Eye scans provide early warning of chronic disease
A quick and efficient eye scan could soon provide early warning of serious chronic diseases such as diabetes, hypertension, dementia and stroke thanks to EU-funded research combining the latest advances in computer vision and mathematical modelling.
© YakobchukOlena - fotolia.com
The work, conducted over four years in the EU-funded REVAMMAD project, has advanced understanding of how different diseases trigger early-stage changes to the vasculature, the network of veins and arteries throughout the human body.
By photographing the retina at the back of the eye with a specialised camera and using advanced image analysis software it is possible to detect tiny changes to the blood vessels, enabling early diagnosis and improving treatment for a variety of chronic diseases.
Georgios Leontidis, an early-stage researcher on REVAMMAD at the University of Lincoln in the UK, says complications arising from diabetes are among several debilitating and chronic conditions that could be detected through retinal scans sooner than with current techniques.
Specifically, retinal imaging can be used to detect diabetic retinopathy, a common complication of diabetes that occurs when high blood sugar levels damage the cells in the retina, eventually leading to impaired vision and blindness.
Here at the University of Lincoln, our efforts have focused on analysing images of diabetic patients before the first stage of diabetic retinopathy, Leontidis says. We wanted to see what changes diabetes caused to the retinal vessels and how these changes progressed to retinopathy.
Accurately measuring blood vessels for signs of change
Because changes to the bodys vasculature system are often subtle in the early stages of diabetes and other diseases, REVAMMAD researchers developed sophisticated computer algorithms to locate the blood vessels in a retinal image and recognise the structure of the vascular network.
The software is capable of analysing retinal images and taking accurate measurements to detect narrowing or bulging of the blood vessels or changes in shape. For example blood vessels could become more convoluted than normal. In this way routine retinal scans of the same patient can be used to build up a retinal model that provides early warning of blood vessel changes that could be indicative of disease.
By applying clinical knowledge of what types of blood vessel changes are caused by which diseases, it is possible to generate an accurate and early diagnosis.
We correlated the information from these images with data on functional changes, such as abnormal blood pressure, blood flow volume and blood flow velocity, as well as associating them with some risk factors like age, type of diabetes, duration of diabetes, gender and smoking, Leontidis says.
The researchers analysed studies of 24 patients over three years to determine statistically significant changes in some blood flow parameters associated with the development of diabetic retinopathy, as well as developing probabilistic models relating the location of retinal lesions to clinical risk, using a database of 900 images from 60 patients recorded over 10 years.
This in turn has enabled the REVAMMAD team to develop risk models for populations of patients screened annually for diabetic retinopathy, the leading cause of blindness in the developed world with 4 200 people diagnosed as being at risk every year in England alone.
The vasculature plays a key role in chronic medical conditions that account for an increasing proportion of EU healthcare costs, including Alzheimers, diabetes, stroke and coronary heart disease, says project coordinator and lead investigator Andrew Hunter, head of the College of Science at the University of Lincoln. These issues have ignited considerable interest in computerised analysis of vascular images to support scientific enquiry, diagnosis, prognosis and screening.
The results of the research are expected to feed into the future development of computerised screening systems for various vascular diseases, ultimately allowing routine early diagnosis, improved treatment and better outcomes for patients, with some of the project partners actively exploring options to commercially exploit the software.
Funded through the EUs Marie Curie Networks for Initial Training initiative, REVAMMAD has helped train early-stage researchers in the field and related technologies, many of whom have gone on to establish successful academic careers.