A Novel Approach using Deep Neural Network Vessel Segmentation & Retinal Disease Detection

Machine Learning has sparked tremendous interest in medical healthcare over past 3 years at multiple level of abstraction to process large amount of structured data like 3-dimensional medical scans and imaging and non-structured data like healthcare prescriptions and notes with less manual interventions. In ophthalmology, machine learning framework has the amazing potential to fasten screening programs with higher accuracies by providing instantaneous feedback and early diagnostics to increase patient compliance and improved health care. Machine Learning has similarly been applied to ocular imaging, using Optical Coherence Tomography (OCT) to detect retinal diseases like diabetic retinopathy, glaucoma, age-related macular degeneration, and retinopathy of prematurity. To reduce manual interventions and faster early diagnostics, machine learning coupled with deep learning will be a potential long-term solution to screen and monitor patients within primary eye care settings. In this study, we proposed a novel 2D Gabor Wavelets using Gradient Boosting trees approach for retinal vessel segmentation and detection of retinal diseases with higher accuracies in Optical Coherence Tomography (OCT) scans which can be extended further in real time environment at clinical settings at different pathologies.