Nothing Special   »   [go: up one dir, main page]

CN107993204B - MRI image reconstruction method based on image block enhanced sparse representation - Google Patents

MRI image reconstruction method based on image block enhanced sparse representation Download PDF

Info

Publication number
CN107993204B
CN107993204B CN201711211072.0A CN201711211072A CN107993204B CN 107993204 B CN107993204 B CN 107993204B CN 201711211072 A CN201711211072 A CN 201711211072A CN 107993204 B CN107993204 B CN 107993204B
Authority
CN
China
Prior art keywords
image
coefficient
image block
reconstruction
matrix
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.)
Active
Application number
CN201711211072.0A
Other languages
Chinese (zh)
Other versions
CN107993204A (en
Inventor
刘书君
曹建鑫
沈晓东
唐明春
张新征
王品
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chengdu Guoyi Technology Co ltd
Original Assignee
Chengdu Guoyi Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chengdu Guoyi Technology Co ltd filed Critical Chengdu Guoyi Technology Co ltd
Priority to CN201711211072.0A priority Critical patent/CN107993204B/en
Publication of CN107993204A publication Critical patent/CN107993204A/en
Application granted granted Critical
Publication of CN107993204B publication Critical patent/CN107993204B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/90—Dynamic range modification of images or parts thereof
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10088—Magnetic resonance imaging [MRI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20048—Transform domain processing
    • G06T2207/20064—Wavelet transform [DWT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30016—Brain
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30168—Image quality inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/513—Sparse representations

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

The invention discloses an MRI image reconstruction method based on image block enhancement sparse representation. Belongs to the technical field of digital image processing. The method is a method for improving coefficient sparsity and estimation performance by utilizing pixel sequencing and non-convex norm constraint in an image block. Firstly, extracting a target image block from an MRI image, then establishing a sequencing training model based on the image block, establishing a reconstruction model of the MRI image by combining coefficient non-convex constraint, then iteratively solving a sequencing matrix and a sparse coefficient in the model by adopting an alternating direction method, and reconstructing a final MRI image by using the estimated sparse coefficient; the invention improves the performance of sparse transformation by sequencing pixels in the image blocks, and carries out non-convex norm minimum constraint on the coefficient, so that the estimated coefficient is closer to a real coefficient.

