Enhancing Readability and Detection of Age-Related Macular Degeneration Using Optical Coherence Tomography Imaging:...

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Artificial intelligence (AI) shows promise in medical diagnosis.
• Optical coherence tomography (OCT) is crucial for analyzing retinal conditions.
• Accurate diagnosis of age-related macular degeneration (AMD) is vital for patient outcomes.

Purpose of the Study:

• To evaluate a collective AI model for diagnosing AMD stages using OCT images.
• To compare the performance of an ensemble AI model against individual AI architectures (ResNet, EfficientNet, Attention).
• To demonstrate the effectiveness of weighted fusion in combining AI model predictions.

Main Methods:

• Utilized the Noor dataset (16,822 OCT images) for training and validation.
• Developed an ensemble AI model employing weighted fusion of predicted probabilities.
• Compared the ensemble model's performance against ResNet, EfficientNet, and Attention models using standard metrics (precision, recall, F1 score, ROC curves).

Main Results:

• The collective AI model achieved superior accuracy (91.88%) in classifying AMD stages compared to individual models.
• Ensemble model performance metrics included 92.54% precision, 92.01% recall, and 92.03% F1 score.
• Refinement of misclassified cases improved the model's accuracy to 97%.

Conclusions:

• Ensemble AI models with trainable weights offer enhanced accuracy for diagnosing retinal conditions like AMD.
• Model fusion is beneficial for complex medical image analysis in ophthalmology.
• AI presents a valuable tool for improving diagnostic capabilities in ophthalmology.