4093

Improved efficiency and quantitative accuracy in hyperpolarized ¹²⁹Xe ventilation imaging using 2D spiral acquisition

Riaz Hussain¹, Joseph W Plummer^1,2, Abdullah S Bdaiwi^1,2, Matthew M. Willmering¹, Laura L Walkup^1,2,3,4, and Zackary I Cleveland^1,2,3,4
¹Center for Pulmonary Imaging Research, Division of Pulmonary Medicine, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, ²Department of Biomedical Engineering, University of Cincinnati, Cincinnati, OH, United States, ³Imaging Research Center, Department of Radiology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, United States, ⁴Department of Pediatrics, University of Cincinnati, Cincinnati, OH, United States

Synopsis

Keywords: Data Acquisition, Hyperpolarized MR (Gas)

Despite yielding high signal-to-noise ratio, conventional gradient recalled echo (GRE) sequence requires a relatively long breath-hold for ¹²⁹Xe ventilation imaging. 2D-spiral sequence enables faster imaging and offers possibility to correct regional B₁/flip-angle inhomogeneities and signal decay using keyhole reconstruction without making assumptions about bias texture that can obscure the physiology of interest. Here, GRE and 2D-spiral showed comparable image quality and revealed regional ventilation impairment in cystic fibrosis. Furthermore, flip-angle correction preserved signal variation due to underlying lung pathophysiology, including gravitation ventilation gradients and subtle defects. Thus, flip-angle corrections in 2D-spiral sequence may detect early and reversible disease-induced ventilation impairment.

Introduction:

Hyperpolarized ¹²⁹Xe MRI quantifies regional ventilation with high resolution, and has been applied to various obstructive lung diseases including asthma^1,2, lymphangioleiomyomatosis³, Chronic Obstructive Pulmonary Disease^4-6, and Cystic Fibrosis (CF)^7-10. Conventionally, slice-selective gradient recalled echo (GRE) was preferred for ventilation MRI, providing high signal-to-noise (SNR) and online image reconstruction. However, GRE requires ~10s for full lung coverage resulting in T₁-induced signal decay in the last encoded slices, and challenging breath-holds for children and patients with severe disease. Fortunately, equivalent lung coverage can be achieved in <3s using efficient spiral sequences^11-15.
However, GRE and spiral suffer coil-dependent B₁-inhomogeneities that cause spatially varying sensitivity—and worse—flip-angles (FAs) that cause spatially heterogeneous, RF-induced signal decay. These signal variations are independent of physiology and both obscure and mimic impaired ventilation. These biases must be corrected before images can be quantified—e.g., calculating ventilation defect percentage (VDP) after bias-field correction using the N4 algorithm in ANTS^16,17. Unfortunately, N4 correction assumes all low-frequency signal variations is bias and can obscure the ventilation heterogeneity, MRI is intended to measure¹⁸.
Here, we mitigate slow acquisition and non-physiological signal variation using 2D-spiral imaging. First, N4-bias-corrected 2D-spiral is shown to be comparable to conventional 2D-GRE for assessing ventilation in CF. A Bloch-equation-based analytical model was then combined with the spiral-keyhole reconstruction to map and correct regional FA variations due to heterogeneous B₁^19-22. FA-corrections mitigate coil-dependent signal variations, while preserving ventilation heterogeneity obscured by N4.

Methods:

