VIDEO: AI model provides valuable insight into meibomian gland morphology
Click Here to Manage Email Alerts
INDIANAPOLIS — An AI model reliably quantified meibomian gland morphological features, but could not distinguish between contact lens wearers and non-contact lens wearers, according to a presenter at Academy 2024.
“We have leveraged the power of artificial intelligence to provide a deeper understanding of human meibomian gland morphology and its clinical implications,” Meng C. Lin, OD, PhD, director of UC Berkeley Clinical Research Center and professor at Herbert Wertheim School of Optometry & Vision Science at University of California, Berkeley, said in this Healio video. “Clinicians can easily visualize these glands with infrared camera. However, there is no quick and accurate way to quantify these gland features.”
To address this challenge and determine whether contact lens wear affects gland morphology, Lin and colleagues input 2,233 meibography images into an AI model.
According to Lin, the AI model was able to reliably quantify gland length, width, tortuosity, contrast and atrophy, as well as demographic differences, but it was not able to differentiate meibography images between contact lens wearers and non-contact lens wearers.
“We have learned a lot about morphology,” she said. “The meibomian gland morphology actually encodes a lot more information than we previously thought.”