Comparison of air space measurement imaged by CT, small-animal CT, and hyperpolarized Xe MRI

Lung disease is the third leading cause of death in the western world. Lung air volume measurements are thought to be early indicators of lung disease and markers in pharmaceutical research. The purpose of this work is to develop a lung phantom for assessing and comparing the quantitative accuracy of hyperpolarized xenon 129 magnetic resonance imaging (HP 129 Xe MRI), conventional computed tomography (HRCT), and highresolution small-animal CT ( μ CT) in measuring lung gas volumes. We developed a lung phantom consisting of solid cellulose acetate spheres (1, 2, 3, 4 and 5 mm diameter) uniformly packed in circulated air or HP 129 Xe gas. Air volume is estimated based on simple thresholding algorithm. Truth is calculated from the sphere diameters and validated using μ CT. While this phantom is not anthropomorphic, it enables us to directly measure air space volume and compare these imaging methods as a function of sphere diameter for the first time. HP 129 Xe MRI requires partial volume analysis to distinguish regions with and without 129 Xe gas and results are within %5 of truth but settling of the heavy 129 Xe gas complicates this analysis. Conventional CT demonstrated partial-volume artifacts for the 1mm spheres. μ CT gives the most accurate air-volume results. Conventional CT and HP 129 Xe MRI give similar results although non-uniform densities of 129 Xe require more sophisticated algorithms than simple thresholding. The threshold required to give the true air volume in both HRCT and μ CT, varies with sphere diameters calling into question the validity of thresholding method.