Human Subjects: 24 subjects (12M/12F) with CF (8–22 years) underwent ¹²⁹Xe MRI after providing written informed consent. Studies were approved by our local Institutional Review Board and FDA IND-123577. ¹²⁹Xe was administered under supervision of a registered nurse with blood-oxygenation/heart-rate monitored throughout scanning.
MRI: A 9820A polarizer (Polarean plc, Durham, NC) polarized ¹²⁹Xe to 20-35% and MRI was performed using Philips-Ingenia 3T scanner and flexible chest coil (Clinical MR solutions, Brookfield, WI). MRI parameters: (2D-GRE) resolution=3x3mm², slice thickness=15mm, TR/TE=7.73/3.75ms, FOV=320×240mm², flip-angles=6-12⁰, acquisition window=3.7ms. (2D-spiral) resolution=3x3mm², slice thickness=15mm, TR/TE=12.6/1.52ms, FOV=320×320mm², spiral-interleaves=13-29, flip-angles=10-30^o, acquisition window=6.25ms. GRE images were reconstructed online. Spiral images were reconstructed and FA-corrected using custom Python scripts¹⁹.
Image Analysis: Ventilation was quantified in MATLAB (MathWorks, Natick, MA) via signal thresholding⁹. Voxels with signal <60% of whole-lung mean were included in ventilation defect percentage (VDP); voxels with signal >200% were included in hyperventilated percentage (HVP). VDP/HVP calculations were performed after N4 bias-field correction¹⁶ (shrink factor=1, iterations=100,50,50,50, tolerance=1x10^-10, spline parameter=200), or FA correction^{18, 19, 22}. In spiral–FA–correction, T₁ was treated as negligible (<0.2s for spiral-interleaves $$$N_s$$$<30), and regional FA, $$$\theta$$$, was extracted from two, temporally-resolved images reconstructed from a single k-space dataset via keyhole reconstruction. The fully sampled image was then scaled voxel-by-voxel by
$$ \gamma = \frac{1}{\theta} \frac{1}{sin(\theta)} \left[ \frac{1-cos(\theta))}{1- cos^{N_s}(\theta)} \right] $$
to correct spatially varying sensitivity and hyperpolarization decay. Gravitationally-dependent signal gradients were calculated from best-fit lines (mean signal vs position) in subjects with VDP<3%.
Statistical Analysis: Differences in VDP, SNR and gravitational gradients were assessed with Wilcoxon signed-rank test (significance: P<0.05). Skew between GRE and 2D-spiral data was assessed with Bland-Altman and Spearman’s correlation tests in RStudio.

Results and Discussion:

GRE and 2D-spiral provided similar sensitivity to regional lung function in CF, including homogeneous ventilation (e.g., circle, Figure 1a), hyperventilation (red arrows, Figure 1b and 1c), and impaired ventilation (white arrows, Figure 1b and 1c). There was no significant difference in signal with each sequence before bias correction (Figure 1d, P=0.16), but SNR was higher for GRE (Figure 1f, P<0.001), because of increased noise-like background artifacts in spiral (Figure 1e, P<0.001).
For N4 bias-field corrected GRE and 2D-spiral, there were no significant differences in either VDP or HVP (Figure 2a,d; P=0.51 for VDP; 0.69 for HVP). VDP and HVP from GRE and 2D-spiral strongly correlated (Figure 2c,f; ρ²=0.89 for VDP; ρ²=0.69 for HVP). Bland-Altman analysis (Figure 2b,e) showed no systemic skew (mean difference=0.33% for VDP and 0.0% for HVP).
Beyond providing more rapid—but comparable —ventilation sensitivity to GRE, 2D-spiral enables bias-correction using local FAs. N4 bias-field correction mitigates coil-dependent signal inhomogeneities (e.g., lung apex, Figure 3a) but spuriously eliminates physiological variability. For example, slice-by-slice analysis of uncorrected 2D-spiral images shows signal decreases posterior-to-anterior (slope=-0.025 signal arb. units/cm). This gradient originates in-part from the well-known gravitational dependence of ventilation²³, but is eliminated by N4-bias-field correction (Figure 3b). In contrast, gravitational dependence in signal was preserved following FA correction (Figure 4a: N4 slope=0; FA slope=–0.0024 signal arb. units/cm).
Moreover, N4 correction can mask pathophysiology. Figure 5a shows an example of pathologically hyper- and hypo-ventilated areas (red and white arrows, respectively) which were eliminated by N4 correction. In Figure 5b, N4 removed apical ventilation defects and in Figure 5c eliminated most defect volume in the right lung.

Conclusions:

2D-spiral yields comparable ¹²⁹Xe ventilation images to GRE in <1/3 of time, thus avoiding T₁-induced signal loss and enabling <3s breath-holds. Both sequences suffer from coil-dependent inhomogeneities but the N4 algorithm commonly used to alleviate this bias, obscures physiological heterogeneity due to gravitational ventilation gradients—and more importantly—small hyper- and hypo-ventilated regions, which are markers of early disease. In contrast, FA-correction mitigates coil-dependent signal variability without any additional data collection, while preserving underlying ventilation inhomogeneity.

Acknowledgements

The authors acknowledge research funding from Cincinnati Children’s Research Foundation, the NIH (R01HL151588, R01HL143011, R00HL138255), and the Cystic Fibrosis Foundation (NAREN19R0).

