A Novel Meibomian Gland Morphology Analytic System Based on a Convolutional Neural Network

Meibomian glands dysfunction (MGD) is the main cause of dry eyes. Biological parameters meibomian gland (MG) such as height, tortuosity and degree atrophy are closely related to its function. However, Thus, an effective quantitative diagnostic tool needed for clinical diagnosis. Automatic quantification MGs' morphological features could be a challenging task play important role in MGD diagnosis classification. Our objective develop artificial intelligence (AI) system evaluating morphology explore relationship between functions. We proposed novel MGs extraction method based on convolutional neural network (CNN) with enhanced mini U-Net. A prospective study was conducted, 120 subjects were included taken meibography. The training validation sets encompassed 60 subjects; test set consisted other comprehensive examinations ocular surface disease index questionnaire (OSDI), tear meniscus height (TMH), break-up time (TBUT), corneal fluorescein staining (CFS), lid margin score, meibum expressibility score. algorithm effectively extracted from meibography even this small sample. As result, while intersection over union (IoU) achieved 0.9077, repeatability 100%. processing each image 100ms. Using method, investigators identified significant linear correlation MG parameters. This provided new obtained by meibography, which has advantages reducing analysis time, improving efficiency, assisting ophthalmologists limited expertise.