CN109597012A - A kind of single sweep space-time code imaging reconstruction method based on residual error network - Google Patents
A kind of single sweep space-time code imaging reconstruction method based on residual error network Download PDFInfo
- Publication number
- CN109597012A CN109597012A CN201811580826.4A CN201811580826A CN109597012A CN 109597012 A CN109597012 A CN 109597012A CN 201811580826 A CN201811580826 A CN 201811580826A CN 109597012 A CN109597012 A CN 109597012A
- Authority
- CN
- China
- Prior art keywords
- space
- time code
- residual error
- imaging
- network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R33/00—Arrangements or instruments for measuring magnetic variables
- G01R33/20—Arrangements or instruments for measuring magnetic variables involving magnetic resonance
- G01R33/44—Arrangements or instruments for measuring magnetic variables involving magnetic resonance using nuclear magnetic resonance [NMR]
- G01R33/48—NMR imaging systems
- G01R33/54—Signal processing systems, e.g. using pulse sequences ; Generation or control of pulse sequences; Operator console
- G01R33/56—Image enhancement or correction, e.g. subtraction or averaging techniques, e.g. improvement of signal-to-noise ratio and resolution
- G01R33/5608—Data processing and visualization specially adapted for MR, e.g. for feature analysis and pattern recognition on the basis of measured MR data, segmentation of measured MR data, edge contour detection on the basis of measured MR data, for enhancing measured MR data in terms of signal-to-noise ratio by means of noise filtering or apodization, for enhancing measured MR data in terms of resolution by means for deblurring, windowing, zero filling, or generation of gray-scaled images, colour-coded images or images displaying vectors instead of pixels
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Radiology & Medical Imaging (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Artificial Intelligence (AREA)
- High Energy & Nuclear Physics (AREA)
- Condensed Matter Physics & Semiconductors (AREA)
- General Physics & Mathematics (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
A kind of single sweep space-time code imaging reconstruction method based on residual error network, is related to the MR image reconstruction technology based on deep learning network.Offer obtains two dimensional image in single sweep operation, reuses deep learning method and rebuilds to obtain a kind of single sweep space-time code imaging reconstruction method based on residual error network.Driving pulse is replaced with into linear frequency sweep pulse, is effective against the scalloping as caused by non-uniform magnetic field and chemical shift, while obtaining image taking speed similar with EPI, resolution ratio and signal-to-noise ratio.SPEN imaging is all lack sampling along phase-encoding direction.Although space-time code imaging signal itself is not necessarily to rebuild the profile for being just able to reflect imaging object, the intrinsic resolution of the profile is usually very low.SPEN image is rebuild from the signal space of low resolution using deep learning, greatly improves image resolution ratio, proton density distribution is presented, while obtaining resolution ratio similar with traditional deconvolution method for reconstructing, obtains high signal-to-noise ratio.
Description
Technical field
The present invention relates to the MR image reconstruction technologies based on deep learning network, particularly say, are to be related to a kind of base
In the single sweep space-time code imaging reconstruction technology of residual error network.
Background technique
Magnetic resonance imaging (MRI) is a kind of imaging technique of nondestructive analysis internal body tissues structural information.In clinic,
MRI suffers from extremely important effect in neuroimaging, cardiovascular imaging and functional mri.Magnetic resonance imaging is logical
The radio-frequency pulse for applying certain frequency to proton is crossed, detects generated signal in combination with additional three-dimensional gradient field.Often
More scanning magnetic resonance sequences of rule obtain a width magnetic resonance image and generally require a few minutes or even dozens of minutes, this exists for MRI
Application in clinic is inappropriate.Therefore, ultra-fast imaging techniques can not only shorten the sampling time, improve sampling efficiency,
Picture quality can be more improved to a certain extent, weaken influence of the motion artifacts to image.It is fast with ultra-fast imaging techniques
Development is read, wherein can obtain in single sweep operation based on Mansfield echo planar imaging (EPI) acquisition method proposed ultrafastly
Whole k-space [1], is widely applied in Perfusion Imaging, dynamic imaging and functional imaging.In recent years it has been proposed that a kind of
It is new based on space-time code (Spatiotemporal Encoding, SPEN) imaging technique, can with speed is imaged similar in EPI
Degree obtains two dimensional image [2].In the k-space of SPEN, the imaging dimension of the phase code of script is substituted by direct position space
Dimension, that is, only need to be obtained with locational space information by one-dimensional Fourier transform.Available and EPI is imaged in SPEN
Similar sampling time, resolution ratio, signal-to-noise ratio etc., simultaneously because packet of the signal of SPEN coding dimension directly from locational space
Network, therefore the ability for resisting non-uniform field is improved, especially under the Pulse Design situation that full weight is gathered.
