A Novel Automatic Morphologic Analysis of Eyelids Based on Deep Learning Methods.

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Objective evaluation of eyelid morphology is crucial for diagnosing various conditions.
• Current manual measurement methods can be subjective and time-consuming.
• Automated analysis of eyelid features using digital images remains an area for development.

Purpose of the Study:

• To develop and validate a deep learning approach for automated, objective assessment of eyelid morphologic features from 2D digital photographs.
• To evaluate the agreement between automated measurements and manual assessments by ophthalmologists.
• To identify potential age-related trends in eyelid morphologic features.

Main Methods:

• A deep learning model was trained and validated using 2D photographs of 1378 participants.
• Segmentation of eyelids and corneas was performed using a specialized network.
• Manual measurements of Margin Reflex Distance 1 (MRD1) and Margin Reflex Distance 2 (MRD2) were compared with automated measurements for 8 eyelid features.
• Statistical analyses including Spearman's correlation, ICC, and Bland-Altman analyses were employed.

Main Results:

• The deep learning model achieved high accuracy in eyelid (Dice coefficient 0.922) and cornea (Dice coefficient 0.974) segmentation.
• Strong correlations and excellent reliability were observed between manual and automated measurements for MRD1 (r=0.993, ICC=0.996) and MRD2 (r=0.950, ICC=0.974).
• Automated measurements revealed age-related increases in 8 eyelid features, peaking between ages 21-30.

Conclusions:

• The proposed deep learning-based integrative analysis scheme demonstrates performance comparable to human experts.
• The automated approach exhibits excellent reliability and reproducibility for measuring eyelid morphologic features.
• This method holds significant potential for automated diagnosis and remote monitoring of eyelid-related diseases.