Description

MRI image reconstruction method based on image block enhanced sparse representation
Technical Field
The invention belongs to the technical field of digital image processing, and particularly relates to a method for enhancing sparsity of transform domain coefficients by ordering pixels in an image block and reconstructing an image by restricting the coefficients by a non-convex norm, which is used for high-quality recovery of a medical image.
Background
Magnetic Resonance Imaging (MRI) is an imaging technique developed by using the principle that nuclei having magnetic moments can generate energy level transitions under the action of a magnetic field. The technology has the advantages of no biological damage, high imaging resolution and the like, and is an important detection means in clinical medicine. However, MRI has a very important disadvantage that the imaging time is too slow, which results in that the patient needs to be still for a long time, and the movement of the patient during the period also results in imaging blurring, which causes the image quality to be degraded, which also becomes the maximum resistance limiting the development of MRI technology. Therefore, solving the problem of MRI imaging speed is a popular direction for the MRI technology research. The traditional MRI imaging needs to sample the original data K space according to the Nyquist sampling theorem, the number of sampled data is large, and the problem of long sampling time is inevitably caused. The compressed sensing theory developed later enables the original data K space to be subjected to down-sampling, sampling data are greatly reduced, sampling time is shortened, and the key point is how to effectively reconstruct an MRI image from the down-sampled data, which also becomes a hotspot of research in the aspect.
The MRI image reconstruction method based on compressed sensing generally utilizes the sparsity of an image, selects a sparse basis (such as wavelet, discrete cosine or a learning-based redundant dictionary) to perform sparse representation on the image, and then estimates the obtained coefficient to reconstruct the image. The method for utilizing the image sparsity is mainly researched on the selection of a dictionary or coefficient estimation, the sparsity is improved by neglecting the utilization of the self structure information of the image, and recent research shows that the representation performance of sparse bases can be improved by improving the arrangement and the phase of pixels in the image, so that the reconstruction effect is improved.
Disclosure of Invention
The invention aims to provide an MRI image reconstruction method based on image block enhancement sparse representation, aiming at the defects of the existing MRI image reconstruction method. The method fully considers the sparsity of the image in a transform domain, improves the sparsity of transform domain coefficients by reordering pixels in an image block, and solves an image reconstruction model by using an alternating direction method. The method specifically comprises the following steps:
(1) inputting an MRI original K space observation data y, and taking a result obtained after zero filling and Fourier inversion are carried out on the y as an initial reconstruction image x;
(2) extracting target image block x by using image block extraction matrix in reconstructed image xiAnd establishing each target image block xiAnd ordering the transformed coefficients alphaiThe model (2) is as follows:
Figure BDA0001484594700000021
where Ψ represents the wavelet transform,
Figure BDA0001484594700000022
representing a matrix, theta, ordering the pixels within the image blockiIs a sorted sequence, and θiEach element in (1) represents
Figure BDA0001484594700000023
The column index of each row in which the non-zero element is located is calculated by using the value of thetaiTo RiAfter the pixels in x are sorted, the pixels are reused
Figure BDA0001484594700000024
Adjusting the phase of each pixel after sorting, RiExtracting a matrix for the image block;
(3) according to the target image block xiCoefficient alpha obtained by sequencing transformationiAnd carrying out non-convex p (0 < p < 1) norm constraint on the MRI image, and establishing a reconstruction model of the MRI image:
Figure BDA0001484594700000025
wherein FuIs a down-sampling Fourier transform matrix, and lambda and beta are regularization parameters;
(4) for the reconstruction model in (3), iterative solution is carried out by using an alternating direction method:
(4a) given x and αiIn respect of the ordering matrix
Figure BDA0001484594700000026
The sub-problems of (1) are:
Figure BDA0001484594700000027
then solving the Cauchy inequality and the sequencing inequality;
(4b) in order to obtain the ordering matrix
Figure BDA0001484594700000028
Then, with respect to the coefficient αiThe sub-problems of (1) are:
Figure BDA0001484594700000029
solving the subproblem by using a generalized iterative soft threshold method;
(4c) in order to obtain the ordering matrix
Figure BDA00014845947000000210
And coefficient alphaiLater, the sub-questions about image x are:
Figure BDA00014845947000000211
solving by using a conjugate gradient method to obtain a reconstruction result;
(4d) and (4) repeating the steps (4a) to (4c) until convergence or the iteration number reaches a preset upper limit.
