Introduction

Age is the single most important risk factor for stroke and cerebrovascular disease. In fact, up to 28% of the healthy ageing population >65years are identified with subclinical features of stroke¹. Therefore, the clinical relevance of cerebral perfusion imaging is increasing. Aside from providing non-invasive measures of perfusion, free-lunch time-encoded ASL (te-ASL)² offers the ability to measure arterial transit time (ATT), allowing predictions of ischemic tissue outcome^3,4, and arterial blood volume (aBV)⁵.
Even after the consensus paper⁶, different (2D or 3D, single- or multi-time-point) implementations of ASL are still being employed. However, it is unknown whether perfusion maps acquired with these different measurement strategies differ among young and elderly populations. We hypothesised that since cerebral hemodynamics change with age, that different ASL-implementations would provide qualitatively different perfusion maps, particularly at more superior slices where the post-label delay is longer for 2D acquisitions. Furthermore, we assessed the added value of ATT maps from te-ASL.

Methods

Imaging
ASL MR imaging was performed in 5 young volunteers (mean age: 35±3 years, 4 women) and 5 older volunteers (mean age: 70±1 years, 5 women) at 3T (Philips, Best, NL) with a 32-channel receive head-coil. 3D pCASL data were acquired with a multi-shot 3D GRASE sequence with TSE/EPI factor 18/15, FOV=240x240x120 mm³, voxel size=3.75x3.75x4 mm³, 3 dynamics including one dynamic for magnetization equilibrium (M0). 2D pCASL data were acquired using a single-shot FFE gradient-echo EPI sequence with SENSE=2.5, FOV=220x220x120 mm³, voxelsize=3.06x3.06x6 mm³, 30 dynamics. Time-encoded ASL (te-ASL) data were acquired using a 2D single-shot FFE gradient-echo EPI read-out with a SENSE factor=2.5, FOV=220x220x120 mm³, voxelsize=3.06x3.06x6 mm³, 7x8 Hadamard matrix with labelling durations and PLDs shown in FIG1a. 3D/2D/te-ASL sequences were matched on TR=4127/4600/4127, TE=10/12/10, flip angle=90°, labelling duration=1800ms (for te-ASL the first (perfusion-)block), PLD=2000ms, background suppression and total scan duration 5:22/5:07/5:52 mins. The 2D M0 scan had a TR/TE=2000/10ms, no background suppression and no labelling, and a total scan duration of 0:14 minutes.
Quantification and post-processing
Images were analysed in MATLAB (The MathWorks, Inc., Natick, Massachusetts, USA). Quantification of CBF including M0 and slice timing correction was done according to the ASL consensus paper⁶; this was also done for the perfusion-block of the te-ASL (which will be called single PLD te-ASL from now on). For te-ASL, CBF, ATT and aBV were fitted voxel-wise to the multi-PLD time-course(FIG1b). A macrovascular model was fitted for aBV-estimation⁷ and the resulting signal removed. Then a general kinetic model was fitted for CBF- and ATT-estimation⁸ (fminsearch, MATLAB9).
CBF images were co-registered to each subject’s 2D-pCASL CBF image using SPM12 (www.fil.ion.ucl.ac.uk/spm/software/spm12). The 2D volumes were smoothed (2mm FWHM 2D-Gaussian kernel) to visually match the blurring in the 3D volume(FIG1c). Grey matter (GM) was defined in each subject as voxels >30 mL/100g/min in the 2D-pCASL CBF.
To compare sequences, voxel-wise scatter-plots of GM CBF-estimates were made for each participant with colour-coding of sequential slices. Least squares polynomial fits of the scatterplots yielded slopes, and Pearson’s correlation coefficients were compared for the following pairs: 1) 2D-pCASL vs single-PLD te-ASL), 2) 3D-pCASL vs single-PLD te-ASL, and 3) 3D-pCASL vs multi-PLD te-ASL.

Results

2D-pCASL versus single-PLD te-ASL CBF was comparable as shown by the scatterplots(FIG2a). Overall, both groups showed similar correlations and similar slopes for all acquired slices(FIG2b). We found similar whole-brain CBF for te-ASL (mean 35±26mL/100g/min) and 2D-pCASL (mean 35±27mL/100g/min) indicating excellent agreement.
3D-pCASL versus single-PLD te-ASL (1800ms) showed a larger discrepancy in older subjects, as indicated by the steeper slopes particularly in 3 elderly participants(FIG3b).
Multi-PLD te-ASL fitted CBF showed higher CBF than single-PLD, but agreement with 3D pCASL was similar to single-PLD te-ASL as indicated by the similar Pearson’s correlation coefficients (>0.85) and slopes (>1)(FIG4). Whole-brain multi-PLD te-ASL CBF was not significantly different in older participants (43 ±30mL/100g/min) versus younger participants (43±33mL/100g/min, P=0.65). However, ATT was longer in the older (1.45±0.52s) versus younger participants (1.29 ± 0.54s, P<0.01)(FIG5).

Discussion

3D-,2D-, and te-ASL showed concordant results and slice-wise correlations differentiated older from younger individuals, particularly in 3D-pCASL versus te-ASL, which may be explained by the use of a single, on average shorter, PLD for the 3D GRASE readout. In elderly subjects with longer arrival times, this PLD may be too short to guarantee total delivery of labelled spins. In the 2D sequences, the effective longer PLD for the higher slices mitigate such effects. The almost identical CBF obtained from single-PLD te-ASL and 2D-pCASL indicates that te-ASL can offer additional parameters at no cost to CBF as proposed by Günther². Higher vascular signal captured in lower slices may explain the steeper slopes observed for 3D-pCASL versus te-ASL, again showing that both sequences show distinct information. A limitation is that 3D-pCASL was not compared to 3D GRASE te-ASL. Second, low temporal resolution could underestimate aBV and ATT^10,11, but since prolonged ATT is easier to detect, te-ASL should still be suitable for older clinical populations.
Conclusion
In elderly populations single-PLD pCASL may underestimate CBF if the PLD is not known a priori. Te-ASL provides additional ATT information, which we found to be prolonged in older volunteers, even though CBF was the same. Hence, ATT could be a more sensitive marker for risk profiling in clinical settings.

Acknowledgements

This research was supported by the EU under the Horizon2020 program (project: CDS-QUAMRI), and by Philips, Best, The Netherlands.