Prediction and Characterization of Lung Tissue Motion during Quiet Respiration

Purpose: The purpose of this dissertation is to quantitatively characterize and predict lung tissue motion with the goal of improving the local control of lung cancer. This is accomplished by producing a biomechanical model of lung tissue motion during quiet respiration. This dissertation proposes the development of algorithms and protocols for the analysis of motion information in 4DCT images.Methods: A cohort of 50 patients was acquired with a 16-slice CT scanner. This data was used throughout the dissertation. Based on the law of volume conservation, a relationship between the tidal volume and the geometric expansion of the torso was devised and used to improve breathing motion studies. The breathing patterns of these patients were used to characterize breathing patterns based on the measured external surrogate information with the aim of improving the efficiency of linear accelerator gating windows. A characteristic breath was defined as an average breath for use in generating patterns representative of realistic motion for breathing motion studies. A prospective gating algorithm was developed to allow the acquisition of user specified breathing phases with a relatively simple model to accurately predict respiratory phase occurrence in order to reduce the number of scans necessary to obtain sufficient data for breathing motion modeling. A new term to the breathing motion model to account for cardiac induced lung tissue motion was developed to improve the accuracy of the model.Results: Breathing studies can be optimized by placing the surrogate device between the third and fourth lumbar vertebra. Three types of breathing patterns were observed in the patient cohort. The hysteresis component of lung tissue trajectories was shown to be between 8 - 18 % of the volume filling component of motion. A simple prediction algorithm was shown to be a significant improvement over commercially available software. An additional term was devised to account for cardiac-induced lung motion and was shown to be accurate. Conclusion: This dissertation has demonstrated new quantitative methods to characterize lung tissue motion. Future work includes incorporating the work described in this dissertation into a new fast helical CT image acquisition protocol for breathing motion modeling