Vision Sensing-Driven Intelligent Ocular Disease Detection Using Conformer-Based Dual Fusion.

Area of Science:

• Ophthalmology
• Computer Vision
• Artificial Intelligence

Background:

• Deep vision sensing is crucial for early disease detection, particularly in recognizing ocular diseases.
• Extracting fine-grained ocular features for accurate diagnosis remains a significant challenge in the field.

Purpose of the Study:

• To propose an intelligent ocular disease detection system using a Conformer-based dual fusion model.
• To leverage the combined strengths of convolution and Transformer architectures for enhanced feature fusion.

Main Methods:

• Developed a novel vision sensing-driven model integrating convolution and visual Transformer (Conformer).
• Implemented a dual fusion mechanism to combine local subtle features and global image representations.
• Optimized model depth and width to improve accuracy and robustness in ocular disease detection.

Main Results:

• The proposed Conformer-based dual fusion model demonstrated superior performance on real-world ocular disease image datasets.
• Achieved detection accuracy improvements of 1% to 3.7% compared to several mainstream baseline methods.
• The model exhibited enhanced accuracy and robustness in ocular disease identification.

Conclusions:

• The Conformer-based dual fusion approach represents a significant advancement in vision sensing for ocular disease detection.
• This research provides more reliable technical support for the accurate diagnosis of ophthalmic diseases.
• The findings contribute to the development of AI-driven diagnostic tools in ophthalmology.