CN109741409A - Echo-planar imaging eddy current artifacts without reference scan bearing calibration - Google Patents
Echo-planar imaging eddy current artifacts without reference scan bearing calibration Download PDFInfo
- Publication number
- CN109741409A CN109741409A CN201811453665.2A CN201811453665A CN109741409A CN 109741409 A CN109741409 A CN 109741409A CN 201811453665 A CN201811453665 A CN 201811453665A CN 109741409 A CN109741409 A CN 109741409A
- Authority
- CN
- China
- Prior art keywords
- phase
- eddy current
- image
- epi
- neural 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.)
- Pending
Links
Abstract
Echo-planar imaging eddy current artifacts without reference scan bearing calibration, the parameter of neural network is trained first: utilizing the random Mass production analog sample of simulation softward, it is then introduced into the EPI sequence write in advance to sample these analog samples, obtains the K space data for corresponding to each sample.According to the phase feature of eddy current effect, phase modification is carried out to collected signal, then carry out the simulation EPI image that Fourier transformation obtains being influenced by eddy current effect.The phase information of modification is preserved as label.Neural network model is trained using these images as training dataset, when below the threshold value that training error is reduced to setting, saves network parameter.After the completion of neural network model training, the EPI image really acquired input neural network model is rebuild, the EPI image for eliminating eddy current artifacts can be obtained.The present invention realizes the removal of EPI image eddy current artifacts in the case where no reference scan, can provide help for the clinical application of EPI.
Description
Technical field
This application involves nuclear magnetic resonance image method for reconstructing, more particularly, to a kind of echo wave plane neural network based at
As eddy current artifacts are without reference scan bearing calibration.
Background technique
Magnetic resonance imaging (Magnetic Resonance Imaging, MRI) is without ionising radiation and has more imaging parameters,
It plays an important role in clinical diagnosis and scientific research, such as open source literature: E.Yacoub, N.Harel, and
K.U"High-field fMRI unveils orientation columns in humans,"Proceedings
of the National Academy of Sciences of the United States of America,vol.105,
no.30,pp.10607-10612,2008；And J.Budde, G.Shajan, J.Hoffmann, K.and
R.Pohmann,"Human imaging at 9.4T using T2 *-,phase-,and susceptibility-weighted
contrast,"Magnetic Resonance in Medicine,vol.65,no.2,pp.544-550,2011.
Traditional magnetic resonance imaging, which needs to carry out multiple radio-frequency pulse excitation, could obtain complete K spacing wave, referring to
R.W.Brown,Y.C.N.Cheng,E.M.Haacke,M.R.Thompson,and R.Venkatesan,Magnetic
Resonance Imaging:Physical Principles and Sequence Design,Second
Edition.J.Wiley&Sons,2013.To make the nuclear spin being stimulated return to equilibrium state, usually needed between scanning twice in succession
The time interval of several seconds is wanted, therefore the experimental period of more scanning magnetic resonance imagings is longer.Longer experimental period is to test object
Discomfort is brought, and it may be moved during the scanning process, image is caused motion artifacts, image quality decrease occur.
Therefore, ultra-fast imaging techniques have important value, especially total to the function magnetic for needing high temporal resolution
Vibration imaging (R.D.Hoge, M.A.Franceschini, R.J.Covolan, T.Huppert, J.B.Mandeville, and
D.A.Boas,"Simultaneous recording of task-induced changes in blood
oxygenation,volume,and flow using diffuse optical imaging and arterial spin-
Labeling MRI, " Neuroimage, vol.25, no.3, pp.701-707,2005) and diffusion tensor imaging (S.Mori
and J.Zhang,"Principles of diffusion tensor imaging and its applications to
Basic neuroscience research, " Neuron, vol.51, no.5, pp.527-539,2006) etc..Single sweep echo
Planar imaging (Echo Planar Imaging, EPI) is that a kind of excitation of single radio-frequency pulse can acquire entire K spacing wave
The supper-fast MR imaging method of typical case.EPI is using the positive negative gradient read output signal being switched fast and fills the space K.Gradient
System it is imperfect, such as eddy current effect or time domain delay, will lead to K space data phase error, these errors appear alternatively in K
In the ranks, be reflected in image area is artifact to the odd even in space.The artifact is presented as test object along phase-encoding direction in image
The half in the domain mobile visual field (Field of View, FOV), commonly referred to as " Nyquist artifact " or " N/2 artifact ".
