Editorial for “Comparison of Myocardial Blood Flow Quantification Models for Double ECG Gating Arterial Spin Labeling MRI : Reproducibility Assessment”

Imaging myocardial perfusion, particularly under stress conditions is clinically useful for detecting potential areas of ischemia in patients with suspected coronary artery disease.1 Arterial spin labeling (ASL) magnetic resonance imaging (MRI) allows quantitative measurement of myocardial perfusion using arterial blood water as an endogenous diffusible tracer2 instead of any external contrast agent injection-based techniques. In ASL technique, blood flowing to the myocardium has been typically labeled using a flow-sensitive alternating inversion recovery (FAIR) approach.3 This MRI technique is based on the acquisition of label and control images after an inversion time (TI), following alternating non-selective (NS) and slice-selective (SS) inversions, respectively. The perfusion-weighted signal within the myocardium results from the pairwise subtraction of these images. However, the presence of motion during data acquisition introduces artifacts in the subtracted images that can compromise quantification of myocardial perfusion.4 During cardiac imaging, it is crucial to synchronize the acquisition with the cardiac cycle and adjust the imaging parameters to minimize motion artifacts. One particularity of myocardial ASL is the importance of performing both labeling and readout at the same cardiac phase. Using single-ECG gating (SG) based triggering image acquisition at the same cardiac phase is made difficult by the presence of HRV. Double-ECG gating (DG) was introduced, as an attractive alternative to SG, to trigger labeling and readout independently, ensuring the same cardiac phase at the expense of introducing TI variability.3, 5, 6 Previous studies have utilized different strategies for the quantification for myocardial blood flow (MBF). However, it remains unclear which of these methods leads to higher temporal signal to noise ratio (tSNR) and more reproducible MBF measurements. In these studies, either breath-hold or synchronized breathing strategies were used to minimize the degree of motion caused by respiration, which have been found to be effective in both healthy subjects and patients.7 Evaluating perfusion reproducibility in DG acquisitions is crucial before this technique can be used more extensively in clinical research and ultimately in clinical practice. Recent study by Aramendía-Vidaurreta et al8 assessed the performance of various MBF quantification approaches for DG and compared to SG. Moreover, in this study, the effects of presaturation pulses such as hyperbolic secant (HS) adiabatic and C-shape frequency offset corrected inversion (FOCI) adiabatic inversion pulses were evaluated. This study also reported the reproducibility of DG and SG myocardial ASL. Aramendía-Vidaurreta et al8 have used Monte Carlo simulations for comparing three DG MBF quantification models for FAIR ASL considering the effects of HRV. Numerical simulations were conducted to investigate the robustness of HS and C-shape FOCI adiabatic pulses to field inhomogeneity. In addition, cardiac MRI data of 16 healthy volunteers were acquired at a 1.5-T scanner. A subgroup of 10 subjects was scanned twice to collect ASL data for assessing intra- and inter-session reproducibility. Postprocessing of images and data analysis were carried out for computation of MBF, standard error of the perfusion measurement (SEMBF), and tSNR. The within-subject coefficient of variation (wsCV)9 and Bland–Altman plots were used to evaluate intra- and inter-session reproducibility of MBF measurements. In this study, it was observed that DGq1a provided higher tSNR compared to the other quantification models. Moreover, MBF measurements remained consistent across different noise levels, with values showing a slight overestimation in comparison to the reference perfusion signal. tSNR increased with perfusion and tSNR in DGq1a was significantly higher than in the other quantification models. DG sequences showed lower SD of the series of label and control images separately for a subject with high HRV, which reflected a reduced degree of motion prior to registration. Despite the occurrence of reduced motion artifacts in DG sequences, the identified number of outliers was similar between SG and DG sequences. Intrasession reproducibility was similar across DG quantification methods, but higher than the reproducibility of the SG technique. In this study, no differences were observed between acquisitions with FOCI and HS pulses in vivo despite the slice profile benefits encountered in the simulations under various B0 and B1 conditions. This study has some limitations such as the small sample size and the time between scans was quite large. Further work is required to evaluate the performance of DG with free-breathing strategies, which could be especially useful in clinical practice. In conclusion, this study demonstrated that reproducibility values were enhanced with the use of three-parameter fitting quantification model in DG ASL data. Furthermore, it revealed that the number of ASL image pairs needed can be reduced without increasing the error of the MBF results.