Fast and accurate PET image reconstruction on parallel architectures
Abstract/Contents
- Abstract
- Positron Emission Tomography (PET) has become an invaluable modality for in vivo imaging of molecular signatures of disease. The reconstruction algorithm for PET and the resulting image quality, however, have been limited by the large scale nature of the inverse problem, prohibiting sophisticated statistical algorithms from being used in practice. This work addresses the challenges in PET image reconstruction using state-of-the-art techniques in parallel computing, computer graphics, image processing, and convex optimization. More specifically, we reformulate the image reconstruction algorithm so that it can run efficiently on the massively parallel architecture of the Graphics Processing Unit (GPU). Memory access patterns and execution controls are optimized to maximize device resource utilization. Furthermore, strategies for extending the parallel processing to a cluster of GPUs is developed, taking the inter-node communication overhead into consideration. Finally, the drastic reduction of the computational cost enables practical applications of three methods for image quality and accuracy improvement, including using Point Spread Function (PSF) for resolution modeling, utilizing Compton scattered events in the detectors, and combining multiple reconstructions to achieve higher Signal-to-Noise Ratio (SNR) in motion correction tasks. Experiments show that the proposed methods are able to improve image quality and accuracy while reducing execution time. Part of the work has been integrated into the next generation Philips Time-of-Flight (ToF) PET systems.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2013 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Cui, Jingyu |
|---|---|
| Associated with | Stanford University, Department of Electrical Engineering. |
| Primary advisor | Levin, Craig |
| Thesis advisor | Levin, Craig |
| Thesis advisor | Nishimura, Dwight George |
| Thesis advisor | Xing, Lei |
| Advisor | Nishimura, Dwight George |
| Advisor | Xing, Lei |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Jingyu Cui. |
|---|---|
| Note | Submitted to the Department of Electrical Engineering. |
| Thesis | Thesis (Ph.D.)--Stanford University, 2013. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2013 by Jingyu Cui
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...