Keywords:
Not applicable, Arteriosclerosis, Screening, Computer Applications-Detection, diagnosis, CT, Computer applications, Cardiovascular system, Artificial Intelligence, Artificial Intelligence and Machine Learning, Performed at one institution
Authors:
N. Martini1, C. Fabbri1, A. Ripoli2, S. Chiappino1, A. Aimo2, F. Avogliero1, M. Emdin2, D. Chiappino1, D. Della Latta1; 1Massa/IT, 2Pisa/IT
DOI:
10.26044/ecr2020/C-11638
Purpose
Since 2009 In a small town in Tuscany (Fig. 1), the Montignoso HEart and Lung Project (MHELP) offers to the general population a screening for the prevention of lung and cardiovascular diseases (CVD).
1826 subjects aged between 45 and 75 years, joined the program (52% of the target population), undergoes a series of clinical tests (Fig. 2) including a low-dose chest CT to assess lung and CVD risks.
To date only 86 deaths and 67 cardiac events have occurred in the enrolled population, These numbers are too low to be able to define predictive models based on traditional statistics algorithms.
A first analysis of the acquired variables was carried out by implementing Machine Learning based algorithms (random forest) with the aim of defining a predictive model capable of stratifying the cardiovascular risk of each subject.
The above implementation has allowed us to observe that among all the most relevant variables acquired in the CVD risk definition are the coronary calcium score and the pericardial fat volume. Both features are associated with arterosclerotic disease.
The atherosclerosis process is closely related to aging.
Moreover, aging causes structural changes in the heart such as an increase in the chambers volume, in the valves thickness as well as in the layer of pericardial fat and in the coronary
calcifications.
CT image matrix therefore hides all the information that would allow to estimate the subject's age however, while extracting this information, the radiologist cannot estimate biological age of the heart.
We investigated whether using deep learning it is possible extract features from CT scans that can estimate the patient’s biological age of the heart as a new biomarker of CVD risk.