Fast motion-robust magnetic resonance imaging
Abstract/Contents
- Abstract
- Magnetic resonance imaging (MRI) is a non-invasive imaging modality with high value in medical imaging. MRI provides multi-contrast structural and functional information for accurate and efficient clinical diagnoses. However, compared to other medical imaging modalities, MRI has a relatively long acquisition process, thus it is usually sensitive to motion, such as respiratory motion, cardiac motion, flow motion, bowel motion, and bulk body motion. Apart from resulting in high motion-sensitivity, long scan times also impact the clinical workflow and patient comfort. Therefore, it is highly desirable to accelerate MRI and improve its motion-robustness. To enable fast motion-robust MRI, approaches for fast motion-robust T1-weighted and T2-weighted imaging were developed. Among these approaches, the proposed T2-weighted single-shot fast spin echo (SSFSE) imaging has been clinically deployed and thoroughly evaluated in clinical practice since 2017, providing useful diagnostic information for over 1,000 clinical patients. A deep-learning-based reconstruction technique was first developed for fast and robust image reconstruction of standard 2D Cartesian variable-density (VD) SSFSE acquisitions. Variational networks (VN) were trained using images reconstructed from 130 abdominal patients with standard parallel imaging and compressed sensing (PICS) and evaluated on another 27 abdominal patients. Image quality was evaluated in an independent blinded fashion by three radiologists in terms of overall image quality, perceived signal-to-noise ratio, image contrast, sharpness, and residual artifacts. Results showed that VN achieved improved perceived signal-to-noise ratio (P = 0.01) and improved sharpness (P< 0.001), with no difference in image contrast (P = 0.24) and residual artifacts (P = 0.07). In terms of overall image quality, VN performed better than did PICS (P = 0.02). Average reconstruction time±standard deviation was 5.60±1.30 seconds per section for PICS and 0.19±0.04 seconds per section for VN. This reconstruction time allows real-time image reconstruction of VD-SSFSE sequences for practical clinical deployment. Wave-encoded VD-SSFSE imaging with self-calibrating wave-encoding waveforms and reconstruction was developed to enable high acceleration and full k-space coverage. Wave-encoded variable-density sampling with self-refocusing waveforms was implemented to improve acquisition efficiency. Self-calibrated estimation of wave-encoding point-spread-function (PSF) and coil sensitivity was developed to improve motion-robustness. PICS reconstruction was used to reconstruct the highly accelerated datasets. The proposed method was tested on 20 consecutive patients and compared with standard Cartesian acquisition independently and blindly by two radiologists for noise, contrast, confidence, sharpness, and artifacts. Wave-encoded variable-density SSFSE significantly reduced the noise and improved the sharpness of the abdomen wall and the kidneys compared with standard Cartesian acquisition (P< 0.003). A 17% reduction in scan time was achieved using the proposed approach. Overall, the proposed approach achieves improved image quality with clinically relevant echo time and reduced scan time, thus enabling fast and robust 2D SSFSE imaging. A data-driven self-calibration and reconstruction technique of wave-encoded SSFSE was then proposed for computation time reduction and quality improvement. Data-driven calibration of wave-encoding PSF was developed using a trained deep neural network. Data-driven reconstruction was developed with another set of neural networks. Training of the calibration and reconstruction networks was performed on 15,783 2D wave-encoded SSFSE abdominal images. The proposed method was assessed on 29 consecutive adult patients. Image quality of the proposed data-driven approach was compared independently and blindly with conventional self-calibration and PICS reconstruction by two radiologists for noise, contrast, sharpness, artifacts, and confidence. Results showed that the proposed approach significantly reduced the perceived noise level (P< 0.0001) while maintaining non-inferior image contrast, sharpness, artifacts, and confidence compared to the conventional approach. An average 2.1-fold speedup in computation was achieved using the proposed method. Overall, this data-driven approach improves wave-encoded VD-SSFSE by reducing the perceived noise and improving the reconstruction speed. Finally, a comprehensive 3D wave encoding method for fast motion-robust free-breathing abdominal imaging was developed. Auto-calibration for wave encoding was designed to avoid extra scan for coil sensitivity measurement. Intrinsic butterfly navigators were used to track respiratory motion and perform localized rigid motion correction. Variable-density sampling was included to enable compressed sensing. Golden-angle hybrid radial-Cartesian view-ordering was incorporated to improve motion robustness. The proposed method was tested on six subjects and image quality was compared with standard accelerated Cartesian acquisition both with and without respiratory triggering. Inverse gradient entropy (IGE) and normalized gradient squared (NGS) metrics were calculated. For respiratory-triggered scans, wave-encoding significantly reduced residual aliasing and blurring compared with standard Cartesian acquisition (P = 0.04). For non-respiratory-triggered scans, the proposed method yielded significantly better motion correction compared with standard motion-corrected Cartesian acquisition (P = 0.001). Overall, the proposed method reduces motion artifacts and improves overall image quality of fast 3D free-breathing abdominal MRI. In conclusion, the proposed techniques have the potential to improve the speed and motion-robustness of MRI. Some of these techniques have been clinically applied to T2-weighted SSFSE imaging, and they have brought much benefit to patients by improving the scan efficiency and diagnostic accuracy of MRI.
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 | Chen, Feiyu |
|---|---|
| Degree supervisor | Pauly, John (John M.) |
| Thesis advisor | Pauly, John (John M.) |
| Thesis advisor | Nishimura, Dwight George |
| Thesis advisor | Vasanawala, Shreyas |
| Degree committee member | Nishimura, Dwight George |
| Degree committee member | Vasanawala, Shreyas |
| Associated with | Stanford University, Department of Electrical Engineering. |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Feiyu Chen. |
|---|---|
| Note | Submitted to the Department of Electrical Engineering. |
| Thesis | Thesis Ph.D. Stanford University 2019. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2019 by Feiyu Chen
- 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...