Comparison of CT-based lobar ventilation models with helium-3 MRI ventilation measurements in asthmatics

Rationale: CT models of ventilation can be computed from expiratory and inspiratory scans and the relative lung expansion at a lobar level can be used as a surrogate for ventilation. This ventilation model has previously been validated against SPECT [1]. Here we compare models of regional percentage lobar ventilation from CT with 3He MRI measurements of ventilation using multi-modality image registration in a population of 29 asthma patients. Methods: 29 patients with moderate-to-severe asthma underwent breath-hold CT at TLC and FRC and 3He MRI at FRC+1 liter. In the same breath, 1H MRI was acquired [2]. CT was used to segment the lobes by identification of major fissures in both lungs. These lobar masks were used to calculate percentage regional ventilation per lobe using the inspiration and expiration CT data [1]. 1H MRI was registered to inspiration CT via a diffeomorphic transform using ANTs [3]. As 1H and 3He MRI were acquired in the same breath, the transformation from 1H MRI to CT registration was applied directly to 3He MRI allowing direct comparison of lobar ventilation from CT and 3He. Figure. Scatter plot of 3He MRI and CT lobar ventilation measurements. Results: 3He MRI was successfully warped to inspiration CT using the transform derived from the 1H MRI and CT registration. The mean absolute difference between CT and 3He MRI ventilation across lobes was 4.8% (mean±SD for each lobe: RUL+RML=6.2±4.3%; RLL=5.1±5.4%; LUL=4.3±4.1%; LLL=4.8±3.7%). Greater ventilation was seen in upper lobes with 3He and in lower lobes with the CT model (see figure). This is confirmed by Bland-Altman analysis, 95% limits of agreement: RUL+RML (-4.5, 15.7), RLL (-16.3, 6.7), LUL (-5.5, 13.0), LLL (-12.5, 3.4). Discussion: The mean absolute difference of 4.8% observed is larger than the 2.1% reported previously in a smaller cohort of mild asthmatics against SPECT [1]. The larger difference here may be due to increased asthma severity and differences in the nature of the ventilation measurements (e.g. different gases, image resolutions and registration). Conclusion: This study compares percentage regional ventilation per lobe calculated by CT models with a more direct measure of gas ventilation with hyperpolarized 3He MRI. Acknowledgements: The EU FP7 AIRPROM consortium and Novartis provided funding for patient data. BT is funded by the University of Sheffield’s James Morrison Fund and Weston Park Hospital Cancer Charity. References: [1] Radiol 2010;257:854-862, [2] NMR Biomed 2011;24:130-134, [3] Neuroimage 2011;54:2033-2044