^{Advanced glaucoma disease segmentation and classification with grey wolf optimized U -Net++ and capsule networks}

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Glaucoma is a leading cause of irreversible blindness worldwide.
• Current screening methods are time-consuming and require expert interpretation, delaying diagnosis and intervention.
• Early detection is crucial for preserving vision in glaucoma patients.

Purpose of the Study:

• To develop an automated glaucoma diagnostic system integrating an optimized U-Net++ segmentation model with a Capsule Network (CapsNet) classifier.
• To enhance segmentation of optic disc and cup regions using the Grey Wolf Optimization Algorithm (GWOA).
• To achieve accurate glaucoma classification from retinal fundus images.

Main Methods:

• A two-phase computer-assisted diagnosis (CAD) framework was proposed.
• An enhanced U-Net++ model, optimized by GWOA, was used for segmenting optic disc and cup regions.
• A CapsNet architecture was employed for classifying images as glaucomatous or normal.

Main Results:

• The GWOA-UNet++ and CapsNet framework achieved 95.1% accuracy in segmentation and classification.
• The system outperformed existing benchmark models in accuracy, sensitivity, specificity, precision, and F1-score.
• The model demonstrated robustness against image quality variations and optic disc size differences.

Conclusions:

• The automated system offers enhanced diagnostic accuracy, efficiency, and reliability for early glaucoma detection.
• This tool can serve as valuable clinical decision support for ophthalmologists.
• Future work includes validation on diverse datasets and integration into clinical workflows for scalable deployment.