Whether the method for correction EPI eddy current artifacts can be by being divided into two class (V.B.Xie, M.Lyu, and using model at present
E.X.Wu,"EPI Nyquist ghost and geometric distortion correction by two-frame
phase labeling,"Magnetic Resonance in Medicine,vol.77,no.5,pp.1749-1761,
2017).First kind method is the bearing calibration based on model, and eddy current effect is usually modeled as x-k by this methodyThe one dimensional line in domain
Property phase offset (H.Bruder, H.Fischer, H.E.Reinfelder, and F.Schmitt, " Image
reconstruction for echo planar imaging with non-equidistant k-space
sampling,"Magnetic Resonance in Medicine,vol.23,no.2,pp.311-323,2010；And
S.Skare,D.Clayton,R.Newbould,M.Moseley,and R.Bammer,"A fast and robust
minimum entropy based non-interactive Nyquist ghost correction algorithm,"in:
Proceedings of the 14th Annual Meeting of ISMRM,Seattle,Washington,USA,2006,
P.2349), one-dimensional nonlinear phase error (H.P.D.Xiaoping and T.H.Le, " Artifact reduction in
EPI with phase-encoded reference scan,"Magnetic Resonance in Medicine,vol.36,
No.1, pp.166-171,2010) or two-dimensional phase error (N.K.Chen and A.M.Wyrwicz, " Removal of EPI
Nyquist ghost artifacts with two-dimensional phase correction,"Magnetic
Resonance in Medicine,vol.51,no.6,pp.1247-1253,2004；D.Xu,K.F.King,Y.Zur,and
R.S.Hinks,"Robust 2D phase correction for echo planar imaging under a tight
field-of-view,"Magnetic Resonance in Medicine,vol.64,no.6,pp.1800-1813,2010；
And N.Chen, A.V.Avram, and A.W.Song, " Two-dimensional phase cycled reconstruction
for inherent correction of EPI Nyquist artifacts,"Magnetic Resonance in
Medicine,vol.66,no.4,pp.1057-1066,2011).Phase error by reference to scan obtained image information into
Row estimation and modeling and then elimination, but more times and cost can be expended using reference scan, and test object moves
It may cause correction failure.Second class is alternative manner, this method using Multiple-Scan (B.A.Poser, M.Barth,
P.E.Goa,W.Deng and V.A.Stenger,“Single-shot echo-planar imaging with Nyquist
ghost compensation:interleaved dual echo with acceleration(IDEA)echo-planar
Imaging (EPI), " Magnetic Resonance in Medicine, vol.69, no.1, pp.37-47,2013) or when
Domain interval scan (W.van der Zwaag, J.P.Marques, H.X.Lei, N.Just, T.Kober, and
R.Gruetter,"Minimization of Nyquist ghosting for echo-planar imaging at
ultra-high fields based on a"negative readout gradient"strategy,"Journal of
Magnetic Resonance Imaging, vol.30, no.5, pp.1171-1178,2010), to same target along x-axis forward direction
Repeated sampling is carried out with negative sense, by rearranging the data positively and negatively sampled along x-axis, elimination is positively and negatively adopted
Inconsistency between sample data, and then remove artifact.However, since K space data needs to sample in a reverse direction twice
It obtains, this method echo train length is longer or time resolution lower (V.B.Xie, M.Lyu, and E.X.Wu, " EPI
Nyquist ghost and geometric distortion correction by two-frame phase
labeling,"Magnetic Resonance in Medicine,vol.77,no.5,pp.1749-1761,2017)。
In conclusion needing to find a kind of method of more efficient removal EPI image eddy current artifacts.
Summary of the invention
It is a primary object of the present invention to overcome drawbacks described above in the prior art, a kind of echo-planar imaging whirlpool is proposed
Flow artifact without reference scan artifact correction method.This method can correct the image artifacts as caused by eddy current effect well,
And it does not need additional scanning and reference picture is provided.
