Purpose

Multi-band (MB) echo planar imaging (EPI) for pseudo-continuous arterial spin labeling (PCASL) imaging has been recently developed^1,2. Particularly, high-resolution whole brain MB-EPI PCASL imaging study indicates that MB-EPI PCASL provides superior spatial and temporal perfusion signal-to-noise ratio (sSNR and tSNR) efficiencies and improves the accuracy of cerebral blood flow (CBF) quantification over regular EPI-PCASL³. However, dynamic B0 changes, caused by respiration and scanner frequency drift, leads to dynamic EPI image artifacts, such as pixel intensity changes and pixel location shifts in phase-encoding direction. Furthermore, dynamic frequency shift may potentially introduce variations in fat saturation and reduce labeling efficiency of PCASL. Due to simultaneous multi-slice (SMS) acquisition and increased scanner duty cycle of MB-EPI, these dynamic B0-induced artifacts can be more severed on MB-EPI PCASL with a high MB factor, which will negatively impact sSNR, tSNR, and perfusion quantification of MB-EPI PCASL. Dynamic B0 changes can be detected and corrected using navigator signals⁴ and methods such as correction of the dynamic off-resonance in k-space⁵. Real-time dynamic frequency correction has been demonstrated to improve data quality of single-voxel MR spectroscopy⁶. Dynamically updating excitation, fat saturation, and labeling RF pulse frequency of MB-EPI PCASL using real-time frequency feedback (FB) detected from navigator signals may help mitigate problems associated with dynamic frequency changes. In this study, we tested the hypothesis that dynamic frequency feedback can improve sSNR, tSNR, and CBF measurement reproducibility of MB-EPI PCASL.

Methods

With IRB approval and informed consent, MB-EPI PCASL imaging studies in 5 healthy volunteers were performed on a 3T MRI scanner (MAGNETOM Prisma, Siemens Healthcare, Erlangen, Germany) using a 32-channel head coil. Three MB-EPI PCASL protocols were compared: (1) No dynamic frequency feedback (B0 shimming volume covering both imaging and labeling planes) (2) Dynamic frequency feedback (B0 shimming volume covering both imaging and labeling planes) (3) Dynamic frequency feedback (B0 shimming volume covering only image planes). Imaging parameters include 3800 ms/14 ms TR/TE, multiband factor 6, 3.5x3.5x3.5 mm3 resolution, 36 slices, 20% slice gap, 2.0 sec delay time, fat saturation, 40 parried pCASL measurements plus 2 additional M0 measurements. Within each ASL series scan, additional 200 noise images were collected for noise measurement. Perfusion sSNR and tSNR were calculated from each subject. Reproducibility study was performed in 3 subjects. MB-EPI PCASL with and without dynamic frequency feedback were executed twice. Coefficient of variation (CV) between the two CBF maps was calculated. sSNR, tSNR, and CV measurements within grey and white matters between MB-EPI PCASL protocols were compared using paired t-tests with a=0.05.

Results

For all MB-EPI PCASL measurements, dynamic frequency feedback with B0 shimming volume covering imaging and labeling planes has the highest sSNR and tSNR. sSNR ratio and tSNR ratio normalized to dynamic frequency feedback with B0 shimming volume covering imaging and labeling planes are shown in Figure 1. For MB-EPI PCASL with B0 shimming volume covering imaging and labeling planes, sSNR is significantly higher in frequency feedback protocol than no frequency feedback protocol (P = 0.015 in grey matters and P < 0.002 in white matters (all n =5)). tSNR also shows significant improvement with dynamic frequency feedback over without frequency feedback when pooling data of grey and white matters together (P = 0.013, n =10), although it is not statistically significant for grey and white matters individually (P = 0.131 in grey matters and P = 0.061 in white matters (all n =5)). For MB-EPI PCASL with dynamic frequency feedback, sSNR and tSNR of measurements with B0 shimming volume covering only image planes are significantly less than those of measurements with B0 shimming volume covering both imaging and labeling planes (sSNR: P < 0.008 in grey matters and P < 0.005 in white matter tSNR: P = 0.038 in grey matters and P = 0.032 in white matters (all n= 5)). The calculated CBF maps with and without frequency feedback from one subject show comparable CBF distributions (Figure. 2). From the reproducibility study of CBF measurement, MB-EPI PCASL with dynamic frequency feedback shows significantly better reproducibility than no frequency feedback, i.e. smaller CV between repeated measurements, when grouping samples of grey and white matters together (5.1% vs 9.1%, P = 0.031, n = 6).

Discussions

Our data indicates dynamic frequency feedback benefits sSNR, tSNR, and CBF measurement reproducibility of MB-EPI PCASL. Frequency feedback helps maintain optimal labeling efficiency, achieve stable fat saturation, and correct scanner drift.

Conclusions

Dynamic frequency feedback to multiband (MB) EPI PCASL excitation, fat saturation, and labeling RF pulse frequency can improve sSNR, tSNR, and cerebral blood flow (CBF) measurement reproducibility of MB-EPI PCASL.

Acknowledgements

P41 EB015894, P30 ICC and Human Connectome Project (1U54 MH091657), UMF0003900, and UL1TR000114. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

References

1. Li et al. ISMRM 2014:4557. 2. Li et al. ISMRM 2014: 995. 3. Li et al. Neuroimage 2015. 4. Hu et al. MRM 1994. 5. Pfeuffer et al. MRM 2002. 6. Keating et al. MRM 2012.