Eye provides clues to insidious vascular disease
Researchers from the University and University Hospital of Bonn have developed a method that could be used to diagnose atherosclerosis. Using self-learning software, they were able to identify vascular changes in patients with peripheral arterial disease (PAD), often at an early stage. Although these early stages do not yet cause symptoms, they are nevertheless already associated with increased mortality. The algorithm used photos of an organ not normally associated with PAD: the eye. The results have just been published in the journal Scientific reports.
Poets regard the eyes as a window to the soul. But more prosaically, we could also call them windows on our ships. The fundus is very well irrigated. It’s necessary for the more than 100 million photoreceptors in the retina and the nerve cells connected to them to do their job. At the same time, arteries and veins can be observed and photographed through the pupil without much effort.
It may be possible to detect the first signs of atherosclerosis (hardening of the arteries) with such an examination in the future. In this case, chronic remodeling processes lead to narrowing of the vessels and hardening of the affected arteries. It is the leading cause of heart attacks and strokes, the most common causes of death in Western industrialized countries, as well as peripheral arterial disease (PAD).
More than four million people in this country suffer from PAD. “As it usually causes no symptoms in the first few years, the diagnosis is often only made when secondary damage has already occurred,” explains Dr. Nadjib Schahab, head of the angiology section and one of the authors of the study. ‘study. “The consequences can be dramatic. In the long term, progressive circulatory problems in the legs and arms can even lead to amputation. In addition, the risk of a fatal heart attack or stroke is greatly increased – even in the early stages of the disease.
Early diagnosis is therefore very important in order to be able to treat those affected in time. The interdisciplinary project of the Department of Informatics at the University of Bonn and the Department of Ophthalmology and Heart Center at the University Hospital Bonn begins exactly there. “We photographed 97 eyes of women and men who suffered from PAD,” says Dr. Maximilian Wintergerst of the University Eye Hospital Bonn. “In more than half of them, the disease was still at a stage where it caused no symptoms.” Additionally, the team took background images of 34 eyes of healthy control subjects.
Neural network detects early vascular changes
They then used the images to feed a convolutional neural network (CNN). It is a software that is inspired by the human brain in its operation. If such a CNN is formed with photos whose content is known to the computer, it can later recognize the content of unknown photos. However, for this to work with sufficient certainty, it normally takes several tens of thousands of training photos – far more than was available in the study.
“We therefore first carried out a pre-training with another disease that attacks the eye vessels,” explains Professor Dr. Thomas Schultz of the International Center for Information Technology Bonn-Aachen (b-it ) and the Institute of Computer Science. II at the University of Bonn. To do this, the researchers used a dataset of more than 80,000 additional photos. “In a sense, the algorithm learns from them what to pay particular attention to,” explains Schultz, who is also a member of the transdisciplinary research areas “Modeling” and “Life and Health” at the University of Bonn. “So we’re also talking about transfer learning.”
The CNN trained in this way was able to diagnose with remarkable accuracy whether the pictures of the eyes were from a PAD patient or a healthy person. “A good 80% of all affected individuals were correctly identified, if we considered 20% false positives, i.e. healthy individuals that the algorithm incorrectly classified as sick,” Schultz explains. “It’s amazing because even for trained ophthalmologists, PAD cannot be detected from fundus images.”
In further analyses, the researchers were able to show that the neural network pays particular attention to the large vessels at the back of the eye when evaluating it. However, to obtain the best possible result, the method required digital images with a sufficiently high resolution. “A lot of CNNs work with very low-resolution photos,” Schultz says. “This is sufficient to detect major changes. For our PAD classification, on the other hand, we need a resolution at which details of vascular structures remain discernible.”
The researchers hope to further improve the performance of their method in the future. To do this, they plan to cooperate with ophthalmology and vascular medicine centers around the world who will provide them with additional fundus images of those affected. The long-term goal is to develop a simple, rapid and reliable diagnostic method that does not require concomitant procedures such as the administration of eye drops.
The B-IT and the Institute for Informatics II of the University of Bonn, the University Eye Hospital Bonn and the Clinic for Cardiology and Angiology of the University Hospital Bonn participated in the study.
Publication: Mueller, S., Wintergerst, MWM, Falahat, P. et al. Multi-instance learning detects peripheral arterial disease from high-resolution color fundus photography. Sci Rep 12, 1389 (2022).
Teacher. Dr. Thomas Schultz
Bonn-Aachen International Center for Information Technology (B-IT) and Institute for Computer Science II at the University of Bonn
Email: [email protected]
Dr. Maximilian WM Wintergerst
Bonn University Eye Hospital
Email: [email protected]
Priv.-Doz. Dr. Nadjib Shahab
Head of Angiology Section
Bonn University Hospital Cardiac Center
Email: [email protected]
The title of the article
Multi-instance learning detects peripheral arterial disease from high-resolution color fundus photography.
Publication date of articles
Warning: AAAS and EurekAlert! are not responsible for the accuracy of press releases posted on EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.