Retinal Disease Identification from OCT Images Using Dictionary Learning and YOLOv8.

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Automated identification of retinal diseases is crucial for timely diagnosis and treatment.
• Optical Coherence Tomography (OCT) is a key imaging modality for visualizing retinal structures.

Purpose of the Study:

• To develop and validate a novel AI-driven methodology for automated retinal disease identification from OCT images.
• To integrate dictionary learning with deep neural networks for enhanced diagnostic accuracy.

Main Methods:

• Preprocessing of OCT images (normalization, resizing).
• Minibatch online dictionary learning to extract representative features.
• Transformation of data into lower-dimensional sparse encoded images.
• Fine-tuning a YOLOv8n classification model on transformed data.

Main Results:

• Achieved high accuracies of 0.97 and 0.96 on two independent OCT datasets.
• Demonstrated superior performance compared to existing methods.
• Significantly reduced computational costs.

Conclusions:

• The proposed method offers an efficient and accurate approach for automated retinal disease detection.
• The lightweight design facilitates clinical deployment and telemedicine applications.
• This AI tool can serve as an intelligent assistant for ophthalmologists, aiding in early detection of macular diseases.