Simulating Supine and Upright Lung Tissue Mechanics using CT and Tomosynthesis Imaging

Abstract

Lung tissue is deformable, which enables it to in ate when we breathe, thus delivering oxygen to the blood supply. Damage of the lung tissue because of disease or the ageing process can a ect the lung function by altering the mechanical properties of lung tissue. However, changes to the distribution of lung tissue can also occur between di erent postures. This makes assessment of lung function from imaging di cult as very di erent conditions, with di erent implications for the patient can present similarly in imaging, and high-resolution imaging, which is acquired in the supine posture, cannot be directly related to normal function, which is generally upright. The availability of volumetric supine and upright lung imaging, presents new opportunities for lung analysis through reconciling the imaging data across postures. A methodological framework was developed towards the reconciliation of lung images obtained from computed tomography and tomosynthesis imaging in the supine and up- right postures. The lungs were extracted from each imaging modality through existing and custom segmentation algorithms. The supine lung shape was derived from tting a nite element method volume mesh to surface coordinates. Limitations of tomosyn- thesis imaging resolution meant a host mesh deformation was implemented to derive the upright lung shape from the supine lung shape. The majority of lung shape change between the supine and upright postures was found to be diaphragmatic changes. Lung tissue mechanics was simulated in both postures to determine if there were func- tional di erences between the two postures resultant from this shape change. The grav- itational density gradients from the upright posture tissue mechanics simulations were shown to be steeper than the ones seen in the supine simulations due to higher upright volumes. This con rms di erences in function between the supine and upright postures. The model also suggests that di erences may not be entirely due to volume change, as isotropic expansion of the lung between supine and upright volume does not result in the same tissue density distribution as simulations incorporating a shape change. The results will help with future image registration and reconciliation of imaging techniques, with the purpose of long term monitoring of the lung between postures.