AI Retinal Image Analysis
A window to effective detection of cardiovascular diseases.
In 2024, Our Health Journeys partnered with Saint Kentigern College in Auckland and challenged a number of students to conduct research into an aspect of the medical history of Aotearoa New Zealand. The students, ranging from Years 8-13, produced their research in written, oral, or video format and the top projects were chosen for publication to Our Health Journeys. A new project was published weekly following the completion of the project, until late October 2024.
AI Retinal Image Analysis
Cardiovascular diseases (CVDs) are the leading cause of death all across the globe. In New Zealand, every 90 minutes, one person dies from CVDs, which is 5840 deaths every year. Hence, in 2019, Toku Eyes was founded by University of Auckland scientist Associate Professor Ehsan Vaghefi using AI retinal image analysis as the solution.
By analysing retinal images with their AI, they can examine and report on overall cardiovascular health through a wide range of key markers in a patient’s eyes, like Diabetic Retinopathy (the damage of blood vessels in the eye because of poorly controlled blood sugar levels), maculopathy (a disease affecting the back of the retina that warps your vision), age-related macular degeneration (degeneration of your central vision), glaucoma (damage to the optical nerve). Because the eyes are the only external body part with visible blood vessels, Toku Eyes can analyse these blood vessels. Hence, regular eye exams will shield against CVDs.
Currently, Toku Eyes has two products available on the market: BioAge and CLAiR. By imaging a patient’s eyes and identifying the eye's condition, BioAge estimates a patient’s “biological” age and the rate of cell deterioration. Therefore, they can more accurately assess your health. Though most people think of age as a linear factor; however, cells do not age in this way. Hence, BioAge allows for a better understanding of a patient's health. CLAiR, the second product, uses age, gender and a retinal image to determine the likelihood of a cardiovascular event in the next ten years. Unlike the current test, CLAiR is faster, cheaper and more accurate. CLAiR gives the patient risk in percentages; 0-5% to 5-7.4% are considered no elevated risks, and 7.5-20+ is considered as an elevated risk, and further action is needed to lower this risk. They are also developing one more product, My Kidney AI, and THEIA, which is available only in New Zealand. All these products utilise AI screening on retinal images to report on one's health and likelihood of having a cardiovascular event.
All the AI products were trained using routinely collected medical data, including entire medical history and a retinal image. THIEA and their other AIs look for markers such as hemorrhage, fluid, thickening of retinal arterioles from chronic hypertension and even damage to the optical nerve. In conjunction with a medical history, Toku Eyes can provide a comprehensive report on one’s cardiovascular health. THEIA demonstrates promising results, not missing more than mild to slightly threatening causes. AI-based algorithms like THEIA and their other AIs significantly increase the efficiency and availability of these exams.
The innovative and revolutionary solution by Toku Eyes is leading the way to the future. Current practices are inconvenient and costly, some taking weeks to get results. Providing instantaneous reports with high accuracy could significantly improve the detection of Diabetic Retinopathy and other CVDs and is vital to preventing blindness, which will have massive impacts on healthcare worldwide. Fast and accurate detection will reduce the cost and price of such exams, allowing for more readily available and affordable non-intrusive examinations. This is also a powerful tool for healthcare workers as it speeds up the process of evaluation. Toku Eyes’ solutions allow for faster, cheaper and more readily available examinations, leading to better health for everyone. The integration of AI is vital to our ever-changing world.