2012

Velocity Selective Arterial Spin Labeling at 7T

Emiel C.A. Roefs¹, Lydiane Hirschler¹, Natalia Petridou², and Matthias J.P. van Osch¹
¹Radiology, C.J. Gorter MRI Center, Leiden Univeristy Medical Center, Leiden, Netherlands, ²Radiology, Center for Image Sciences, Univeristy Medical Center Utrecht, Utrecht, Netherlands

Synopsis

Keywords: Arterial Spin Labelling, Arterial spin labelling

Motivation: Velocity selective (VS)-ASL can be of great added benefit for fMRI at ultra-high field because of its specificity and ATT-insensitivity.

Goal(s): Our goal was to show the feasibility of VS-ASL at 7T with minimal TR, which involves minimizing SAR and timing parameters within the sequence.

Approach: VS-ASL was implemented using BIR4 labeling module and compared to the robust FAIR ASL. Image quality was measured based on quantified ASL-signal and tSNR-efficiency.

Results: Good agreement between FAIR and VSS_inv-ASL was found. VSS_inv-ASL was used to create intrinsic BGS without additional SAR. A shortest TR of 4500ms was achieved which could potentially be shorted with pTx.

Impact: We showed the feasibility of Velocity Selective (VS)-ASL at UHF and the application in fMRI. VS-ASL showed comparable but less homogeneous tSNR as FAIR. VS-ASL is of added benefit for fMRI at ultra-high field because of its specificity and ATT-insensitivity.

Background:

Despite the longer T1 and therefore less severe decay of label, arterial spin labeling (ASL) at ultra-high field (UHF: 7T and above) has been found challenging mainly due to magnetic field inhomogeneities, B1-inhomogeneities, increase in physiological noise, shorter T2, and high specific absorption rate (SAR)⁽¹⁾. Over the last years, pseudo continuous (pC)ASL⁽²⁾ and pulsed (P)ASL using FAIR⁽³⁾ have shown to provide robust perfusion maps at 7T. Velocity selective (VS)-ASL creates label within the imaging volume by saturating the blood flowing above a certain cut-off velocity (Vcut) and is therefore insensitive to arterial-transit-time (ATT) delays which enables scanning with short PLDs⁽⁴⁾. Moreover, the fact that label is also created within the sweet spot of the transmit coil, could be the essential advantage of VS-ASL over other ASL-approaches. In this abstract, we show the first feasibility of 3D VS-ASL at 7T and compared it to current state-of-the-art FAIR and show its use in fMRI.

Methods:

Velocity selective saturation (VSS)-ASL was implemented using a BIR4 labeling module (Figure 1) that consists of adiabatic excitation and refocusing to guarantee proper performance in the presence of severe B1- and B0-inhomogeneties. Meta-materials were placed around the head to homogenize B1⁽⁵⁾. Because of the short T2 of blood (54ms) at 7T, single refocusing (BIR4) with its short TE was chosen over a double-refocused hyperbolic-secant (DRHS). Further design criteria were minimal SAR and thereby TR, especially to support neuroscience applications of VS-ASL.

In this abstract, we first compared VSS-ASL to the current state-of-the-art implementation of FAIR. Secondly, we investigated the effect of background suppression (BGS) on the VSS-ASL signal by comparing VSS-ASL to VSS_inv-ASL⁽⁶⁾. In the latter, a π/2 phase-offset is added to the last BIR4 segment resulting in an inversion of the static tissue providing intrinsic BGS without additional SAR. The influence of CSF diffusion weighting was studied by short TI experiments. Lastly, we studied whether the recommendations from the VS-ASL grey-paper⁽⁴⁾ are also applicable at 7T by varying Tsat (750:250:1500ms) and Vcut (1,2 and 4cm/s), because we hypothesized that these are most dependent on B1-coverage (Figure 1).

Volunteers were scanned on a Philips 7T Achieva (Best, Netherlands) equipped with a 32R-2Tx Nova head coil. The read-out (3D Gradient-echo EPI, FOV=210x210x51mm, voxel-size=3x3x3mm (reconstructed at 1.8mm inplane), SENSE_RL=3, FA=20°, TR/TE=26/8.3ms, SPIR fat-suppression) was kept the same for VSS-ASL and FAIR. FAIR scanning parameters: PLD=1800ms, QUIPPSII=1000ms, 2 BGS-pulses at 900ms and 1500ms, TR=7000ms (Figure 1A). Base parameters of VSS_(inv)-ASL: TI=1500ms, Vcut=2cm/s, Tsat=1500ms and vascular-crushing-module (VCM) (Figure 1B,C). For all scans, 20 averages were acquired at ≈85% of the SAR-limit.

Results & Discussion

Figure 2 shows ASL-maps without and with VCM, as well as quantitative perfusion maps of VSS_inv-ASL. Mean GM perfusion of this participant was 41.1ml/100g/min which is in line with average literature values. Comparable mean GM perfusion values were found (42.7ml/100g/min) between VSS_inv and FAIR (Figure 3). Mean GM tSNR was similar for FAIR (1.73±0.5) and VSASL (1.77±0.3). However, for VSS-ASL it should be noted that the tSNR is especially good in the posterior part of the brain and that the FAIR tSNR is more homogeneous within the cortex and would therefore be preferable. The high tSNR in the posterior lobe could be explained by the meta-materials placement. The use of pTx would be a good solution to improve B1-homogeneity.

