generated reliable quantitative measurements of gland morphology across a wide pediatric age range.
The CAMG dataset comprises 1,114 quality-controlled infrared images collected from 730 children and adolescents aged 4–18 years. Each image underwent rigorous preprocessing and multi-level expert screening before being analyzed by an AI model based on a U-Net architecture. This approach enabled precise segmentation of meibomian glands and extraction of key morphological parameters, including gland length, width, area, gland count, and gland dropout ratio.
The AI model achieved high performance, with an accuracy exceeding 97% and strong agreement with expert annotations, demonstrating its suitability for automated gland analysis. Quantitative results revealed that major gland parameters remained remarkably stable across childhood and adolescence, suggesting that meibomian gland morphology reaches a developmental plateau early in life. Subtle but consistent sex-related differences were observed: females tended to have slightly wider and larger glands, whereas males showed a higher number of glands overall.
Importantly, the study focused on upper eyelid glands, which account for the majority of meibomian tissue and show stronger associations with dry eye symptoms. By providing transparent documentation of image acquisition, quality control, and annotation procedures, the dataset addresses a major limitation of earlier studies and offers a reproducible benchmark for future AI development and epidemiological research.
“Reliable AI in medicine starts with reliable data,” said one of the study's senior investigators. “For pediatric eye health, we have lacked standardized reference values for normal gland development. This dataset fills that gap by combining strict quality control with quantitative analysis. It allows researchers and clinicians to distinguish between normal developmental variation and early signs of dysfunction, which is essential for both clinical decision-making and the next generation of AI diagnostic tools.”
The CAMG dataset provides a critical foundation for precision diagnostics in pediatric ophthalmology. By offering age- and sex-specific reference values, it enables earlier identification of abnormal gland development that may predispose children to dry eye disease later in life. Beyond clinical use, the open-access nature of the dataset supports global collaboration, algorithm benchmarking, and transparent AI validation. It may also facilitate future studies on environmental and lifestyle factors—such as screen exposure—that increasingly affect children's eye health. Ultimately, this work demonstrates how carefully curated data can transform AI from a promising technology into a dependable clinical partner.
###
References
DOI
10.1186/s40662-025-00460-2
Original Souce URL
https://doi.org/10.1186/s40662-025-00460-2
Funding information
The work was supported by Natural Science Foundation of Fujian Province, China (Grant No. 2024J011019) to LL.
About Eye and Vision
Eye and Vision is an open access, peer-reviewed journal for ophthalmologists and visual science specialists. It welcomes research articles, reviews, commentaries, case reports, perspectives and short reports encompassing all aspects of eye and vision. Topics of interest include but are not limited to: current developments of theoretical, experimental and clinical investigations in ophthalmology, optometry and vision science which focus on novel and high-impact findings on central issues pertaining to biology, pathophysiology and etiology of eye diseases as well as advances in diagnostic techniques, surgical treatment, instrument updates, the latest drug findings, results of clinical trials and research findings. It aims to provide ophthalmologists and visual science specialists with the latest developments in theoretical, experimental and clinical investigations in eye and vision.