Retinal imaging based glaucoma detection using modified pelican optimization based extreme learning machine.

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Glaucoma is a progressive optic neuropathy leading to irreversible blindness.
• Accurate and timely diagnosis is crucial for managing glaucoma.
• Automated computer-aided diagnosis (CAD) systems can aid in early detection using fundus images.

Purpose of the Study:

• To develop an improved automated computer-aided diagnosis (CAD) model for binary classification of glaucoma from fundus images.
• To enhance the accuracy and reduce computational time for glaucoma detection.
• To investigate the interpretability of the developed model using Explainable AI (XAI) techniques.

Main Methods:

• Feature extraction using Fast Discrete Curvelet Transform with Wrapping (FDCT-WRP) to capture curve-like features.
• Feature reduction by combining Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA).
• Classification using a hybrid model: Modified Pelican Optimization Algorithm (MOD-POA) optimizing an Extreme Learning Machine (ELM).

Main Results:

• The proposed MOD-POA+ELM model achieved high accuracy: 93.25% on the G1020 dataset and 96.75% on the ORIGA dataset.
• The FDCT-WRP method effectively extracted relevant features for glaucoma classification.
• Seven Explainable AI (XAI) methodologies were employed for model interpretability and validation.

Conclusions:

• The developed automated CAD system demonstrates high efficacy in glaucoma detection from fundus images.
• The combination of advanced feature extraction, reduction, and optimization techniques leads to superior classification performance.
• The use of XAI enhances the credibility and trustworthiness of the automated glaucoma diagnosis system for clinical application.