Blepharoptosis AI Platform (Xception CNN)
Automated screening for functionally significant blepharoptosis
▸ More attributes
Brief overview
A GAN-augmented Xception convolutional neural network was developed to classify functionally significant blepharoptosis from cropped external periocular photographs. Using 771 eyes from a single tertiary oculoplastic center, the baseline model achieved sensitivity 0.68, specificity 0.89, and AUC 0.87, while augmentation with synthetic StyleGAN2-generated images improved sensitivity to 0.95 and AUC to 0.91 at the expense of specificity (0.67). Functional ptosis ground truth incorporated both MRD1 measurements and Humphrey visual field testing. The model was internally validated only, with no external testing, deployment, released code, or pretrained weights. The study is among the first in oculoplastics to explore GAN-generated synthetic periocular data augmentation for ptosis classification.
Baseline model: AUC 0.87; sensitivity 0.68; specificity 0.89; accuracy ~90%.
Baseline model: AUC 0.87; sensitivity 0.68; specificity 0.89; accuracy ~90%. GAN-augmented model: AUC 0.91; sensitivity 0.95; specificity 0.67; accuracy ~90%.
| Study design | Diagnostic accuracy study |
|---|---|
| Center type | Single-center |
| Patients | Not reported |
| Images / eyes | 771 eyes |
| Split | ~70/10/20 |
| Reporting frameworks | None stated |
| Age range | >21 |
|---|---|
| Sex (% female) | Not reported |
| Race / ethnicity reported | N |
| Subgroup performance | N |
- Concept
- Prototype
- Internal val.
- External val.
- Regulator
- Clinical
▸ Why this score?
The APPRAISE-AI tool grades AI/ML medical studies across 24 items in 6 subdomains, scored independently by two reviewers. The total reflects the methodological transparency of the publication — not the underlying clinical value of the model.
How to read this: high scores in clinical relevance and reporting with low scores in methodological conduct, robustness, or reproducibility is the dominant pattern in ophthalmic AI today.
Quality bands: Low <40 · Moderate 40–59 · High 60–79 · Very High 80+. Most published ophthalmic AI sits in the Low–Moderate band; studies that break into High typically combine multi-center datasets, prospective evaluation, and open-source release.
Bibliographic record ▸
| Model ID | oculoplastics_tan_2025 |
|---|---|
| Full author list | Lim, G., Leong, Y. Y., Elangovan, K., Ting, D., Chan, A. S., & Shen, S. |
| Journal | Orbit |
| DOI | 10.1080/01676830.2025.2497460 |
| Conditions | Blepharoptosis |
Model internals ▸
| Architecture family | CNN |
|---|---|
| Architecture detail | Xception-based convolutional neural network with and without GAN-generated synthetic images for data augmentation. |
Disclosures ▸
| Funding | Ophthalmology and Visual Sciences Academic Clinical Programme Grant, Singapore National Eye Centre, Singapore (05/FY2019/P1/09-A2) |
|---|---|
| Conflicts of interest | No conflicts |
Status, version, reviewers ▸
| Card status | Published |
|---|---|
| Card version | v1.0.0 |
| First added | 2026-04-28 |
| Last reviewed | 2026-04-28 |
| Reviewers | attribution opt-in pending |
Card changelog ▸
- 2026-04-28 initial publication (v1.0.0)