Skip to content
DraconiticResearch
Search Sources API
197,768 studies31 sourcesUpdated 37 min ago
Peer-reviewedCC BY 4.0PLOS One

Automated quantification of central corneal epithelial thickness for the diagnosis of limbal stem cell deficiency

  • Micah Vinet1
  • Nathan Siu2
  • Clemence Bonnet3
  • Craig Choi4
  • Sarah Wu5
  • Celso Nguyen6
  • Corey Arnold7
  • William Speier
  • Sophie X. Deng
7 affiliations
  1. 1Jules Stein Eye Institute, University of California Los Angeles, Los Angeles, California, United States of America
  2. 2Department of Bioengineering, University of California Los Angeles, Los Angeles, California, United States of America
  3. 3Biomedical AI Research Lab, University of California Los Angeles, Los Angeles, California, United States of America
  4. 4Medical Informatics Home Area, Graduate Programs in Bioscience, UCLA, Los Angeles, California, United States of America
  5. 5Department of Computational and Systems Biology, Graduate Programs in Bioscience, UCLA, Los Angeles, California, United States of America
  6. 6Department of Molecular and Cell Biology, University of California, Berkeley, California, United States of America
  7. 7Department of Radiology, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, California, United States of America
Open at sourcePDFDOICite

Abstract

Limbal Stem Cell Deficiency (LSCD) is a corneal disorder characterized by conjunctivalization of the cornea. Central corneal epithelial thickness, measured using Anterior Segment Optical Coherence Tomography (AS-OCT), is a biomarker for LSCD. Manual measurement of epithelial thickness remains the gold standard for diagnosis because existing automated commercial software is unreliable in severe cases. Standard machine learning approaches have limited applicability for automating this measurement due to the difficulty of distinguishing corneal epithelium from scar tissue, and imprecise clinical labels. The goal of this study is to develop a deep learning pipeline to automate measurement of central epithelial thickness in AS-OCT images of eyes with LSCD. To accomplish this goal, we developed AI-CE, a model based on Multi-scale U-Net (MSU-Net) trained to accurately delineate the epithelial layer. The primary dataset comprised of 134 patient eyes, including 47 healthy normal eyes (35.0%), 34 with Stage 1 (25.4%), 34 with Stage 2 (25.4%), and 19 with Stage 3 (14.2%) LSCD. Performance was evaluated by calculating the average mean absolute error between model segmentations and manually measured clinical assessments. AI-CE model performance was compared against a commercially available segmentation software to demonstrate its improvement over the current state-of-the-art. The proposed model achieved a mean absolute error of 5.10 μm, significantly improving (p < 0.01) over clinical software (12.20 μm). This result was verified in a secondary retrospective dataset which similarly showed significant improvement over commercial software. We also demonstrate that clinical epithelial thickness values provided by AI-CE can distinguish between LSCD stages similarly to manual evaluations.

Related

  1. Sulfonylurea receptor 1 deficiency alleviates acute lung injury by activating PARP1 in epithelial cells
    Xiaojuan Yin, Qiaoqiao Zhu, Li Zhao +9
    iSciencePeer-reviewedCC BY 4.0Full textGeneral medicineCrossrefEurope PMC
  2. Gut epithelial aryl hydrocarbon receptor deficiency aggravates fungal dysbiosis-mediated neutrophilic allergic airway disease in mice
    Ting Wang, Chao Lei, Zhishan Xu +10
    Nature CommunicationsPeer-reviewedCC BY-NC-ND 4.0Full textPhysiologyCrossref
  3. Automated quantification of spinal axis distortion for behavioral assessment after spinal cord injury in mice
    Natsuki Oe, Akiko Uyeda, Yasuyuki Ishikawa +3
    Scientific ReportsPeer-reviewedCC BY-NC-ND 4.0Full textGeneral medicineCrossref
  4. Phase-adaptive automated whole-liver quantification of signal intensity-derived liver indices from gadoxetic acid-enhanced MRI
    Felix Barajas Ordonez, Simon Westfechtel, Fadi Matar +7
    European radiology experimentalPeer-reviewedCC BYFull textMedical imaging & radiologyEurope PMC
  5. Automated CMR Quantification of Pulmonary Artery Size-Based and Slow-Flow Metrics in Pulmonary Hypertension
    Khalid Alghamdi, Alireza Hokmabadi, Rob van der Geest +8
    Diagnostics (Basel, Switzerland)Peer-reviewedCC BYFull textGeneral medicineEurope PMC

Cite

BibTeXRIS
BibTeX
@article{vinet2026automated,
  title = {{Automated quantification of central corneal epithelial thickness for the diagnosis of limbal stem cell deficiency}},
  author = {Micah Vinet and Nathan Siu and Clemence Bonnet and Craig Choi and Sarah Wu and Celso Nguyen and Corey Arnold and William Speier and Sophie X. Deng},
  journal = {PLOS One},
  year = {2026},
  month = oct,
  doi = {10.1371/journal.pone.0358258},
  url = {https://doi.org/10.1371/journal.pone.0358258},
  note = {CC BY 4.0},
}
RIS
TY  - JOUR
TI  - Automated quantification of central corneal epithelial thickness for the diagnosis of limbal stem cell deficiency
AU  - Micah Vinet
AU  - Nathan Siu
AU  - Clemence Bonnet
AU  - Craig Choi
AU  - Sarah Wu
AU  - Celso Nguyen
AU  - Corey Arnold
AU  - William Speier
AU  - Sophie X. Deng
PY  - 2026
DA  - 2026/10/05
JO  - PLOS One
DO  - 10.1371/journal.pone.0358258
UR  - https://doi.org/10.1371/journal.pone.0358258
AB  - Limbal Stem Cell Deficiency (LSCD) is a corneal disorder characterized by conjunctivalization of the cornea. Central corneal epithelial thickness, measured using Anterior Segment Optical Coherence Tomography (AS-OCT), is a biomarker for LSCD. Manual measurement of epithelial thickness remains the gold standard for diagnosis because existing automated commercial software is unreliable in severe cases. Standard machine learning approaches have limited applicability for automating this measurement due to the difficulty of distinguishing corneal epithelium from scar tissue, and imprecise clinical labels. The goal of this study is to develop a deep learning pipeline to automate measurement of central epithelial thickness in AS-OCT images of eyes with LSCD. To accomplish this goal, we developed AI-CE, a model based on Multi-scale U-Net (MSU-Net) trained to accurately delineate the epithelial layer. The primary dataset comprised of 134 patient eyes, including 47 healthy normal eyes (35.0%), 34 with Stage 1 (25.4%), 34 with Stage 2 (25.4%), and 19 with Stage 3 (14.2%) LSCD. Performance was evaluated by calculating the average mean absolute error between model segmentations and manually measured clinical assessments. AI-CE model performance was compared against a commercially available segmentation software to demonstrate its improvement over the current state-of-the-art. The proposed model achieved a mean absolute error of 5.10 μm, significantly improving (p < 0.01) over clinical software (12.20 μm). This result was verified in a secondary retrospective dataset which similarly showed significant improvement over commercial software. We also demonstrate that clinical epithelial thickness values provided by AI-CE can distinguish between LSCD stages similarly to manual evaluations.
SN  - 1932-6203
ER  -