Prediction of revascularization after myocardial perfusion SPECT by machine learning in a large population

Reza Arsanjani, Damini Dey, Tigran Khachatryan, Aryeh Shalev, Sean W. Hayes, Mathews Fish, Rine Nakanishi, Guido Germano, Daniel S. Berman, Piotr Slomka

Research output: Contribution to journalArticlepeer-review

91 Scopus citations


Objective: We aimed to investigate if early revascularization in patients with suspected coronary artery disease can be effectively predicted by integrating clinical data and quantitative image features derived from perfusion SPECT (MPS) by machine learning (ML) approach. Methods: 713 rest 201Thallium/stress 99mTechnetium MPS studies with correlating invasive angiography with 372 revascularization events (275 PCI/97 CABG) within 90 days after MPS (91% within 30 days) were considered. Transient ischemic dilation, stress combined supine/prone total perfusion deficit (TPD), supine rest and stress TPD, exercise ejection fraction, and end-systolic volume, along with clinical parameters including patient gender, history of hypertension and diabetes mellitus, ST-depression on baseline ECG, ECG and clinical response during stress, and post-ECG probability by boosted ensemble ML algorithm (LogitBoost) to predict revascularization events. These features were selected using an automated feature selection algorithm from all available clinical and quantitative data (33 parameters). Tenfold cross-validation was utilized to train and test the prediction model. The prediction of revascularization by ML algorithm was compared to standalone measures of perfusion and visual analysis by two experienced readers utilizing all imaging, quantitative, and clinical data. Results: The sensitivity of machine learning (ML) (73.6% ± 4.3%) for prediction of revascularization was similar to one reader (73.9% ± 4.6%) and standalone measures of perfusion (75.5% ± 4.5%). The specificity of ML (74.7% ± 4.2%) was also better than both expert readers (67.2% ± 4.9% and 66.0% ± 5.0%, P < .05), but was similar to ischemic TPD (68.3% ± 4.9%, P < .05). The receiver operator characteristics areas under curve for ML (0.81 ± 0.02) was similar to reader 1 (0.81 ± 0.02) but superior to reader 2 (0.72 ± 0.02, P < .01) and standalone measure of perfusion (0.77 ± 0.02, P < .01). Conclusion: ML approach is comparable or better than experienced readers in prediction of the early revascularization after MPS, and is significantly better than standalone measures of perfusion derived from MPS.

Original languageEnglish
Pages (from-to)877-884
Number of pages8
JournalJournal of Nuclear Cardiology
Issue number5
StatePublished - 26 Oct 2015
Externally publishedYes


  • Machine learning
  • coronary artery disease
  • myocardial perfusion SPECT
  • revascularization
  • total perfusion deficit

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging
  • Cardiology and Cardiovascular Medicine


Dive into the research topics of 'Prediction of revascularization after myocardial perfusion SPECT by machine learning in a large population'. Together they form a unique fingerprint.

Cite this