Synopsis

Free-breathing non-contrast enhanced ¹H MRI techniques show promise as a surrogate for ventilation imaging but more validation against methods which directly image lung ventilation is needed in patients. The aim of this work was to investigate the relationship between ¹H free-breathing and ³He breath-hold ventilation images in patients with CF. Phase-resolved functional lung (PREFUL) ¹H and ³He ventilation images were compared in 10 patients with CF. Qualitative similarities were observed between techniques to varying degrees. Ventilation defect percentage showed strong correlation between methods and with FEV₁, although median and inter-quartile range fractional ventilation values were not related.

Introduction

Free-breathing non-contrast enhanced ¹H MRI techniques based on Fourier decomposition show promise as a surrogate for ventilation imaging^1,2. However, there has been little validation performed against methods which directly image lung ventilation in patients.

Ventilation defect percentage (VDP) calculated from free-breathing ¹H ventilation imaging has shown moderate correlation with ³He VDP in patients with COPD³ and patients with asthma⁴. Fractional ventilation (FV) measured using free-breathing ¹H ventilation imaging has shown strong regional correlation with ¹⁹F dynamic washout MRI in patients with COPD⁵.

The aim of this work was to investigate the relationship between ¹H free-breathing and ³He breath-hold ventilation images in patients with cystic fibrosis (CF).

Methods

Ten patients with CF were scanned using a 1.5T whole body MRI system (GE HDx). Patients also underwent spirometry.

³He MRI: A mix (500-1000mL) of hyperpolarized ³He (25% polarization, 120-150mL) and N₂ was inhaled from functional residual capacity (FRC). Sequence parameters were: 3D coronal steady state free precession (SSFP), full lung coverage, voxel size=3.3x3.3x5-4x4x5mm, θ=10°, TE/TR=0.6/1.9ms, BW=167kHz. Coil: ³He transmit-receive vest coil (CMRS). ¹H anatomical images of the same imaging volume were acquired during the same 14s breath-hold.

Free-breathing ¹H MRI: 2D spoiled gradient echo (SPGR) dynamic images of a single coronal slice were acquired during 75 seconds of free-breathing. The slice was positioned where ventilation abnormality was evident on the ³He images. Sequence parameters were: voxel size=5x5x15mm, θ=5°, TE/TR=0.7/2.3ms, BW=167kHz, temporal resolution=301ms. Coil: 8-element chest coil (GE). Phase-resolved functional lung (PREFUL) analysis⁶ was performed, including registration, low-pass filtering and calculation of fractional ventilation.

Analysis: Three adjacent ³He image slices matching the thicker PREFUL image slice were combined by square-root sum of squares. ³He and PREFUL images were segmented using spatial fuzzy c-means thresholding⁷ with 5 clusters to calculate ventilation defect percentage, and 4 clusters defined ventilation. PREFUL fractional ventilation was calculated as FV_insp=(S_exp-S_insp)/S_exp for inspiration and values outside the lungs were excluded using the ¹H image segmentation. ³He FV was calculated as FV_i,j,k=V_IS_i,j,k/V_voxel∑S_i,j,k, where V_I=inhaled volume and V_voxel=voxel volume⁸, and the mean FV of the ³He matching image slices was used for comparison. Median and inter-quartile range (IQR) of FV were calculated. To test for similarity between FV distributions, the 2-sample Kolmogorov-Smirnov test was performed once median histogram values were matched to normalize for translation along the FV axis. Data were tested for normality and the appropriate correlation analyses were performed between measures. Bland-Altman analysis was used to assess agreement between ³He and PREFUL metrics.

Results and Discussion

Visual similarities were observed to varying degrees between methods (figure 1). Table 1 summarizes patient demographics and quantitative metrics.

PREFUL VDP correlated strongly with ³He VDP (r=0.89, p=0.0006, figure 2a). PREFUL VDP was lower than ³He VDP with a bias of 14.7% (figure 2b), suggesting that PREFUL VDP may be less sensitive to lung abnormalities than ³He VDP. Differences between PREFUL and ³He VDP were greater with increasing mean VDP (figure 2b).

