Unsupervised learning to characterize patients with known coronary artery disease undergoing myocardial perfusion imaging

Michelle C. Williams, Bryan P. Bednarski, Konrad Pieszko, Robert J. H. Miller, Jacek Kwiecinski, Aakash Shanbhag, Joanna X. Liang, Cathleen Huang, Tali Sharir, Sharmila Dorbala, Marcelo F. Di carli, Andrew J. Einstein, Albert J. Sinusas, Edward J. Miller, Timothy M. Bateman, Mathews B. Fish, Terrence D. Ruddy, Wanda Acampa, M. Timothy Hauser, Philipp A. KaufmannDamini Dey, Daniel S. Berman, Piotr J. Slomka

Research output: Contribution to journalArticlepeer-review

Abstract / Description of output

Patients with known coronary artery disease (CAD) comprise a heterogenous population with varied clinical and imaging characteristics. Unsupervised machine learning can identify new risk phenotypes in an unbiased fashion. We use cluster analysis to risk-stratify patients with known CAD undergoing single-photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI).

From 37,298 patients in the REFINE SPECT registry, we identified 9,221 patients with known coronary artery disease. Unsupervised machine learning was performed using clinical (23), acquisition (17), and image analysis (24) parameters from 4,774 patients (internal cohort) and validated with 4,447 patients (external cohort). Risk stratification for all-cause mortality was compared to stress total perfusion deficit (
Three clusters were identified, with patients in Cluster 3 having a higher body mass index, more diabetes mellitus and hypertension, and less likely to be male, have dyslipidemia, or undergo exercise stress imaging (p
Our unsupervised cluster analysis in patients with known CAD undergoing SPECT MPI identified three distinct phenotypic clusters and predicted all-cause mortality better than ischemia alone.
Original languageEnglish
JournalEuropean Journal of Nuclear Medicine and Molecular Imaging
Early online date17 Apr 2023
Publication statusE-pub ahead of print - 17 Apr 2023

Keywords / Materials (for Non-textual outputs)

  • Machine learning
  • SPECT myocardial perfusion
  • Coronary artery disease
  • Cluster analysis


Dive into the research topics of 'Unsupervised learning to characterize patients with known coronary artery disease undergoing myocardial perfusion imaging'. Together they form a unique fingerprint.

Cite this