Introduction

Hyperpolarized helium-3 (HP 3He) MRI is a useful tool for assessing disease related alterations in lung function and structure associated with the development and progression of chronic obstructive pulmonary disease (COPD)(1–3). HP 3He MR imaging can be used to obtain measurements of fractional ventilation (FV), which is defined as the ratio of inspired gas entering a region of the lung during inhalation to the total volume of gas present at end inhalation. Despite the fact that CT is considered to provide superior structural information about the lung compared to MRI, its capacity to provide functional information regarding gas filling and ventilation has never been directly compared to that of HP 3He MRI. Here, we used subjects from the national COPDGene study to explore the possibility of deriving CT-based FV measurements that could then be used to validate those same measurements obtained via HP 3He MRI.

Methods

We previously presented a multibreath hybrid HP 3HeMR imaging protocol which simultaneously provides the structural and functional parameters of ADC, FV and PAO2 in under 2 minutes (4). Subjects were recruited from the national COPDGene study, and completed pulmonary function testing (PFT) prior to the multibreath HP 3He MRI protocol in this study. MR imaging was performed in a whole body 1.5-T MRI system (Siemens MAGNETOM) using a flexible 8-channel (2×4 phased array) chest coil (Stark, Germany) and a 2D multi-slice GRE image set was acquired with 6 ~25mm coronal slices (adjusted to cover the whole lung with a 20% inter-slice gap), MS=6x48x36 (~8×8 mm2 planar resolution), α≈6°, TR/TE=7.0/3.3ms. Imaging acceleration was performed using GRAPPA (~2×). The FV image series consisted of 6 normoxic breaths (3He:N2:O2 ~1:3:1) at 5.5s intervals and a 7th breath for PAO2/ADC acquisition. Next, EE and EI CT images were obtained for each subject from COPDGene and the EI image was registered to the EE image using the ANTs toolbox. To derive FV measurements from these images, we utilized equations that have been previously published by Guerrero et al. and Mistry et al., which effectively calculate FV as the change in tissue density (expressed as Hounsfield units) within a voxel from EE to EI, while also correcting for respiration-induced changes in cardiac output and blood vessel distension within the lung (5,6).

Results & Discussion

Panels A and B of figures 1 and 2 show spin density maps as well as FV maps obtained during HP 3He MR imaging progressing anterior to posterior, while panel C shows FV measurements derived from the EE and EI CT images from one representative COPD GOLD 1 subject (FEV1/FVC= 70%)(Figure 1) and one COPD GOLD 3 subject (FEV1/FVC=40%)(Figure 2). In the COPD GOLD1 HP 3He MR images in figure 1a & b, ventilation is apparent across all regions of the lung in both spin density and FV maps, with some regions showing a degree of hyperinflation. A qualitative comparison of these images to the CT-derived FV images in Figure 1c shows that CT-derived FV appears to underestimate regions of hyperinflation (Figure 1b and 1c green arrow), while also incorrectly presenting properly ventilated regions as hypoinflated (Figure 1c. red arrow). In the HP 3He MR images from the COPD GOLD 3 subject in Figures 2 a and 2b, widespread ventilation defects can be seen across many lung regions (Figure 2a and 2b red, green, and purple arrow). Strikingly, these same ventilation defects are also seen in the CT-derived FV images in Figure 1c. However, although regions of hypoinflation can clearly be seen in both MR and CT images, regions of hyperinflation are once again underestimated throughout the lung in the CT images. Possible reasons for this discrepancy include potential CT image registration errors, as well as the different breathing maneuvers associated with EE/EI CT imaging vs. MR imaging. Additionally, the MRI and CT images generate two different sets of images with significantly different resolutions and operating characteristics, thus making a meaningful comparison between these two modalities challenging as cross-modality registration proves difficult. Finally, we note that the intensity-based assessment of FV in airways fails, since the HU value is unchanged despite near-complete gas replacement during the breath. Thus, a bias toward hypoinflation is expected, and may be present even in parenchymal regions containing unresolved small airways.

Conclusion

HP 3He MRI-derived FV measurements appear to depict gas distribution throughout the lungs more accurately than CT-derived FV measurements, which as a result do not prove effective for validating the former. This is an ongoing study whose ultimate conclusion is contingent upon comparing a greater number of subjects and investigating possible issues with image registration.

Acknowledgements

This work was funded by NIH R01-HL127969 04.