Deep-Learning for Epicardial Adipose Tissue Assessment with Computed Tomography: Implications for Cardiovascular Risk Prediction

ORFAN Investigators, Henry W. West, Muhammad Siddique, Michelle C Williams, Lucrezia Volpe, Ria Desai, Maria Lyasheva, Sheena Thomas, Katerina Dangas, Christos P. Kotanidis, Pete Tomlins, Ciara Mahon, Attila Kardos, David M Adlam, John Graby, Jonathan C.L. Rodrigues, Cheerag Shirodaria, John E Deanfield, Nehal N. Mehta, Stefan NeubauerKeith M. Channon, Milind Y. Desai, Edward D. Nicol, David E Newby, Charalambos Antoniades*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract / Description of output

BACKGROUND: Epicardial adipose tissue volume (EAT) is a marker of visceral obesity that can be measured in coronary CT angiograms (CCTA). The clinical value of integrating this measurement in routine CCTA interpretation has not been documented.
OBJECTIVE: To develop a deep-learning network (DLN) for automated quantification of EAT from CCTA, test it in technically challenging patients, and validate its prognostic value in routine clinical care.
METHODS: The DLN was trained and validated to auto-segment EAT in 3720 CCTA scans from the Oxford Risk Factors And Non-invasive imaging cohort. The model was tested in patients with challenging anatomy and scan artefacts and applied to a longitudinal cohort of 253 post-cardiac surgery patients and 1558 patients from the SCOT-HEART trial, to investigate its prognostic value.
RESULTS: External validation of the DLN yielded a concordance correlation coefficient of 0.970 for machine vs human. EAT was associated with coronary artery disease (OR[95%CI] per SD increase in EAT 1.13[1.04-1.30] p=0.01), and atrial fibrillation (AF) (1.25[1.08-1.40], p=0.03), after correction for risk factors (including BMI). EAT predicted all-cause mortality (HR[95%CI] per SD = 1.28[1.10-1.37], p=0.02), myocardial infarction (1.26[1.09-1.38] p=0.001) and stroke (1.20[1.09-1.38] p=0.02) independently of risk factors in SCOT-HEART (5y follow-up). It also predicted in-hospital (HR[95%CI] = 2.67[1.26-3.73], p=< 0.01) and long-term post-cardiac surgery AF (7y follow-up; HR[95%CI] = 2.14[1.19-2.97], p=< 0.01).
5
CONCLUSIONS: Automated assessment of EAT is possible in CCTA, including in challenging patients; it forms a powerful marker of metabolically unhealthy visceral obesity, which could be used for cardiovascular risk stratification.
Original languageEnglish
JournalJACC: Cardiovascular Imaging
Early online date8 Feb 2023
DOIs
Publication statusE-pub ahead of print - 8 Feb 2023

Keywords / Materials (for Non-textual outputs)

  • Computed tomography
  • deep-learning
  • adipose tissue
  • visceral fat
  • atherosclerosis
  • atrial fibrillation

Fingerprint

Dive into the research topics of 'Deep-Learning for Epicardial Adipose Tissue Assessment with Computed Tomography: Implications for Cardiovascular Risk Prediction'. Together they form a unique fingerprint.

Cite this