A collaborative team of Cedars-Sinai investigators proposes that artificial intelligence can effectively evaluate the volume and density of heart fat, both of which are associated with cardiovascular risk.
New research from Cedars-Sinai shows that people with a larger volume of heart fat, and those with more dense heart fat, are at higher risk for cardiovascular disease. Image Credit: Image by Getty
A collaborative team of researchers employed low-dose computed tomography (CT) scanning as part of a standard test to swiftly and precisely measure the amount of fat surrounding the heart. The method, which was discovered by Cedars-Sinai’s Biomedical Imaging Research Institute and the Division of Artificial Intelligence in Medicine, can assist medical professionals in better understanding and treating patients' risks for heart disease.
The researchers claim that their AI measurement tool provides novel assistance in the prediction of heart attacks and cardiovascular disease. Their findings were published in the peer-reviewed journal npj Digital Medicine.
Our research shows that people with a larger volume of heart fat - a key indicator of metabolic health - and those with more dense heart fat - a marker of inflammation - are at higher risk for cardiovascular disease. This novel technology uses AI to more quickly and accurately assess the volume and density of heart fat, offering valuable insight beyond traditional methods.
Piotr J. Slomka PhD, Director and Professor, Innovation in Imaging, Division of Artificial Intelligence and Cardiology in Medicine, Smidt Heart Institute, Cedars-Sinai
By providing physicians and other clinicians with this specialized software, Slomka and colleagues hope to expand the application of these findings in clinical settings.
Slomka says, “These results underscore the efficiency and clinical importance of AI in heart disease risk assessment, offering a fast and reliable tool for predicting cardiovascular risk.”
8,781 patients from four clinical sites were included in the study by the investigators. At the time of the study, none of the patients had been diagnosed with coronary artery disease, a form of heart disease, and they had cardiac imaging.
Key results include:
- Even after accounting for additional risk factors such as age, medical history, prior heart imaging results, and coronary artery calcium scores, the correlation with cardiovascular risk remained.
- Patients with both elevated heart fat volumes and density had an almost threefold increased risk of heart attacks and cardiovascular-related death.
- Artificial intelligence (AI) measures heart fat in less than two seconds, while manual measurement usually takes fifteen minutes.
- Heart-related problems and death were more common in those with higher volumes and densities of heart fat over the course of the 2.7-year follow-up period.
The study’s findings support earlier findings while also providing fresh perspectives thanks to AI.
These findings validate the use of AI for quick and accurate heart fat measurement, highlighting a potential shift toward more AI-assisted diagnostic methods in cardiology.
Sumeet Chugh MD, Director and Associate Director, Division of Artificial Intelligence in Medicine, Smidt Heart Institute, Cedars-Sinai
Chugh, who was not involved in the study, stated that the findings support future investigations into the role of heart fat as a major risk factor for heart disease by demonstrating a direct correlation between heart fat and cardiovascular risk.
This approach serves as a practical, real-world test of the application’s effectiveness when implemented in clinical settings.
Sumeet Chugh MD, Director and Associate Director, Division of Artificial Intelligence in Medicine, Smidt Heart Institute, Cedars-Sinai
As a next step, investigators plan to further test their algorithm in a more diverse patient population to ensure the accuracy and generalizability of the findings.
Other authors involved in the study include Robert J. H. Miller, Aakash Shanbhag, Aditya Killekar, Mark Lemley, Bryan Bednarski, Serge D. Van Kriekinge, Paul B. Kavanagh, Joanna X. Liang, Valerie Builoff, Attila Feher, Edward J. Miller, Andrew J. Einstein, Terrence D. Ruddy, Daniel S. Berman, and Damini Dey.
This research was supported in part by grants R01HL089765 and R35HL161195 from the National Heart, Lung, and Blood Institute at the National Institutes of Health.
Journal Reference
Miller, H. J. R., et.al. (2024) AI-derived epicardial fat measurements improve cardiovascular risk prediction from myocardial perfusion imaging. npj Digital Medicine volume. doi.org/10.1038/s41746-024-01020-z.
Source: https://www.cedars-sinai.org/home.html