The innovation point of the method is that the image blocks are sparsely represented in a transform domain by using the sparsity of the images; in order to enhance the sparse representation performance, the pixels in the image block are sequenced and phase-adjusted; non-convex norm minimization constraint is carried out on the sparse coefficient, so that the constraint capacity on the coefficient sparsity is improved; the reconstructed image is solved iteratively by an alternating direction method and applied to Magnetic Resonance Image (MRI) reconstruction.
The invention has the beneficial effects that: the pixels in the image block are sequenced and phase-adjusted, so that the sparsity of the image block in a transform domain is enhanced; the non-convex norm is adopted to constrain the sparse coefficient, so that the constraint capability of the coefficient is improved; the coefficient is estimated by utilizing the generalized iterative soft threshold method, and the precision of the estimated coefficient is improved, so that the finally estimated image not only has good overall visual effect, but also retains a large amount of details in the image, and the whole estimation result is closer to a true value.
The invention mainly adopts a simulation experiment method for verification, and all steps and conclusions are verified to be correct on MATLAB 8.0.
Drawings
FIG. 1 is a workflow block diagram of the present invention;
FIG. 2 is an original image of an MRI human brain image used in simulation of the present invention;
FIG. 3 is the result of reconstruction of a human brain image with a sampling rate of 20% using the RecPF method;
FIG. 4 is the result of reconstructing a human brain image with a sampling rate of 20% using the PANO method;
fig. 5 is a result of reconstructing an image of a human brain with a sampling rate of 20% by the NLR method;
fig. 6 shows the result of reconstructing a human brain image with a sampling rate of 20% using the method of the present invention.
Detailed Description
Referring to fig. 1, the invention relates to an MRI image reconstruction method based on image block enhancement sparse representation, which comprises the following specific steps:
step 1, establishing an ordering of image blocks and an image reconstruction model.
(1a) Inputting an MRI original K space observation data y, and taking a result obtained after zero filling and Fourier inversion are carried out on the y as an initial reconstruction image x;
(1b) extracting target image block x by using image block extraction matrix in reconstructed image xiAnd establishing each target image block xiCoefficient alpha corresponding theretoiThe order transformation model of (1):
Figure BDA0001484594700000031
where Ψ represents the wavelet transform,
Figure BDA0001484594700000032
representing a matrix, theta, ordering the pixels within the image blockiIn order to order the sequence of the sequence,
Figure BDA0001484594700000033
as a sequence of adjustment factors for the phase, RiExtracting a matrix for the image block;
(1c) according to the target image block xiCoefficient alpha obtained by sequencing transformationiAnd carrying out non-convex p (0 < p < 1) norm minimization constraint on the MRI image, and establishing a reconstruction model of the MRI image:
Figure BDA0001484594700000034
wherein FuFor downsampling the Fourier transform matrix, λ and β are regularization parameters.
And 2, iteratively solving the reconstructed total model by using an alternating direction method.
(2a) Given x and αiIn respect of the ordering matrix
Figure BDA0001484594700000041
The sub-problems of (1) are:
Figure BDA0001484594700000042
then solving the Cauchy inequality and the sequencing inequality by using the following equations:
(2a1) will be described in (2a) with respect to
Figure BDA0001484594700000043
The subproblem expression of (a) is equivalently transformed into:
Figure BDA0001484594700000044
the minimization constraint in (2a) is equivalent to the maximization constraint
Figure BDA0001484594700000045
An item;
(2a2) on the basis of (2a1), according to the Cauchy inequality and the ordering inequality:
Figure BDA0001484594700000046
wherein (Ψ)Hαi)*Represents (Ψ)Hαi) Will be (Ψ)Hαi) And (R)ix) after ascending and arranging according to the absolute value, the index position of each element absolute value corresponding to the original sequence is marked as u and v;
(2a3) according to the condition that the inequality and the Cauchy inequality in (2a2) take equal signs, the following can be obtained:
θi(k)＝v(u-1(k))，
Figure BDA0001484594700000047
(2a4) determining thetai(k) And
Figure BDA0001484594700000048
then, substitute it into
Figure BDA0001484594700000049
In the generalized expression of (a):
Figure BDA00014845947000000410
wherein
Figure BDA00014845947000000411
The final sorting matrix can be obtained
Figure BDA00014845947000000412
(2b) In order to obtain the ordering matrix
Figure BDA00014845947000000413
Then, with respect to the coefficient αiThe sub-problems of (1) are:
Figure BDA00014845947000000414
the sub-problem can be solved using a generalized iterative soft threshold method:
(2b1) due to alphaiEach coefficient being independent of the other, and thus with respect to alphaiIs equivalent to αiThe scalar form for each coefficient, where the expression for the kth coefficient is:
Figure BDA0001484594700000051
