3D fusion of functional cardiac magnetic resonance imaging and computed tomography coronary angiography. Accuracy and added clinical value

Purpose: To evaluate the accuracy and added diagnostic value of 3-dimensional (3D) image fusion of computed tomography coronary angiography (CTCA) and functional cardiac magnetic resonance (CMR) for assessing hemodynamically relevant coronary artery disease (CAD). Methods: Twenty-seven patients with significant coronary stenoses on prospectively electrocardiography-gated dual-source CTCA, confirmed by catheter angiography and perfusion defects on CMR at 1.5 T were included. Surface representations and volume-rendered images from 3D-fused CTCA/ CMR data were generated using a software prototype. Fusion accuracy was evaluated by calculating surface distances of blood pools and Dice similarity coefficients. Two independent, blinded readers assigned myocardial defects to culprit coronary arteries with side-by side analysis of CTCA and CMR and using fused CTCA/CMR. Added value of fused CTCA/CMR was defined as change in assignment of culprit coronary artery to myocardial defect compared with side-by-side analysis. Results: 3D fusion of CTCA/CMR was feasible and accurate (surface distance of blood pools: 4.1 +/- 1.3 mm, range: 2.4-7.1 mm; Dice similarity coefficients: 0.78 +/- 0.08, range: 0.51-0.86) in all patients. Side-by-side analysis of CTCA and CMR allowed no assignment of a single culprit artery to a myocardial defect in 6 of 27 (22%) patients. Fused CTCA/CMR allowed further confinement of culprit coronary arteries in 3 of these 6 patients (11%). Myocardial defects were reassigned in 2 of 27 (7%) patients using fused CTCA/CMR, whereas the results remained unchanged in 22 of 27 (81%) patients. Interobserver agreement for assignment of culprit arteries to myocardial defects increased with fused CTCA/CMR (k = 0.66-0.89). Conclusion: 3D fusion of low-dose CTCA and functional CMR is feasible and accurate, and adds, at a low radiation dose, diagnostic value for the assessment of hemodynamically relevant CAD as compared with side-byside analysis alone. This technique can be clinically useful for the following: planning of surgical or interventional procedures in patients having a high prevalence of CAD and for improved topographic assignment of coronary stenoses to corresponding myocardial perfusion defects