Dundee, Scotland – Adding new meaning to the scriptural adage that “the light of the body is the eye”, a research team from the University of Dundee’s School of Computing will be using their high definition image analysis application called “Vampire” to see if various maladies can be early detected from images of the human eye. The project has received government funding to the tune of £1.1 million over the next three years. It is part of a much broader £8 million funding project across 11 universities in the United Kingdom by the Engineering and Physical Science Research Council (EPSRC).
Among the other maladies that researchers hope to detect are cardiovascular diseases, strokes, and dementia. Specifically, Vampire will analyze the ocular images of numerous patients to determine if there is a link between ocular vein patterns and the aforementioned maladies. The promise is that an affordable machine can be developed to perform a simple non-invasive eye exam and discover the risk level a person has for certain medical conditions including dementia. If successful, Vampire might even be able to discriminate between forms of dementia. Such careful analysis would be painstaking to perform by the aid of the human eye alone. Vampire is able to scour thousands of images and automate the analytical process. The city’s Ninewells Hospital is providing a trove of ocular images from prior patients to feed into Vampire.
Project coordinator Emanuele Trucco, who is also the computation professor at the university, explained that the retinal veins sometimes reflect internal changes in the human body. How thick, thin, or wiggly the retinal veins become may be well reflections of ongoing problems that can be used to diagnose certain health conditions. ESPRC Chief Executive Professor Philip Nelson explained that with the aging UK population, certain health conditions pose a major challenge for the government. This is where a program such as Vampire might help improve quality of life for individuals and benefit the government as well.