The present invention adopts the following technical scheme:
Echo-planar imaging eddy current artifacts without reference scan bearing calibration, which comprises the steps of:
1) magnetic resonance imager is used, load EPI sequence samples test object, saves K space data, and record
Sampling parameter；
2) Fourier transformation is carried out to K space data and obtains EPI image, its amplitude is normalized and is saved；
3) analog sample with similar features is generated according to the textural characteristics of EPI image, and is joined according to the sampling of record
The EPI sequential parameter that number setting simulation uses, samples analog sample with the EPI sequence, generates corresponding K spatial simulation
Data simultaneously are handled to save after obtaining analog image；
4) phase modification is carried out to the analog image that step 3) obtains, obtains the analog image influenced by eddy current effect, made
It is saved for input picture, vortex field phase information is saved as label；
5) neural network model is built, the hyper parameter of neural network model is set, the analog image obtained using step 4)
It is label training neural network model to input, being vortexed field phase, training error is observed, when training error is restrained and is lower than setting
Threshold value when, then the neural network model has trained, terminate training simultaneously saves network parameter；
6) by EPI image input step 5 obtained in step 2)) in trained neural network model carry out vortex field phase
Position information extraction, then phase is carried out to EPI image obtained in step 2) by the method for step 4) using the vortex field phase information
Position amendment, obtains the EPI image of removal eddy current artifacts.
Preferably, it in step 1), after test object to be sent into the test chamber of magnetic resonance imager, is first tuned, is even
Field, frequency correction and capability correction, reload EPI sequence and sample to test object.
Preferably, in step 1), the EPI sequence includes: 90 ° of sinc radio frequency excitation pulses, layer choosing gradient Gs, frequency
Dimension biasing gradient GaWith readout gradient Gro, phase dimension biasing gradient GeWith phase encoding gradient Gpe, frequency dimension biasing gradient Ga's
Area is first readout gradient GroThe half of area, direction is in contrast；Phase dimension biasing gradient GeArea be all phases
Position coding GpeThe half of area sum, it is contrary；90 ° of sinc radio-frequency pulse combination layer choosing gradient GssIt carries out space and selects layer；It reads
Gradient GroWith phase encoding gradient GpeIt combines, realizes the acquisition to spin-magnetic resonance signal.
Preferably, in step 3), Fourier transformation is carried out to the K spatial simulation data and obtains amplitude figure, to its width
Value is normalized to obtain analog image.
Preferably, in step 3), the analog sample is using computer random Mass production；Analog sampling process
In, the stochastic variable factor is added to improve network model to the robustness of undesirable experimental situation.
Preferably, according to eddy current effect the characteristics of, by phase offset caused by vortex field be reduced to first-order linear phase and
The combination of zeroth order phase indicates are as follows:
Wherein a is the coefficient of first-order linear phase, and b is the coefficient of zeroth order phase, and x is the seat of image area frequency coding dimension
Mark, y are the coordinate of image area phase code dimension, and odd, even row is tieed up with phase code and distinguished；
In step 4), the phase modification includes: the image area signal S (x, y) the simulation of preservation along phase code
Dimension does one-dimensional Fourier inversion, obtains x-kySignal I (x, the k in domainy), wherein kyDimension coordinate, k are encoded for K space phasexFor K
Space freqluncy coding ties up coordinate.Then phase theta (x) is introduced I (x, ky), result I ' (x, k after obtaining phase modificationy):
I′(x,ky)=I (x, ky)·e-iθ(x)
Again to I ' (x, ky) tieed up along phase code and be the analog image T after one-dimensional Fourier transform obtains phase modification
(x,y)。
Preferably, in step 5), the loss function of the neural network model are as follows:
Wherein M is the number of samples for participating in training every time, and L and H are that input picture is tieed up in frequency coding peacekeeping phase code
Pixel number, W and b are network parameters, and X is the label of image area, Y representing input images, and g indicates network to the work of input picture
Use function.
By the above-mentioned description of this invention it is found that compared with prior art, the invention has the following beneficial effects:
The present invention first uses EPI sequential sampling to obtain an EPI image, then rebuilds to obtain one using deep learning method
EPI image without eddy current artifacts.It is provided by the invention it is neural network based to EPI image eddy current artifacts without reference scan school
Correction method will combine the characteristics of single sweep EPI sequence fast imaging with learning ability that neural network is powerful, guarantee ultrafast
In the case where rapid-result picture, the EPI image without eddy current artifacts of high quality is reconstructed.
Detailed description of the invention
Fig. 1 is the single sweep asymmetric blipped echoplanar single pulse technique that the present invention uses.
Fig. 2 is the sample track for the single sweep asymmetric blipped echoplanar single pulse technique that the present invention uses.
Fig. 3 is the representative neural network model rebuilding no eddy current artifacts image and using.
Fig. 4 is the EPI image after the EPI image that actual acquisition obtains and use the application method correction.In Fig. 4, a-c
For the human brain image of the different layers of EPI sequence acquisition, d-f is EPI corresponding with the a-c figure corrected using the application method
Picture.
