Introduction

Since its introduction in 1985, diffusion MRI (dMRI) has enabled the visualisation and characterisation of brain white matter, allowing scientists and clinicians to study and diagnose diseases including multiple sclerosis, stroke and dementia¹ and to study the brain’s structural connectivity (“wiring”). One of the main challenges of dMRI is its intrinsic low signal-to-noise ratio (SNR), which can be improved by scanning at ultra-high field strength². However, the increased RF transmit (B₁⁺) non-uniformity at 7T often results in spatially non-uniform signal magnitude across the brain, including signal dropouts in lower brain regions³. This is particularly problematic in dMRI since typical sequences rely on a high flip angle (nominally 180°) refocusing pulse. In this abstract, we tackle the challenge of B₁⁺ non-uniformity using RF parallel transmission (pTx) by demonstrating the use of dynamic pTx spokes pulses in dMRI (for the first time to our knowledge) and comparing performance against dMRI using only pulses played in the Circularly Polarised (CP+) mode that is equivalent to a traditional single-channel transmit system.

Methods

Data acquisition: One healthy volunteer (female, age = 28 years old) gave informed consent and was scanned in a 7T Terra scanner (Siemens) with an 8Tx/32Rx head coil (Nova Medical). A T₁ -weighted whole brain structural image was obtained with a MP2RAGE sequence at 0.75 mm isotropic resolution. dMRI data were collected with a single‐shot, spin‐echo, echo planar imaging (EPI) sequence modified to enable pTx pulses. Diffusion-weighting gradients were applied in 6 directions with a b value of 1000 s/mm²; TE = 60 ms; FOV = 225 × 225 mm²; voxel size = 1.5 × 1.5 × 1.5 mm³; 80 slices with 20% gap covering the entire brain; bandwidth = 1852 Hz/Px; GRAPPA acceleration factor R=3 and 6/8 partial Fourier. For each diffusion‐weighted acquisition, one image without diffusion weighting (b = 0) was also acquired with matching imaging parameters. The entire diffusion protocol was acquired using CP excitation/refocusing, and then repeated with dynamic pTx spokes for both excitation and refocusing. In CP mode, all 80 slices were acquired in one scan, however, due to the conversative SAR limits imposed in pTx mode, the 80 slices were acquired in 4 groups of 20 slices each. On top of that, to limit central brightening artifact in CP acquisition, the flip angles used in both CP and pTx acquisitions were reduced to 80° and 140° for excitation and refocusing respectively. An additional pair of EPI images were acquired for distortion correction purposes with matching imaging parameters but opposite phase encoding directions.

Pulse design: For the pTx acquisition, spokes pulses were designed separately for excitation and refocusing. B₀ field-maps and per-channel B₁⁺ field-maps were acquired using the default vendor provided sequences embedded in the pTx framework. The field-maps were exported and used for offline MATLAB pulse design. For the two superior imaging slabs, one set of pulses (excitation and refocusing) were designed per slab. For the two inferior imaging slabs, four sets of pulses (excitation and refocusing) were designed per slab to combat the greater variation in B₁⁺ in the inferior brain. Thus, a total of 10 sets of pulses (10 excitation pulses and 10 refocusing pulses) were designed across the 80 slices. i.e. one pair of pulses for each 20 slices in the superior slabs, and one pair of pulses for each 5 slices in the inferior half of the imaging volume. Each pulse consists of 3 spokes with positions determined using the inverse Fourier transform method. Channel weightings were optimised by a magnitude least square algorithm4. The pulses were then converted to variable-rate selective excitation (VERSE) pulses⁵ to reduce both SAR and pulse duration (Figure 1).

Data analysis: The DTI data were analysed in FSL. They were first distortion-corrected using ‘topup’⁶ before measures derived from the diffusion MR data, including frational anisotropy (FA) and mean diffusitivity (MD), were calculated using ‘dtifit’. For ROI analysis, the JHU white-matter tractography atlas⁷ was used and registration between the MNI 152 common space and the DTI images was performed via the T1 structural image using an affine transformation (using ‘flirt’)⁸.

Results

CP pulses yielded lower signal intensity and signal dropouts in the brain, especially in the cerebellum, inferior temporal lobes and brainstem (Figure 2). Tissue contrasts in these areas are subsequently recovered with pTx pulses. Violin plots also show increased signal intensities across all ROIs (Figure 3).

The signal dropouts in the CP acquisition translate into noisy depiction of the FA distribution in these regions (Figure 4). By contrast, the diffusion tracts in the pTx acquisition are much more defined, especially in the brainstem. The average FA values are also higher when using spokes pulses compared to CP pulses across all ROIs (Figure 5).

Discussion & Conclusions

The preliminary results from this experiment show the potential value of dynamic pTx pulses for dMRI. Compared to the standard CP pulses, pTx pulses increases signal across the whole brain potentially leading to a better definition of fibre orientations, especially in problematic areas such as the temporal lobes or cerebellum. This could allow improved visualisation of white matter architecture in those basal areas with ultra-high field strength.

Acknowledgements

BD is supported by Gates Cambridge Trust. CTR is funded by a Sir Henry Dale Fellowship from the Wellcome Trust and the Royal Society [098436/Z/12/B]. This study was funded by the NIHR Cambridge Biomedical Research Centre and MRC Clinical Research Infrastructure Award for 7T and has also received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 801075.