서브메뉴
검색
Optimizing Signal Sampling Strategies for Magnetic Resonance Imaging- [electronic resource]
Optimizing Signal Sampling Strategies for Magnetic Resonance Imaging- [electronic resource]
상세정보
- 자료유형
- 학위논문
- Control Number
- 0016935691
- International Standard Book Number
- 9798380374026
- Dewey Decimal Classification Number
- 004
- Main Entry-Personal Name
- Wang, Guanhua.
- Publication, Distribution, etc. (Imprint
- [S.l.] : University of Michigan., 2023
- Publication, Distribution, etc. (Imprint
- Ann Arbor : ProQuest Dissertations & Theses, 2023
- Physical Description
- 1 online resource(128 p.)
- General Note
- Source: Dissertations Abstracts International, Volume: 85-03, Section: B.
- General Note
- Advisor: Fessler, Jeffrey A.;Noll, Douglas C.
- Dissertation Note
- Thesis (Ph.D.)--University of Michigan, 2023.
- Restrictions on Access Note
- This item must not be sold to any third party vendors.
- Restrictions on Access Note
- This item must not be added to any third party search indexes.
- Summary, Etc.
- 요약Magnetic resonance imaging (MRI) is an important imaging modality in modern medicine that provides various biomarkers without harmful radiation. However, the low acquisition speed of MRI limits its spatiotemporal resolution, affordability, and patient experience. Efficient data acquisition schemes can enhance the quality and speed of MRI, thereby improving its scientific and clinical values.Designing efficient and optimal sampling schemes is an important yet challenging task due to the scale and complexity of the problem. Current research broadly adopts oversimplified models and heuristics-driven methods, which have effectively sped up MRI, but hinder the design of even more efficient sampling schemes. Therefore, there is a need to explore new methods to automatically design or tailor sampling trajectories.This dissertation presents new methods for optimizing MRI sampling trajectories using optimization-based and data-driven methods. We proposed gradient methods for optimizing non-Cartesian sampling trajectories, as well as Bayesian methods for optimizing Cartesian sampling trajectories. The proposed optimization methods can simultaneously improve image quality, hardware conformity, and patient comfort. Notably, these methods can automatically learn efficient sampling patterns from raw datasets, and the results are tailored to specific anatomical structures, scanning protocols, and hardware. To solve this large-scale, non-convex optimization problem, we introduced several computational methods, such as accurate Jacobian approximations for system matrices, parallel computing, and high-efficiency solvers.We evaluated the proposed methods in multiple MRI applications, such as structural imaging, functional imaging, and dynamic imaging. In both simulation and prospective in-vivo studies, our methods improved image quality by 2-5 dB (in PSNR) and increased acquisition speed by 8-10x. The optimization results also enhanced the patient experience by minimizing uncomfortable peripheral nerve stimulation. Notably, our learning-based approaches exhibited strong generalization ability and robustness to the shift between in-silico training and real-world prospective experiments, which can be explained by simple signal and system theories. These promising preliminary results highlight the potential of our methods to practically accelerate MRI and contribute to advancements in medical imaging research. To promote reproducible research, the accompanying codes, as well as an open-source reconstruction toolbox MIRTorch, are publicly available.
- Subject Added Entry-Topical Term
- Computer science.
- Subject Added Entry-Topical Term
- Medical imaging.
- Subject Added Entry-Topical Term
- Electrical engineering.
- Index Term-Uncontrolled
- Magnetic resonance imaging
- Index Term-Uncontrolled
- Machine learning
- Index Term-Uncontrolled
- Data-driven methods
- Index Term-Uncontrolled
- Dynamic imaging
- Index Term-Uncontrolled
- System matrices
- Added Entry-Corporate Name
- University of Michigan Biomedical Engineering
- Host Item Entry
- Dissertations Abstracts International. 85-03B.
