Objective assessment of eyelid metrics for Occuloplastic surgery using an Al model
Author: Mohsan Malik
Base Hospital / Institution: Moorfields Eye Hospital
Rapid fire oral presentation
Abstract ID: 24-377
Purpose
We aimed to develop a tool to assess eyelid and ocular kinematics for oculoplastic surgery objectively
Methods
We used a CE-Marked headset (BulbiCam, Bulbitech, Norway) with near-infrared eye tracking (400-600 fps) and dual, independent screens approved for acuity and pupil function assessment. We modified and calibrated the headset to extend video capture to include periocular areas. We recorded participants’ ocular movements in primary position, elevation, depression, levoversion, and dextroversion. We developed automated measurements using an advanced neural network to extract static and dynamic measures of eyelid function.
Results
We recruited 1000 control participants to develop the neural network model and evaluated the findings with pathological conditions including 10 with thyroid eye disease. The neural network approach enabled automated assessment of eyelid function in the static phase, measuring pupil size, lid position – marginal reflex distance (MRD) 1 and 2, eyelid contour, and scleral show. Additionally, the algorithm assessed sequential frames to determine dynamic changes in lid position in different gaze positions, calculating levator function, ocular and lid vectors and velocity, ocular motility, and blink lagophthalmos for each eye. This allowed for a comparative assessment of asymmetric disease. The model demonstrated internal validity and reliability for repeated evaluation.
Conclusion
Using an automated device, we produced objective, reproducible measurements of lid and ocular function that offer a quantitative approach to assessing clinical parameters for diagnosis. Further work is required to determine the sensitivities of longitudinal assessment.
Additional Authors
| First name | Last name | Base Hospital / Institution |
|---|---|---|
| Jimmy | Uddin | Moorfields Eye Hospital |