References

1. Svenningsen, S., et al., Hyperpolarized ³He and ¹²⁹Xe MRI: differences in asthma before bronchodilation. Journal of Magnetic Resonance Imaging, 2013. 38(6): p. 1521-1530.

2. Ebner, L., et al., Hyperpolarized ¹²⁹Xenon magnetic resonance imaging to quantify regional ventilation differences in mild to moderate asthma: a prospective comparison between semiautomated ventilation defect percentage calculation and pulmonary function tests. Investigative radiology, 2017. 52(2): p. 120-127.

3. Walkup, L.L., et al., Cyst ventilation heterogeneity and alveolar airspace dilation as early disease markers in lymphangioleiomyomatosis. Annals of the American Thoracic Society, 2019. 16(8): p. 1008-1016.

4. Matin, T.N., et al., Chronic obstructive pulmonary disease: lobar analysis with hyperpolarized ¹²⁹Xe MR imaging. Radiology, 2017. 282(3): p. 857-868.

5. Virgincar, R.S., et al., Quantitative analysis of hyperpolarized ¹²⁹Xe ventilation imaging in healthy volunteers and subjects with chronic obstructive pulmonary disease. NMR in Biomedicine, 2013. 26(4): p. 424-435.

6. Marshall, H., et al., A comparison of ¹²⁹Xe MRI and advanced lung function testing in patients with asthma and/or COPD: The NOVELTY ADPro substudy. European Respiratory Journal, 2021. 58(suppl 65): p. PA1872.

7. Kanhere, N., et al., Correlation of lung clearance index with hyperpolarized ¹²⁹Xe magnetic resonance imaging in pediatric subjects with cystic fibrosis. American journal of respiratory critical care medicine, 2017. 196(8): p. 1073-1075.

8. Rayment, J.H., et al., Hyperpolarised ¹²⁹Xe magnetic resonance imaging to monitor treatment response in children with cystic fibrosis. European Respiratory Journal, 2019. 53(5).

9. Thomen, R.P., et al., Hyperpolarized ¹²⁹Xe for investigation of mild cystic fibrosis lung disease in pediatric patients. Journal of Cystic Fibrosis, 2017. 16(2): p. 275-282.

10. Smith, L.J., et al., The assessment of short- and long-term changes in lung function in cystic fibrosis using ¹²⁹Xe MRI. European Respiratory Journal, 2020. 56(6): p. 2000441.

11. Abdullah S. Bdaiwi, M.L.C., Joseph W. Plummer, Matthew M. Willmering, Laura L. Walkup, and Zackary I. Cleveland, B₁-inhomogeneity Correction of Hyperpolarized ¹²⁹Xe Lung Ventilation Imaging Using Spiral, in International Society for Magnetic Resonance in Medicine. 2022, ISMRM: London.

12. Brandon Zanette, S.M., Yonni Friedlander, Felix Ratjen, and Giles Santyr, Comparison of Spiral and Cartesian Acquisitions for Rapid Hyperpolarized ¹²⁹Xe Ventilation Mapping in Pediatric Cystic Fibrosis Lung Disease, in International Society for Magnetic Resonance in Medicine. 2022, ISMRM: London.

13. Brandon Zanette, Y.F., Samal Munidasa, and Giles Santyr, Comparison of 3D Stack-of-Spirals and 2D Gradient Echo for Ventilation Mapping using Hyperpolarized ¹²⁹Xe, in International Society for Magnetic Resonance in Medicine. 2020.

14. Willmering, M.M., et al., Improved pulmonary ¹²⁹Xe ventilation imaging via 3D-spiral UTE MRI. Magnetic Resonance in Medicine, 2020. 84(1): p. 312-320.

15. Salerno, M., et al., Dynamic spiral MRI of pulmonary gas flow using hyperpolarized ³He: Preliminary studies in healthy and diseased lungs. Magnetic Resonance in Medicine, 2001. 46(4): p. 667-677.

16. Tustison, N.J., et al., N4ITK: Improved N3 Bias Correction. IEEE Transactions on Medical Imaging, 2010. 29(6): p. 1310-1320.

17. Niedbalski, P.J., et al., Protocols for multi-site trials using hyperpolarized ¹²⁹Xe MRI for imaging of ventilation, alveolar-airspace size, and gas exchange: A position paper from the 129Xe MRI clinical trials consortium. Magnetic Resonance in Medicine, 2021. 86(6): p. 2966-2986.

