Design and reconstruction of conical trajectories for motion-robust coronary magnetic resonance angiography
Abstract/Contents
- Abstract
- Coronary artery disease is the leading cause of death in the United States with more than half a million Americans suffering their first heart attack every year. X-ray catheterization is the standard method for disease diagnosis but is invasive and uses ionizing radiation. Magnetic resonance imaging (MRI) provides a noninvasive method for assessing the presence of stenoses in coronary arteries without the use of ionizing radiation. One of the major challenges with free-breathing coronary magnetic resonance angiography (CMRA) is respiratory motion. This dissertation introduces various technical innovations for advancing CMRA toward clinical relevance. These methods include a 3D phyllotaxis-cones trajectory design used for improved motion-robust cardiac imaging when using a balanced steady state free precession (bSSFP) sequence, an improved 3D image-based navigator (iNAV) design and reconstruction that reduces coherent aliasing artifacts caused by undersampling, and a novel deep learning (DL) approach for the accelerated reconstruction of 3D undersampled non-Cartesian datasets. For CMRA applications, bSSFP sequences are typically used due to the produced T2/T1-weighted signal. This leads to high contrast between blood and myocardium. For free-breathing CMRA, respiratory motion artifacts can be reduced by using view-ordering techniques where k-space coverage is well distributed for every heartbeat. Unfortunately for bSSFP sequences, eddy currents and associated artifacts are introduced when the k-space position is drastically varied between excitations. The proposed trajectory solves this problem by using a 3D phyllotaxis-cones design to sample a more distributed region of k-space during each heartbeat to improve motion robustness without introducing noticeable eddy current artifacts. The results from point spread function analysis, moving phantom studies, and in vivo scans demonstrated improved robustness to motion and superior coronary vessel sharpness when using the proposed phyllotaxis-cones design compared to the standard sequential-cones acquisition. Further techniques for respiratory motion artifact reduction involve monitoring and retrospectively correcting for motion. This can be achieved by using iNAVs, which are collected before and/or after segmented high-resolution data over multiple heartbeats. To obtain localized motion information, 3D iNAVs are acquired which allow for more sophisticated motion-correction techniques such as autofocusing. When collecting 3D iNAVs, a short temporal resolution and limited acquisition window create design challenges, and thus require using an undersampled variable-density (VD) approach. Furthermore, iterative techniques are then used to remove the aliasing artifacts that arise due to undersampling. The proposed technique in this dissertation uses an improved design and reconstruction for undersampled 3D (cones) iNAVs that reduces coherent aliasing artifacts. The 3D iNAV design was compared to a prior method by using point spread function analysis, simulated phantoms, and in vivo scans. The results showed decreased coherent aliasing artifacts for the proposed technique. Additionally in the in vivo motion-corrected images, coronary image quality was superior (or similar) after motion correction using the improved 3D iNAVs. DL has the potential for accelerating iterative reconstruction techniques used for undersampled reconstruction. Unfortunately, DL architectures are highly dependent on large amounts of training data which are not always readily available. If DL architectures incorporate physics specific to the application, the model has the potential for improved results with a finite amount of training data. The proposed DL approach uses an unrolled model architecture to accommodate non-Cartesian 3D k-space datasets by incorporating a non-uniform Fast Fourier Transform (NUFFT) operator with a convolutional neural network (CNN) to improve the results beyond the typical "black box" CNN approach. This technique decreases the sparse reconstruction time by one-twentieth on CPU and one-third on GPU when using previous state-of-the-art iterative reconstruction methods. Also, when applying motion correction using model-based 3D iNAVs compared to previous iterative reconstructed 3D iNAVs, coronary image quality remained the same. This confirmed that the unrolled model can properly generalize the iterative reconstruction algorithm while substantially reducing the reconstruction time. The new motion-robust k-space trajectory design, improved undersampled 3D iNAV design and reconstruction, and accelerated DL reconstruction techniques address several of the challenges with CMRA by reducing respiratory motion artifacts, reducing coherent aliasing artifacts, and accelerating reconstruction time. These advances can help put CMRA one step closer toward clinical relevance.
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource. |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2019; ©2019 |
| Publication date | 2019; 2019 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Malavé, Mario Octavio |
|---|---|
| Degree supervisor | Nishimura, Dwight George |
| Thesis advisor | Nishimura, Dwight George |
| Thesis advisor | Hu, Bob |
| Thesis advisor | Pauly, John (John M.) |
| Degree committee member | Hu, Bob |
| Degree committee member | Pauly, John (John M.) |
| Associated with | Stanford University, Department of Electrical Engineering. |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Mario Octavio Malavé. |
|---|---|
| Note | Submitted to the Department of Electrical Engineering. |
| Thesis | Thesis Ph.D. Stanford University 2019. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2019 by Mario Octavio Malave
- 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...