Inline quantitative myocardial perfusion flow mapping
Journal of Cardiovascular Magnetic Resonance volume 18, Article number: W8 (2016)
Quantification of myocardial blood flow (MBF) using first-pass perfusion MRI has potential to objectively evaluate ischemic heart disease. While considerable development and clinical research has been conducted during the past two decades on this topic, there lacks a perfusion flow mapping solution that runs on the MRI scanner and produces MBF maps immediately following the perfusion scan. To be effective in a clinical environment we identified following capabilities are needed: fully automated workflow, pixel-wise flow calculation, single-bolus contrast injection, complete free-breathing acquisition, rapid processing, and inline deployment.
Multi-slice saturation recovery (SR) imaging was performed during the first pass of a single bolus of Gd injection with normal free breathing. A low resolution blood pool image was acquired each heart beat for the measurement of arterial input function (AIF). AIF imaging used a dual-echo FLASH sequence and correction of T2* during the first pass. Higher resolution myocardial imaging used either FLASH or SSFP protocols. Proton density (PD) weighted images were acquired prior to contrast, used for correction of surface coil intensity and conversion to Gd concentration units (mmol/L). Figure 1a illustrates the fully automated processing workflow: (1) image reconstruction, (2) motion correction (MOCO) of both AIF and perfusion images and co-registration with PD images, (3) automated segmentation of the AIF LV blood pool signal, (4) surface coil intensity correction, (5) conversion to [Gd] units from SR and PD signal intensities using Bloch equation calculations including correction of T2* loss in AIF from dual echo signal, and (6) calculation of pixel-wise MBF maps by deconvolution processing and display using a custom colormap. Three methods were implemented inline: (1) model-free method using a novel L1-norm based optimization, (2) a constrained Fermi function and (3) a two compartment model with interstitial volume estimation. The whole process was implemented in C++ via the Gadgetron framework  and integrated inline so that the motion corrected Gd concentration images and flow maps were calculated without any user interaction.
The pixel-wise MBF maps for all slices were created within 1.5 mins after the end of scan. Mean MBF values (Table 1) were computed from a ROI drawn in the myocardium. Rest ROIs include all myocardium. Stress positive (8 of 16 studies) ROIs was drawn remote from the hypoperfused region. The performance of motion correction and surface coil correction was visually assessed to be good to excellent for all cases. Inline MBF mapping for a stress positive study are shown in Figure 1b-h.
Fully automated pixel-wise MBF mapping was integrated inline on a clinical scanner. Myocardial perfusion flow maps were computed without any user interaction. This technique could promote the clinical usage of fully quantitative perfusion flow mapping.
Hansen MS, et al: MRM. 2013, 69: 1768-1776.
About this article
Cite this article
Xue, H., Hansen, M.S., Nielles-Vallespin, S. et al. Inline quantitative myocardial perfusion flow mapping. J Cardiovasc Magn Reson 18 (Suppl 1), W8 (2016). https://doi.org/10.1186/1532-429X-18-S1-W8