0406

A time-efficient OGSE sequence with spiral readout for an improved depiction of diffusion dispersion

Eric Seth Michael¹, Franciszek Hennel¹, and Klaas Paul Pruessmann¹
¹Institute for Biomedical Engineering, ETH Zurich and University of Zurich, Zurich, Switzerland

Synopsis

Oscillating gradient spin-echo (OGSE) sequences typically suffer low diffusion sensitization and long echo times, thereby compromising the diffusion-to-noise ratio (DNR). To address these issues, recently proposed OGSE shapes providing increased diffusion sensitization were combined with single-shot spiral readout trajectories using a high-performance gradient system. This implementation was used to study the frequency dependence of diffusivity in the in-vivo human brain for OGSE frequencies up to 125 Hz and b-values up to 1000 s/mm², yielding an improved depiction of diffusion dispersion. The results indicate the significance of higher b-values in characterizing this relationship.

Introduction

Oscillating gradient spin-echo (OGSE) sequences offer additional insight into tissue microstructural properties compared to their pulsed gradient counterparts but are limited by relatively low diffusion sensitization. Recent works have overcome this sensitivity disadvantage by implementing OGSE with high-performance gradient systems,^1,2 but diffusion information remains susceptible to noise due to the lengthy echo time (TE) required (typically, 115 ms or more) to achieve both adequate diffusion sensitization and high OGSE frequencies.

Spiral readouts can reduce TE with respect to EPI readouts (the latter have exclusively been used in OGSE studies to date) owing to the center-out design of the former; as such, spirals provide a substantial signal-to-noise ratio (SNR) benefit.³ However, spirals are more sensitive to static B₀ inhomogeneities and gradient imperfections, which is likely why spirals are rarely used in practice⁴ and have not yet been used in OGSE acquisitions. In this work, spiral readouts were utilized in OGSE imaging of the in-vivo human brain for the first time, in an implementation that utilized a high-performance gradient insert⁵ and a recently developed diffusion sensitization scheme,² which was shown to improve sensitivity and specificity in OGSE measurements. This methodology was used to investigate the frequency dependence of diffusivity and the impact of flow compensation in OGSE waveforms for frequencies up to 125 Hz and b-values up to 1000 s/mm².

Methods

Experiments were performed with a 3T Philips Achieva system (Philips Healthcare, Best, the Netherlands) equipped with the abovementioned high-performance gradient insert. Two healthy volunteers and a bottle-shaped phantom were scanned with PGSE DTI and OGSE DTI sequences with spiral readouts at b-values of 300, 500, and 1000 s/mm² and frequencies up to 125, 100, and 75 Hz, respectively (10 slices, 3 mm slice thickness, 2 mm gap, 2 mm in-plane resolution, TR/TE = 5600/95 ms, five b = 0 acquisitions, 16 DTI directions at b > 0). OGSE scans were performed with improved oscillating gradient waveforms providing increased diffusion sensitivity.² Because these waveforms introduce a residual flow sensitivity, OGSE scans were repeated for a further modification to the shapes providing flow compensation (FC) (see Figure 1). The flow-compensated 50 Hz, 1000 s/mm² sequence was also performed using EPI readouts with the minimum achievable TE (128 ms) for an equivalent nominal resolution.

Following scanning, the readout k-space trajectories of each sequence were monitored using a field camera⁶ (Skope Magnetic Resonance Technologies, Zurich, Switzerland). Images were reconstructed using the resulting information, as well as off-resonance maps, in a reconstruction algorithm accounting for a third-order spherical harmonic model of the spatiotemporal magnetic field dynamics.⁷ Reconstructed images were smoothed with a Hamming filter and co-registered, and mean diffusivity (MD) was calculated voxel-wise based on diffusion tensor fits. Scaling factors were applied to MD values to account for possible deviations between the intended and applied b-values, determined empirically based on the phantom data. Among PGSE/OGSE scans at the same b-value, diffusion dispersion maps were computed via voxel-wise fitting of the power law relationship $$$D(\omega)=\Lambda\omega^\theta+D_{\omega=0}$$$, where $$$\Lambda$$$ is the diffusion dispersion rate, $$$D(\omega)$$$ is MD at frequency $$$\omega$$$, $$$D_{\omega=0}$$$ is the 0 Hz diffusivity, and structural disorder parameter $$$\theta$$$ = 0.5 is assumed.^8,⁹ The power law relationship was also fitted to average MD values of several white matter tracts.

Results

Figure 2 illustrates the TE and subsequent diffusion-to-noise ratio (DNR) advantage of spiral readouts with respect to EPI. Figure 3 depicts MD maps corresponding to OGSE acquisitions without FC; MD clearly increases with rising frequency and with decreasing b-value. Figure 4 captures these trends, fitted to the diffusion dispersion model, for select white matter regions and compares the diffusivity trends for OGSE shapes with and without FC. Figure 5 shows maps of the diffusion dispersion rate across b-values, again for OGSE sequences without FC; the clearest maps are seen for b = 1000 s/mm².

Discussion

In terms of resolving diffusion dispersion, the data indicate a clear advantage of higher b-values in the tradeoff between maximizing b-value or OGSE frequency for a given TE. The underlying MD maps at higher b-values exhibit a clear DNR advantage, thereby benefitting the stability of region-wise and, more significantly, voxel-wise fits of the diffusion dispersion model (see Figures 4-5) despite fitting to fewer sampled frequencies.

