Imaging of the retina’s vascular network enabled by AI can predict cardiovascular disease and death.

Without the need for blood tests or blood pressure monitoring, AI-enabled imaging of the retina’s network of veins and arteries can reliably predict cardiovascular disease and death, according to research published online in the British Journal of Ophthalmology.

As a result, it paves the way for a highly effective, non-invasive screening test for those at moderate to high risk of cardiovascular illness that does not require a clinic visit, according to the researchers.

Circulatory illnesses, such as cardiovascular disease, coronary heart disease, heart failure, and stroke, account for one in four deaths in the United Kingdom alone.

Researchers note that despite the existence of numerous risk frameworks, these are not always able to effectively predict who will develop or die from circulatory disorders. According to previously published research, the breadth of the retina’s microscopic veins and arteries (arterioles and venules) may serve as an accurate early diagnostic of circulatory disease. The retina is the area of the eye that receives and organises visual information. However, it is unclear whether these findings apply equally and consistently to men and women.

Therefore, the researchers developed a fully automated artificial intelligence (AI)-enabled algorithm (QUantitative Analysis of Retinal vessels Topology and siZe, or QUARTZ for short) to assess the potential of retinal vasculature imaging in conjunction with known risk factors for predicting vascular health and death.

They applied QUARTZ to retinal pictures of 88,052 UK Biobank individuals aged 40-69, analysing the breadth, vascular area, and degree of curvature (tortuosity) of the arterioles and venules in the retina in order to construct prediction models for stroke, heart attack, and mortality from circulatory disease.

They then applied these models to the retinal scans of 7,411 European Prospective Investigation of Cancer (EPIC)-Norfolk patients aged 48 to 92. Together and separately, the performance of QUARTZ was compared to the widely utilised Framingham Risk Scores framework.

All participants’ health was monitored for an average of 7 to 9 years, during which time 327 circulatory disease deaths occurred among 64,144 UK Biobank participants (average age 56) and 201 circulatory disease deaths occurred among 5862 EPIC-Norfolk individuals (average age 67).

Arteriolar and venular width, tortuosity, and width variation emerged as significant indicators of mortality due to cardiovascular disease in men. Variation in arteriolar and venular area and breadth, as well as venular tortuosity and width, contributed to risk prediction in females.

The prognostic influence of retinal vasculature on circulatory disease mortality interacted with smoking, hypertension medications, and a prior heart attack. Overall, these predictive models, which were based on age, smoking, medical history, and retinal vasculature, captured between fifty percent and two-thirds of circulatory disease-related deaths among the highest-risk individuals.

Retinal vasculature models caught around 5% more instances of stroke in UK Biobank men and 8% more cases in UK Biobank women, as well as 3% more cases among the most at-risk EPIC-Norfolk men and roughly 2% less cases among EPIC-Norfolk women. Framingham Risk Scores identified higher heart attack cases among those at the highest risk.

The addition of retinal vasculature to the Framingham Risk Scores had minimal effects on the prediction of stroke and heart attack. Researchers report that a simpler, noninvasive risk score based on age, sex, smoking, medical history, and retinal vasculature functioned as well as the Framingham Risk Scores, without the requirement for blood tests or blood pressure monitoring.

They admit that both groups of research participants enjoy healthier lifestyles than other middle-aged adults in the same geographic region, and that the majority of them are white. They explain that this is the largest population-based research of retinal vasculature and that the prediction models were externally evaluated in a large number of individuals.

Retinal imaging is already commonplace in the United Kingdom and the United States, according to the researchers, who conclude that “AI-enabled vasculometry risk prediction is fully automated, low cost, non-invasive, and has the potential to reach a greater proportion of the community’s population due to ‘high street’ availability and the absence of blood sampling or blood pressure measurement.”

Leave a Reply

error: Content is protected !!
Open chat
WhatsApp Now