Ocular Diseases Detection using Recent Deep Learning Techniques.

Area of Science:

• Ophthalmology
• Artificial Intelligence
• Medical Imaging Analysis

Background:

• Early fundus screening is crucial for reducing blindness caused by ophthalmic diseases.
• Manual evaluation of fundus images is time-consuming and labor-intensive.
• Current deep learning models often focus on detecting only a single ophthalmic disease.

Purpose of the Study:

• To investigate and optimize various deep learning models for the detection of multiple eye diseases.
• To improve the efficiency and accuracy of ophthalmic disease detection compared to manual methods.
• To address the limitation of single-disease detection in existing deep learning approaches.

Main Methods:

• Exploration of diverse deep learning architectures for ophthalmic disease identification.
• Implementation of several optimization techniques to enhance model performance.
• Evaluation of model efficacy across a range of eye conditions.

Main Results:

• The optimized deep learning model achieved a high Area Under the Curve (AUC) of 98.31% for detecting six eye diseases.
• The model demonstrated strong performance with an AUC of 96.04% for detecting eight eye diseases.
• Significant improvements in detection accuracy and efficiency were observed.

Conclusions:

• Deep learning models, when optimized, offer a powerful and efficient tool for multi-disease ophthalmic screening.
• This approach has the potential to significantly reduce blindness through early and accurate detection of eye conditions.
• The study highlights the effectiveness of advanced AI in advancing ophthalmology diagnostics.