Voxel-wise Quantitative assessment of Myocardial Perfusion: a Comparison of Four Different Deconvolution Algorithms Using Real Flow Values

Background Accurate assessment of relative reductions in regional myocardial blood flow (MBF) is clinically important in identifying haemodynamically significant coronary artery disease. First-pass perfusion magnetic resonance imaging allows for a non-invasive and radiation free quantification of MBF. One of the advantages of MR perfusion over nuclear medicine techniques is better spatial resolution. Previous quantification methods used to asses MBF have been derived from segmental analysis of the myocardium. Here we propose the use of voxel-wise myocardium perfusion analysis, which potentially includes useful diagnostic information, the theoretical trade-off however is a lower SNR. In this study we compared four different regularized deconvolution methods by using voxel-wise and segmental analysis. We took advantage of a hardware perfusion phantom and thus we were able to compute absolute errors. We aimed to evaluate both the robustness of quantification methods to noise and accuracy of tissue transfer function estimates and the physiological parameters which can be deduced from it using segmental and voxel-wise analysis . Methods Myocardial blood flow in both a 10-patient series and a hardware MR-perfusion phantom was quantified using Fermi function modelling[1], AutoRegressive Moving Average model (ARMA)[2], model-independent approach using B-spline basis[1] and exponential basis deconvolution [3](with non-negativity and non-increasing constraints). Patients perfusion data were acquired in three LV short axis slices with a saturation recovery gradient echo method (TR/TE 3.0ms/1.0ms, flip-angle 15°; effective k-t SENSE acceleration 3.8, spatial resolution 1.2x1.2x10mm) during adenosine-induced hyperaemia (140μg/kg/min) using 0.05mmol/kg Gd-DTPA (Magnevist, Schering, Germany) at 4ml/minute followed by a 20 ml saline flush. Voxel-wise perfusion estimates were compared to the results obtained with different levels of spatial averaging (10 voxels -1 segment – whole slice). Accuracy was tested using the same sequence to image first pass perfusion (5 different experiments) in a hardware phantom that was recently developed by our group. We validated the estimated perfusion values for the perfusion phantom with true perfusion values, as measured by means of precision flow-meters.