Labeling efficiency (LabEff) is one of the most important parameters for CBF quantification of ASL perfusion scans, although usually a constant value is assumed in current clinical practice¹. However, LabEff may vary for pCASL due to differences in flow velocity and off-resonance effects, indicating the necessity to measure LabEff for each pCASL perfusion scan. Previously, a method was proposed to measure LabEff by a fast, additional scan². However, even the limited additional time demanded by this LabEff scan would seriously threaten its clinical acceptation. In this study, we propose a method that integrates the LabEff-measurement into a normal pCASL scan without SNR-penalty.

Sequence: The combined perfusion&LabEff sequence exploits time-encoded ASL by including an additional shorter labeling block and a single-slice Look-locker EPI readout into the post-labeling delay of a normal pCASL perfusion acquisition (without background suppression). This enables the monitoring of blood signal in the large arteries directly after the labeling. The two labeling blocks of the combined sequence are arranged in a Hadamard-like encoding scheme (Fig. 1A). Three types of LabEff signal are acquired: Control, Label and M0 (with pCASL RF switched off to measure the equilibrium blood signal). A thin saturation slab (5mm) is applied 1mm above the LabEff imaging slice and performed before each EPI readout, in order to saturate signal from venous blood (Fig. 1B).

In vivo experiments: Five healthy volunteers (age 23-28y; 3m) were scanned at a 3T scanner (Philips Healthcare) after providing written informed consent. They underwent the following scans: a separate perfusion scan without background suppression and a separate LabEff scan (both using pCASL flip angle (FA) of 21°), four combined perfusion&LabEff scans using FAs of 8°, 15°, 21° and 25° respectively, brain M0 scan, vessel-encoded pCASL scan for vascular territory mapping, 3D T1W, and a phase-contrast (PC) quantitative flow scan at the location of labeling slab. All non-vessel-encoded perfusion scans: labeling duration/PLD 1800/1800ms, single-shot EPI, 19 slices, voxel size 3×3×6mm³, TR/TE 4209/9.6ms, 30 control-label pairs. All LabEff scans: labeling duration/PLD 600/7ms, voxel size 2×2×3mm³, 26 time points with interval of 39.7ms.

Postprocessing: Total GM CBF and territorial GM CBF were calculated using the recommended model in ASL white paper assuming a constant LabEff of 0.85¹. The ASL time signal of LabEff scan was normalized wrt M0 signal and then used to calculate LabEff with a previously proposed model²:$$\overline{\alpha}=\frac{1}{2M_{0}}\int_{0}^{V_{max}} \alpha\left(\nu\right)M\left(\nu\right)d\nu\tag{1}$$$$ASL\left(t\right)=\int_{0}^{V_{max}} \alpha\left(\nu\right)M\left(\nu\right)D\left(\nu\right)E\left(\nu,t\right)d\nu\tag{2}$$LabEff was calculated for all 4 arteries (i.e. RICA, LICA, RVA, LVA) and was also simulated for a range of flow velocities³. The CBF and LabEff with FAs of 8°, 15°, and 21° were normalized with their concurrent 25° ones, before correlating CBF­_norm with LabEff_norm.

The combined perfusion&LabEff sequence yielded similar CBF maps as the separate scan except that the lower slices were affected due to the close proximity of the LabEff imaging and saturation slab (Fig. 2). No significant differences were found between the combined and the separate sequence with regard to GM CBF (51.2±12.2 vs. 51.2±11.5ml/100g/min, p=0.99) and GM temporal-SNR (1.03±0.21 vs. 0.98±0.18, p=0.15), indicating that the incorporated LabEff sequence had little to no influence on the perfusion scan. However, LabEff acquired with the combined sequence was significantly lower than the efficiency from the separate LabEff scan (0.715±0.062 vs. 0.741±0.067, p=0.049). This may be attributed to MT effects or imperfect Hadamard decoding due to the influence of cardiac pulsations. The measured in vivo LabEff decreased sharply for the lower FA’s (Fig. 3A). Besides, the relationship between in vivo LabEff and FA was similar to simulation (Fig. 3B). There was a strong correlation between the quantified CBF and LabEff data for the different FA’s (Fig. 4). In one subject we found a remarkably lower CBF in the RICA- as compared to the LICA-territory. This could largely be attributed to a lower labeling efficiency in the RICA. This capability to distinguish true RICA-territory hypoperfusion from technical failure due to sub-optimal labeling clearly demonstrates the clinical potential of the proposed approach. Current limitations include: 1) No background suppression was included; 2) A slight increase in total scan time occurred due to the need of additional preparation phases.We demonstrated that LabEff could be measured simultaneously with perfusion imaging by efficiently using the PLD of the pCASL sequence, with almost no compromise on SNR of the perfusion images and almost no additional time cost. The measured LabEff is artery-specific and could potentially be used for calibration of territorial perfusion.