Type
Text
Type
Dissertation
Advisor
Mueller, Klaus , Gu, Xianfeng | Mueller, Klaus | Gu, Xianfeng | Kaufman, Arie | Helm, Patrick
Date
2012-12-01
Keywords
ant colony optimization, Computed Tomography, GPU, verification, view suggestion, volume visualization | Computer science--Medical imaging and radiology
Department
Department of Computer Science
Language
en_US
Source
This work is sponsored by the Stony Brook University Graduate School in compliance with the requirements for completion of degree.
Identifier
http://hdl.handle.net/11401/71476
Publisher
The Graduate School, Stony Brook University: Stony Brook, NY.
Format
application/pdf
Abstract
Cone-beam CT (Computed Tomography) has become a major imaging technique thanks to its image-fidelity and scanning time. Scientists and practitioners frequently utilize volume visualization tools for diagnosis and decision-making. The thesis work presented here seeks to improve on the volume visualization pipeline for CT generated data. We summarize our contributions into three categories. Cone-beam CT scanners typically use analytical algorithms to reconstruct volumetric data. We studied the interpolation error of visualization tools and built a verifiable visualization tool and efficient data structure to enable users to enjoy interactive rendering speed to freely examine the high resolution data at minimal error. For the recently developed low-dose CT which suffers from either noisy or an insufficient number of X-ray projections, we proposed an optimization framework to determine effective parameters for the data denoising and volume reconstruction stage. We have devised an efficient method to optimize various parameters for iterative CT reconstruction using an ant colony optimization algorithm. We also developed an interactive user interface to visually explore various acquisition settings. Our preliminary results show that the learned parameters can be readily applied to similar scans with promising results. Lastly, we provide visual guidance which can boost user efficiency when exploring the data. For guided visualization, we propose a view suggestion framework rooted in high-dimensional feature space which does not rely on particular transfer functions or volume segmentations as an initial input. | 112 pages
Recommended Citation
Zheng, Ziyi, "Efficient Reconstruction and Visualization of CT Data" (2012). Stony Brook Theses and Dissertations Collection, 2006-2020 (closed to submissions). 682.
https://commons.library.stonybrook.edu/stony-brook-theses-and-dissertations-collection/682