Approaches for respiratory motion compensation in positron emission tomography imaging
Clark, John W., Jr.
Doctor of Philosophy thesis
Respiratory patient motion often causes image blur and inaccurate quantitation results in positron emission tomography (PET) imaging. Traditional gated acquisition approach corrects motion artifacts by generating a snapshot sequence. Each snapshot, however, is of poor image quality due to insufficient photon counts. This dissertation addresses the problem of how to correct motion artifacts without sacrificing image quality or extending scan duration. To this end, a motion-incorporated model of the PET imaging system has been developed which enables utilization of data acquired during the entire scanning period in an image reconstruction process, thereby improving photon statistics while eliminating motion artifacts. Three related approaches within this framework are presented. In the first approach, motion information is first derived from a corresponding four-dimensional computed tomography image sequence, and then incorporated into the PET system model to get a motion-free image. The second approach develops a region of interest extension to the motion-incorporated system model to improve efficiency and accuracy of motion estimation. Finally, the motion-incorporated system model is integrated with an anatomy based image reconstruction algorithm, thereby enabling utilization of mismatched anatomical information during the image reconstruction process to improve image quality and quantitation performance.