Since SPEN scan method introduces a quadratic phase distribution related with spatial position by scanning frequency pulse, in conjunction with
Stable phase angle approximation theorem can obtain magnetic resonance image, but the side for passing through Fourier transformation with EPI by simple modulus
Method is compared, and the spatial resolution decline of modulus value image is serious.2010, Ben-Eliezer et al. discovery, SPEN method sampled
To signal there are redundancy, i.e., there is overlapping [3] in the phase stabilization region that samples every time.It is existing using signal redundancy into
Capable super-resolution rebuilding algorithm, including conjugate gradient algorithm for reconstructing, partial Fourier algorithm for reconstructing [4], deconvolution, which is rebuild, calculates
Method etc..Although existing method can be carried out Super-resolution Reconstruction, Riming time of algorithm is longer, and denoising effect is unobvious.
Deep learning, a kind of Nonlinear Mapping effectively learnt between input data and output data by multiple hidden layers
Algorithm achieving more and more concerns in recent years with GPU performance fast lifting.Deep learning network, especially convolution
Neural network is widely applied in the medical image analysis of various problems, including classification, detection and segmentation.It is existing the result shows that,
It is much better than the method [5] of traditional rarefaction representation in terms of image super-resolution rebuilding based on the method for convolutional neural networks.As
The representative of deep neural network model, residual error network can solve the problems, such as that gradient explodes/disappears in network number of plies chin-deep.
SPEN reconstruction is carried out using trained deep learning network, the mistake of image reconstruction can be completed in a short time
Journey effectively removes picture noise, while obtaining resolution ratio similar with traditional deconvolution method for reconstructing.
Bibliography:
[1]Stehling M K,Turner R,Mansfield P.Echo-Planar Imaging:Magnetic
Resonance Imaging in a Fraction of a Second[J].Science,1991,254(5028):43-50.
[2]Shrot Y,Frydman L.Spatially encoded NMR and the acquisition of 2D
magnetic resonance images within a single scan[J].Journal of Magnetic
Resonance,2005,172(2):179-190.
[3]Ben-Eliezer N,Shrot Y,Frydman L,et al.Parametric analysis of the
spatial resolution and signal-to-noise ratio in super-resolved
spatiotemporally encoded(SPEN)MRI[J].Magnetic Resonance in Medicine Official
Journal of the Society of Magnetic Resonance in Medicine,2014,72(2):418-429.
[4]Chen Y,Li J,Qu X,et al.Partial Fourier transform reconstruction
for single-shot MRI with linear frequency-swept excitation[J].Magnetic
Resonance in Medicine Official Journal of the Society of Magnetic Resonance
in Medicine,2013,69(5):1326-1336.
[5]Baumgartner C F,Oktay O,Rueckert D.Fully Convolutional Networks in
Medical Imaging:Applications to Image Enhancement and Recognition[J].2017.
Summary of the invention
The purpose of the present invention is to provide complete two dimensional image is obtained in single sweep operation, deep learning side is then used
Method rebuilds to obtain a kind of more high-resolution single sweep space-time code imaging reconstruction method based on residual error network.
The present invention the following steps are included:
1) code that space-time code imaging sequence is write in the operating software of magnetic resonance imager is debugged and is compiled logical
It crosses.
