Investigation of lung mechanics using CT scan analysis.

Student thesis: Master Thesis and HD Thesis

  • Mads Holm Andersen
Patients with Acute Lung Injury and Acute Respiratory Distress Syndrome depend on ventilator therapy to survive. Finding correct ventilator settings for these patients is complicated as a balance must be obtained between ensuring sufficient gas exchange whilst preventing Ventilator Induced Lung Injury. The static PV curve and CT scanning can offer information about the evolution of the disease. However, the shape of the PV curve is poorly understood and CT scanning cannot be performed repeatedly under clinical conditions to gain a better understanding. Several approaches have been taken to model the static PV curve. Amongst them a model based on alveoli compartments that is able to simulate the distribution of open, collapsed and overdistended alveoli when fitted to PV data. The purpose of this project is to investigate the relationship between the alveolar states extracted from CT scans and simulated alveolar states from the model. Both CT scans and PV curves are obtained from different pigs, with and without OA damaged lungs and with different applied PEEP levels. An image segmentation algorithm which extracts information regarding the alveolar states from CT scans is developed and presented. Results show that there is no relationship between the optimal pressure range identified by the model and extracted from the CT scans. Results furthermore show that there is no relationship between the progression of collapsed alveoli along the entire static PV curve identified by the model and extracted from the CT scans. It has furthermore been shown that not all of the obtained values of the parameters used to obtain the best fit of the model to PV data were within the normal physiological range. It has also been shown that the model has a lack of ability to simulate the true shape of the static PV curve which may be considered as a major limitation. It can be concluded that there is no relationship between the simulated alveolar states and the extracted alveolar states from the algorithm.
Publication date2008
Number of pages112
Publishing institutionInstitut for sundhedsteknologi
ID: 14381140