-
Highly accelerated multishot echo planar imaging through synergistic machine learning and joint reconstruction
- Back
Metadata
Document Title
Highly accelerated multishot echo planar imaging through synergistic machine learning and joint reconstruction
Author
Bilgic B, Chatnuntawech I, Manhard MK, Tian QY, Liao CY, Iyer SS, Cauley SF, Huang SY, Polimeni JR, Wald LL, Setsompop K
Name from Authors Collection
Affiliations
Harvard University; Harvard Medical School; Harvard University; Massachusetts Institute of Technology (MIT); National Science & Technology Development Agency - Thailand; National Nanotechnology Center (NANOTEC); Massachusetts Institute of Technology (MIT)
Type
Article
Source Title
MAGNETIC RESONANCE IN MEDICINE
Year
2019
Volume
82
Issue
4
Page
1343-1358
Open Access
Green Accepted
Publisher
WILEY
DOI
10.1002/mrm.27813
Format
Abstract
Purpose: To introduce a combined machine learning (ML)- and physics-based image reconstruction framework that enables navigator-free, highly accelerated multishot echo planar imaging (msEPI) and demonstrate its application in high-vresolution structural and diffusion imaging. Methods: Single-shot EPI is an efficient encoding technique, but does not lend itself well to high-resolution imaging because of severe distortion artifacts and blurring. Although msEPI can mitigate these artifacts, high-quality msEPI has been elusive because of phase mismatch arising from shot-to-shot variations which preclude the combination of the multiple-shot data into a single image. We utilize deep learning to obtain an interim image with minimal artifacts, which permits estimation of image phase variations attributed to shot-to-shot changes. These variations are then included in a joint virtual coil sensitivity encoding (JVC-SENSE) reconstruction to utilize data from all shots and improve upon the ML solution. Results: Our combined ML + physics approach enabled R-inplane x multiband (MB) = 8-x 2-fold acceleration using 2 EPI shots for multiecho imaging, so that whole-brain T-2 and T-2* parameter maps could be derived from an 8.3-second acquisition at 1 x 1 x 3-mm(3) resolution. This has also allowed high-resolution diffusion imaging with high geometrical fidelity using 5 shots at Rinplane x MB = 9- x 2-fold acceleration. To make these possible, we extended the state-of-the-art MUSSELS reconstruction technique to simultaneous multislice encoding and used it as an input to our ML network. Conclusion: Combination of ML and JVC-SENSE enabled navigator-free msEPI at higher accelerations than previously possible while using fewer shots, with reduced vulnerability to poor generalizability and poor acceptance of end-to-end ML approaches.
Industrial Classification
Knowledge Taxonomy Level 1
Knowledge Taxonomy Level 2
Knowledge Taxonomy Level 3
Funding Sponsor
National Institute of Biomedical Imaging and Bioengineering [P41 EB015896, R01 EB017337, R01 EB019437, R01 EB020613, U01 EB025162]; National Institute of Neurological Disorders and Stroke [K23 NS096056]; National Institute of Mental Health [R01 MH116173, R24 MH106096]; Center for Biomedical Imaging [S10-RR023401, S10-RR023043]; NVIDIA GPU grant; NATIONAL INSTITUTE OF BIOMEDICAL IMAGING AND BIOENGINEERING [P41EB015896, F32EB026304, U01EB026996, R01EB017337, R01EB019437, R01EB020613, U01EB025162] Funding Source: NIH RePORTER; NATIONAL INSTITUTE OF MENTAL HEALTH [R01MH111419, R01MH116173] Funding Source: NIH RePORTER; NATIONAL INSTITUTE OF NEUROLOGICAL DISORDERS AND STROKE [K23NS096056] Funding Source: NIH RePORTER
License
N/A
Rights
N/A
Publication Source
WOS