Specific embodiment
Below by way of specific embodiment, the invention will be further described.
Each step that the present invention provides the eddy current artifacts of EPI without reference scan bearing calibration, in specific implementation process
It is as follows:
1) magnetic resonance imager operating software is opened in station, test object is sent into magnetic resonance imager by experimental bed
Test chamber.It is tuned, shimming, frequency correction and capability correction.EPI sequence is loaded, the parameters of sequence are set
It sets and checks.
EPI sequence includes: 90 ° of sinc radio frequency excitation pulses, layer choosing gradient Gs, frequency dimension biasing gradient GaAnd readout gradient
Gro, phase dimension biasing gradient GeWith phase encoding gradient Gpe.Frequency dimension biasing gradient GaArea be first readout gradient Gro
The half of area, direction is in contrast；Phase dimension biasing gradient GeArea be all phase encoding gradient GpeThe one of area sum
Half, it is contrary.90 ° of sinc radio-frequency pulse combination layer choosing gradient GssIt carries out space and selects layer；Readout gradient GroWith phase code ladder
Spend GpeIt combines, realizes the acquisition to spin-magnetic resonance signal.
Test object is sampled using EPI sequence, after the completion of sampling, saves K space data, and records sampling ginseng
Number.
2) Fourier transformation is carried out to the K space data of acquisition and obtains EPI image, its amplitude is normalized simultaneously
It saves.
3) analog sample with similar features is generated according to the textural characteristics of EPI image, and is joined according to the sampling of record
The EPI sequential parameter that number setting simulation uses, samples analog sample with the EPI sequence, generates corresponding K spatial simulation
Data simultaneously are handled to save after obtaining analog image.
The analog sample is according to the textural characteristics of magnetic resonance image using computer random Mass production.Analog sampling
In the process, the stochastic variable factor is added to improve network model to the robustness of undesirable experimental situation, such as increases by one at random
Quadric perturbation.
4) phase modification is carried out to the analog image that step 3) obtains, obtains the analog image influenced by eddy current effect, made
It is preserved for input picture.Vortex field phase information is saved as label.
Phase offset caused by vortex field is reduced to first-order linear phase and zeroth order phase by the characteristics of according to eddy current effect
Combination, indicate are as follows:
Wherein a is the coefficient of first-order linear phase, and b is the coefficient of zeroth order phase, and x is the seat of image area frequency coding dimension
Mark, y are the coordinate of image area phase code dimension, and odd, even row is tieed up with phase code and distinguished.
Phase modification includes: that the image area signal S (x, y) of the simulation of preservation is one-dimensional Fourier along phase code dimension is anti-
Transformation, obtains x-kySignal I (x, the k in domainy), wherein kyDimension coordinate, k are encoded for K space phasexIt ties up and sits for K space freqluncy coding
Mark.Then phase theta (x) is introduced I (x, ky), result I ' (x, k after obtaining phase modificationy):
I′(x,ky)=I (x, ky)·e-iθ(x)
Again to I ' (x, ky) tieed up along phase code and be the image T (x, y) after one-dimensional Fourier transform obtains phase modification.
5) neural network model is built, the hyper parameter of neural network model is set.
The loss function of neural network model are as follows:
Wherein M is the number of samples for participating in training every time, and L and H are that input picture is tieed up in frequency coding peacekeeping phase code
Pixel number, W and b are network parameters, and X is the label of image area, Y representing input images, and g indicates network to the work of input picture
Use function.
Training neural network model, observes training error, when training error is restrained and is lower than the threshold value of setting, it is believed that should
Neural network model has trained, and terminates training and saves network parameter.
For trained neural network model due to using the analog sample being randomly generated to be trained, generalization is stronger, fits
Eddy current artifacts removal and image reconstruction for measurement data under various experimental situations.
6) by EPI image input step 5 obtained in step 2)) in trained neural network model carry out vortex field phase
Position information extraction, then phase is carried out to the EPI image that step 2) obtains by the method for step 4) using the vortex field phase information
It corrects to get the EPI image of removal eddy current artifacts is arrived.
The single sweep EPI sequence that the present invention uses is referring to Fig. 1.The sample track for the EPI sequence that the present invention uses is referring to figure
2.The representative neural network model used without eddy current artifacts image is rebuild referring to Fig. 3.The network model mainly includes four portions
Point: input network, coding network, decoding network and output network.Network is inputted using the real and imaginary parts of EPI image as defeated
Enter.Coding network is to correct image eddy current artifacts.Decoding network improves image to rebuild to the image corrected
Precision.It exports network and exports the EPI image without eddy current artifacts.
