Article Text

Download PDFPDF

021 Perfusion cardiovascular magnetic resonance (CMR) – can david (resolution) take on goliath (coverage) again?
  1. Joy S Shome1,
  2. Kerem C Tezcan2,
  3. Sohaib Nazir1,
  4. Markus Henningsson1,
  5. Adriana Villa1,
  6. Imran Rashid1,
  7. David Snell1,
  8. Antonis Pavlidis4,
  9. Brian Clapp4,
  10. Simon Redwood3,4,
  11. Kamran Baig5,
  12. Reza Razavi1,
  13. Tevfik F Ismail1,4,
  14. Amedeo Chiribiri1,
  15. Divaka Perera3,4,
  16. Sebastian Kozerke2,
  17. Sven Plein1,6
  1. 1Division of Imaging Sciences, The Rayne Institute, King’s College London, St. Thomas’ Hospital, London, UK
  2. 2Department of Biomedical Engineering, ETH Zurich, Zurich, Switzerland
  3. 3Cardiovascular Division, The Rayne Institute, King’s College London, St. Thomas’ Hospital, London, UK
  4. 4Department of Cardiology, St. Thomas’ Hospital, London, UK
  5. 5Department of Cardiothoracic Surgery, St. Thomas’ Hospital, London, UK
  6. 6Leeds Institute of Cardiovascular and Metabolic Medicine, University of Leeds, Leeds, UK


Background Both 3D and high-­resolution 2D-­perfusion CMR accurately detect coronary artery disease (CAD). 3D provides whole-­heart coverage whereas 2D better detects sub-­endocardial ischaemia. We compared the diagnostic accuracy of both techniques to detect flow limiting CAD as measured by fractional flow reserve (FFR). We also investigated the relative accuracy of these tools in identifying prognostically significant myocardial ischaemic burden (MIB).

Methods Patients with suspected angina underwent high spatial resolution 2D k-­t SENSE (3 slices, in-­plane spatial resolution 1.3 × 1.3×8 mm) and 3D k-­t PCA whole heart (12 slices, in plane spatial resolution 2.3 × 2.3×5 mm) myocardial perfusion CMR during adenosine stress in a single sitting. Invasive coronary angiography with FFR (for stenoses of 50-­80% severity visually) was performed in all patients prior to revascularisation. Perfusion defects were contoured using circleCVI software and MIB was calculated for both 2D and 3D-­CMR. The anatomical and functional BCIS-­1 Jeopardy Scores (BCIS-­JS) scores were calculated from the invasive angiograms.

Results Forty-­seven patients were included in the analysis. Per-­patient sensitivity, specificity, diagnostic accuracy, PPV, and NPV in identifying flow-­limiting CAD were 76%, 100%, 85%, 100%, 72% for 2D

and 69%, 89%, 77%, 91%, 64% for 3D perfusion CMR (AUC of 0.88 versus 0.79, p=0.27). In 24 patients with confirmed CAD, the MIB by 2D and 3D was 15.7 ±­ 8.6% and 19.6% ±­ 11.7% respectively (p=0.088), with a trend towards 3D underestimating MIB by a mean of 3.8% (Figure 1). In these patients, considering 2D as the gold standard, the diagnostic accuracy of 3D CMR, anatomical BCIS-­JS, and functional BCIS-­JS in identifying prognostically significant MIB was 79%, 75%, and 87.5% respectively. (See Figures 2 and 3 for case examples)

Conclusion In this first head-­to-­head comparison with invasive angiography and FFR, high-­resolution 2D and whole-­heart 3D perfusion CMR had comparable diagnostic performance in detecting flow-­limiting CAD on a per–patient basis. Both 3D and functional BCIS-­JS identify prognostically significant MIB well as determined by 2D. 2D estimates of MIB tend to be higher than 3D, however both methods have limitations (resolution versus coverage). In this contest, superior resolution may satisfactorily offset the lack of myocardial coverage.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.