Deep Learning Image Analysis of Macular Optical Coherence Tomography Angiography Images for Detection of Progression in Glaucoma

Optical Coherence Tomography Angiography (OCTA) is a new, non-invasive imaging modality for visualizing retinal capillaries for the diagnosis and management of multiple eye-related diseases. Glaucoma is one such disease, caused by the loss of retinal ganglion cells, which if progressed and not treated in time, will eventually lead to blindness. In this study, we present an innovative approach based on Convolutional Neural Networks (CNNs) to detect glaucoma progression. Our progression detection method is composed of five major steps: vessel segmentation, thick vessel masking, Otsu thresholding, super-pixel extraction, and measurement of macular microvascular density. Microvascular density measurements within predefined super pixels are compared between baseline and follow-up OCTA scans of 10 glaucoma patients. Our analysis demonstrates the effectiveness of our approach in distinguishing patients with progressive glaucoma from those whose glaucoma is stable. Our study offers a comprehensive assessment of microvascular vessel density dropouts observed in glaucomatous eyes over time, which enables early detection of disease progression. With proper management and timely treatment for patients with progressive glaucoma, vision loss can be slowed or prevented.