The utility of artificial intelligence in analysis of retinal tomography scans

It is known that foveal shape can change with ageing, and it is likely that changes occur with systemic diseases such as diabetes and neurodegeneration. However, there have been no systematic large-scale population level studies using a comprehensive assessment of foveal shape. As a result, the biomarkers linked to vision loss, ageing, and occurrence of systemic diseases remain unknown. This gap in our understanding could be closed with an application of deep learning models, as they could analyse the retinal images and extract the main predictive biomarkers.

In this talk, I will briefly describe the OCT dataset and discuss how fovea parameters vary across the population. Then, I will present my initial results with respect to image processing and retinal layers segmentation. Finally, I will discuss the machine learning techniques used to predict some patient characteristics such as sex or age.