Purpose: This study aims to develop and validate a parametric response mapping (PRM) methodology to accurately identify diseased regions of the lung by using variable thresholds to account for alterations in regional lung function between the gravitationally-independent (anterior) and gravitationally-dependent (posterior) lung in CT images acquired in the supine position.
Methods: 34 male Sprague-Dawley rats (260-540 g) were imaged, 4 of which received elastase injection (100 units/kg) as a model for emphysema (EMPH). Gated volumetric CT was performed at end-inspiration (EI) and end-expiration (EE) on separate groups of free-breathing (n = 20) and ventilated (n = 10) rats in the supine position. To derive variable thresholds for the new PRM methodology, voxels were first grouped into 100 bins based on the fractional distance along the anterior-to-posterior direction. Lower limits of normal (LLN) for x-ray attenuation in each bin were set by determining the smallest region that enclosed 98% of voxels from healthy, ventilated animals.
Results: When utilizing fixed thresholds in the conventional PRM methodology, a distinct posterior-anterior gradient was seen, in which nearly the entire posterior region of the lung was identified as HEALTHY, while the anterior lung was labeled as significantly less so (t(29) = -3.27, p = 0.003). In both cohorts, %SAD progressively increased from posterior to anterior, while %HEALTHY lung decreased in the same direction. After applying our PRM methodology with variable thresholds to the same rat images, the posterior-anterior trend in %SAD quantification was removed from all rats and the significant increase of diseased lung in the anterior was removed.
Conclusions: The PRM methodology using variable thresholds provides regionally specific markers of %SAD and %EMPH by correcting for alterations in regional lung function associated with the naturally occurring vertical gradient of dependent vs. non-dependent lung density and compliance.
Keywords: Lung CT; PRM; Variable Thresholds; Volumetric Lung CT.
Copyright © 2021 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved.