| Notes: | Advisor: Xiaochuan Pan.
Thesis (Ph.D.)--The University of Chicago, Division of the Biological Sciences, and The Pritzker School of Medicine, Committee on Medical Physics, 2009.
Dissertation Abstracts International, Volume: 70-12, Section: B, page: 7519.
|
|---|
| Summary: | X-ray computed tomography (CT) is a dominant imaging tool in medical diagnosis, monitoring, and assessment. In many of the medical applications of CT, one is only interested in information about certain regions and/or organs within the imaged subject. Therefore, it is desirable to have scanning configurations, including x-ray source trajectories and illumination coverages, and reconstruction algorithms that yield images only in targeted regions of interest (ROIs) within the subject from reduced projection data. Most currently available CT imaging techniques accommodate a limited set of scanning configurations and, more importantly, require complete x-ray-illumination coverage of the subject cross sections, they are not applicable to ROI imaging. In this thesis, we have developed and investigated innovative scanning configurations and novel reconstruction algorithms for obtaining ROI images from reduced projection data in CT, which can reduce the scanning time, lower the radiation dose and scatter, and improve. Based upon the concept of chord, we have developed and evaluated algorithms for image reconstruction from reduced data acquired in circular fan-beam and helical cone-beam scans, which can exactly reconstruct an image within an ROI from reduced data. Moreover, the chord-based algorithms have been extended to the scans with some pratically useful trajectories and detector shapes. We have also studied the noise properties of chord-based image reconstructions from fan- and cone-beam projection data, and a rebinned algorithm has been proposed for improving the noise properties of our proposed algorithms. We have tailored these algorithms to the clinical data acquired from an advanced CT scanner with state-of-the-art technologies. We have tailored these algorithms to the clinical data acquired from an advanced CT scanner with state-of-the-art technologies. Finally, we have proposed a chordless algorithm for reconstructing images within an ROI from a reduced-scan circular sinusoidal scan.
|
|---|