2) laboratory sample is got out, is fixed on the sample cell of magnetic resonance imager, magnetic resonance imager is put into
Among cavity;
3) magnetic resonance imager operating software is opened on the station of magnetic resonance imager, uses spin-echo sequence first
It is positioned, finds suitable imaging region, determine the size of layer choosing information and area-of-interest.Then conventional tune is carried out
The operation of humorous, shimming, frequency correction and capability correction;
4) space-time code imaging sequence described in step 1), setting sampling bandwidth and relevant parameter are executed, in step 3)
Area-of-interest sampled, obtain K space data;
5) K space data that step 4) obtains is normalized, zeroized and one-dimensional Fourier transform, obtain image area
Data;
6) sampling for generating random sample in digital simulation software, and being simulated in software to it obtains original
SPEN K space data, K space data is normalized, is zeroized, and one-dimensional Fourier is carried out to frequency coding dimension
Transformation, obtains the training data for training deep learning network;
7) residual error network model is built using TensorFlow deep learning frame and Python, sets trained correlation
Parameter;The training data input network that step 6) obtains is trained, up to residual error network convergence and reaches stable, is instructed
Then the network model perfected is rebuild the image domain data that step 5) obtains using trained network model, is obtained
The reliably single sweep space-time code imaging based on residual error network.
In step 1), the structure of the space-time code imaging sequence is successively are as follows: the linear frequency sweep that flip angle is 90 ° excites
Pulse, the reunion pulse that flip angle is 180 °, displacement gradient and sampled echo chain.
In step 6), the random sample can experimental sample to be tested feature distribution using computer batch with
Machine generates, and random sample includes all features of experiment sample, and unstable factor is added, and improves network model to undesirable experiment
The robustness of environment.
In step 7), the trained network can rebuild piece image and only need a few tens of milliseconds, calculate much smaller than tradition
The reconstruction time of method.
Driving pulse is replaced with linear frequency sweep pulse by the present invention, can be effective against due to non-uniform magnetic field and chemical potential
Scalloping caused by shifting, while obtaining image taking speed similar with EPI, resolution ratio and signal-to-noise ratio.SPEN imaging is compiled along phase
Code direction is all lack sampling.Although space-time code imaging signal itself is not necessarily to rebuild the profile for being just able to reflect imaging object,
Be the profile intrinsic resolution be usually it is very low.The present invention utilizes signal space weight of the depth learning technology from low resolution
SPEN image is built, to greatly improve the resolution ratio of image, proton density distribution is more accurately presented, is being obtained and traditional warp
While product method for reconstructing similar resolution ratio, higher signal-to-noise ratio is obtained.
Detailed description of the invention
Fig. 1 is the single sweep space-time code imaging sequence figure that the present invention uses.
Fig. 2 is the residual error network model rebuilding space-time code image and using.In Fig. 2, residual error network model includes 3 sons
Network: input network, residual error learning network and reconstruction network, the input network make the real and imaginary parts of space-time code image
For input;The residual error learning network is to carry out the core of image reconstruction;The reconstruction network is used to multiple characteristic patterns rebuild
For a high-resolution two dimensional image.
Fig. 3 is obtained respectively to the data reconstruction of water mould, lemon and mouse brain with traditional deconvolution method and residual error network method
The image arrived.
Specific embodiment
Below in conjunction with the drawings and specific embodiments, the invention will be further described.
As illustrated in fig. 1 and 2, steps are as follows by each in specific implementation process of the present invention:
1) code that space-time code imaging sequence is write in the operating software of magnetic resonance imager is debugged and is compiled logical
It crosses.
