A Hybrid Swarm Algorithm for optimizing glaucoma diagnosis.

Area of Science:

• Ophthalmology
• Medical Imaging
• Computational Intelligence

Background:

• Glaucoma is a leading cause of irreversible blindness.
• Early detection is crucial but challenging due to asymptomatic initial stages.
• Automated screening methods are needed for mass population diagnosis.

Purpose of the Study:

• To develop an automated system for glaucoma diagnosis using fundus images.
• To enhance diagnostic accuracy through optimized feature extraction and classification.
• To improve optimization algorithms for robust automated medical diagnosis.

Main Methods:

• Image pre-processing and feature extraction using hyper analytic wavelet transformation (HWT).
• Classification of extracted features using Support Vector Machine (SVM) with a radial basis function (RBF) kernel.
• Simultaneous optimization of wavelet filter coefficients and SVM parameters using a hybrid Particle Swarm Optimization (PSO) algorithm, incorporating Group Search Optimizer (GSO) for enhanced exploration.

Main Results:

• The hybrid PSO algorithm demonstrated superior diagnostic accuracy compared to conventional PSO in 10-fold cross-validation tests.
• The proposed method effectively extracted informative features from fundus images using HWT.
• The optimized SVM classifier achieved high accuracy in distinguishing glaucoma cases.

Conclusions:

• The developed automated system, utilizing hybrid PSO for optimization, shows significant potential for accurate and efficient glaucoma screening.
• The integration of GSO within PSO enhances the optimization process, preventing premature convergence and improving diagnostic performance.
• This approach offers a promising tool for early glaucoma detection, contributing to the prevention of vision loss.