Doctor of Philosophy
Dr. Ting-Yim Lee & Dr. Jerry Battista
Lung cancer has both a high incidence and death rate. A contributing factor to these high rates comes from the difficulty of treating lung cancers due to the inherent mobility of the lung tissue and the tumour. 4D-CT imaging has been developed to image lung tumours as they move during respiration. Most 4D-CT imaging methods rely on data from an external respiratory surrogate to sort the images according to respiratory phase. However, it has been shown that respiratory surrogate 4D-CT methods can suffer from imaging artifacts that degrade the image quality of the 4D-CT volumes that are used to plan a patient's radiation therapy.
In Chapter 2 of this thesis a method to investigate the correlation between an external respiratory surrogate and the internal anatomy was developed. The studies were performed on ventilated pigs with an induced inconsistent amplitude of breathing. The effect of inconsistent breathing on the correlation between the external marker and the internal anatomy was tested using a linear regression. It was found in 10 of the 12 studies performed that there were significant changes in the slope of the regression line as a result of inconsistent breathing. From this study we conclude that the relationship between an external marker and the internal anatomy is not stable and can be perturbed by inconsistent breathing amplitudes.
Chapter 3 describes the development of a image based 4D-CT imaging algorithm based on the concept of normalized cross correlation (NCC) between images. The volumes produced by the image based algorithm were compared to volumes produced using a clinical external marker 4D-CT algorithm. The image based method produced 4D-CT volumes that had a reduced number of imaging artifacts when compared to the external marker produced volumes. It was shown that an image based 4D-CT method could be developed and perform as well or better than external marker methods that are currently in clinical use.
In Chapter 4 a method was developed to assess the uncertainties of the locations of anatomical structures in the volumes produced by the image based 4D-CT algorithm developed in Chapter 3. The uncertainties introduced by using NCC to match a pair of images according to respiratory phase were modeled and experimentally determined. Additionally, the assumption that two subvolumes could be matched in respiratory phase using a single pair of 2D overlapping images was experimentally validated. It was shown that when the image based 4D-CT algorithm developed in Chapter 3 was applied to data acquired from a ventilated pig with induced inconsistent breathing the displacement uncertainties were on the order of 1.0 millimeter.
The results of this thesis show that there exists the possibility of a miscorrelation between the motion of a respiratory surrogate (marker) and the internal anatomy under inconsistent breathing amplitude. Additionally, it was shown that an image based 4D-CT method that operates without the need of one or more external respiratory surrogate(s) could produce artifact free volumes synchronous with respiratory phase. The spatial uncertainties of the volumes produced by the image based 4D-CT method were quantified and shown to be small (~ 1mm) which is an acceptable accuracy for radiation treatment planning. The elimination of the external respiratory surrogates simplifies the implementation and increases the throughput of the image based 4D-CT method as well.
Pierce, Greg, "Assessing and Improving 4D-CT Imaging for Radiotherapy Applications" (2011). Electronic Thesis and Dissertation Repository. 197.