2) laboratory sample is got out, is fixed on the sample cell of magnetic resonance imager, magnetic resonance imager is put into
Among cavity;
3) magnetic resonance imager operating software is opened on the station of magnetic resonance imager, uses spin-echo sequence first
It is positioned, finds suitable imaging region, determine the size of layer choosing information and area-of-interest.Then conventional tune is carried out
The operation of humorous, shimming, frequency correction and capability correction;
4) space-time code imaging sequence described in step 1), setting sampling bandwidth and relevant parameter are executed, in step 3)
Area-of-interest sampled, obtain K space data;
5) K space data that step 4) obtains is normalized, zeroized and one-dimensional Fourier transform, obtain image area
Data;
6) sampling for generating random sample in digital simulation software, and being simulated in software to it obtains original
SPEN K space data, K space data is normalized, is zeroized, and one-dimensional Fourier is carried out to frequency coding dimension
Transformation, obtains the training data for training deep learning network;
7) residual error network model is built using TensorFlow deep learning frame and Python, sets trained correlation
Parameter;The training data input network that step 6) obtains is trained, up to residual error network convergence and reaches stable, is instructed
Then the network model perfected is rebuild the image domain data that step 5) obtains using trained network model, is obtained
The reliably single sweep space-time code imaging based on residual error network.
Specific embodiment is given below:
It is tested with the single sweep space-time code magnetic resonance reconstruction method based on residual error network, respectively to water mould, lemon
Lemon and mouse brain have carried out imaging test, for verifying feasibility of the invention.Experiment is in nuclear magnetic resonance 7T small animal imaging instrument
Lower progress.On the sample bed that ready sample is placed on instrument, sample bed is put into the line of 7T magnetic resonance imager
Circle is intermediate;On the station of magnetic resonance imager open magnetic resonance imager operating software, first with spin-echo sequence into
Row positioning, finds suitable imaging region, determines the size of layer choosing information and area-of-interest.Then it is tuned, shimming, frequency
Rate correction and capability correction;Simultaneously compilation run space-time code imaging sequence, compiling are imported in magnetic resonance imager operating software
By rear, the sampling bandwidth and other parameters of pulse train are set.Then it is sampled with SPEN sequence.
After the completion of data sampling, according to step 5)~7) data are rebuild.The experimental data input that sampling is obtained
In trained residual error network, is rebuild, obtain reliable image.Meanwhile SPEN sampled result is carried out traditional
It is as shown in Figure 3 that the image for reconstructing and to make comparisons is rebuild in deconvolution.
Space-time code sequence is applied to the signal sampling process in magnetic resonance imaging by the present invention.Then sampled signal is passed through
Cross normalize, zeroize is input to rebuild in trained residual error network and obtains with after the one-dimensional Fourier transform of frequency dimension
High-resolution space-time code image.It is compared with traditional Deconvolution Method, has and rebuild speed and higher noise faster
Than.
Claims (3)
1. a kind of single sweep space-time code imaging reconstruction method based on residual error network, it is characterised in that the following steps are included:
1) code that space-time code imaging sequence is write in the operating software of magnetic resonance imager, debugs and compiles and pass through;
2) laboratory sample is got out, is fixed on the sample cell of magnetic resonance imager, the cavity of magnetic resonance imager is put into
It is intermediate;
3) magnetic resonance imager operating software is opened on the station of magnetic resonance imager, is carried out first with spin-echo sequence
Positioning, finds suitable imaging region, determines the size of layer choosing information and area-of-interest;Then conventional tuning, even is carried out
The operation of field, frequency correction and capability correction;
4) space-time code imaging sequence described in step 1), setting sampling bandwidth and relevant parameter are executed, to the sense in step 3)
Interest region is sampled, and K space data is obtained;
5) K space data that step 4) obtains is normalized, zeroized and one-dimensional Fourier transform, obtain the number of image area
According to;
6) sampling for generating random sample in digital simulation software, and being simulated in software to it obtains original
The K space data of SPEN, is normalized K space data, zeroizes, and carries out one-dimensional Fourier's change to frequency coding dimension
It changes, obtains the training data for training deep learning network;
7) residual error network model is built using TensorFlow deep learning frame and Python, sets trained related ginseng
Number;The training data input network that step 6) obtains is trained, up to residual error network convergence and reaches stable, is trained
Good network model, then rebuilds the image domain data that step 5) obtains using trained network model, obtaining can
The single sweep space-time code imaging based on residual error network leaned on.
