Advanced Imaging Algorithms with Position-Sensitive Gamma-Ray Detectors

Compton imaging;Cadmium Zinc Telluride;image reconstruction;filtered back-projection;maximum-likelihood;Nuclear Engineering and Radiological Sciences;Engineering;Nuclear Engineering & Radiological Sciences

Chu, JiyangMartin, William R ;

University of Michigan

关键词: Compton imaging;    Cadmium Zinc Telluride;    image reconstruction;    filtered back-projection;    maximum-likelihood;    Nuclear Engineering and Radiological Sciences;    Engineering;    Nuclear Engineering & Radiological Sciences;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/146036/chujy_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

The 3D position-sensitive Cadmium Zinc Telluride (CZT) detector can provide position and energy information of gamma-ray interactions, enabling imaging capability. Previous imaging methods are able to reconstruct distinguishable source locations. However, there exists a major contradiction in imaging problem eternally: information rate and reconstruction speed. The information rate represents how much information can be extracted from a measurement, including resolution, variance, and noise level. The reconstruction speed relies on the computing hardware heavily, but can be greatly accelerated by sophisticated reconstruction algorithms.To improve the combined performance of information rate and reconstruction speed, several imaging algorithms were investigated. Among them, the Simple Back- Projection (SBP) provides fastest reconstruction speed but least information due to its biased and blurred estimation. The Filtered Back-Projection (FBP) is almost as fast as SBP, but much more informative. However, better understanding of FBP is required to let it work for real data without too much manual intervention, thus the adjusted FBP and the adaptive FBP were developed in this thesis to make FBP practical. On the contrary, the Energy-Imaging Integrated Deconvolution (EIID) is a very informative reconstruction algorithm, but too slow to be applicable in many time-sensitive scenarios. The convergence rate of EIID should be accelerated without too much degradation to the information, thus the Energy Decremental Integrated Deconvolution (EDID) was developed in this thesis. Finally 3D image reconstruction is proposed, which provides a new dimension to interpret source distribution compared to traditional 4π spherical image reconstruction, and is more natural for people to understand. The image reconstruction code University of Michigan Imaging System (UMIS) was designed and developed during the research period. The principle of UMIS is to keep the code fast, simple, reusable, and suitable for research.