(2b2) according to the cauchy inequality, the coefficient expression in (4b1) is:
Figure BDA0001484594700000052
the conditions for obtaining equal sign are as follows:
Figure BDA0001484594700000053
then the (2b1) neutron problem can be translated as follows for | αi(k) The optimization problem of | is as follows:
Figure BDA0001484594700000054
(2b3) and (3) obtaining the threshold of the optimization problem in (2b2) by using a generalized iterative soft threshold method as follows:
Figure BDA0001484594700000055
(2b4) when 0 < p < 1 and p ≠ 0.5, the following iterative shrinkage operator is used to solve the subproblems in (2b2) according to generalized iterative soft thresholding:
Figure BDA0001484594700000056
wherein delta(l)Is delta(l)＝γ-p(δ(l-1))p-1The result of the/2 β first iteration;
(2b5) when p is 0.5, the subproblem in (2b2) can have its closed solution obtained by the following fast operator:
Figure BDA0001484594700000057
(2b6) combining the solutions obtained in both cases (2b4) and (2b5), the final coefficients can be found to be:
Figure BDA0001484594700000058
(2c) in order to obtain the ordering matrix
Figure BDA0001484594700000059
And coefficient alphaiThe reconstructed model for image x is then:
Figure BDA00014845947000000510
the reconstruction result can be solved by using a conjugate gradient method:
Figure BDA0001484594700000061
and 3, repeating the processes from (2a) to (2c) until the result converges or the iteration number reaches a preset value.
The effect of the invention can be further illustrated by the following simulation experiment:
experimental conditions and contents
The experimental conditions are as follows: the experiment used a pseudo-radial sampling matrix; the experimental image adopts a real human brain MRI image, as shown in figure 2; and the experimental result evaluation index objectively evaluates the reconstruction result by adopting a peak signal-to-noise ratio (PSNR) and a Structural Similarity (SSIM), wherein RLNE is defined as:
Figure BDA0001484594700000062
wherein
Figure BDA0001484594700000063
For the reconstruction result, x is the original image, μx，
Figure BDA0001484594700000064
σx，
Figure BDA0001484594700000065
And
Figure BDA0001484594700000066
respectively represent x and
Figure BDA0001484594700000067
mean, variance and covariance of C1And C2Constants for avoiding numerical problems, PSHigher values of NR and SSIM indicate better reconstruction results and are closer to real images.
The experimental contents are as follows: under the above experimental conditions, the reconstruction results were compared with the method of the present invention using the NCTV method, DLMRI method, and PBDW method, which are currently representative in the field of MRI image reconstruction.
Experiment 1: the images sampled in the figure 2 are reconstructed by the method of the invention, the NCTV method, the DLMRI method and the PBDW method respectively. The NCTV method is a method for improving the constraint global sparsity in a finite difference domain by combining total variation constraint and non-convex constraint, and the reconstruction result is shown in FIG. 3; the DLMRI method is a reconstruction method based on image blocks and utilizing a learning dictionary to adapt to sparse representation, and the reconstruction result is shown in FIG. 4; the PBDW method is a method of ordering image blocks by using directional wavelets and then performing norm constraint on coefficients, and the reconstruction result is shown in fig. 5. The method of the invention sets the size of the image block in the experiment
Figure BDA0001484594700000068
Constraint parameter λ 106(ii) a The final reconstruction result is fig. 6.
As can be seen from the reconstruction results and the enlarged partial areas of the respective methods of fig. 3, 4, 5 and 6, comparing the NCTV method, the DLMRI method, the PBDW method and the method of the present invention, it can be seen that the method of the present invention is higher in the detail part of the reconstruction results than the other comparison methods.
TABLE 1 PSNR indicators for different reconstruction methods
Image of a person NCTV method DLMRI method PBDW process The method of the invention
Human brain picture 31.15 32.43 32.59 34.39
Table 1 shows PSNR indexes of the reconstruction results of the methods, where higher PSNR values indicate better reconstruction effects; compared with other methods, the method has the advantages that the PSNR value is greatly improved, the reconstruction result of the method is closer to a real image, and the result is consistent with a reconstruction effect graph.
TABLE 2 SSIM index for different reconstruction methods
Image of a person NCTV method DLMRI method PBDW process The method of the invention
Human brain picture 0.7480 0.8351 0.8114 0.8701
Table 2 shows the SSIM index condition of the reconstruction result of each method, wherein a higher SSIM value indicates that the reconstruction result is closer to the real image; compared with other methods, the reconstruction result of the method is more similar to the original image, and the result is consistent with the reconstruction effect graph.
The experiment shows that the reconstruction method of the invention has obvious reduction effect, rich image content after reconstruction and better visual effect and objective evaluation index, thereby being effective to medical image reconstruction.

