AI Glaucoma Detection Using Dual Deep Learning Networks

The system enables automatic self-service through AI algorithms that autonomously extract features from fundus images without requiring manual intervention by domain experts. The deep learning models automatically identify and extract relevant features such as cup-to-disc ratio, optic disc boundaries, and other glaucoma-indicative parameters, eliminating the need for time-consuming manual assessment while maintaining diagnostic accuracy.

The patent replaces the mechanical manual process of feature extraction by ophthalmologists with an automated AI-based system. The mechanical action of manual measurement and analysis is substituted by computational algorithms including convolutional neural networks and other machine learning models that automatically process fundus images and extract diagnostic features, significantly reducing time and cost while preserving accuracy.

The system performs preliminary action by pre-processing fundus images to enhance contrast and reduce noise before the main detection process. This preliminary enhancement of image quality allows subsequent algorithms to work more efficiently with simpler computational requirements while maintaining high positioning accuracy for the optic disc and cup structures.

The patent employs parameter changes by adjusting image processing parameters such as contrast enhancement levels, threshold values, and normalization factors to optimize the balance between detection accuracy and computational complexity. By dynamically tuning these parameters, the system achieves high precision in optic disc positioning without requiring excessively complex algorithms or hardware resources.