Specific embodiment is given below:
Fig. 4 illustrate the human brain image that EPI sequential sampling obtains and using the application method rebuild as a result, wherein a-c
For the human brain image of the different layers of EPI sequence acquisition, d-f is EPI corresponding with the a-c figure corrected using the application method
Picture.
The present embodiment sweep without reference by using eddy current artifacts of the U-shaped neural network to the EPI human brain image acquired
Correction is retouched, to verify feasibility of the invention.Experiment carries out on 7T magnetic resonance imager.Imager operation is opened in station
Software makes test object lie low on experimental bed, and experimental bed is sent into the test chamber of imager, is tuned, shimming, frequency
Correction and capability correction.According to specific experimental conditions, the parameters of pulse train are set.The experiment parameter of the present embodiment is set
Set as follows: visual field FOV is 40mm × 40mm.The sampled echo time of EPI sequence is 65.94ms, and pulse-recurrence time is
32.97ms, total sweep time are 692.37ms, and the sampling number of frequency coding peacekeeping phase code dimension is 64.Will more than
Experiment parameter starts to sample after setting.
After the completion of data acquisition, according to above-mentioned steps 2)~step 6) rebuilds data, in reconstructed results such as Fig. 4
D, shown in e, f figure.
The present invention is believed under the excitation of 90 ° of sinc pulse by the entire space K of the positive and negative switching Quick Acquisition of gradient
Number, then image is obtained by carrying out two-dimensional Fourier transform to sampled signal.In sampling process, due to shadows such as gradient eddies
It rings, the parity rows phase in the space K is caused to mismatch, reconstruction image is then made to generate eddy current artifacts.The input of neural network comes from
The real and imaginary parts of reconstruction image.Network training uses simulated data sets, and the simplified process of training is: being first randomly generated mould
Quasi- EPI sampled signal produces the figure with eddy current effect then according to the feature of eddy current effect based on analog signal
As being used as input data.Analogue data abundant is generated using simulation softward batch, and network parameter is finely adjusted, to ensure
Training error converges to given relatively fractional value.After the completion of network training, input measured data, which can be rebuild, eliminates vortex
The image of artifact.
The above is only a specific embodiment of the present invention, but the design concept of the present invention is not limited to this, all to utilize this
Design makes a non-material change to the present invention, and should all belong to behavior that violates the scope of protection of the present invention.
Claims (7)
1. echo-planar imaging eddy current artifacts without reference scan bearing calibration, which comprises the steps of:
1) magnetic resonance imager is used, load EPI sequence samples test object, saves K space data, and record sampling
Parameter；
2) Fourier transformation is carried out to K space data and obtains EPI image, its amplitude is normalized and is saved；
3) analog sample with similar features is generated according to the textural characteristics of EPI image, and is set according to the sampling parameter of record
The EPI sequential parameter that simulation uses is set, analog sample is sampled with the EPI sequence, generates corresponding K spatial simulation data
And it is handled to save after obtaining analog image；
4) phase modification is carried out to the analog image that step 3) obtains, the analog image influenced by eddy current effect is obtained, as defeated
Enter image to be saved, vortex field phase information is saved as label；
5) neural network model is built, the hyper parameter of neural network model is set, the analog image for using step 4) to obtain is defeated
Entering, being vortexed field phase is label training neural network model, observes training error, when training error is restrained and is lower than the threshold of setting
When value, then the neural network model has trained, and terminates training and saves network parameter；
6) by EPI image input step 5 obtained in step 2)) in trained neural network model carry out vortex field phase letter
Breath extracts, then carries out phase to EPI image obtained in step 2) by the method for step 4) using the vortex field phase information and repair
Just, the EPI image of removal eddy current artifacts is obtained.
2. echo-planar imaging eddy current artifacts as described in claim 1 without reference scan bearing calibration, which is characterized in that
In step 1), after test object to be sent into the test chamber of magnetic resonance imager, first it is tuned, shimming, frequency correction and power
Correction, reloads EPI sequence and samples to test object.
