Synopsis

We apply a multi-TE ASL technique in the mouse brain, to separate intravascular and extravascular components of the ASL signal, as a non-invasive assessment of vascular permeability. Methodological development enabled the technique to reliably capture ASL signal compartmentation in the mouse brain, despite inherently low SNR. We report a significant decrease in intravascular fraction of the ASL signal from 0.66 (± 0.17) to 0.35 (± 0.10) as inflow time increases from 1000ms to 1500ms. This technique can be applied to transgenic mouse models of neurodegeneration, and a kinetic model can extract vascular permeability parameters from the ASL signal distribution.

Introduction

Methods

Images were acquired using an Agilent 9.4T imaging system with a two channel array surface coil (Rapid Biomedical). ASL data were acquired in female C57/B6 wild type (WT) mice at 18 weeks (n=6), using a flow-alternating inversion recovery (FAIR) sequence with a two shot segmented SE-EPI readout method. A multiple TE protocol was implemented at inflow times (TI) = 1000 ms and 1500 ms. Imaging parameters were: TE = 10, 12, 15, 18, 23, 30, 40, 50, 65 ms; FOV = 25 x 25 mm; matrix size = 32x32; TR = 5000 ms; repetitions = 20. Anaesthesia was induced at 2% isoflurane in a mixture of 1.0L/min medical air, and adjusted to maintain the respiration rate at ~100 bpm throughout the scans.

Analysis was performed in Matlab R2015a (Mathworks), with mean and standard error being reported across all subjects. The mean ASL signal intensity, ΔM, from a manually defined cortical region of interest, was extracted from the images at each TE. The data was evaluated using two models: a) a monoexponential model, $$$\triangle M= \exp(- \frac{TE}{T2^{app}})$$$ and b) a bi-exponential model, $$$\triangle M= \triangle M_{IV}\exp(- \frac{TE}{T2_{fast}}) + \triangle M_{EV}\exp(- \frac{TE}{T2_{control}})$$$, where T2app is apparent transverse decay, T2fast is the ‘fast-decaying’ compartment of the transverse decay that derives from the IV compartment and T2control is the transverse decay derived from the control signal (static tissue), ΔMIV is the intravascular component and ΔMEV is the extravascular component of the ASL signal. As such, for the bi-exponential model, all four fitted parameters are estimated from the acquired data 7.

Results

The mean apparent transverse relaxation, T2^app, measured by the monoexponential model, increases from 25.5 (± 1.4 ms at TI = 1000 ms) to 29.9 (± 1.7 ms at TI = 1500 ms) (p = 0.0023), reflecting the shift of labelled water from the IV to EV space at longer TI. The T2^app values shows a decrease compared to measurements in the rat cortex ⁷, which may reflect subtle physiological differences of the mouse brain.

Figure 1A shows that the transverse decay of the ASL signal in the mouse brain is well described by the bi-exponential model and provides evidence to support the feasibility of this approach to extract the intravascular, ΔMIV, and extravascular, ΔMEV, contributions to the ASL signal. Methodological improvements for the mouse brain include a larger range of TEs, to more reliably capture the fast decaying component of the bi-exponential model. Figure 1B shows that there is a reduction in the intravascular fraction of the ASL signal, from 0.66 ± 0.17 to 0.35 ± 0.10 as the TI increases from 1000 to 1500 ms (p = 0.0087). This reflects the gradual exchange of the labelled blood water into the tissue, giving a surrogate index of vascular permeability. The IV and EV signal contributions will be used to extract the tissue transit time using the kinetic model 8, to evaluate the permeability further. Figure 1B suggests that the technique is able to estimate the compartmental distribution of the ASL signal with good between-subject reproducibility and thus may provide a sensitive non-invasive biomarker of BBB pathology.

Conclusion

Acknowledgements

References