Introduction

Image registration of lung CT images acquired at different inflation levels has been proposed as a surrogate method to map lung ‘ventilation’, based on the assumption that expansion and change in signal intensity of a given lung unit equates to ventilation. However, this technique requires validation against established ventilation modalities. Initial work has indicated the potential for clinically validating CT ventilation against hyperpolarised (HP) gas MRI¹, a direct measure of gas ventilation. However, variations in lung inflation have been shown to affect the ventilation distribution observed in HP gas images^2,3.

Purpose

To evaluate the impact of lung inflation levels when comparing CT ventilation imaging to gold-standards of ventilation from HP gas MRI.

Methods

7 patients with moderate-to-severe asthma underwent breath-hold CT at total lung capacity (TLC) and functional residual capacity (FRC). In addition, ³He-MRI and a same breath ¹H-MRI⁴ were acquired at two lung inflation levels, FRC+1L and TLC, and spirometry was performed. MRI was performed within 6 days of spirometry and CT with a mean±SD interval of 1.5±0.8 days and range of 1 to 4 days. For the CT ventilation calculation, FRC-CT was registered to TLC-CT via a diffeomorphic transform⁵. Next, TLC-CT and registered FRC-CT were used to compute a surrogate ventilation image from voxel-wise intensity differences in Hounsfield unit values⁶. The lung ventilation surrogate was then thresholded at the 10^th and 20^th percentiles, above which the lung was assumed to be ventilated. For direct comparison of CT and ³He-MRI ventilation, FRC+1L and TLC ³He-MRI were registered to TLC-CT indirectly via the corresponding same-breath ¹H-MRI data⁷ (figure 1). To test for variations in FRC+1L and TLC ³He-MRI, the percentage ventilated volume (%VV) was calculated for each patient by taking the ratio of binary segmented ³He-MRI and ¹H-MRI volumes⁸. %VVs for FRC+1L and TLC ³He-MRI and both CT ventilation percentiles were correlated via the Spearman correlation coefficient against the clinically established measure of forced expiratory volume in 1 second (FEV1) reported as percentage of predicted value. For spatial comparison of ³He-MRI and CT ventilation, Dice similarity coefficients (DSCs) were computed separately for the binary segmentations of CT ventilation with those of FRC+1L and TLC ³He-MRI, respectively. Statistical significance of differences was tested using the Wilcoxon signed-ranks test.

Results

The median (range) of %VV for FRC+1L and TLC ³He MRI were 90.5 (54.9-93.6) and 91.8 (67.8-96.2), respectively (p=0.018) (figures 2 and 3). Spearman correlation coefficients between FEV1 and %VV were 0.75 (p=0.05) for FRC+1L ³He-MRI, 0.82 (p=0.023) for TLC ³He-MRI, 0.86 (p=0.014) for the 10-100^th CT percentile and 0.86 (p=0.014) for the 20-100^th CT percentile. For the MRI versus CT ventilation comparison, visual evaluation of ventilation distributions (figure 4) and quantitative analysis via DSCs (figure 5) demonstrated improved correlation between CT and TLC ³He MRI when compared with FRC+1L ³He MRI with median (range) values of 0.93 (0.86-0.93) and 0.86 (0.68-0.92), respectively, for the 10-100^th percentile (p=0.017) and 0.87 (0.83-0.88) and 0.81 (0.66-0.87), respectively, for the 20-100^th percentile (p=0.027).

Discussion

Qualitative visual evaluation and quantitative analysis of ³He-MRI demonstrated increased and more homogenous ventilation at TLC when compared with FRC+1L. In several cases, ventilation defects depicting airway closures at FRC+1L opened up at TLC. Inflation-related variations in ventilation led to improved spatial agreement of CT ventilated volumes with ³He MRI at TLC. CT ventilation purports to measure local air volume change between two lung inflation states, whilst static ventilation-weighted ³He MR images represent a snapshot of the concentration of ³He gas within the air in the lungs at a given inflation level. As the CT-based surrogates of ventilation were derived from breath-hold CT acquired at FRC and TLC, they more closely depicted the distribution of ³He gas at TLC. High correlations with FEV1 were observed for all four MRI- and CT-derived %VV measurements studied. The slightly higher correlation of %VV derived from TLC ³He MR images when compared to FRC+1L images (0.82 vs 0.75) may be due to the breathing manoeuvre used for imaging at TLC which was more similar to that performed during measurement of FEV1, leading to similar patterns of airway opening. The higher correlation of CT ventilation with FEV1 than MRI with FEV1 may be attributable to the fact that PFTs were performed on the same day as CT.

Conclusion

This study demonstrates that comparison of CT ventilation and HP gas MRI varies with inflation state during HP gas MRI. If CT is acquired at FRC and TLC, a higher degree of spatial correlation with HP gas ventilation imaging can be achieved if the latter is acquired at TLC.

Acknowledgements

University of Sheffield James Morrison Fund, Sheffield Hospitals Charity, Weston Park Hospital Cancer Charity, National Institute of Health Research, Airprom and Novartis.

References

[1] Mathew L et al. Acad Radiol 2012;19:1546-53, [1] Sun Y et al. Respir Physiol Neurobiol 2009;169:44-9, [3] Marshall H et al. ATS 2013, [4] Wild J et al. NMR Biomed 2011;24:130-4, [5] Avants B et al. Neuroimage 2011;54:2033-44, [6] Int J Radiat Oncol Biol Phys 2005;62:630-634, [7] Tahir B et al. Phys Med Biol 2014;59:7267-77, [8] J Magn Reson Imaging 2017 doi: 10.1002/jmri.25804.