Tidal changes on CT and progression of ARDS

Cereda, Maurizio; Xin, Yi; Hamedani, Hooman; Bellani, Giacomo; Kadlecek, Stephen; Clapp, Justin; Guerra, Luca; Meeder, Natalie; Rajaei, Jennia; Tustison, Nicholas J; Gee, James C; Kavanagh, Brian P; Rizi, Rahim R

The characterisation of lung injury as 'severe' using a threshold of −300 HU is arbitrary; indeed, other studies chose different thresholds (eg, −100 HU) to indicate fixed loss of aeration due to oedema and/or non-recruitable atelectasis. 6 However, we chose an empiric approach in animals, aiming to indicate locally worse disease in the context of relatively mild early injury. 12 In addition, the methodology relies on the accuracy of image alignment, which is problematic in ARDS due to significant tidal deformation. 9 Image warping may distort the lungs and lead to overestimation of a density range at the expense of another. Misaligning the border between inflated and non-inflated regions can thereby introduce bias increasing the prominence of unstable inflation. However, our measurements of percent aeration distribution were virtually identical in warped and non-warped images (online supplementary table 1), and this suggests limited bias. Furthermore, the image warping technique seems adequate to maintain positions of major airway landmarks in animals and humans (online supplementary figures 3 and 6) and to generate realistic displacement maps (online supplementary figure 4) ; nonetheless, validation is required to further develop our methodology.