Introduction

There is a clinical & pre-clinical need for high SNR quantitative ASL. We introduce a quantitative model for a high SNR ASL sequence (mbASL) in rats and mice, which uses a train of adiabatic pulses to label multiple boli of arterial water¹. mbASL can be described as a hybrid of PASL and CASL: either as a multiple PASL with limited labelling thickness or a large thickness CASL. We have been able to fully quantify this technique in mice and rats by describing mbASL signal as a sum of limited thickness PASL signals². Using a modification of Buxton’s kinetic model (Figure 1) to account for multiple boli slice-selective labelling, we were able to reproduce a range of mbASL signal distributions.

Methods

Variable labelling slice thickness mbASL experiments were performed using CD1 mice (n=6) & Wistar rats (n=6). The distance between labelling and imaging slice was kept constant. Exploration of the labelled bolus arrival at different post-label delay times used 1-shot mbASL with an inversion time (CI) of 500ms and a time after inversion TI= 50:500:4050ms (NA=10). CBF maps were produced using a T1 map and a 4-shot mbASL (NA=10, CI = 5000ms, TI = 50ms). mbASL was compared with FAIR, NA= 10, TI = 1750ms.

Results & Discussion

The number of pulses for optimal signal decreased as the thickness of the labelling slice increased, with the optimal values dependent on the animal strain. The arrival of the labelled bolus changes distribution as the slice thickness increases shifting from a CASL-like distribution to a PASL-like distribution. Figure 2 demonstrates this, with the peak signal changing from 50ms for a 4mm slice to 550ms and 1050ms for the larger thicknesses which is mirrored in the rat results. For the larger slice thickness in rats, we see the signal peak at multiple points which we believe is due to blood circulation from the heart. The signal seen in Figure 2 reflects the theoretical predictions (Figure 3) from the model that the ΔM distribution would change at different post label delays with increasing thickness. This confirms the hybrid nature of the mbASL sequence. The sequence has shown a large increase in overall signal and SNR compared to the standard FAIR sequence (Figure 4). The model has been used to produce CBF maps using acquired mbASL images and T₁ maps (Figure 5) in rats and mice. Early analysis is producing CBF values that agree with FAIR values and the literature^3,4. There is high potential for this sequence to be adapted for different animal strains and clinical use. With its adaptable nature for signal distribution, mbASL has potential to measure CBF non-invasively with a higher SNR and accuracy compared to the standard ASL sequences.

Conclusion

We have been able to fully quantify the mbASL sequence and produce quantitative CBF maps for mice and rats which agree with literature values^3,4. The sequence has shown significantly higher signal and SNR over the standard FAIR sequence in both mice and rats.

Acknowledgements

Many thanks to Jim Mullin for his technical help with the MRI scanning, and to the University of Glasgow and EPSRC for my funding.

References

¹Vallatos, A et al. (2017), ²Buxton, R. B. et. al. (1998), ³Muir, E. R. et al. (2008), ⁴Sicard, K. M et al. (2005).