Wave-Encoded Model-Based Deep Learning for Highly Accelerated Imaging with Joint Reconstruction
A recently introduced model-based deep learning (MoDL) technique successfully incorporates convolutional neural network (CNN)-based regularizers into physics-based parallel imaging reconstruction using a small number of network parameters. Wave-controlled aliasing in parallel imaging (CAIPI) is an e...
Main Authors: | , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-11-01
|
Series: | Bioengineering |
Subjects: | |
Online Access: | https://www.mdpi.com/2306-5354/9/12/736 |
_version_ | 1797461395410780160 |
---|---|
author | Jaejin Cho Borjan Gagoski Tae Hyung Kim Qiyuan Tian Robert Frost Itthi Chatnuntawech Berkin Bilgic |
author_facet | Jaejin Cho Borjan Gagoski Tae Hyung Kim Qiyuan Tian Robert Frost Itthi Chatnuntawech Berkin Bilgic |
author_sort | Jaejin Cho |
collection | DOAJ |
description | A recently introduced model-based deep learning (MoDL) technique successfully incorporates convolutional neural network (CNN)-based regularizers into physics-based parallel imaging reconstruction using a small number of network parameters. Wave-controlled aliasing in parallel imaging (CAIPI) is an emerging parallel imaging method that accelerates imaging acquisition by employing sinusoidal gradients in the phase- and slice/partition-encoding directions during the readout to take better advantage of 3D coil sensitivity profiles. We propose wave-encoded MoDL (wave-MoDL) combining the wave-encoding strategy with unrolled network constraints for highly accelerated 3D imaging while enforcing data consistency. We extend wave-MoDL to reconstruct multicontrast data with CAIPI sampling patterns to leverage similarity between multiple images to improve the reconstruction quality. We further exploit this to enable rapid quantitative imaging using an interleaved look-locker acquisition sequence with T<sub>2</sub> preparation pulse (3D-QALAS). Wave-MoDL enables a 40 s MPRAGE acquisition at 1 mm resolution at 16-fold acceleration. For quantitative imaging, wave-MoDL permits a 1:50 min acquisition for T<sub>1</sub>, T<sub>2</sub>, and proton density mapping at 1 mm resolution at 12-fold acceleration, from which contrast-weighted images can be synthesized as well. In conclusion, wave-MoDL allows rapid MR acquisition and high-fidelity image reconstruction and may facilitate clinical and neuroscientific applications by incorporating unrolled neural networks into wave-CAIPI reconstruction. |
first_indexed | 2024-03-09T17:18:49Z |
format | Article |
id | doaj.art-074f26bdaea24584b18e78774154e6d7 |
institution | Directory Open Access Journal |
issn | 2306-5354 |
language | English |
last_indexed | 2024-03-09T17:18:49Z |
publishDate | 2022-11-01 |
publisher | MDPI AG |
record_format | Article |
series | Bioengineering |
spelling | doaj.art-074f26bdaea24584b18e78774154e6d72023-11-24T13:20:06ZengMDPI AGBioengineering2306-53542022-11-0191273610.3390/bioengineering9120736Wave-Encoded Model-Based Deep Learning for Highly Accelerated Imaging with Joint ReconstructionJaejin Cho0Borjan Gagoski1Tae Hyung Kim2Qiyuan Tian3Robert Frost4Itthi Chatnuntawech5Berkin Bilgic6Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA 02129, USADepartment of Radiology, Harvard Medical School, Boston, MA 02115, USAAthinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA 02129, USAAthinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA 02129, USAAthinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA 02129, USANational Nanotechnology Center, Khlong Nueng, Klong Luang, Pathum Thani 12120, ThailandAthinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA 02129, USAA recently introduced model-based deep learning (MoDL) technique successfully incorporates convolutional neural network (CNN)-based regularizers into physics-based parallel imaging reconstruction using a small number of network parameters. Wave-controlled aliasing in parallel imaging (CAIPI) is an emerging parallel imaging method that accelerates imaging acquisition by employing sinusoidal gradients in the phase- and slice/partition-encoding directions during the readout to take better advantage of 3D coil sensitivity profiles. We propose wave-encoded MoDL (wave-MoDL) combining the wave-encoding strategy with unrolled network constraints for highly accelerated 3D imaging while enforcing data consistency. We extend wave-MoDL to reconstruct multicontrast data with CAIPI sampling patterns to leverage similarity between multiple images to improve the reconstruction quality. We further exploit this to enable rapid quantitative imaging using an interleaved look-locker acquisition sequence with T<sub>2</sub> preparation pulse (3D-QALAS). Wave-MoDL enables a 40 s MPRAGE acquisition at 1 mm resolution at 16-fold acceleration. For quantitative imaging, wave-MoDL permits a 1:50 min acquisition for T<sub>1</sub>, T<sub>2</sub>, and proton density mapping at 1 mm resolution at 12-fold acceleration, from which contrast-weighted images can be synthesized as well. In conclusion, wave-MoDL allows rapid MR acquisition and high-fidelity image reconstruction and may facilitate clinical and neuroscientific applications by incorporating unrolled neural networks into wave-CAIPI reconstruction.https://www.mdpi.com/2306-5354/9/12/736parameter mappingmodel-based deep learningwave-encodingwave-MoDL |
spellingShingle | Jaejin Cho Borjan Gagoski Tae Hyung Kim Qiyuan Tian Robert Frost Itthi Chatnuntawech Berkin Bilgic Wave-Encoded Model-Based Deep Learning for Highly Accelerated Imaging with Joint Reconstruction Bioengineering parameter mapping model-based deep learning wave-encoding wave-MoDL |
title | Wave-Encoded Model-Based Deep Learning for Highly Accelerated Imaging with Joint Reconstruction |
title_full | Wave-Encoded Model-Based Deep Learning for Highly Accelerated Imaging with Joint Reconstruction |
title_fullStr | Wave-Encoded Model-Based Deep Learning for Highly Accelerated Imaging with Joint Reconstruction |
title_full_unstemmed | Wave-Encoded Model-Based Deep Learning for Highly Accelerated Imaging with Joint Reconstruction |
title_short | Wave-Encoded Model-Based Deep Learning for Highly Accelerated Imaging with Joint Reconstruction |
title_sort | wave encoded model based deep learning for highly accelerated imaging with joint reconstruction |
topic | parameter mapping model-based deep learning wave-encoding wave-MoDL |
url | https://www.mdpi.com/2306-5354/9/12/736 |
work_keys_str_mv | AT jaejincho waveencodedmodelbaseddeeplearningforhighlyacceleratedimagingwithjointreconstruction AT borjangagoski waveencodedmodelbaseddeeplearningforhighlyacceleratedimagingwithjointreconstruction AT taehyungkim waveencodedmodelbaseddeeplearningforhighlyacceleratedimagingwithjointreconstruction AT qiyuantian waveencodedmodelbaseddeeplearningforhighlyacceleratedimagingwithjointreconstruction AT robertfrost waveencodedmodelbaseddeeplearningforhighlyacceleratedimagingwithjointreconstruction AT itthichatnuntawech waveencodedmodelbaseddeeplearningforhighlyacceleratedimagingwithjointreconstruction AT berkinbilgic waveencodedmodelbaseddeeplearningforhighlyacceleratedimagingwithjointreconstruction |