Claims (3)

1. An MRI image reconstruction method based on image block enhancement sparse representation comprises the following steps:
(1) inputting an MRI original K space observation data y, and taking a result obtained after zero filling and Fourier inversion are carried out on the y as an initial reconstruction image x;
(2) extracting target image block x by using image block extraction matrix in reconstructed image xiAnd establishing each target image block xiAnd ordering the transformed coefficients alphaiThe model (2) is as follows:
Figure FDA0003273437120000011
where Ψ denotes a wavelet transform,
Figure FDA0003273437120000012
representing a matrix, theta, ordering the pixels within the image blockiIs a sorted sequence, and θiEach element in (1) represents
Figure FDA0003273437120000013
The column index of each row in which the non-zero element is located is calculated by using the value of thetaiTo RiAfter the pixels in x are sorted, the pixels are reused
Figure FDA0003273437120000014
Adjusting the phase of each pixel after sorting, RiExtracting a matrix for the image block;
(3) for the ith target image block xiCarrying out sequencing transformation to obtain a coefficient alpha corresponding to the ith target image blockiAnd carrying out non-convex p norm constraint on the MRI image, wherein the value range of p is more than 0 and less than 1, and establishing a reconstruction model of the MRI image:
Figure FDA0003273437120000015
wherein FuIs a down-sampling Fourier transform matrix, and lambda and beta are regularization parameters;
(4) for the reconstruction model in (3), iterative solution is carried out by using an alternating direction method:
(4a) given x and αiIn respect of the ordering matrix
Figure FDA0003273437120000016
The sub-problems of (1) are:
Figure FDA0003273437120000017
then solving the Cauchy inequality and the sequencing inequality;
(4b) in order to obtain the ordering matrix
Figure FDA0003273437120000018
Then, with respect to the coefficient αiThe sub-problems of (1) are:
Figure FDA0003273437120000019
solving the subproblem by using a generalized iterative soft threshold method;
(4c) in order to obtain the ordering matrix
Figure FDA00032734371200000110
And coefficient alphaiLater, the sub-questions about image x are:
Figure FDA00032734371200000111
solving by using a conjugate gradient method to obtain a reconstruction result;
(4d) and (4) repeating the steps (4a) to (4c) until convergence or the iteration number reaches a preset upper limit.
2. The image block enhancement sparse representation-based MRI image reconstruction method according to claim 1, wherein in step (4a), the ordering matrix is obtained by solving the Cauchy inequality and the ordering inequality
Figure FDA0003273437120000021
Comprises the following steps:
(4a1) will be described in (4a) with respect to
Figure FDA0003273437120000022
The subproblem expression of (a) is equivalently transformed into:
Figure FDA0003273437120000023
then minimize
Figure FDA0003273437120000024
Equivalence is maximization
Figure FDA0003273437120000025
(4a2) On the basis of (4a1), according to the Cauchy inequality and the ranking inequality:
Figure FDA0003273437120000026
where n represents the number of pixels in the image block, theta (k) represents the kth element in theta,
Figure FDA0003273437120000027
to represent
Figure FDA0003273437120000028
The k-th element in (j) represents an imaginary unit, (Ψ)Hαi)*Denotes ΨHαiIs conjugated of (2) will beHαiAnd Rix is arranged according to the ascending of the absolute value, the index positions of the absolute values of all elements corresponding to the original sequence are recorded as u and v, u (k) represents the kth element in u, and v (k) represents the kth element in v;
(4a3) according to the condition that the inequality and the Cauchy inequality in (4a2) take equal signs, the following can be obtained:
θi(k)＝v(u-1(k))，
Figure FDA0003273437120000029
wherein theta isi(k) Denotes thetaiThe k-th element of (a) the first,
Figure FDA00032734371200000210
to represent
Figure FDA00032734371200000211
The kth element;
(4a4) determining thetai(k) And
Figure FDA00032734371200000212
then, substitute it into
Figure FDA00032734371200000213
In a generalized expression of：
Figure FDA00032734371200000214
Wherein
Figure FDA00032734371200000215
The final sorting matrix can be obtained
Figure FDA00032734371200000216
3. The MRI image reconstruction method based on image block enhanced sparse representation as claimed in claim 1, wherein the coefficient α is solved in step (4b) by using generalized iterative soft threshold methodiThe sub-problem steps are:
(4b1) due to alphaiEach coefficient being independent of the other, and thus with respect to alphaiIs equivalent to αiIn scalar form with respect to each coefficient, wherein with respect to αiMiddle k coefficient alphai(k) The expression of (a) is:
Figure FDA0003273437120000031
(4b2) according to the cauchy inequality, the coefficient expression in (4b1) is:
Figure FDA0003273437120000032
the condition for obtaining the equal sign is as follows:
Figure FDA0003273437120000033
then the (4b1) neutron problem can be translated into the following with respect to | αi(k) The optimization problem of | is as follows:
Figure FDA0003273437120000034
(4b3) obtaining the threshold value tau of the optimization problem in (4b2) by using a generalized iterative soft threshold methodp,βThe expression is as follows:
Figure FDA0003273437120000035
(4b4) when p is more than 0 and less than 1 and p is not equal to 0.5, solving the optimization problem in (4b2) by adopting the following iterative shrinkage operator according to a generalized iterative soft threshold method:
Figure FDA0003273437120000036
where gamma is an iterative contraction operator
Figure FDA0003273437120000037
Independent variable of delta(l)As a result of the first iteration, i.e.
Figure FDA0003273437120000038
(4b5) When p is 0.5, the optimization problem in (4b2) can be obtained by the following fast operator:
Figure FDA0003273437120000039
(4b6) combining the solutions obtained in both cases (4b4) and (4b5), the final coefficients can be obtained as:
Figure FDA00032734371200000310
CN201711211072.0A 2017-11-28 2017-11-28 MRI image reconstruction method based on image block enhanced sparse representation Active CN107993204B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711211072.0A CN107993204B (en) 2017-11-28 2017-11-28 MRI image reconstruction method based on image block enhanced sparse representation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711211072.0A CN107993204B (en) 2017-11-28 2017-11-28 MRI image reconstruction method based on image block enhanced sparse representation