2. a kind of single sweep space-time code imaging reconstruction method based on residual error network as described in claim 1, it is characterised in that
In step 4), the structure of the space-time code imaging sequence is successively are as follows: flip angle be 90 ° linear frequency sweep excitation pulse, turn over
Reunion pulse, displacement gradient and the sampled echo chain that corner is 180 °.
3. a kind of single sweep space-time code imaging reconstruction method based on residual error network as described in claim 1, it is characterised in that
In step 6), the feature distribution of the experimental sample to be tested of random sample is generated using computer batch is random, with
Press proof sheet includes all features of experiment sample, and unstable factor is added, and improves network model to the Shandong of undesirable experimental situation
Stick.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811580826.4A CN109597012B (en) | 2018-12-24 | 2018-12-24 | Single-scanning space-time coding imaging reconstruction method based on residual error network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811580826.4A CN109597012B (en) | 2018-12-24 | 2018-12-24 | Single-scanning space-time coding imaging reconstruction method based on residual error network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109597012A true CN109597012A (en) | 2019-04-09 |
CN109597012B CN109597012B (en) | 2020-08-04 |
Family
ID=65964262
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811580826.4A Active CN109597012B (en) | 2018-12-24 | 2018-12-24 | Single-scanning space-time coding imaging reconstruction method based on residual error network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109597012B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110109036A (en) * | 2019-05-24 | 2019-08-09 | 厦门大学 | Two-dimension time-space coding sweeps the sampling of magnetic resonance imaging non-Cartesian and method for reconstructing |
WO2021082103A1 (en) * | 2019-10-31 | 2021-05-06 | 清华大学 | Distortionless single-shot echo planar imaging method and apparatus employing deep learning |
CN112924913A (en) * | 2021-02-02 | 2021-06-08 | 厦门大学 | Space-time coding magnetic resonance imaging super-resolution reconstruction method and system |
CN113077527A (en) * | 2021-03-16 | 2021-07-06 | 天津大学 | Rapid magnetic resonance image reconstruction method based on undersampling |
CN113391251A (en) * | 2020-03-12 | 2021-09-14 | 上海联影医疗科技股份有限公司 | Magnetic resonance image reconstruction method, device and equipment |
CN115494439A (en) * | 2022-11-08 | 2022-12-20 | 中遥天地(北京)信息技术有限公司 | Space-time coding image correction method based on deep learning |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090231996A1 (en) * | 2005-08-01 | 2009-09-17 | Siemens Aktiengesellschaft | Method for the phase-related scheduling of data flow in switched networks |
CN107945146A (en) * | 2017-11-23 | 2018-04-20 | 南京信息工程大学 | A kind of space-time Satellite Images Fusion method based on depth convolutional neural networks |
CN108010100A (en) * | 2017-12-07 | 2018-05-08 | 厦门大学 | A kind of single sweep magnetic resonance based on residual error network quantifies T2Imaging reconstruction method |
-
2018
- 2018-12-24 CN CN201811580826.4A patent/CN109597012B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090231996A1 (en) * | 2005-08-01 | 2009-09-17 | Siemens Aktiengesellschaft | Method for the phase-related scheduling of data flow in switched networks |
CN107945146A (en) * | 2017-11-23 | 2018-04-20 | 南京信息工程大学 | A kind of space-time Satellite Images Fusion method based on depth convolutional neural networks |
CN108010100A (en) * | 2017-12-07 | 2018-05-08 | 厦门大学 | A kind of single sweep magnetic resonance based on residual error network quantifies T2Imaging reconstruction method |
Non-Patent Citations (2)
Title |
---|
SHROT, Y等: "《Spatially encoded NMR and the acquisition of 2D》", 《JOURNAL OF MAGNETIC RESONANCE》 * |
马红强等: "基于改进栈式稀疏去噪自编码器的自适应图像去噪", 《光学学报》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110109036A (en) * | 2019-05-24 | 2019-08-09 | 厦门大学 | Two-dimension time-space coding sweeps the sampling of magnetic resonance imaging non-Cartesian and method for reconstructing |
