New AI tool accurately diagnoses eye conditions, aids Parkinson’s detection.
New AI tool accurately diagnoses eye conditions, aids Parkinson's detection.
AI Technology Revolutionizes Vision and Systemic Health Diagnosis
Imagine a future where artificial intelligence (AI) can swiftly detect and treat eye, heart, and neurological disorders. Well, that future is closer than ever, thanks to an innovative AI program developed by researchers in the United Kingdom. This groundbreaking technology, called RETFound, utilizes retinal images to accurately diagnose and treat various health conditions.
A Leap Forward in Healthcare
RETFound is a pioneering AI foundation model specifically designed for ophthalmology. It has been trained on millions of eye scans, making it one of healthcare’s first AI models to assist in detecting and treating blindness. In multiple tests, RETFound has demonstrated superior performance compared to existing AI systems and clinical experts in various complex diagnostic functions. Notably, this transformative technology goes beyond conventional scans and current AI systems by accounting for diverse populations and rare diseases often overlooked.
Moreover, RETFound significantly reduces the workload of human experts involved in analyzing and labeling retinal imaging. This breakthrough brings hope for improved diagnostic efficiency, especially in the realm of systemic health issues such as stroke, heart attacks, and Parkinson’s disease.
The Power of AI in Diagnosing Eye Disorders
The researchers at Moorfields Eye Hospital and the UCL Institute of Ophthalmology in England are at the forefront of utilizing AI to revolutionize eye examinations. Their recent study on RETFound, a world-first foundation model, utilized millions of eye scans from the UK’s National Health Service (NHS). This open-source initiative could pave the way for more effective AI-assisted detection and treatment of blindness.
As we observe World Retina Day on September 27, World Sight Day in October, and Diabetic Eye Disease Awareness Month in November, these remarkable advancements in AI technology arrive at a crucial time.
- Increased risk of heart attack and stroke in women with migraine he...
- Increase in Cases of Babies Born With Syphilis in Mississippi
- Suppressing negative thoughts can be healthy, according to a study.
According to Professor Pearse Keane, senior author at the UCL Institute of Ophthalmology: “This is another big step towards using AI to reinvent the eye examination for the 21st century, both in the UK and globally. We show several exemplar conditions where RETFound can be used, but it has the potential to be developed further for hundreds of other sight-threatening eye diseases that we haven’t yet explored.”
Training RETFound: A Transformative Approach
AI foundation models like RETFound utilize vast amounts of data, making them transformative technologies in the healthcare landscape. The launch of ChatGPT in November 2022 demonstrated the potential of AI models in developing adaptable language tools. RETFound follows a similar approach in its training process, using millions of retinal images to construct a versatile model with unlimited potential applications.
Typically, AI models heavily depend on human expertise and effort for training, which can be a highly demanding and time-consuming process. However, RETFound stands out by matching the performance of other AI programs while using only 10% of human labels in its dataset. This accomplishment is attributed to RETFound’s self-supervised learning (SSL) approach, where the model learns to predict missing parts of an image by masking and analyzing the available data. By eliminating the need for extensive data labeling, this technology significantly reduces the burden on healthcare professionals and offers more efficient and cost-effective solutions.
Dr. Steve Frank, a technology developer, emphasized the value of self-supervised learning in healthcare AI, stating: “This approach is of particular value for healthcare AI because the cost of labeling is so high — doctors are already busy saving lives, and their time is quite precious.”
The Promise of Oculomics and Disease Detection
The retina is often referred to as “a window to the body,” and the field of oculomics explores the correlations between retinal image characteristics and various diseases using deep learning techniques. The authors of the current study firmly believe that RETFound holds immense potential in improving the diagnosis of sight-threatening eye diseases, such as diabetic retinopathy and glaucoma. Additionally, RETFound has the ability to predict systemic disorders like heart failure, stroke, and Parkinson’s disease.
Dr. Brigham Hyde, co-founder of Atropos Health, believes that AI and deep learning techniques can play a crucial role in disease detection. He highlights that AI-aided imaging techniques can reveal diseases that a human eye might miss, while combining digital, medical, and experiential data can uncover digital biomarkers for early disease diagnosis. Risk scoring algorithms can also assist care teams in identifying patients at high risk promptly.
Demonstrating Superior Performance and Efficiency
The researchers evaluated RETFound, an SSL-based foundation model for retinal images, and compared its performance to other existing models. In their assessment of ocular diseases, disease prognosis, and systemic diseases, RETFound consistently exhibited superior performance and label efficiency. Although the accuracy achieved by RETFound may not yet be sufficient for clinical use, it outperformed most conventional systems.
Dr. Frank applauded RETFound’s results, stating: “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.”
Embracing Diversity in Healthcare AI
The UCL-Moorfields experts found that RETFound demonstrated equal effectiveness in detecting diseases across diverse ethnic groups. By training RETFound with datasets representing the ethnic diversity of London, the researchers have established a valuable foundation for global researchers to develop their own healthcare systems for ocular disease diagnosis and systemic disease prediction. The importance of patient diversity in model development cannot be overstated, as it ensures accurate and comprehensive diagnoses for a broad range of populations.
Dr. Tyler Wagner, Vice President of Biomedical Research at Anumana, commended the study, stating: “While RETFound performs better than the other models compared in the manuscript during external evaluation on a set of patients with different demographics, the authors note the decrease in performance, highlighting the importance of the patient diversity in model development.”
To encourage further research and advancements in the field of oculomics, the study authors have made RETFound publicly available, allowing other researchers to build upon their efforts and facilitate diverse ocular and oculomic research.
In conclusion, the development of RETFound is a significant milestone in the integration of AI technology into healthcare. By utilizing retinal images, this innovative system offers a revolutionized approach to diagnosing eye, heart, and neurological disorders. As researchers continue to explore the potentials of oculomics and AI, the future of healthcare looks brighter, faster, and more accurate than ever before.
