Arterial spin labeling of the heart has been shown to estimate myocardial perfusion and perfusion reserve for coronary artery disease assessment.¹ However, current spatial labeling methods suffer from transit delay effects when imaging is extended to more than a single slice. Velocity selective (VS) labeling is a promising alternative that does not suffer from transit delay effects.²Images were acquired using a 3T GE Signa Excite HD scanner with an 8-channel cardiac coil in 12 healthy volunteers. Myocardial ASL measurements were made at a single short axis slice using both VSASL and conventional flow alternating inversion recovery (FAIR) ASL as a reference.^1,2 Figure 1 shows the VS pulse, which consists of a symmetric BIR4 for reduced eddy current sensitivity³ and bipolar gradients to prevent spatial signal modulation in static tissue.⁴ The VS labeling pulse was executed during mid-diastole when coronary blood flow is high (>15-40 cm/s)⁵ and myocardial velocity is low (< 2 cm/s)⁶ to increase labeling efficiency and avoid spurious myocardial tagging. The cutoff velocity was 10 cm/s and applied in the through-slice direction. An additional triple inversion recovery background suppression preparation was used to null myocardial T1s between 1250 ms and 1450 ms. Each scan consisted of 6 breath-held labeled/control image pairs. In a single volunteer, an additional FAIR experiment was performed with a thick inversion slab to simulate whole heart coverage with increased transit delay. Myocardial blood flow (MBF), physiological noise (PN), and temporal SNR (TSNR = MBF/PN) were measured within the left ventricular myocardium ROI.

Figure 2 illustrates successful VS labeling of blood within the right coronary artery in two subjects. VSASL performance in individual subjects are summarized in Table 3. In 3 subjects, VSASL performance was poor (TSNR < 1) while in the other 9, VSASL performed well with results comparable to the FAIR reference scan. In the poor performers, MBF and PN measurements for VSASL and FAIR were -0.13 ± 1.35 ml/g/min and 1.86 ± 0.42 ml/g/min respectively while in good performers, they were 2.13 ± 0.48 ml/g/min and 1.97 ± 0.38 ml/g/min. We speculate that the high PN in poor performers is from inconsistent pulse performance, possibly due to variations in B1 and off-resonance, but have yet to test this. Poor performance may also be from spurious labeling of myocardium, which can be further reduced by more consistent background suppression. Low TSNR will be addressed by further sequence improvements that explore different cutoff velocities and velocity labeling directions.

In the volunteer in whom we performed FAIR with a thicker inversion slab (FAIR-TS), MBF and PN measurements from VSASL, FAIR, and FAIR-TS were 1.38 ± 0.58 ml/g/min, 0.88 ± 0.37 ml/g/min, and -0.04 ± 0.24 ml/g/min respectively. FAIR-TS suffered from a large transit delay and was unable to estimate MBF while VSASL did not suffer from transit delay effects.

VS labeling has several important advantages over spatial labeling sequences, notably its insensitivity to transit delay and its compatibility with whole heart coverage. We have demonstrated the feasibility of VS labeling of coronary blood and demonstrated that VSASL is sensitive to myocardial perfusion. Performance of VSASL was comparable to FAIR in 75% of the subjects scanned. In the other 25% of subjects, resolving the inconsistent performance is still a work in progress. American Heart Association 13GRNT13850012; Wallace H. Coulter Foundation Clinical Translational Research Award.