CN110109036B (en) * | 2019-05-24 | 2020-07-03 | 厦门大学 | Two-dimensional space-time coding multi-scan magnetic resonance imaging non-Cartesian sampling and reconstruction method |
WO2021082103A1 (en) * | 2019-10-31 | 2021-05-06 | 清华大学 | Distortionless single-shot echo planar imaging method and apparatus employing deep learning |
CN113391251A (en) * | 2020-03-12 | 2021-09-14 | 上海联影医疗科技股份有限公司 | Magnetic resonance image reconstruction method, device and equipment |
CN112924913A (en) * | 2021-02-02 | 2021-06-08 | 厦门大学 | Space-time coding magnetic resonance imaging super-resolution reconstruction method and system |
CN113077527A (en) * | 2021-03-16 | 2021-07-06 | 天津大学 | Rapid magnetic resonance image reconstruction method based on undersampling |
CN115494439A (en) * | 2022-11-08 | 2022-12-20 | 中遥天地(北京)信息技术有限公司 | Space-time coding image correction method based on deep learning |
Also Published As
Publication number | Publication date |
---|---|
CN109597012B (en) | 2020-08-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109597012A (en) | A kind of single sweep space-time code imaging reconstruction method based on residual error network | |
CN108132274B (en) | Echo-planar imaging is without reference scan pattern distortion antidote under non-uniform magnetic field | |
CN108010100B (en) | Single-scanning magnetic resonance quantitative T based on residual error network2Imaging reconstruction method | |
CN108663644B (en) | Single-scan quantitative magnetic resonance T based on residual error network reconstruction2Imaging method | |
Lam et al. | Constrained magnetic resonance spectroscopic imaging by learning nonlinear low-dimensional models | |
Ramzi et al. | NC-PDNet: A density-compensated unrolled network for 2D and 3D non-Cartesian MRI reconstruction | |
US10338178B2 (en) | System and method for high-resolution spectroscopic imaging | |
CN106772167B (en) | Magnetic resonance imaging method employing and device | |
Ramos‐Llordén et al. | High‐fidelity, accelerated whole‐brain submillimeter in vivo diffusion MRI using gSlider‐spherical ridgelets (gSlider‐SR) | |
CN105785298A (en) | High-precision three-dimensional chemical shift imaging method | |
WO2019148610A1 (en) | Multi-excitation diffusion-weighted magnetic resonance imaging method based on data consistency | |
Zhang et al. | Accelerated MRI reconstruction with separable and enhanced low-rank Hankel regularization | |
Piredda et al. | Fast and high‐resolution myelin water imaging: Accelerating multi‐echo GRASE with CAIPIRINHA | |
Mani et al. | Model-based deep learning for reconstruction of joint kq under-sampled high resolution diffusion MRI | |
CN116597037A (en) | Physical generation data-driven rapid magnetic resonance intelligent imaging method | |
Wang et al. | High‐fidelity direct contrast synthesis from magnetic resonance fingerprinting | |
Qian et al. | Physics-informed deep diffusion MRI reconstruction: Break the bottleneck of training data in artificial intelligence | |
Mani et al. | Multi‐band‐and in‐plane‐accelerated diffusion MRI enabled by model‐based deep learning in q‐space and its extension to learning in the spherical harmonic domain | |
Lam et al. | High-dimensional mr spatiospectral imaging by integrating physics-based modeling and data-driven machine learning: Current progress and future directions | |
Sun et al. | Joint cross-attention network with deep modality prior for fast MRI reconstruction | |
Levac et al. | Accelerated motion correction with deep generative diffusion models | |
Zhang et al. | Towards Better Generalization Using Synthetic Data: A Domain Adaptation Framework for T 2 Mapping via Multiple Overlapping-Echo Acquisition | |
Vaish et al. | TL-HARDI: transform learning based accelerated reconstruction of HARDI data | |
Shimron et al. | CORE‐PI: Non‐iterative convolution‐based reconstruction for parallel MRI in the wavelet domain | |
CN113920211A (en) | Rapid magnetic sensitivity weighted imaging method based on deep learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |