Cross-sectional analysis demonstrates improved accuracy in detecting diabetic retinopathy and macular holes using CNN, suggesting enhanced outcomes for early diagnosis.
Key Points
The proposed CNN model achieves superior accuracy in classifying retinal diseases compared to a traditional DNN model.
Both models were evaluated using standard metrics, with the CNN showing better performance particularly in detecting subtle pathologies.
Deep learning techniques like data augmentation and normalization were essential for improving model robustness and generalization.
This research highlights the utility of AI in enhancing automated diagnostic systems in ophthalmology and real-world clinical settings.