Accelerated Acquisition of High-resolution Diffusion-weighted Imaging of the Brain with a Multi-shot Echo-planar Sequence: Deep-learning-based Denoising

Kawamura, Motohide; Tamada, Daiki; Funayama, Satoshi; Kromrey, Marie‐Luise; Ichikawa, Shintaro; Oda, Hiroshi

doi:10.2463/mrms.tn.2019-0081

Cited by 29 publications

(33 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…15 To address this issue, further learning trained using undersampled data may be warranted. [7][8][9][10] In the current study, motion artifact played no role because the study participants were healthy. Therefore, the potential resistance of DLR-applied accelerated sequence to motion artifact may result in less artifact overall in the clinical situations.…”

Section: Discussionmentioning

confidence: 95%

“…Recent advances in deep learning have improved the computational cost, training time, and amount of data required, and have led to its application to medical imaging denoising, 1,2 as reported in several publications that have achieved robust denoising performances in magnetic resonance (MR) images. [3][4][5][6][7][8][9][10][11][12][13] However, most have been trained using brain images acquired on a 3T machine and applied to brain images also on a 3T machine. [3][4][5][6][7][8][9][10][11][12][13] However, clinically, the contrast of MR images can vary because of differences in the scan protocol, field strength, and anatomical location, and it is unclear whether this variability affects the performance of these denoising methods.…”

Section: Introductionmentioning

confidence: 99%

“…[3][4][5][6][7][8][9][10][11][12][13] However, most have been trained using brain images acquired on a 3T machine and applied to brain images also on a 3T machine. [3][4][5][6][7][8][9][10][11][12][13] However, clinically, the contrast of MR images can vary because of differences in the scan protocol, field strength, and anatomical location, and it is unclear whether this variability affects the performance of these denoising methods. 14 To ensure robustness against these variations, deep learning-based reconstruction (DLR) methods have been proposed to perform denoising only for high-frequency components that contain detailed information about structures and most of the noise, while leaving low-frequency components containing the image contrast information.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Applicability of deep learning-based reconstruction trained by brain and knee 3T MRI to lumbar 1.5T MRI

Kashiwagi

Tanaka

Yamashita

et al. 2021

Acta Radiologica Open

View full text Add to dashboard Cite

Background Several deep learning-based methods have been proposed for addressing the long scanning time of magnetic resonance imaging. Most are trained using brain 3T magnetic resonance images, but is unclear whether performance is affected when applying these methods to different anatomical sites and at different field strengths. Purpose To validate the denoising performance of deep learning-based reconstruction method trained by brain and knee 3T magnetic resonance images when applied to lumbar 1.5T magnetic resonance images. Material and Methods Using a 1.5T scanner, we obtained lumber T2-weighted sequences in 10 volunteers using three different scanning times: 228 s (standard), 119 s (double-fast), and 68 s (triple-fast). We compared the images obtained by the standard sequence with those obtained by the deep learning-based reconstruction-applied faster sequences. Results Signal-to-noise ratio values were significantly higher for deep learning-based reconstruction-double-fast than for standard and did not differ significantly between deep learning-based reconstruction-triple-fast and standard. Contrast-to-noise ratio values also did not differ significantly between deep learning-based reconstruction-triple-fast and standard. Qualitative scores for perceived signal-to-noise ratio and overall image quality were significantly higher for deep learning-based reconstruction-double fast and deep learning-based reconstruction-triple-fast than for standard. Average scores for sharpness, contrast, and structure visibility were equal to or higher for deep learning-based reconstruction-double-fast and deep learning-based reconstruction-triple-fast than for standard, but the differences were not statistically significant. The average scores for artifact were lower for deep learning-based reconstruction-double-fast and deep learning-based reconstruction-triple-fast than for standard, but the differences were not statistically significant. Conclusion The deep learning-based reconstruction method trained by 3T brain and knee images may reduce the scanning time of 1.5T lumbar magnetic resonance images by one-third without sacrificing image quality.

show abstract

Section: Discussionmentioning

confidence: 95%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Applicability of deep learning-based reconstruction trained by brain and knee 3T MRI to lumbar 1.5T MRI

Kashiwagi

Tanaka

Yamashita

et al. 2021

Acta Radiologica Open

View full text Add to dashboard Cite

show abstract

“…Intraoperative verification of tractography with or without SMS acceleration in patients requiring surgery is not trivial to obtain, but it could be the subject of a future study. There are additional methods for accelerating diffusion acquisitions that are being translated into clinical practice, including compressed sensing 27 and image denoising with 28 or without deep learning. 29 These techniques are not readily available for clinical implementation, but direct comparison of these diffusion-acceleration techniques and/or their combination also could be the subject of future study.…”

