View Abstract

A4671 - Quantifying the Effect of Sub-resolution Ventilation Heterogeneity on Oxygen Enhanced Specific Ventilation Measurements
Author Block: W. Kang1, G. Prisk2, R. C. Sa3, M. H. Tawhai4, K. S. Burrowes5; 1Auckland Bioengineering Institute, The University of Auckland, Auckland, New Zealand, 2Medicine, Univ of California At San Diego, La Jolla, CA, United States, 3Medicine, University of California, San Diego, La Jolla, CA, United States, 4Auckland Bioengineering Institute, Auckland, New Zealand, 5Department of chemical and materials engineering, University of Auckland, Auckland, New Zealand.
Rationale
Specific ventilation imaging (SVI) uses MRI to detect the rate of signal change during cyclical wash-in/out of O2 to measure specific ventilation (SV) in the lung. The spatial resolution of this technique is approximately 1 cm3, encompassing 5-10 acini per voxel. Thus, the potential for heterogeneity at a sub-voxel scale exists in such measurements. We applied a computational model to quantify the impact of sub-resolution ventilation () heterogeneity on SVI measurements.
Method
We used an anatomically-based airway geometry with 27 acini, stacked together to resemble a cuboid section of lung volume (~2.5 cm3 in volume). Each acinus had identical end-expiratory lung volume, capillary blood volume, perfusion, and membrane properties, but with differing tidal volume, resulting in different SV. A gas exchange model was used to predict the time course of O2 concentration (proportional to MR signal) following the SVI protocol of alternately breathing air and O2 (J Appl Physiol;116(8):1048-56, 2014). Acini were prescribed random values that resulted in different coefficients of variation in (COV: 0.1, 0.2, 0.3, 0.4 and 0.5), while keeping tidal volume constant for the overall model. Monte-carlo simulations with 1000 iterations were performed for each COV value. In each iteration, one 1 cm3 “imaging voxel” was randomly positioned, within the lung model, the true SV calculated, and “measured” SV calculated within the voxel using 50 distinct bins (ranging from 0.01-10) evenly spaced on a log-scale, matching the typical experimental analysis.
Results
Comparison of true SV with “measured” SV showed that there was a consistent underestimation of SV in the presence of increasing sub-resolution heterogeneity. The SVI analysis relies on the selection of the most appropriate SV based on a discrete number of bins. The extent of underestimation increased for units with higher SV, but remained within 1 SV bin up to a COV of 30%. This COV of 30% is representative of the range of V heterogeneities found in normal human lungs as measured by Musch et al. (J Appl Physiol; 93(5):1841-51), at a coarser resolution than the current study. SV underestimation did not exceed 2 SV bins even for the most extreme case of physiologically relevant heterogeneity considered (COV = 0.5).
Conclusions
The presence of sub-resolution heterogeneity in favors a slight underestimation in SV measured via the SVI technique.
This work was supported by a BRP NIH grant, number HL119263.