Connecting properties of the micro- and macro-vasculature from multimodal imaging through genetics and deep learning to better understand vascular pathomechanisms and predict disease risk
Approved Research ID: 90947
Approval date: October 27th 2022
Each year cardiovascular diseases (CVD) cause 3.9 million deaths in Europe, amounting to 45% of all deaths. While certain risk factors like age, smoking and hypertension have been well documented, the impact of blood vessel characteristics is poorly understood. Vascular properties - such as shape and size of the blood vessels - can be investigated with non-invasive and inexpensive imaging methods in some organs, such as the retina. For example, the bendiness (also known as "tortuosity") of retinal vessels has been shown to be associated with increased risk of CVD. However, the extent to which vascular properties extracted from retinal images reflect those of other body parts has not been studied systematically.
Our project aims at providing an extensive analysis of vascular properties of the retina and link them to those from other organs, such as the brain or the heart. We will use genetic information about such properties to investigate the biological mechanisms that link vascular properties to CVD. Using machine learning, we will test whether we can predict CVD risk in the general population, based on vascular properties derived from multi-organ imaging and genetic information.
Our findings have the potential to increase our understanding of the pathological mechanisms leading to CVD and provide tools for the detection of presymptomatic vascular modifications that can support early diagnosis of vascular diseases. We propose to complete this project in 36 months.