Because of the selective pre-saturation and limited extent of the transmit coil, the Tsat could potentially be shorter than 1500ms since a clear decrease in ASL-signal was only found for Tsat=750ms (Figure 4A), however already a small penalty in ASL-signal was found at Tsat=1250ms. Besides the decreased ASL-signal, it can also be observed that the ventricles become more visible, which could be partly explained by difference in BGS.

To investigate diffusion-weighting, the TI and PLD were shortened, since it is expected that diffusion-weighting will remain the same (as can be seen in Figure 4B in the ventricles), but perfusion signal will decrease. This proves that while it is possible to measure perfusion at 7T, CSF diffusion-weighting is also present, although smaller than the signal due to perfusion. Additionally, in Figure 4C variations in ASL-signal can be seen as a function of Vcut, where also at 7T, with the small transmit coverage, a Vcut around 1-2cm/s seems optimal.

Lastly, Figure 5 shows a proof of concept of the use of VSS_inv-ASL at UHF for fMRI with a visual stimulation (27s on-27s off, 11 repeats) showing both activation curves in the visual cortex for ASL and BOLD.

Conclusion:

We showed the feasibility of Velocity Selective (VS)ASL at UHF and its potential use in fMRI.

Acknowledgements

This project is funded by: TTW-ZonMW-SGF-LHS Human Measurement models grant #18969-Virtual Cerebrovascular Responses.

References

1. Teeuwisse WM, Webb AG, van Osch MJ. Arterial spin labeling at ultra‐high field: all that glitters is not gold. International Journal of Imaging Systems and Technology. 2010;20(1):62-70.

2. Zhao C, Shao X, Shou Q, Ma SJ, Gokyar S, Graf C, et al. Whole-Cerebrum distortion-free three-dimensional pseudo-Continuous Arterial Spin Labeling at 7T. NeuroImage. 2023:120251.

3. Kashyap S, Ivanov D, Havlicek M, Huber L, Poser BA, Uludağ K. Sub-millimetre resolution laminar fMRI using Arterial Spin Labelling in humans at 7 T. PLOS ONE. 2021;16(4):e0250504.

4. Qin Q, Alsop DC, Bolar DS, Hernandez‐Garcia L, Meakin J, Liu D, et al. Velocity‐selective arterial spin labeling perfusion MRI: A review of the state of the art and recommendations for clinical implementation. Magnetic Resonance in Medicine. 2022;88(4):1528-47.

5. Koloskov V, Brink W, Webb A, Shchelokova A. Flexible metasurface for improving brain imaging at 7T. arXiv preprint arXiv:231020019. 2023.

6. Guo J. Robust dual‐module velocity‐selective arterial spin labeling with velocity‐selective saturation and inversion. Magnetic Resonance in Medicine. 2023;89(3):1026-40.

Figures

Figure 1. Sequence diagrams and Mz simulations. A). FAIR ASL with PLD the time between labeling and imaging. B). VSS_(inv)-ASL with Tsat the time between pre-saturation and labeling in which the blood within the FOV should be refreshed, TI the time between labeling and the vascular crushing module (VCM), and PLD the time between the VCM and imaging. Note the difference in PLD definition. C). BIR4 module used in this study focusing on the difference between VSS and VSS_inv.

Figure 2. VSS_inv perfusion quantification results. A). raw ASL signal from imaging with and without VCM. Clearly visible that the venous signal is removed after the VCM also proving the BIR4 labeling module used in this study performs well B.) quantified ASL signal in ml/100g/min using equation 1. from the VS-ASL grey paper (4).

Figure 3. Comparison of VSS, VSS_inv and FAIR ASL in terms of A.) ASL-signal B). and C). tSNR. Mean GM tSNR for VSS_inv and FAIR were comparable 1.77±0.3 and 1.73±0.5 respectively. For VSS_inv-ASL this mean tSNR is mainly because of high tSNR in the posterior lobe where for FAIR the tSNR is more constant across the cortex and is therefore preferable still despite similar tSNR. The BGS-level for VSS_inv-ASL (≈70%) was slightly lower than for FAIR (≈85%). However, the tSNRefficiency defined as tSNR/√TR would be better for VSS_inv-ASL compared to FAIR because the TR’s were 4500ms and 7000ms resp.

Figure 4. A.) Varying Tsat from 1500ms-750ms (steps of 250ms), visible that for Tsat=750ms the ASL-signal decreases because of not fully relaxed blood in the imaging volume during labeling. B.) Varying TI, difference originating from perfusion (since TI+PLD=constant) and CSF-diffusion remains. C.) Varying Vcut, 1cm/s, 2cm/s and 4cm/s. What stands out is that the ASL-signal especially posteriorly does not change much between 1cm/s and 2cm/s. However, more diffusion weighting can be seen in the ventricles with a Vcu =1cm/s. For a Vcut=4cm/s, a clear drop in signal can be observed.

Figure 5. fMRI response: 27son-27soff visual task (8Hz flickering checkerboard, 11 averaged repeats, scantime=10min) for both BOLD (orange, derived using surround mean) and ASL (blue, derived using surround subtraction. Individual ROI drawn based on BOLD activation. An increase of more than 20% can be observed using VSS_inv-ASL where the BOLD response is around 1.3% for the first participants (left). For the second participant (right) the response is less clear for the VSS_inv-ASL signal. Shaded area represents 95% confidence interval across voxels.

Proc. Intl. Soc. Mag. Reson. Med. 32 (2024)

2012

DOI: https://doi.org/10.58530/2024/2012