- Host Item Entry
- Dissertation Abstract International
- Electronic Location and Access
- 로그인을 한후 보실 수 있는 자료입니다.
- Control Number
- joongbu:642037
MARC
008240221s2023 ulk 00 kor■001000016935691
■00520240214102006
■006m o d
■007cr#unu||||||||
■020 ▼a9798380374026
■035 ▼a(MiAaPQ)AAI30748362
■035 ▼a(MiAaPQ)umichrackham005029
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a004
■1001 ▼aWang, Guanhua.
■24510▼aOptimizing Signal Sampling Strategies for Magnetic Resonance Imaging▼h[electronic resource]
■260 ▼a[S.l.]▼bUniversity of Michigan. ▼c2023
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2023
■300 ▼a1 online resource(128 p.)
■500 ▼aSource: Dissertations Abstracts International, Volume: 85-03, Section: B.
■500 ▼aAdvisor: Fessler, Jeffrey A.;Noll, Douglas C.
■5021 ▼aThesis (Ph.D.)--University of Michigan, 2023.
■506 ▼aThis item must not be sold to any third party vendors.
■506 ▼aThis item must not be added to any third party search indexes.
■520 ▼aMagnetic resonance imaging (MRI) is an important imaging modality in modern medicine that provides various biomarkers without harmful radiation. However, the low acquisition speed of MRI limits its spatiotemporal resolution, affordability, and patient experience. Efficient data acquisition schemes can enhance the quality and speed of MRI, thereby improving its scientific and clinical values.Designing efficient and optimal sampling schemes is an important yet challenging task due to the scale and complexity of the problem. Current research broadly adopts oversimplified models and heuristics-driven methods, which have effectively sped up MRI, but hinder the design of even more efficient sampling schemes. Therefore, there is a need to explore new methods to automatically design or tailor sampling trajectories.This dissertation presents new methods for optimizing MRI sampling trajectories using optimization-based and data-driven methods. We proposed gradient methods for optimizing non-Cartesian sampling trajectories, as well as Bayesian methods for optimizing Cartesian sampling trajectories. The proposed optimization methods can simultaneously improve image quality, hardware conformity, and patient comfort. Notably, these methods can automatically learn efficient sampling patterns from raw datasets, and the results are tailored to specific anatomical structures, scanning protocols, and hardware. To solve this large-scale, non-convex optimization problem, we introduced several computational methods, such as accurate Jacobian approximations for system matrices, parallel computing, and high-efficiency solvers.We evaluated the proposed methods in multiple MRI applications, such as structural imaging, functional imaging, and dynamic imaging. In both simulation and prospective in-vivo studies, our methods improved image quality by 2-5 dB (in PSNR) and increased acquisition speed by 8-10x. The optimization results also enhanced the patient experience by minimizing uncomfortable peripheral nerve stimulation. Notably, our learning-based approaches exhibited strong generalization ability and robustness to the shift between in-silico training and real-world prospective experiments, which can be explained by simple signal and system theories. These promising preliminary results highlight the potential of our methods to practically accelerate MRI and contribute to advancements in medical imaging research. To promote reproducible research, the accompanying codes, as well as an open-source reconstruction toolbox MIRTorch, are publicly available.
■590 ▼aSchool code: 0127.
■650 4▼aComputer science.
■650 4▼aMedical imaging.
■650 4▼aElectrical engineering.
■653 ▼aMagnetic resonance imaging
■653 ▼aMachine learning
■653 ▼aData-driven methods
■653 ▼aDynamic imaging
■653 ▼aSystem matrices
■690 ▼a0574
■690 ▼a0544
■690 ▼a0984
■71020▼aUniversity of Michigan▼bBiomedical Engineering.
■7730 ▼tDissertations Abstracts International▼g85-03B.
■773 ▼tDissertation Abstract International
■790 ▼a0127
■791 ▼aPh.D.
■792 ▼a2023
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T16935691▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
■980 ▼a202402▼f2024