Kandakji, Lynn;
Balal, Shafi;
Stupnicki, Aleksander;
Liu, Siyin;
Leucci, Marcello;
Gore, Dan;
Allan, Bruce;
(2025)
Data-driven detection of subclinical keratoconus via semi-supervised clustering of multi-dimensional corneal biomarkers.
Ophthalmology Science
, Article 100998. 10.1016/j.xops.2025.100998.
Preview |
Text
1-s2.0-S2666914525002969-main.pdf - Accepted Version Download (2MB) | Preview |
Abstract
Purpose To objectively identify subclinical keratoconus (SKC) from a large sample of healthy and keratoconus (KC) patients via a data-driven framework on corneal imaging data from an anterior optical tomography (AS-OCT) device (MS-39, CSO Italia, Florence, Italy). Design Retrospective cohort study Subjects 25,816 corneal scans from 5,005 patients, including 3,605 with keratoconus and 1,400 healthy control patients, acquired between 2020 and 2024 at two sites within the Moorfields Eye Hospital network in London, UK. Methods Principal Component Analysis (PCA) followed by Gaussian Mixture Modeling (GMM) was applied to AS-OCT derived data, including 20 keratoconus indices and patient age, to identify SKC eyes which were then statistically compared against healthy, and KC eyes. SKC eyes were also validated against external systems including same-day Pentacam (Oculus Optikgeräte, Wetzlar, Germany) scans, Belin-Ambrosio’s ABCD system, KC progression criteria determined by a panel of corneal specialists, and the Moorfields Corneal Cross-linking (CXL) Risk Calculator. Main Outcome Measures Detection of SKC and progression of these eyes to clinically diagnosable keratoconus over time Results The GMM identified 166 eyes from 161 patients with distinct structural differences to healthy and KC eyes. These eyes clustered in the morphometric transition zone in PCA space and were predominantly classified as ABCD Stage 0. However, they demonstrated asymmetry with their fellow eye, higher predicted CXL risk at 1–4 years (p < 0.001) and faster progression to KC (log-rank p < 0.0001) compared to healthy eyes. Among SKC eyes with longitudinal data, 72.7% met Global Consensus criteria for progression. Conclusions SKC remains challenging to detect, and while classic staging such as ABCD retain clinical utility, they are insufficient for early disease detection. PCA followed by GMM classification on a multidimensional AS-OCT dataset identifies a distinct and high-risk subclinical keratoconus group. This semi-supervised framework offers a complementary tool for early risk stratification and can be applied to new patients via projection into the learned PCA space and computation of KC probability. Threshold values corresponding to the 25th and 75th percentiles of KC probability for each parameter may serve as clinical context for flagging eyes when multiple features fall in the atypical range.
| Type: | Article |
|---|---|
| Title: | Data-driven detection of subclinical keratoconus via semi-supervised clustering of multi-dimensional corneal biomarkers |
| Open access status: | An open access version is available from UCL Discovery |
| DOI: | 10.1016/j.xops.2025.100998 |
| Publisher version: | https://doi.org/10.1016/j.xops.2025.100998 |
| Language: | English |
| Additional information: | Under a Creative Commons license https://creativecommons.org/licenses/by/4.0/ |
| Keywords: | Keratoconusa, symptomatic diseases, diagnosis, tomographyoptical coherence, artificial intelligence, machine learning, cluster analysis |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Brain Sciences UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Brain Sciences > Institute of Ophthalmology |
| URI: | https://discovery.ucl.ac.uk/id/eprint/10217952 |
Archive Staff Only
 |
View Item |