3. echo-planar imaging eddy current artifacts as described in claim 1 without reference scan bearing calibration, which is characterized in that
In step 1), the EPI sequence includes: 90 ° of sinc radio frequency excitation pulses, layer choosing gradient Gs, frequency dimension biasing gradient GaAnd reading
Gradient Gro, phase dimension biasing gradient GeWith phase encoding gradient Gpe, frequency dimension biasing gradient GaArea be that first reading is terraced
Spend GroThe half of area, direction is in contrast；Phase dimension biasing gradient GeArea be all phase code GpeThe one of area sum
Half, it is contrary；90 ° of sinc radio-frequency pulse combination layer choosing gradient GssIt carries out space and selects layer；Readout gradient GroWith phase code ladder
Spend GpeIt combines, realizes the acquisition to spin-magnetic resonance signal.
4. echo-planar imaging eddy current artifacts as described in claim 1 without reference scan bearing calibration, which is characterized in that in step
It is rapid 3) in, to the K spatial simulation data carry out Fourier transformation obtain amplitude figure, its amplitude is normalized to obtain
Analog image.
5. echo-planar imaging eddy current artifacts as described in claim 1 without reference scan bearing calibration, which is characterized in that in step
It is rapid 3) in, the analog sample is using computer random Mass production；During analog sampling, the stochastic variable factor is added and comes
Network model is improved to the robustness of undesirable experimental situation.
6. echo-planar imaging eddy current artifacts as described in claim 1 without reference scan bearing calibration, which is characterized in that according to
Phase offset caused by vortex field is reduced to the combination of first-order linear phase and zeroth order phase by the characteristics of eddy current effect, is indicated
Are as follows:
Wherein a is the coefficient of first-order linear phase, and b is the coefficient of zeroth order phase, and x is the coordinate of image area frequency coding dimension, and y is
The coordinate of image area phase code dimension, odd, even row are tieed up with phase code and are distinguished；
In step 4), the phase modification includes: that the image area signal S (x, y) of the simulation of preservation is done along phase code dimension
One-dimensional Fourier inversion, obtains x-kySignal I (x, the k in domainy), wherein kyDimension coordinate, k are encoded for K space phasexFor the space K
Frequency coding ties up coordinate.Then phase theta (x) is introduced I (x, ky), result I ' (x, k after obtaining phase modificationy):
I′(x,ky)=I (x, ky)·e-iθ(x)
Again to I ' (x, ky) tieed up along phase code and be the analog image T (x, y) after one-dimensional Fourier transform obtains phase modification.
7. echo-planar imaging eddy current artifacts as described in claim 1 without reference scan bearing calibration, which is characterized in that in step
It is rapid 5) in, the loss function of the neural network model are as follows:
Wherein M is the number of samples for participating in training every time, and L and H are the picture that input picture is tieed up in frequency coding peacekeeping phase code
Prime number, W and b are network parameters, and X is the label of image area, Y representing input images, effect letter of the g expression network to input picture
Number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811453665.2A CN109741409A (en) | 2018-11-30 | 2018-11-30 | Echo-planar imaging eddy current artifacts without reference scan bearing calibration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811453665.2A CN109741409A (en) | 2018-11-30 | 2018-11-30 | Echo-planar imaging eddy current artifacts without reference scan bearing calibration |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109741409A true CN109741409A (en) | 2019-05-10 |
Family
ID=66358335
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811453665.2A Pending CN109741409A (en) | 2018-11-30 | 2018-11-30 | Echo-planar imaging eddy current artifacts without reference scan bearing calibration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109741409A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110095742A (en) * | 2019-05-13 | 2019-08-06 | 上海东软医疗科技有限公司 | A kind of echo planar imaging neural network based and device |
