Table 1.
Summary of key literature on AI-based retinal disease diagnosis.
Fig 1.
Proposed workflow of the automated retinal disease classification system.
Fig 2.
Sample from the dataset with all classes.
Table 2.
Complete system specifications used for model training and evaluation.
Table 3.
Training and optimization parameters for all deep learning models.
Table 4.
Performance comparison of different deep learning models for retinal disease classification.
Fig 3.
Heatmap comparison of performance metrics across the four evaluated models: Scratch Model, EfficientNet-B0, EfficientNet-B7, and AlexNet.
Fig 4.
Processing time comparison for the scratch model, EfficientNet-B0, EfficientNet-B7, and AlexNet.
Table 5.
Five-fold cross-validation results for AlexNet.
Fig 5.
Bar chart showing R2 scores across five cross-validation folds for the AlexNet model.
Fig 6.
Radar chart of model metrics.
Fig 7.
Assessment of the accuracy of proposed and compared models.
Fig 8.
Residual values and heatmap analysis for models.
Fig 9.
Visualization of SHAP-based interpretability analysis for three representative retinal images.
Fig 10.
Confusion matrix of the AlexNet model showing classification performance.
Table 6.
Model performance summary.
Table 7.
Wilcoxon signed rank test for classification models.