18. Lu, J., et al., Bias field correction in hyperpolarized ¹²⁹Xe ventilation MRI using templates derived by RF-depolarization mapping. Magnetic Resonance in Medicine, 2022. 88(2): p. 802-816.

19. Joseph W. Plummer, A.S.B., Mariah L. Costa, Matthew M. Willmering, and Zackary I. Cleveland, Laura L. Walkup, Flip-angle map calculation via keyhole reconstruction in uniform and variable-density 2D-spiral hyperpolarized ¹²⁹Xe MRI, in International Society for Magnetic Resonance in Medicine. 2022, ISMRM: London.

20. Miller, G.W., et al., Hyperpolarized ³He lung ventilation imaging with B₁-inhomogeneity correction in a single breath-hold scan. Magnetic Resonance Materials in Physics, Biology and Medicine, 2004. 16(5): p. 218-226.

21. Niedbalski, P.J. and Z.I. Cleveland, Improved preclinical hyperpolarized ¹²⁹Xe ventilation imaging with constant flip angle 3D radial golden means acquisition and keyhole reconstruction. NMR in Biomedicine, 2021. 34(3): p. e4464.

22. Niedbalski, P.J., et al., Mapping and correcting hyperpolarized magnetization decay with radial keyhole imaging. Magnetic Resonance in Medicine, 2019. 82(1): p. 367-376.

23. Hopkins, S.R., et al., Vertical gradients in regional lung density and perfusion in the supine human lung: the Slinky effect. Journal of Applied Physiology, 2007. 103(1): p. 240-248.

Figures

Figure 1: 2D GRE and spiral ¹²⁹Xe ventilation images in CF. (a) Homogenous ventilation (circles), (b) focal defects (white arrows), hyperventilated areas (red arrows), and (c) severely obstructed ventilation are readily observed using both GRE and spiral. (d) There was no significant difference in total lung mean signal between sequences (P=0.16). (e) GRE had significantly lower noise standard deviation (SD) (P<0.001). (f) Reduced noise resulted in higher SNR for GRE (P<0.001).

Figure 2: Comparison of ¹²⁹Xe VDP and HVP from GRE and spiral. Boxplots comparing VDP and HVP in (a) and (d), respectively. No significant differences between measurements from GRE or spiral were observed (VDP: P=0.51, HVP: P=0.69). VDP (b) and HVP (e) from GRE and 2D-spiral are significantly correlated (P<0.001; VDP: ρ²=0.89; HVP: ρ²=0.69). Bland-Altman analysis for (c) VDP and (f) HVP show no obvious skew with most data falling within the 95% confidence interval (shaded region).

Figure 3: Comparison of ¹²⁹Xe signal in 2D-spiral images after N4 and flip-angle (FA) based bias correction (12–year–old female). (a) Gray scale original, N4, and FA slices. White arrows denote signal loss at the lung apex from coil inhomogeneities, which are mitigated by N4 and FA correction. (b) Maps showing SNR distributions of original, N4- and FA-corrected images. Gravitationally dependent ventilation gradients generate higher signal in the posterior slices, but this well-known physiology is eliminated by N4.

Figure 4: Gravity-dependent signal in 2D-spiral. (a) Signal is reduced posterior-to-anterior for uncorrected (slope=-0.025 signal arb. units/cm) and FA-corrected images (slope=-0.0024 signal arb. units/cm), but not after N4-correction (slope=0, signal arb. units/cm). (b) Slope for all subjects with VDP<3% for original, N4, and FA-corrected images. Gradients are significantly reduced following N4 relative to FA-map correction (P<0.001). N4 gradient does not significantly differ from 0 (P=0.11).

Figure 5: 2D-spiral ¹²⁹Xe ventilation images showing defects and hyperventilated regions that are removed by N4 but preserved by FA correction. (a) Signal from hyperventilated areas (red arrow) are eliminated by N4, while defects are underestimated (white arrow). (b) Apical focal defects (white arrow) are overcorrected by N4 but preserved with FA correction. (c) Patient’s right lung ventilation is severely impaired, but defect volume is almost eliminated by N4.

Proc. Intl. Soc. Mag. Reson. Med. 31 (2023)

4093

DOI: https://doi.org/10.58530/2023/4093