Abstract ID: 26-572
Easy-to-use, publicly available software for quantifiable and reproducible reporting of upper eyelid surgery outcomes: new objective shape analysis parameters — what AI needs
Author: Joaquin Alonso Ciria Base Hospital / Institution: Hospital Ramon y Cajal
Presentation Type: ePoster Presentation
Purpose
Reporting of upper eyelid surgery outcomes can be improved in two ways: greater objectivity/reproducibility, and depth beyond MRD1. We aimed to develop a free, publicly accessible, easy-to-use tool based on mathematical curves (Bézier and cubic splines) for quantifiable, objective and comprehensive surgical reporting.
Methods
We built free, fast, intuitive software for upper lid contour analysis from a single clinical photograph, using semi-automatic Bézier/spline curves. It outputs three families of parameters: height (MRD1, radial and vertical mid-pupil-to-lid distances), shape (geometric curvature ‘kappa index’ and contour anomaly detection) and symmetry (bilateral and vs. normative reference). Validated on 184 eyelids from 92 subjects.
Results
For the first time, the tool yields quantifiable parameters of upper lid shape and shape anomalies previously described qualitatively: curvature metrics κ_max, κ_mean and κ_SD — derived from geometric curvature κ (how much the contour bends at each point); the fluctuating Procrustes residual, quantifying shape asymmetry between eyes and vs. the normative reference after optimal alignment, independent of size and position; and three composite 0–1 indices — Contour Shape Index (CSI), Global Normality Index (GNI), Global Symmetry Index (GSI) — classified as within normal limits / borderline / outside normal limits. Validation yielded a normative reference built from the 184 eyelids / 92 subjects.
Conclusion
A free, public, easy-to-use tool delivering quantitative metrics of eyelid height, shape and symmetry improves comparability and objectivity in upper lid surgery, where heterogeneity in outcome reporting hampers cross-technique comparison. As quantitative corneal tomography (e.g. Pentacam) and retinal OCT standardised their fields, these metrics can do so in oculoplastics. It is also a first step toward systematic AI integration: turning qualitative descriptors into quantitative data — what AI needs — keeping the subspecialty at pace with others already benefiting from AI.
Additional Authors
| First name | Last name | Base Hospital / Institution |
|---|---|---|
| Constanza | Barrancos Julian | Hospital Ramon y Cajal |
| Beatriz | Ventas Ayala | Hospital Ramon y Cajal |
| Marco | Sales Sanz | Hospital Ramon y Cajal |