Computer Vision News - January 2023

33 Ruogu Fang The lab uses the open-access and large- scale biomedical database UK Biobank , which has 500,000 subjects from the UK and has been recruited for the past 10 years. Among them, 4,000 people have already been diagnosed with Alzheimer’s disease. The lab uses a subset of the database containing subjects with a retinal fundus image at their baseline visit and then pairs themwith age and gender-matched healthy controls to develop its AI and machine learning platform . It has built a modular machine learning pipeline to classify who has and doesn’t have Alzheimer’s disease from those fundus vasculature images. “ By looking into the eye, which is a window to the brain, we have a big promise to provide a unique, timely, accessible, and low-cost solution to perform Alzheimer’s disease screening for those at risk in the older population, ” Ruogu tells us. “ This work is still in the explorative research stage. When it can be applied to clinical practice with high accuracy, precision, and confidence, that will need our and maybe changes in the brain . The eye and the brain come from the same embryonic stage and share many anatomical, functional, metabolic, hemodynamic similarities. ” The eyes are recognized as a means for understanding diseases of the brain, such as Alzheimer’s. Still, most researchers use a statistical method to show that Alzheimer’s patients and healthy controls have some retinal differences. This method is like a t-test and has no predictive powers. SMILE lab proposes an explainable AI model that can diagnose someone with Alzheimer’s now or say whether they will have the disease in the next five years . It explains the decision using fundus images looking at the retinal vasculature and other anatomical and functional changes in the eye. “ The animal model shows that pathological changes appeared earlier in the retina than the brain, ” Ruogu explains. “ In humans, it’s also likely that changes caused by Alzheimer’s will appear earlier in the eye than can be observed in the brain. ”