Section: Discussionmentioning

confidence: 99%

Simultaneous Multislice for Accelerating Diffusion MRI in Clinical Neuroradiology Protocols

Hoch

Bruno

Pacione

et al. 2021

AJNR Am J Neuroradiol

View full text Add to dashboard Cite

BACKGROUND AND PURPOSE: Diffusion MR imaging sequences essential for clinical neuroradiology imaging protocols may be accelerated with simultaneous multislice acquisitions. We tested whether simultaneous multislice-accelerated diffusion data were clinically equivalent to standard acquisitions. MATERIALS AND METHODS:In this retrospective study, clinical diffusion sequences obtained before and after implementation of 2-slice simultaneous multislice acceleration and an altered diffusion gradient sampling scheme using the same 3T MRI scanner and 20-channel coil (n ¼ 25 per group) were independently and blindly evaluated by 2 neuroradiologists for perceived quality, artifacts, and overall diagnostic utility. Diffusion tractography was performed in 13 patients both with and without 2-slice simultaneous multislice acceleration (b ¼ 0, 1000, 2000 s/mm 2 ; 60 directions). The corticospinal tract and arcuate fasciculus ipsilateral to the lesion were generated using the same ROIs and then blindly assessed by a neurosurgeon for anatomic fidelity, perceived quality, and impact on surgical management. Tract volumes were compared for spatial overlap.RESULTS: Two-slice simultaneous multislice diffusion reduced acquisition times from 141 to 45 seconds for routine diffusion and from 7.5 to 5.9 minutes for diffusion tractography using 3T MR imaging. The simultaneous multislice-accelerated diffusion sequence was rated equivalent for diagnostic utility despite reductions to perceived image quality. Simultaneous multislice-accelerated diffusion tractography was rated clinically equivalent. Dice similarity coefficients between routine and simultaneous multislice-generated corticospinal tract and arcuate fasciculus tract volumes were 0.78 (SD, 0.03) and 0.71 (SD, 0.05), respectively. CONCLUSIONS: Two-slice simultaneous multislice diffusion appeared clinically equivalent for standard acquisitions and diffusion tractography. Simultaneous multislice makes it feasible to acquire higher angular and q-space-resolution diffusion acquisitions required for translating advanced diffusion models into clinical practice.

show abstract

“…Tian et al presented a DeepDTI model that can optimize six diffusion directions to get DTI with less noise and good quality [26]. Kawamura et al developed a convolutional neural network for dMRI denoising [27]. Lin et al used a deep network to learn deep image prior for realizing dMRI denoising [28].…”

Section: Introductionmentioning

confidence: 99%

Investigation of in Vivo Human Cardiac Diffusion Tensor Imaging Using Unsupervised Dense Encoder-Fusion-Decoder Network

Deng

Wang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Diffusion tensor imaging (DTI) is currently the unique imaging technique that can detect the structure of in-vivo human myocardium without invasivity and radiation. However, it is particularly sensitive to motions, especially respiratory motion that results in serious signal loss in diffusion-weighted (DW) images. This makes it impossible to accurately measure cardiac microscopic structural properties. To cope with such problem, this paper proposes an unsupervised dense-encoder-fusion-decoder network (DEFD-net) to compensate for signal loss in cardiac DW images, which allows investigating in-vivo myocardium structure more accurately. The DEFD-net consists of three modules, namely dense-encoder, fusion module and decoder module. The denseencoder and decoder are trained firstly with DW images acquired at different trigger delays in an unsupervised manner for extracting local and global features. A fusion strategy is then designed to fuse the extracted features. Finally, the well-trained decoder is used to reconstruct the fused DW image from the fused features. To validate the superiority of the proposed method, comparison with existing methods such as PCAMIP, WIF and U2Fusion is performed on both simulated and acquired datasets. The experimental results showed that the proposed method effectively compensates for motion-induced signal loss in DW images, thus leading to much better DW image quality with respect to existing methods. Moreover, the subsequently derived myocardium fiber structure is more regular.

show abstract

Accelerated Acquisition of High-resolution Diffusion-weighted Imaging of the Brain with a Multi-shot Echo-planar Sequence: Deep-learning-based Denoising

Cited by 29 publications

References 29 publications

Applicability of deep learning-based reconstruction trained by brain and knee 3T MRI to lumbar 1.5T MRI

Applicability of deep learning-based reconstruction trained by brain and knee 3T MRI to lumbar 1.5T MRI

Simultaneous Multislice for Accelerating Diffusion MRI in Clinical Neuroradiology Protocols

Investigation of in Vivo Human Cardiac Diffusion Tensor Imaging Using Unsupervised Dense Encoder-Fusion-Decoder Network

Contact Info

Product

Resources

About