This is the authors' abstract. We don't add key points for this paper.
Untreated glaucoma, a chronic eye illness, can cause irreversible vision loss if not caught early. The condition begins with abnormalities in the eye's drainage flow, leading to a rise in intraocular pressure. As the disease progresses, the optic nerve head deteriorates, resulting in vision loss. Ophthalmologists need extensive training and expertise to interpret findings accurately during medical follow-ups to examine the retina. To address this challenge, deep learning-based algorithms have been developed to screen for and diagnose glaucoma using images of the optic nerve, retinal structures, and retinal fundus. This research explores the use of classification and segmentation algorithms based on ResNet to identify glaucoma in fundus images. We fine-tuned the classifier using the DuckPack optimizer and employed XGBoost, LightGBM, and CatBoost algorithms for classification. The results were promising. The segmentation model based on ResNet effectively extracted features, aiding the classification models in accurately identifying glaucoma. All three algorithms performed admirably, though further fine-tuning is needed to determine the best one. Enhancing the model's performance was straightforward after using the DuckPack optimizer for fine-tuning. This study highlights the promising applications of deep learning and sophisticated machine learning algorithms in glaucoma detection. Its findings could inform the development of future diagnostic tools.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Manjunath et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: