Assessment of Eyelid Topology and Kinetics Based on Deep Learning Method

Department of Ophthalmology

This study plans to assess eyelid topology (such as margin reflex distance, eyelid contour, and corneal exposure area) and blinking (such as frequency, velocity, and duration), using deep learning method to automatically extract eyelid topological features, and to predict subtypes of levator function, using deep learning method to extract blinking features, in order to provide new ideas and means to assess eyelid topology and kinetics.

Inclusion Criteria:

1. normal volunteers without eyelid diseases
2. patients with blepharoptosis
3. patients with blepharospasm
4. patients with dry eye disease
5. patients with Graves' disease

Exclusion Criteria:

variable ptosis (e.g., myasthenia gravis), entropion, ectropion, enophthalmos, exophthalmos, strabismus, and abnormalities of pupil