Prediction of diabetic diseases using deep learning techniques from Optical Coherence Tomography images
This research is about diagnosis and prediction of diabetic diseases using deep learning techniques from optical coherence tomography (OCT) images.
In Dr. Daniel Bar-David’s doctoral thesis, methods were developed for studying this topic on 2D images and 3D data. In this work, we will investigate the prediction of the development of diabetic eye disease, Diabetic Macular Edema (DME) in diabetic patients, both with and without medication.
Diabetic diseases can cause significant vision impairment and even blindness. Early and accurate diagnosis is crucial in order to treat the disease properly and prevent its rogression.
OCT imaging provides high resolution 2D images of the retinal layers, enabling disease diagnosis and tracking the disease progression.
Today, the diagnosis is a manual, long and complex process performed by highly xperienced specialist doctors.
To improve disease diagnosis, reduce human errors and minimize variation in interpretation, deep learning methods will be explored. Dr. Bar-David’s Preliminary work involved calculations on 2D images and 3D data for diagnosis purposes. The prediction of the disease development was described as a future approach.
In this work, a deep learning method will be developed on 3D information to predict the progression of the disease, with and without providing medical treatment. The goal is to get a diagnosis as accurate as possible, analyze and predict the disease progression over time, so that each patient will receive a personalized treatment. Personalized medicine will maximize the outcomes for each patient.