| CN110244246A (en) * | 2019-07-03 | 2019-09-17 | 上海联影医疗科技有限公司 | MR imaging method, device, computer equipment and storage medium |
| CN110940943A (en) * | 2019-12-06 | 2020-03-31 | 上海联影医疗科技有限公司 | Training method of pulsation artifact correction model and pulsation artifact correction method |
| CN110942496A (en) * | 2019-12-13 | 2020-03-31 | 厦门大学 | Propeller sampling and neural network-based magnetic resonance image reconstruction method and system |
| CN113030813A (en) * | 2021-02-26 | 2021-06-25 | 厦门大学 | Magnetic resonance T2Quantitative imaging method and system |
| CN113534031A (en) * | 2020-04-21 | 2021-10-22 | 上海联影医疗科技股份有限公司 | Image domain data generating method, computer device and readable storage medium |
| CN115359144A (en) * | 2022-10-19 | 2022-11-18 | 之江实验室 | Magnetic resonance plane echo imaging artifact simulation method and system |
| 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 (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1683939A (en) * | 2004-04-13 | 2005-10-19 | 西门子公司 | Movement-corrected multi-shot method for diffusion-weighted imaging in magnetic resonance tomography |
| CN1803092A (en) * | 2005-11-29 | 2006-07-19 | 东南大学 | Artifact correction method based on even marker in plane echo imaging technique |
| CN101484823A (en) * | 2006-07-06 | 2009-07-15 | 皇家飞利浦电子股份有限公司 | Magnetic resonance device and method |
| CN102438518A (en) * | 2010-08-11 | 2012-05-02 | 株式会社东芝 | Magnetic resonance imaging device and magnetic resonance imaging method |
| CN103124517A (en) * | 2010-10-07 | 2013-05-29 | 杜克大学 | Multi-dimensional iterative phase-cycled reconstruction for MRI images |
| EP2793040A1 (en) * | 2013-04-19 | 2014-10-22 | Universitätsklinikum Jena | Method for processing phase information, particularly for susceptibility weighted imaging, obtained from PROPELLER measurements |
| CN104918545A (en) * | 2013-02-19 | 2015-09-16 | 株式会社东芝 | Magnetic resonance imaging device |
| CN108132274A (en) * | 2017-12-21 | 2018-06-08 | 厦门大学 | Echo-planar imaging is without reference scan pattern distortion antidote under non-uniform magnetic field |
-
2018
- 2018-11-30 CN CN201811453665.2A patent/CN109741409A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1683939A (en) * | 2004-04-13 | 2005-10-19 | 西门子公司 | Movement-corrected multi-shot method for diffusion-weighted imaging in magnetic resonance tomography |
| CN1803092A (en) * | 2005-11-29 | 2006-07-19 | 东南大学 | Artifact correction method based on even marker in plane echo imaging technique |
| CN101484823A (en) * | 2006-07-06 | 2009-07-15 | 皇家飞利浦电子股份有限公司 | Magnetic resonance device and method |
| CN102438518A (en) * | 2010-08-11 | 2012-05-02 | 株式会社东芝 | Magnetic resonance imaging device and magnetic resonance imaging method |
| CN103124517A (en) * | 2010-10-07 | 2013-05-29 | 杜克大学 | Multi-dimensional iterative phase-cycled reconstruction for MRI images |
| CN104918545A (en) * | 2013-02-19 | 2015-09-16 | 株式会社东芝 | Magnetic resonance imaging device |
| EP2793040A1 (en) * | 2013-04-19 | 2014-10-22 | Universitätsklinikum Jena | Method for processing phase information, particularly for susceptibility weighted imaging, obtained from PROPELLER measurements |
| CN108132274A (en) * | 2017-12-21 | 2018-06-08 | 厦门大学 | Echo-planar imaging is without reference scan pattern distortion antidote under non-uniform magnetic field |
Non-Patent Citations (4)
| Title |
|---|
| LEE, JUYOUNG ET AL: ""k -Space Deep Learning for Reference-free EPI Ghost Correction"", 《NASA/ADS》 * |
| PU LIAO ET AL: ""Referenceless distortion correction of gradient-echo echo-planar imaging under inhomogeneous magnetic fields based on a deep convolutional neural network"", 《COMPUTERS IN BIOLOGY AND MEDICINE》 * |
| 吴瑞等: ""不均匀场下空间编码超快速高分辨率NMR研究进展"", 《波谱学杂志》 * |
| 陈林: ""时空编码单扫描超快速磁共振成像的超分辨率重建研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110095742A (en) * | 2019-05-13 | 2019-08-06 | 上海东软医疗科技有限公司 | A kind of echo planar imaging neural network based and device |
| CN110095742B (en) * | 2019-05-13 | 2022-02-08 | 上海东软医疗科技有限公司 | Plane echo imaging method and device based on neural network |
| CN110244246A (en) * | 2019-07-03 | 2019-09-17 | 上海联影医疗科技有限公司 | MR imaging method, device, computer equipment and storage medium |
| CN110244246B (en) * | 2019-07-03 | 2021-07-16 | 上海联影医疗科技股份有限公司 | Magnetic resonance imaging method, magnetic resonance imaging apparatus, computer device, and storage medium |