Fractional ventilation maps and histograms provide quantitative values with no requirement for thresholding. There was no significant correlation between PREFUL and ³He median or IQR FV (figure 3e,f), although in some cases FV maps and histograms showed similar features (figure3a-d). FV median and IQR were smaller for PREFUL than ³He (figure 3g,h), consistent with the definition of FV as the ratio of fresh gas entering a volume unit to the end inspiratory size of the unit, and the differing lung volumes during image acquisition (FRC plus 500-1000mL for ³He and FRC plus 193-665mL estimated for PREFUL using a typical tidal volume of 7mL/kg bodyweight⁹).

PREFUL VDP, ³He VDP and ³He FV IQR correlated with FEV₁ z-score (figure 4). Strong correlations of the Kolmogorov-Smirnov statistic with FEV₁ z-score (figure 4d) and VDP suggest that ³He and PREFUL FV histograms become more dissimilar as disease severity increases. Differences between techniques may be due to the fundamentally different sources of image contrast, mechanisms of ¹H signal modulation due to respiratory motion and ³He inhaled gas density respectively. Study limitations are some error in matching planes between ¹H and ³He MRI, and differences in voxel size.

Conclusion

Acknowledgements

This work was funded by the Medical Research Council (MR/M008894/1) and the National Institute of Health Research (NIHR-RP-R3-12-027). The views expressed in this publication are those of the authors and not necessarily those of the National Health Service, the National Institute for Health Research or the Department of Health. Thanks to patients from Sheffield Children's Hospital, Sheffield Teaching Hospitals and University Hospital of South Manchester.

References

[1] Voskrebensev A, Greer M, Gutberlet M, Schonfeld C, Renne J, Hinrichs J, Kaireit T, Welte T, Wacher F, Gottlieb J, Vogel-Claussen J. Detection of lung allograft dysfunction using ventilation-weighted Fourier decomposition MRI. Am J Transplant 2018 18(8):2050-2060.

[2] Nyilas S, Bauman G, Pusterla O, Ramsey K, Singer F, Stranzinger E, Yammine S, Casaulta C, Bieri O, Latzin P. Ventilation and perfusion assessed by functional MRI in children with CF: reproducibility in comparison to lung function. J Cyst Fibros 2018, doi: 10.1016/j.jcf.2018.10.003.

[3] Capaldi DP, Sheikh K, Guo F, Svenningsen S, Etemad-Rezai R, Coxson HO, Leipsic JA, McCormack DG, Parraga G. Free-breathing pulmonary ¹H and hyperpolarized ³He MRI: comparison in COPD and bronchiectasis. Acad Radiol 2015 22(3):320-9.

[4] Capaldi DPI, Sheikh K, Eddy RL, Guo F, Svenningsen S, Nair P, McCormack DG, Parraga G. Free-breathing functional pulmonary MRI: response to bronchodilator and bronchoprovocation in severe asthma. Acad Radiol 2017 24(10):1268-76.

[5] Kaireit TF, Gutberlet M, Voskrebenzev A, Freise J, Welte T, Hohlfeld JM, Wacker F, Vogel-Claussen J. Comparison of quantitative regional ventilation-weighted fourier decomposition MRI with dynamic fluorinated gas washout MRI and lung function testing in COPD patients. J Magn Reson Imaging 2018 47(6):1534-41.

[6] Voskrebenzev A, Gutberlet M, Klimes F, Kaireit TF, Schonfeld C, Rotarmel A, Wacher F, Vogel-Claussen J. Feasiblity of quantitative regional ventilation and perfusion mapping with phase-resolved functional lung (PREFUL) MRI in healthy volunteers and COPD, CTEH, and CF patients. Magn Reson Med 2018 79(4):2306-14.

[7] Hughes PJC, Horn FC, Collier GJ, Biancardi A, Marshall H, Wild JM. Spatial fuzzy c-means thresholding for semiautomated calculation of percentage lung ventilated volume from hyperpolarized gas and 1H MRI. J Magn Reson Imaging 2018 47(3):640-6

[8] Tzeng YS, Lutchen K, Albert M. The difference in ventilation heterogeneity between asthmatic and healthy subjects quantified using hyperpolarized ³He MRI. J Appl Physiol 2009 106(3):813-22.

[9] Beardsell I et al: MCEM Part A: MCQs page 33, Royal Society of Medicine Press, 2009