Optic nerve head segmentation using fundus images and optical coherence tomography images for glaucoma detection

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Glaucoma is a leading cause of blindness, characterized by optic nerve degeneration and visual field loss.
• Early-stage glaucoma is often asymptomatic, making early detection crucial to prevent irreversible vision impairment.
• Elevated intraocular pressure is a primary risk factor associated with glaucoma progression.

Purpose of the Study:

• To present a novel automated method for glaucoma detection.
• To utilize both digital fundus images and optical coherence tomography (OCT) images for enhanced diagnostic accuracy.
• To compare the efficacy of different machine learning classifiers for glaucoma identification.

Main Methods:

• Feature extraction from fundus images, including cup-to-disc ratio (CDR) and ISNT ratio.
• Feature extraction from OCT images, including CDR, cup depth, and retinal thickness.
• Classification of normal versus glaucoma using Back Propagation Neural Network (BPN) and Support Vector Machine (SVM) algorithms.

Main Results:

• The proposed system successfully classified glaucoma using combined fundus and OCT image features.
• Back Propagation Neural Network (BPN) achieved an accuracy of 90.76%.
• Support Vector Machine (SVM) classifier demonstrated a higher accuracy of 96.92%.

Conclusions:

• Automated glaucoma classification is feasible using a combination of fundus and OCT imaging data.
• The SVM classifier provided superior performance in distinguishing between normal and glaucomatous eyes.
• This approach holds potential for early and accurate glaucoma diagnosis, aiding in timely treatment and vision preservation.