| CN110940943B (en) * | 2019-12-06 | 2021-11-02 | 上海联影医疗科技股份有限公司 | Training method of pulsation artifact correction model and pulsation artifact correction method |
| CN110940943A (en) * | 2019-12-06 | 2020-03-31 | 上海联影医疗科技有限公司 | Training method of pulsation artifact correction model and pulsation artifact correction method |
| CN110942496B (en) * | 2019-12-13 | 2022-02-11 | 厦门大学 | Propeller sampling and neural network-based magnetic resonance image reconstruction method and system |
| CN110942496A (en) * | 2019-12-13 | 2020-03-31 | 厦门大学 | Propeller sampling and neural network-based magnetic resonance image reconstruction method and system |
| CN113534031A (en) * | 2020-04-21 | 2021-10-22 | 上海联影医疗科技股份有限公司 | Image domain data generating method, computer device and readable storage medium |
| CN113030813A (en) * | 2021-02-26 | 2021-06-25 | 厦门大学 | Magnetic resonance T2Quantitative imaging method and system |
| CN115359144A (en) * | 2022-10-19 | 2022-11-18 | 之江实验室 | Magnetic resonance plane echo imaging artifact simulation method and system |
| CN115359144B (en) * | 2022-10-19 | 2023-03-03 | 之江实验室 | Magnetic resonance plane echo imaging artifact simulation method and system |
| 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 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109741409A (en) | Echo-planar imaging eddy current artifacts without reference scan bearing calibration | |
| Lee et al. | Reference‐free single‐pass EPI N yquist ghost correction using annihilating filter‐based low rank H ankel matrix (ALOHA) | |
| Dong et al. | Tilted‐CAIPI for highly accelerated distortion‐free EPI with point spread function (PSF) encoding | |
| Doneva et al. | Compressed sensing for chemical shift‐based water–fat separation | |
| Porter et al. | High resolution diffusion‐weighted imaging using readout‐segmented echo‐planar imaging, parallel imaging and a two‐dimensional navigator‐based reacquisition | |
| Frost et al. | Scan time reduction for readout‐segmented EPI using simultaneous multislice acceleration: Diffusion‐weighted imaging at 3 and 7 Tesla | |
| Peng et al. | Accelerated exponential parameterization of T2 relaxation with model‐driven low rank and sparsity priors (MORASA) | |
| Liao et al. | Distortion‐free, high‐isotropic‐resolution diffusion MRI with gSlider BUDA‐EPI and multicoil dynamic B0 shimming | |
| Truong et al. | High‐resolution multishot spiral diffusion tensor imaging with inherent correction of motion‐induced phase errors | |
| Liao et al. | Efficient parallel reconstruction for high resolution multishot spiral diffusion data with low rank constraint | |
| Liu et al. | DWI using navigated interleaved multishot EPI with realigned GRAPPA reconstruction | |
| Arefeen et al. | Scan‐specific artifact reduction in k‐space (SPARK) neural networks synergize with physics‐based reconstruction to accelerate MRI | |
| Chang et al. | Interleaved diffusion‐weighted improved by adaptive partial‐Fourier and multiband multiplexed sensitivity‐encoding reconstruction | |
| Hu et al. | Multi‐shot diffusion‐weighted MRI reconstruction with magnitude‐based spatial‐angular locally low‐rank regularization (SPA‐LLR) | |
| Hu et al. | RUN‐UP: Accelerated multishot diffusion‐weighted MRI reconstruction using an unrolled network with U‐Net as priors | |
| Gao et al. | PCLR: phase‐constrained low‐rank model for compressive diffusion‐weighted MRI | |
| Herbst et al. | Prospective motion correction of segmented diffusion weighted EPI | |
| Moeller et al. | Diffusion imaging in the post HCP era | |
| Koopmans | Two‐dimensional‐NGC‐SENSE‐GRAPPA for fast, ghosting‐robust reconstruction of in‐plane and slice‐accelerated blipped‐CAIPI echo planar imaging | |
| Hu et al. | Phase‐updated regularized SENSE for navigator‐free multishot diffusion imaging | |
| Wald | Ultimate MRI | |
| Lee et al. | k‐Space deep learning for reference‐free EPI ghost correction | |
| Zahneisen et al. | Reverse retrospective motion correction | |
| Mani et al. | Comprehensive reconstruction of multi-shot multi-channel diffusion data using mussels | |
| Chen et al. | 3D‐EPI blip‐up/down acquisition (BUDA) with CAIPI and joint H ankel structured low‐rank reconstruction for rapid distortion‐free high‐resolution T 2* mapping |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190510 |
|
| RJ01 | Rejection of invention patent application after publication |