Evaluation of SLIce Dithered Enhanced Resolution Simultaneous MultiSlice (SLIDER-SMS) for human fMRI at 3T
An T. Vu1,2, Alex Beckett1, Kawin Setompop3, and David A. Feinberg1,2

1Helen Wills Neuroscience Institute, UC Berkeley, Berkeley, CA, United States, 2Advanced MRI Technologies, Sebastopol, CA, United States, 3Martinos Center for Biomedical Imaging, Charlestown, MA, United States

Synopsis

We evaluate the synergistic combination of super-resolution and SMS for high-resolution whole brain fMRI. We find that SLIDER-SMS can acquire high resolution, high CNR fMRI data at 3T which is normally only acquired at 7T. The regularized deblurring/reconstruction of SLIDER yielded 40% more BOLD CNR than normally acquired high resolution (HR) data, while omitting the deblurring step altogether yielded 100% more BOLD contrast with similar high k-space frequency tSNR. Future use of SLIDER for fMRI may enable robust columnar level results at 3T and allow higher spatial resolution fMRI investigations at 7T than currently possible.

Purpose

High isotropic resolution fMRI is challenging primarily due to low SNR, especially at lower field strengths. Recently, Simultaneous Multi-Slice (SMS) imaging with blipped-CAIPI (1, 2) has reduced scan time and improved SNR efficiency of fMRI. Similarly, super-resolution techniques (3, 4) utilizing sub-voxel spatial shifts in the slice direction, have increased both resolution and SNR efficiency. Super-resolution techniques like SLIDER (2) may be particularly promising for fMRI given that the thicker slices employed should enhance sensitivity to BOLD susceptibility changes. Here we evaluate the synergistic combination of super-resolution and SMS for high-resolution whole brain fMRI. Specifically: 1) How does SLIDER fMRI compare to normally acquired high-resolution fMRI data with and without post-processing blurring? 2) How does BOLD CNR and effective resolution (i.e. residual blurring) vary as a function of deblurring regularization parameter λ?

Methods

Data were acquired in three healthy volunteers on a Siemens 3T Trio using a 32 ch head array. During the 2D gradient echo EPI fMRI scans, subjects viewed 3 x 96 sec runs of flashing checkerboard (30 sec period) stimulus per MB-5 (“High res”) and SLIDER-2 MB-5 protocols. Imaging parameters were: 1.25 mm iso (nominal; 2.5 mm excitation thickness for SLIDER-2); FOV = 210x210x137.5 mm; PF = 6/8; TE = 45 ms; TR = 3000 ms (1500 ms per dithered volume); Flip angle = 84o (72o for SLIDER-2); PE direction= AP; axial oblique slices; and no in-plane undersampling. As a “High res blurred” control, the MB-5 (“High res”) data were 2-slice window averaged. For deblurring/reconstruction, the Toeplitz matrix was used as the forward model (T) and Tickhnov regularization was used to calculate the inverse model (Tinv) with λ ranging from 0 to 50% of the largest eigenvalue of T. In Matlab: Tinv = (V*S/(S*S+eye(size(S))*max(diag(S))*λ)*U'); where [U S V] = svd(T). Note larger λ’s result in greater residual blurring.

Results

Fig 1 shows the coronal cross section (of axially acquired slices) of a representative subject’s “High res” (HR), “High res blurred” (HRb), and SLIDER data deblurred with various λ values. Since HRb was blurred in post processing, the original HR image is recovered perfectly without regularization (λ=0). However, as with most super-resolution techniques in the presence of noise, deblurring SLIDER without regularization results in high spatial frequency noise amplification artifacts. SLIDER achieves best results with modest regularization (λ~0.1). Fig 2 shows BOLD CNR (t-values) for the corresponding Fig 1 datasets. SLIDER yields substantially stronger BOLD CNR than HR and HRb. The amount of slice blur as a function of λ is plotted in Fig 3 left (quantified as the spatial correlation of the averaged time series with that of the same image but shifted one slice down). The black vertical line denotes the λ (~0.08) where the effective SLIDER slice resolution matches that of HR. Error bars are SEM across subjects. The mean t-value as a function of λ is shown in Fig 3 right. HRb has the expected sqrt(2) increase over HR. SLIDER with λ~0.08 results in t-values at the level HRb. Without deblurring, SLIDER results in over double the t-values as HR. Given that the slice blur metric of Fig 3 may only be reflecting enhanced low relative to high spatial frequency information, we also calculated the average k-space tSNR in the lowest 50%, middle 25%, and highest 25% slice-axis k-space frequency regions (relative to the low k-space region of HR; Fig 4 left). Notably, SLIDER without deblurring, has significantly higher tSNR (p<0.05, paired-T(2)) in both low and mid k-space frequencies and similar high k-space tSNR compared to HR and HRb (Fig 4 right). This shows that the SNR benefits of thick slice excitation in SLIDER can offset the relative dampening of higher spatial frequencies. This is critical, as prior literature has shown that with high CNR, even methods with moderate amounts of blurring (e.g. in the PE direction due to T2* decay) can yield maps of columnar level structure (5). Fig. 5 shows preliminary results for 0.65 mm resolution ocular dominance mapping at 3T.

Conclusions

We find that SLIDER-SMS can acquire high resolution, high CNR fMRI data at 3T which is normally only acquired at 7T. The BOLD CNR of SLIDER was significantly greater than that of HR and even HRb due to the linear relationship between voxel volume and SNR (as opposed to square root when blurring in post processing) and the greater BOLD sensitivity with thicker slices. Future use of SLIDER for fMRI may enable robust columnar level results at 3T and allow higher spatial resolution fMRI investigations at 7T than currently possible.

Acknowledgements

NIH BRAIN Initiative grant - 1R24MH106096

References

1. Setsompop K. et al, NI 2012 2. Chen, Vu et al, NI 2015 3. Greenspan, H. et al, MRM 2002 4. Setsompop, K. et al, ISMRM 2015 5. Yacoub et al, PNAS 2008 6. Cheng, Waggoner, and Tanaka, Neuron 2001.

Figures

Figure 1. Average EPI images (coronal cross section of axially acquired images).

Figure 2. BOLD CNR.

Figure 3. Blurring (left) and BOLD CNR (right) as a function of deblurring regularization.

Figure 4. k-space tSNR along the slice axis. Left) k-space frequency regions used to calculate k-space tSNR. Right) k-space tSNR.

Figure 5. Single subject 3T fMRI at 0.65 mm isotropic (nominal). Scan parameters: SLIDER-2 (no deblurring), IPAT-2, MB-2, PF=6/8, TR=4000ms, TE=44ms, PE direction=HF, FOV=120x120x39 mm3, coronal slice acquisition. Total fMRI acquisition time: 18 mins. A) Single band reference axial cross section. B) F-value maps (p<0.001). C) Ocular dominance maps. Blue – left eye preference. Orange – right eye. Purple – no preference. Visual stimulus as in Cheng et al 2001.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
0942