Diagnostic accuracy of semi-automatic quantitative metrics as an alternative to expert reading of CT myocardial perfusion in the CORE320 study

J Cardiovasc Comput Tomogr. 2018 May-Jun;12(3):212-219. doi: 10.1016/j.jcct.2018.03.010. Epub 2018 Apr 3.

Abstract

Aims: To determine the diagnostic accuracy of semi-automatic quantitative metrics compared to expert reading for interpretation of computed tomography perfusion (CTP) imaging.

Methods: The CORE320 multicenter diagnostic accuracy clinical study enrolled patients between 45 and 85 years of age who were clinically referred for invasive coronary angiography (ICA). Computed tomography angiography (CTA), CTP, single photon emission computed tomography (SPECT), and ICA images were interpreted manually in blinded core laboratories by two experienced readers. Additionally, eight quantitative CTP metrics as continuous values were computed semi-automatically from myocardial and blood attenuation and were combined using logistic regression to derive a final quantitative CTP metric score. For the reference standard, hemodynamically significant coronary artery disease (CAD) was defined as a quantitative ICA stenosis of 50% or greater and a corresponding perfusion defect by SPECT. Diagnostic accuracy was determined by area under the receiver operating characteristic curve (AUC).

Results: Of the total 377 included patients, 66% were male, median age was 62 (IQR: 56, 68) years, and 27% had prior myocardial infarction. In patient based analysis, the AUC (95% CI) for combined CTA-CTP expert reading and combined CTA-CTP semi-automatic quantitative metrics was 0.87(0.84-0.91) and 0.86 (0.83-0.9), respectively. In vessel based analyses the AUC's were 0.85 (0.82-0.88) and 0.84 (0.81-0.87), respectively. No significant difference in AUC was found between combined CTA-CTP expert reading and CTA-CTP semi-automatic quantitative metrics in patient based or vessel based analyses(p > 0.05 for all).

Conclusion: Combined CTA-CTP semi-automatic quantitative metrics is as accurate as CTA-CTP expert reading to detect hemodynamically significant CAD.

Keywords: Automatic data processing; Coronary artery disease; Multidetector computed tomography; Myocardial perfusion imaging.

Publication types

  • Comparative Study
  • Multicenter Study

MeSH terms

  • Aged
  • Aged, 80 and over
  • Area Under Curve
  • Asia
  • Automation
  • Computed Tomography Angiography / methods*
  • Computed Tomography Angiography / standards
  • Coronary Angiography / methods*
  • Coronary Angiography / standards
  • Coronary Artery Disease / diagnostic imaging*
  • Coronary Artery Disease / physiopathology
  • Coronary Circulation*
  • Coronary Vessels / diagnostic imaging*
  • Coronary Vessels / physiopathology
  • Europe
  • Female
  • Humans
  • Logistic Models
  • Male
  • Middle Aged
  • Multidetector Computed Tomography / methods*
  • Multidetector Computed Tomography / standards
  • Myocardial Perfusion Imaging / methods*
  • Myocardial Perfusion Imaging / standards
  • North America
  • Observer Variation
  • Predictive Value of Tests
  • Prospective Studies
  • ROC Curve
  • Radiographic Image Interpretation, Computer-Assisted / methods*
  • Radiographic Image Interpretation, Computer-Assisted / standards
  • Reference Standards
  • Reproducibility of Results
  • Severity of Illness Index
  • South America
  • Tomography, Emission-Computed, Single-Photon