Regional lung kinematics in an ovine model of acute respiratory distress syndrome under different ventilation strategies

This study characterizes regional lung kinematics through 4DCT in an ovine model of acute respiratory distress syndrome (ARDS) under two ventilation strategies: the ARDSNet low-stretch protocol and positive end-expiratory pressure individualized to maximal respiratory system compliance (PEEP_Cmax). Paired end-expiration (EE) and end-inspiration (EI) 4DCT scans from a surfactant-depletion lung injury sheep model were registered using a symmetric diffeomorphic algorithm in advanced normalization tools (ANTs) to compute voxelwise displacement fields. From these fields, we derived maps of regional volume change (J), a volume-independent distortion index (D), and cranio-caudal displacement. Both strategies produced broadly comparable whole-lung mean J at the whole-lung level, yet their spatial patterns diverged substantially. PEEP_Cmax strategy produced smooth, broadly distributed inflation following the diaphragm arc with coherent cranio-caudal motion, resulting 70.4% of voxels expanding between EE and EI whereas the ARDSNet low-stretch strategy produced fragmented displacement bands with only 60.1% expanding voxels, and elevated distortion concentrated at the boundaries of persistently collapsed tissue (median D = 1.154 versus 1.134 under PEEP_Cmax). Because mean J was nearly identical between strategies, the difference in distortion reveals that the lower PEEP under ARDSNet forced tissue at collapse boundaries into greater shape change to accommodate tidal volume into a smaller recruited fraction. Combining J and D distinguished regions of uniform aeration from those undergoing disproportionate shape change; heterogeneity that whole-lung surrogates such as tidal volume and driving pressure cannot resolve.