AI-based quantification and diagnosis of age-related macular degeneration using OCT and OCT angiography

This dissertation presents deep learning frameworks for automated retinal biomarker quantification and AMD severity staging using volumetric OCT and OCTA. The work includes four contributions: (1) a CNN for segmentation and classification of four drusen subtypes, (2) a deep learning model for geographic atrophy quantification, (3) a volumetric OCTA restoration model for artifact suppression and improved vascular analysis, and (4) a multimodal AMD staging framework comparing biomarker-based, 2D, and 3D architectures.