Deep convolutional neural networks to predict cardiovascular risk from computed tomography

Roman Zeleznik; Borek Foldyna; Parastou Eslami; Jakob Weiss; Ivanov Alexander; Jana Taron; Chintan Parmar; Raza M Alvi; Dahlia Banerji; Mio Uno; Yasuka Kikuchi; Julia Karady; Lili Zhang; Jan-Erik Scholtz; Thomas Mayrhofer; Asya Lyass; Taylor F Mahoney; Joseph M Massaro; Ramachandran S Vasan; Pamela S Douglas; Udo Hoffmann; Michael T Lu; Hugo J W L Aerts

doi:10.1038/s41467-021-20966-2

Deep convolutional neural networks to predict cardiovascular risk from computed tomography

Nat Commun. 2021 Jan 29;12(1):715. doi: 10.1038/s41467-021-20966-2.

Authors

Roman Zeleznik^{1

2

3}, Borek Foldyna^{1

2}, Parastou Eslami², Jakob Weiss^{1

2

3

4}, Ivanov Alexander², Jana Taron^{2

4}, Chintan Parmar^{1

3}, Raza M Alvi², Dahlia Banerji², Mio Uno², Yasuka Kikuchi^{2

5}, Julia Karady^{2

6}, Lili Zhang², Jan-Erik Scholtz², Thomas Mayrhofer^{2

7}, Asya Lyass⁸, Taylor F Mahoney⁹, Joseph M Massaro⁹, Ramachandran S Vasan^{10

11}, Pamela S Douglas¹², Udo Hoffmann^{1

2}, Michael T Lu^{1

2}, Hugo J W L Aerts^{13

14

15

16

17}

Affiliations

¹ Artificial Intelligence in Medicine (AIM) Program, Mass General Brigham, Harvard Medical School, Boston, MA, USA.
² Cardiovascular Imaging Research Center, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA.
³ Department of Radiation Oncology, Brigham and Women's Hospital, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA.
⁴ Department of Diagnostic and Interventional Radiology, Eberhard Karls University of Tübingen, Tübingen, Germany.
⁵ Center for Cause of Death Investigation, Faculty of Medicine, Hokkaido University, Sapporo, Hokkaido, Japan.
⁶ Cardiovascular Imaging Research Group, Heart and Vascular Center, Semmelweis University, Budapest, Hungary.
⁷ School of Business Studies, Stralsund University of Applied Sciences, Stralsund, Germany.
⁸ Department of Mathematics and Statistics, Boston University, Boston, MA, USA.
⁹ Department of Biostatistics, Boston University School of Public Health, Boston, MA, USA.
¹⁰ National Heart, Lung, and Blood Institute and Boston University, Framingham Heart Study, Framingham, MA, USA.
¹¹ Departments of Cardiology and Preventive Medicine, Department of Medicine, Boston University School of Medicine, Boston, MA, USA.
¹² Department of Medicine, Division of Cardiology, Duke University School of Medicine, Duke Clinical Research Institute, Durham, NC, USA.
¹³ Artificial Intelligence in Medicine (AIM) Program, Mass General Brigham, Harvard Medical School, Boston, MA, USA. haerts@bwh.harvard.edu.
¹⁴ Cardiovascular Imaging Research Center, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA. haerts@bwh.harvard.edu.
¹⁵ Department of Radiation Oncology, Brigham and Women's Hospital, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA. haerts@bwh.harvard.edu.
¹⁶ Department of Radiology, Brigham and Women's Hospital, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA. haerts@bwh.harvard.edu.
¹⁷ Radiology and Nuclear Medicine, CARIM & GROW, Maastricht University, Maastricht, The Netherlands. haerts@bwh.harvard.edu.

Abstract

Coronary artery calcium is an accurate predictor of cardiovascular events. While it is visible on all computed tomography (CT) scans of the chest, this information is not routinely quantified as it requires expertise, time, and specialized equipment. Here, we show a robust and time-efficient deep learning system to automatically quantify coronary calcium on routine cardiac-gated and non-gated CT. As we evaluate in 20,084 individuals from distinct asymptomatic (Framingham Heart Study, NLST) and stable and acute chest pain (PROMISE, ROMICAT-II) cohorts, the automated score is a strong predictor of cardiovascular events, independent of risk factors (multivariable-adjusted hazard ratios up to 4.3), shows high correlation with manual quantification, and robust test-retest reliability. Our results demonstrate the clinical value of a deep learning system for the automated prediction of cardiovascular events. Implementation into clinical practice would address the unmet need of automating proven imaging biomarkers to guide management and improve population health.

Publication types

Multicenter Study
Research Support, N.I.H., Extramural
Research Support, Non-U.S. Gov't
Research Support, U.S. Gov't, Non-P.H.S.

MeSH terms

Aged
Asymptomatic Diseases
Calcium / analysis
Cardiovascular Diseases / complications
Cardiovascular Diseases / diagnosis
Cardiovascular Diseases / epidemiology*
Cardiovascular Diseases / pathology
Chest Pain / diagnosis*
Chest Pain / etiology
Coronary Vessels / diagnostic imaging*
Coronary Vessels / pathology
Deep Learning*
Female
Follow-Up Studies
Heart Disease Risk Factors
Humans
Image Processing, Computer-Assisted / methods*
Male
Middle Aged
Reproducibility of Results
Retrospective Studies
Risk Assessment / methods
Tomography, X-Ray Computed

Substances

Calcium

Abstract

Publication types

MeSH terms

Substances

Grants and funding