Noise sensitivity aside, lower b-values are associated with an increased sensitivity to perfusion in OGSE measurements,¹⁰ an effect which appears to have manifested in the data as higher MD values at lower b-values. As such, the b = 1000 s/mm² data likely contains contrast that best represents pure diffusion. A further reduction in perfusion bias could be expected by introducing flow-compensation.¹¹ The data, however, offer no evidence of this effect; there is no clear relationship between the MD values of OGSE data with and without FC. These findings suggest that increasing the b-value has a more significant impact than fully eliminating the first moment for the reduction of perfusion bias in OGSE experiments.

Conclusion

The OGSE implementation employed here permitted both a TE reduction and b-value improvement, forming an altogether time-efficient sequence that boosts DNR over an expanded range of b-values and frequencies. This scheme provided an enhanced depiction of diffusion dispersion and can enable further exploration of human brain microstructure.

Acknowledgements

No acknowledgement found.

References

1. Tan ET, Shih RY, Mitra J, et al. Oscillating diffusion-encoding with a high gradient-amplitude and high slew-rate head-only gradient for human brain imaging. Magn Reson Med. 2020;84(2):950-965.

2. Hennel F, Michael ES, Pruessmann KP. Improved gradient waveforms for oscillating gradient spin-echo (OGSE) diffusion tensor imaging. NMR Biomed. 2020;e4434. DOI: 10.1002/nbm.4434.

3. Lee Y, Wilm BJ, Brunner DO, et al. On the signal-to-noise ratio benefit of spiral acquisition in diffusion MRI. Magn Reson Med. 2020;00:1-14. DOI: 10.1002/mrm.28554.

4. Wilm BJ, Barmet C, Gross S, et al. Single-shot spiral imaging enabled by an expanded encoding model: Demonstration in diffusion MRI. Magn Reson Med. 2017;77(1):83-91.

5. Weiger M, Overweg J, Rösler MB, et al. A high-performance gradient insert for rapid and short-T2 imaging at full duty cycle. Magn Reson Med. 2018;79(6):3256-3266.

6. Dietrich BE, Brunner DO, Wilm BJ, et al. A field camera for MR sequence monitoring and system analysis. Magn Reson Med. 2016;75(4):1831-1840.

7. Wilm BJ, Barmet C, Pavan M, Pruessmann KP. Higher order reconstruction for MRI in the presence of spatiotemporal field perturbations. Magn Reson Med. 2011;65(6):1690-1701.

8. Novikov DS, Jensen JH, Helpern JA, Fieremans E. Revealing mesoscopic structural universality with diffusion. Proc Natl Acad Sci U S A. 2014;111(14):5088-5093.

9. Arbabi A, Kai J, Khan AR, Baron CA. Diffusion dispersion imaging: Mapping oscillating gradient spin-echo frequency dependence in the human brain. Magn Reson Med. 2020;83(6):2197-2208.

10. Wu D, Zhang J. The effect of microcirculatory flow on oscillating gradient diffusion MRI and diffusion encoding with dual-frequency orthogonal gradients (DEFOG). Magn Reson Med. 2017;77(4):1583-1592.

11. Ahlgren A, Knutsson L, Wirestam R, et al. Quantification of microcirculatory parameters by joint analysis of flow-compensated and non-flow-compensated intravoxel incoherent motion (IVIM) data. NMR Biomed. 2016;29(5):640-649.

Figures

Figure 1. Comparison of OGSE gradient waveforms and power spectra. All plots are in arbitrary units, and the gradient plots show effective shapes including the inversion by the refocusing RF pulse. Power spectra are pictured for a standard OGSE shape with a half-period gap (A), a reduced gap OGSE shape² (B), and a further modification to the latter producing FC (C). Note that the power spectra of (B) and (C) have higher peaks and smaller side lobes than that of (A). Also, the zero frequency peak of power spectrum (B) is eliminated in (C) by extending the innermost gradient lobes of (C) by T_FC.

Figure 2. Comparison of EPI and spiral sequences (A) and images (B). A: The minimum TE of an OGSE sequence with a spiral readout is shorter than that of an EPI readout because a spiral readout begins at the k-space center. B: Raw diffusion-weighted images (left), pure diffusion attenuation images (T₂-weighting removed) for a single DTI direction (center left), mean diffusion attenuation images over all DTI directions (center right), and MD maps (right). Note that images were acquired in separate scan sessions and do not show the same slice. Noise reduction is evident in the spiral images.

Figure 3. MD maps of a single slice of one subject for all b-values (different rows) and frequencies (different columns), for OGSE shapes without flow compensation. The 0 Hz images represent PGSE scans. MD is observed to increase as frequency increases (going from left to right) and as b-value decreases (going from bottom to top). A reduction in the graininess of white matter regions (blue/green areas) and a reduction in the extent of CSF regions (red areas) can be seen with rising b-value.

Figure 4. Example diffusion dispersion curves for white matter regions of one subject. Top row: The diffusion dispersion model is fit to region-wise average MD values, corresponding to OGSE shapes without FC. A clear reduction in diffusivity occurs as the b-value increases. Bottom row: The MD curves/points (from top row) are plotted in addition to data corresponding to OGSE shapes with no FC (NFC) for b = 300 and 1000 s/mm². The zero frequency MD values are the same for both FC and NFC. There is no apparent relationship between diffusivities acquired with or without FC.

Figure 5. Diffusion dispersion rate (Λ) maps of a single slice from each subject (different rows) for all b-values (different columns), for OGSE shapes without flow compensation. All sampled frequencies for each respective b-value are accounted for in computation. Voxels with negative Λ are masked. The underlying contrast is similar for all b-values but stabilizes as the b-value increases (going from left to right), as fewer pixels exhibit spurious intensities.

Proc. Intl. Soc. Mag. Reson. Med. 29 (2021)

0406