Physics-based reconstruction methods for magnetic resonance imaging

Wang, Xiaoqing; Tan, Zhengguo; Scholand, Nick; Roeloffs, Volkert; Uecker, Martin

doi:10.1098/rsta.2020.0196

Cited by 26 publications

(26 citation statements)

References 77 publications

(120 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, recent new reconstruction methods that use the high numbers of inversion points (like model-based or subspace) have the potential to reduce the differences in acquisition times between the 2 sequences. 50,51 Consequently, the CS-MP2RAGE sequence is a convincing tool within the panel of T 1 mapping sequences for the clinical MRI community. Another tremendously important feature is the intersubject variations of T 1 , which were extremely low in the current study, and thus may be beneficial for differentiating pathological from healthy tissue.…”

Section: Discussionmentioning

confidence: 99%

“…The MP2RAGE sequence is faster than standard Look-Locker–based T 1 mapping sequences, due to the ETL acceleration. However, recent new reconstruction methods that use the high numbers of inversion points (like model-based or subspace) have the potential to reduce the differences in acquisition times between the 2 sequences 50,51 . Consequently, the CS-MP2RAGE sequence is a convincing tool within the panel of T 1 mapping sequences for the clinical MRI community.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

The Compressed Sensing MP2RAGE as a Surrogate to the MPRAGE for Neuroimaging at 3 T

et al. 2022

View full text Add to dashboard Cite

Objectives: The magnetization-prepared 2 rapid acquisition gradient echo (MP2RAGE) sequence provides quantitative T 1 maps in addition to high-contrast morphological images. Advanced acceleration techniques such as compressed sensing (CS) allow its acquisition time to be compatible with clinical applications. To consider its routine use in future neuroimaging protocols, the repeatability of the segmented brain structures was evaluated and compared with the standard morphological sequence (magnetization-prepared rapid gradient echo [MPRAGE]). The repeatability of the T 1 measurements was also assessed. Materials and Methods: Thirteen healthy volunteers were scanned either 3 or 4 times at several days of interval, on a 3 T clinical scanner, with the 2 sequences (CS-MP2RAGE and MPRAGE), set with the same spatial resolution (0.8-mm isotropic) and scan duration (6 minutes 21 seconds). The reconstruction time of the CS-MP2RAGE outputs (including the 2 echo images, the MP2RAGE image, and the T 1 map) was 3 minutes 33 seconds, using an open-source in-house algorithm implemented in the Gadgetron framework.Both precision and variability of volume measurements obtained from CAT12 and VolBrain were assessed. The T 1 accuracy and repeatability were measured on phantoms and on humans and were compared with literature.Volumes obtained from the CS-MP2RAGE and the MPRAGE images were compared using Student t tests ( P < 0.05 was considered significant). Results: The CS-MP2RAGE acquisition provided morphological images of the same quality and higher contrasts than the standard MPRAGE images. Similar intravolunteer variabilities were obtained with the CS-MP2RAGE and the MPRAGE segmentations. In addition, high-resolution T 1 maps were obtained from the CS-MP2RAGE. T 1 times of white and gray matters and several deep gray nuclei are consistent with the literature and show very low variability (<1%). Conclusions: The CS-MP2RAGE can be used in future protocols to rapidly obtain morphological images and quantitative T 1 maps in 3-dimensions while maintaining high repeatability in volumetry and relaxation times.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

The Compressed Sensing MP2RAGE as a Surrogate to the MPRAGE for Neuroimaging at 3 T

et al. 2022

View full text Add to dashboard Cite

show abstract

“…This leads to a non-linear reconstruction problem which is challenging to solve, but enables efficient qMRI. Wang et al review model-based image reconstruction for different quantitative parameters and discuss several strategies to solve such problems [ 3 ].…”

Section: Contributions In This Issuementioning

confidence: 99%

Synergistic tomographic image reconstruction: part 1

Tsoumpas

Jørgensen

Kolbitsch

et al. 2021

Phil. Trans. R. Soc. A.

View full text Add to dashboard Cite

This special issue focuses on synergistic tomographic image reconstruction in a range of contributions in multiple disciplines and various application areas. The topic of image reconstruction covers substantial inverse problems (Mathematics) which are tackled with various methods including statistical approaches (e.g. Bayesian methods, Monte Carlo) and computational approaches (e.g. machine learning, computational modelling, simulations). The issue is separated in two volumes. This volume focuses mainly on algorithms and methods. Some of the articles will demonstrate their utility on real-world challenges, either medical applications (e.g. cardiovascular diseases, proton therapy planning) or applications in material sciences (e.g. material decomposition and characterization). One of the desired outcomes of the special issue is to bring together different scientific communities which do not usually interact as they do not share the same platforms (such as journals and conferences). This article is part of the theme issue ‘Synergistic tomographic image reconstruction: part 1’.

show abstract

“…In this work, we extend BART with a complete framework for nonlinear operators. The framework builds on our previous work on non-linear calibrationless parallel imaging [21] and physics-based reconstruction [22], and is now extended with automatic differentiation, additional building blocks for neural networks, and new optimization algorithms [23]. In combination with the powerful numerical backend, the non-linear operator framework can then be used to efficiently train neural networks.…”

Section: Introductionmentioning

confidence: 99%

Deep, Deep Learning with BART

Blumenthal¹,

Luo²,

Schilling³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Purpose: To develop a deep-learning-based image reconstruction framework for reproducible research in MRI. Methods: The BART toolbox offers a rich set of implementations of calibration and reconstruction algorithms for parallel imaging and compressed sensing. In this work, BART was extended by a non-linear operator framework that provides automatic differentiation to allow computation of gradients. Existing MRI-specific operators of BART, such as the non-uniform fast Fourier transform, are directly integrated into this framework and are complemented by common building blocks used in neural networks. To evaluate the use of the framework for advanced deep-learning-based reconstruction, two state-of-the-art unrolled reconstruction networks, namely the Variational Network [1] and MoDL [2], were implemented. Results: State-of-the-art deep image-reconstruction networks can be constructed and trained using BART's gradient based optimization algorithms. The BART implementation achieves a similar performance in terms of training time and reconstruction quality compared to the original implementations based on TensorFlow. Conclusion: By integrating non-linear operators and neural networks into BART, we provide a general framework for deep-learning-based reconstruction in MRI.

show abstract

Physics-based reconstruction methods for magnetic resonance imaging

Cited by 26 publications

References 77 publications

The Compressed Sensing MP2RAGE as a Surrogate to the MPRAGE for Neuroimaging at 3 T

The Compressed Sensing MP2RAGE as a Surrogate to the MPRAGE for Neuroimaging at 3 T

Synergistic tomographic image reconstruction: part 1

Deep, Deep Learning with BART

Contact Info

Product

Resources

About