Objective: To detail the principles of using model-based determination of regional myocardial blood flow (MBF) by computed tomography (CT) and demonstrate its in vivo applicability.
Methods: Dual-source CT was performed with a dynamic protocol comprising acquisition with alternating table positions in ECG-triggered end-systolic timing every second for 30 s. The results of two reconstructions were merged into one final image stack (coverage 73 mm), with low spatial frequency components from a 360 degrees reconstruction and high spatial frequency components from a dual-source cardiac partial image reconstruction. A parametric deconvolution technique was used to fit the time-attenuation curves (TAC), the maximum slope of which was used to derive MBF.
Results: One study participant underwent dynamic myocardial stress perfusion imaging (9.6 mSv) followed by invasive coronary angiography and measurement of fractional flow reserve as the gold standard. MBF was 159 ml/100 ml/min in the non-ischaemic anterolateral and 86 ml/100 ml/min in the inferoseptal ischaemic wall.
Conclusion: This first evaluation indicates that mathematical modelling of voxel TACs can potentially be used to quantify differences in MBF in a clinical setting. If confirmed in feasibility studies, cardiac CT may allow for parallel assessment of morphology and haemodynamic relevance of coronary artery disease.