Comparative Evaluation of Multiple Deep Learning Models for Six-Class Keratitis Classification
5 affiliations
- 1School of Future Technology, South China University of Technology, Guangzhou, China
- 2Department of Ophthalmology, Eye Hospital, Nanjing Medical University, Nanjing, China
- 3Beijing Institute of Ophthalmology, Beijing Tongren Eye Center, Beijing Tongren Hospital, Capital Medical University, Beijing, China
- 4Pazhou Lab, Guangzhou, China
- 5Shenzhen Eye Hospital, Shenzhen Eye Medical Center, Southern Medical University, Shenzhen, China
Abstract
Purpose: Keratitis can be classified as infectious or noninfectious, with infectious keratitis being a major cause of blindness. In this study, multiple deep learning (DL) models were applied to perform six-class classification of corneal lesions, and their diagnostic performance was systematically compared.
Methods: We analyzed 3797 slit-lamp images from six categories: normal cornea, bacterial keratitis, fungal keratitis, viral keratitis, Acanthamoeba keratitis, and non-infectious keratitis. Convolutional neural network (CNN), transformer, self-supervised, and vision–language models were evaluated at image and patient levels using patient-level splits, with clinician comparison on a 240-image test subset.
Results: RegNet had the best image-level performance (accuracy, 76.8%; macro-F1 score, 75.7%; macro-AUC, 0.946), and EfficientNet_B0 had the highest patient-level accuracy (77.5%; macro-F1 score, 76.6%; macro-AUC, 0.954). On the same 240-image subset, RegNet accuracy was 75.4% versus 72.7%, 56.3%, and 44.4% for senior corneal specialists, junior corneal specialists, and junior general ophthalmologists.
Conclusions: CNN-based models showed stable performance under the present image-based evaluation pipeline. RegNet achieved the most balanced image-level results, whereas EfficientNet_B0 showed the strongest patient-level performance. On the same 240-image subset, RegNet showed numerically higher accuracy than clinician-group means. However, performance remains insufficient for autonomous diagnosis and requires external and prospective validation.
Translational Relevance: This comparative DL framework could serve as an adjunctive image-based decision-support tool for six-class keratitis classification after further external and prospective validation, which may assist clinicians in early diagnosis and improve decision making in clinical management.
Related
- DeepFusion-HSK: a multimodal deep learning for subtype classification of herpes simplex keratitis with multicenter validation
- Deep Learning-Based Classification of Melanoma Using Dermoscopic Images: Comparative Evaluation of Convolutional Neural Network and DenseNet Models.
- CSwinFK: a bidirectional cross-attention CNN-Swin Transformer fusion model for fungal keratitis classification in in vivo confocal microscopy images
- Comparative evaluation of fractal analysis and canny edge-detection algorithms for the quantification and classification of bone trabecular patterns in cone-beam computed tomography
- Prospective validation of an uncertainty-aware machine learning model for secondary progressive multiple sclerosis classification
BibTeX
@article{tan2026comparative,
title = {{Comparative Evaluation of Multiple Deep Learning Models for Six-Class Keratitis Classification}},
author = {Qian Tan and Yingnan Xu and Yang Zhang and Yuning Wang and Yanwu Xu and Weihua Yang},
journal = {Translational Vision Science \& Technology},
year = {2026},
month = oct,
doi = {10.1167/tvst.15.10.30},
url = {https://doi.org/10.1167/tvst.15.10.30},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Comparative Evaluation of Multiple Deep Learning Models for Six-Class Keratitis Classification
AU - Qian Tan
AU - Yingnan Xu
AU - Yang Zhang
AU - Yuning Wang
AU - Yanwu Xu
AU - Weihua Yang
PY - 2026
DA - 2026/10/21
JO - Translational Vision Science & Technology
DO - 10.1167/tvst.15.10.30
UR - https://doi.org/10.1167/tvst.15.10.30
AB - Purpose: Keratitis can be classified as infectious or noninfectious, with infectious keratitis being a major cause of blindness. In this study, multiple deep learning (DL) models were applied to perform six-class classification of corneal lesions, and their diagnostic performance was systematically compared. Methods: We analyzed 3797 slit-lamp images from six categories: normal cornea, bacterial keratitis, fungal keratitis, viral keratitis, Acanthamoeba keratitis, and non-infectious keratitis. Convolutional neural network (CNN), transformer, self-supervised, and vision–language models were evaluated at image and patient levels using patient-level splits, with clinician comparison on a 240-image test subset. Results: RegNet had the best image-level performance (accuracy, 76.8%; macro-F1 score, 75.7%; macro-AUC, 0.946), and EfficientNet_B0 had the highest patient-level accuracy (77.5%; macro-F1 score, 76.6%; macro-AUC, 0.954). On the same 240-image subset, RegNet accuracy was 75.4% versus 72.7%, 56.3%, and 44.4% for senior corneal specialists, junior corneal specialists, and junior general ophthalmologists. Conclusions: CNN-based models showed stable performance under the present image-based evaluation pipeline. RegNet achieved the most balanced image-level results, whereas EfficientNet_B0 showed the strongest patient-level performance. On the same 240-image subset, RegNet showed numerically higher accuracy than clinician-group means. However, performance remains insufficient for autonomous diagnosis and requires external and prospective validation. Translational Relevance: This comparative DL framework could serve as an adjunctive image-based decision-support tool for six-class keratitis classification after further external and prospective validation, which may assist clinicians in early diagnosis and improve decision making in clinical management.
SN - 2164-2591
ER -