Researchers in the UK have developed a ground-breaking artificial intelligence (AI) tool that uses retinal images to hasten the detection of systemic illnesses such as Parkinson's disease, strokes, and heart attacks. Additionally, this AI technology facilitates a non-invasive view of the nervous system.

The researchers conducted a study using millions of eye scans from the UK's National Health Service (NHS) using RETFound, their ground-breaking foundation model. Their open-source endeavor could be used as a model for initiatives to use AI to diagnose and treat blindness. With retinal pictures, RETFound used a strategy akin to CHatGPT, training on millions of scans. This made it possible to create a versatile model with nearly limitless applications.

The UK researchers claim that RETFound can equal the performance of existing AI methods with just 10% of the dataset's human labeling. With its self-supervising method of masking portions of an image and training itself to predict the missing portions, RETFound was able to attain this increased efficiency. Additionally, when compared to models pre-trained on SL-ImageNet, SSL-ImageNet, and SSL-Retinal, RETFound showed "consistently superior performance and label efficiency."

Dr. Steve Frank remarked that “the RETFound results are especially impressive for the sheer number of tasks their system can perform. The accuracies the researchers achieve aren’t sufficient for clinical use, but the more conventional systems they test against are mostly worse.”

김담은 강남포스트 학생기자  webmaster@ignnews.kr

<저작권자 © 강남포스트, 무단 전재 및 재배포 금지>