Estimation of perfusion and arterial transit time in myocardium using free-breathing myocardial arterial spin labeling with navigator-echo

Abstract

Arterial spin labeling (ASL) provides noninvasive measurement of tissue blood flow, but sensitivity to motion has limited its application to imaging of myocardial blood flow. Although different cardiac phases can be synchronized using electrocardiography triggering, breath holding is generally required to minimize effects of respiratory motion during ASL scanning, which may be challenging in clinical populations. Here a free-breathing myocardial ASL technique with the potential for reliable clinical application is presented, by combining ASL with a navigator-gated, electrocardiography-triggered TrueFISP readout sequence. Dynamic myocardial perfusion signals were measured at multiple delay times that allowed simultaneous fitting of myocardial blood flow and arterial transit time. With the assist of a nonrigid motion correction program, the estimated mean myocardial blood flow was 1.00 ± 0.55 mL/g/min with a mean transit time of ∼ 400 msec. The intraclass correlation coefficient of repeated scans was 0.89 with a mean within subject coefficient of variation of 22%. Perfusion response during mild to moderate stress was further measured. The capability for noninvasive, free-breathing assessment of myocardial blood flow using ASL may offer an alternative approach to first-pass perfusion MRI for clinical evaluation of patients with coronary artery disease.

© 2010 Wiley-Liss, Inc.

Figures

Figure 1

Pulse sequence diagram of free-breathing myocardial ASL which combines ASL (FAIR in this study) with a navigator-gated, ECG-triggered TrueFISP readout sequence. Selective and nonselective inversion pulses are used for label and control acquisitions respectively. A saturation pulse is applied prior to labeling pulses to minimize the variations in heart beat. Navigator-echo is placed on the diaphragm that allows image readout during the end-expiration phase. Image acquisition is always during the mid-diastole cardiac phase.

Figure 2

Slice selective (SS) and nonselective (NS) TrueFISP images, and difference perfusion weight images (PWI or dM) of a representative subject with 5 delays, ROIs of myocardium and ventricle blood are shown. The 3rd and 4th rows show PWIs acquired without and with non-rigid MoCo respectively. Arrows indicate bright and dark bands likely caused by subtraction errors due to motion.

Figure 3

Mean fractional difference perfusion signal (dM/M0) measured in myocardium (with and without non-rigid MoCo respectively), in background muscle (A) and in ventricle blood (B), error bars indicate standard error of the mean (SEM). Note the relatively large error bar for 200ms TI was because data were collected only from 3 subjects.

Figure 4

Model fitting results of myocardial perfusion data acquired with (r=0.94) and without non-rigid MoCo (r=0.98), error bars indicate standard error of the mean (SEM).

Figure 5

Scatter plot of test-retest results using repeated myocardial ASL scans at the TI of 900ms in 8 subjects.

Figure 6

Myocardial ATT and MBF maps of a representative subject (same as Fig. 2) by pixel-by-pixel fitting using data collected at 5 delay times (200–1700ms). The 5 delay dM images were aligned using the Advanced Normalization Tools software (ANTS).