Introduction

Diffusion-weighted (DW) MRI with inhaled hyperpolarized gases (³He and ¹²⁹Xe) is sensitive to changes in lung microstructure through the ADC and mean diffusive length scale (Lm_D) derived from the stretched exponential model (SEM) of gas diffusion in the lungs.^1,2 These parameters have also been shown to be affected by background field inhomogeneities due to the magnetic susceptibility difference at the air-tissue interface. Experiments with ³He DW-MRI demonstrated that background field inhomogeneities will increase derived ³He ADC values³ and are also field strength dependent.⁴ However, to date there have been no experiments exploring the background field inhomogeneity effects on ¹²⁹Xe DW-MRI. This work compares ¹²⁹Xe ADC and SEM diffusion model metrics in healthy volunteers at 1.5T and 3T. We hypothesis that the background field inhomogeneity effects on ¹²⁹Xe will be smaller than those observed with ³He due to the smaller gyromagnetic ratio.

Methods

Five healthy volunteers were imaged on 1.5T (GE HDx) and 3T (GE PET/MR 750W) scanners using quadrature vest coils (Clinical MR Solutions, Brookfield, Wisconsin, USA), tuned to respective Larmor frequencies of ¹²⁹Xe at 1.5T and 3T, with approval of the UK National Research Ethics Committee. For each scan, the same 3D multiple b-value ¹²⁹Xe DW-MRI sequence with compressed sensing⁵ and inhalation of a 1L bag (550mL ¹²⁹Xe, 450mL N₂) from function residual capacity was used. ¹²⁹Xe was polarized to >30% using a regulatory approved SEOP polariser (POLARIS, Sheffield, UK).⁶ Specific ¹²⁹Xe DW-MRI sequence parameters were: voxel size=6.25x6.25x15 mm³; TE/TR=14.0/17.4 ms; diffusion time=8.5 ms; b=[0,12,20,30] s/cm².

ADC and SEM metrics (Lm_D and alpha heterogeneity index) were calculated from reconstructed undersampled images as previously described in MATLAB.⁵ Bland-Altman comparison was used to assess differences between all mean global ¹²⁹Xe ADC (b=12, b=20 and b=30 s/cm²), Lm_D and alpha values at 1.5T and 3T.

Results and Discussion

Table 1 summarizes the mean global ¹²⁹Xe ADC, Lm_D and alpha values for all healthy volunteers at the two field strengths. ¹²⁹Xe ADC (b=12) was larger at 3T, when compared to 1.5T, in four of the five healthy volunteers; while ¹²⁹Xe Lm_D was larger at 3T for three volunteers. The same trend of larger values at 3T were observed for ¹²⁹Xe ADC (b=20) and ADC (b=30). Example ¹²⁹Xe magnitude (b=0), ADC and Lm_D images for one healthy volunteer at 1.5T and 3T are shown in Figure 1. The white arrows in the 3T magnitude image demonstrate more prevalent macroscopic susceptibility artifacts at the diaphragm region related to the air-tissue magnetic susceptibility differences.

Bland-Altman analysis of ¹²⁹Xe ADC and Lm_D found a mean bias of 6.3% towards 3T (95% confidence interval -2.7% to 15.3%) (Figure 2a), and a mean bias of 2.2% towards 3T (95% confidence interval -4.0% to 8.5%) (Figure 2b), respectively. These trends of increased ¹²⁹Xe DW-MRI metrics at 3T corroborate the trends observed previously with ³He ADC and mean chord length (Lm) from the cylinder model.⁴ However, the mean biases obtained for ¹²⁹Xe ADC and Lm_D are much smaller than those observed for ³He ADC (15.5%) and Lm (10.5%). This difference is most likely due to the ~2.7 times smaller gyromagnetic ratio of ¹²⁹Xe to that of ³He, resulting in less microscopic dephasing in the presence of the same background magnetic field inhomogeneity gradients.

While there is a small observed bias for ¹²⁹Xe ADC and Lm_D values at 3T, it is worth noting that these biases are within the same-week inter-scan repeatability 95% confidence interval of ¹²⁹Xe ADC (-6.3% to 7.4%) and Lm_D (-3.8% to 4.8%) reported for healthy volunteers.⁷ Therefore, the background field inhomogeneity effects for ¹²⁹Xe DW-MRI could be considered negligible when comparing DW-MRI results across scanners of different field strengths, and could support direct multi-site quantitative comparative studies across different platforms.

In contrast to ADC and Lm_D, the ¹²⁹Xe alpha value, a biomarker for the underlying acinar airway heterogeneity, had a mean bias of -0.5% towards 3T (95% confidence interval -5.3% to 5.0%) (Figure 2c), and therefore appears not at all field strength dependent. This further supports the previous experimental results that demonstrated ³He alpha value was insensitive to diffusion experiment conditions such as lung inflation level.⁸

This work has utilized a ¹²⁹Xe DW-MRI sequence with a TE of 14 ms that was previously optimized for 1.5T imaging. With a reported ¹²⁹Xe T₂* of 18 ms in partially-inflated lungs⁹, future work will involve optimization of the sequence for 3T to improve image SNR by reducing the TE whilst maintaining the same 8.5 ms ¹²⁹Xe diffusion time.

Conclusion

This work has compared ¹²⁹Xe ADC and Lm_D values at 1.5T and 3T in five healthy volunteers. A small bias towards increased ¹²⁹Xe ADC and Lm_D values at 3T was observed that is similar to ¹²⁹Xe DW-MRI inter-scan repeatability differences. While background field inhomogeneity effects exist for ¹²⁹Xe DW-MRI, they could be considered negligible thus supporting direct multi-site quantitative comparative studies across different scanners.

Acknowledgements

This work was supported by NIHR grant NIHR-RP-R3-12-027 and MRC grant MR/M008894/1. The views expressed in this work are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health