Publications (2)

Publication Number Publication Date
CN107993204A CN107993204A (en) 2018-05-04
CN107993204B true CN107993204B (en) 2022-05-06

Family

ID=62032333

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711211072.0A Active CN107993204B (en) 2017-11-28 2017-11-28 MRI image reconstruction method based on image block enhanced sparse representation

Country Status (1)

Country Link
CN (1) CN107993204B (en)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108830909B (en) * 2018-05-24 2022-08-19 武汉精测电子集团股份有限公司 Image preprocessing system and method for improving compression ratio of periodic texture image
CN109102461B (en) * 2018-06-15 2023-04-07 深圳大学 Image reconstruction method, device, equipment and medium for low-sampling block compressed sensing
CN109191399B (en) * 2018-08-29 2022-01-28 陕西师范大学 Magnetic resonance image denoising method based on improved multi-path matching pursuit algorithm
CN109522971B (en) * 2018-12-18 2023-04-28 重庆大学 CS-MRI image reconstruction method based on classified image block sparse representation
CN111311752B (en) * 2020-02-14 2021-04-27 福州大学 LiDAR data random sampling and reconstruction method based on mapping chart
CN113592728B (en) * 2021-07-01 2024-04-05 温州理工学院 Image restoration method, system, processing terminal and computer medium
CN114004913B (en) * 2021-12-28 2022-03-29 成都理工大学 Kouximab-based TGS image reconstruction method
CN115205835B (en) * 2022-09-08 2022-12-20 山东特联信息科技有限公司 Liquefied petroleum gas label pattern recognition system

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2757775A2 (en) * 2013-01-16 2014-07-23 Inview Technology Corporation Generating modulation patterns for the acquisition of multiscale information in received signals
CN106023293A (en) * 2016-05-26 2016-10-12 北京爱科声科技有限公司 Three-dimensional reconstruction method based on C-scan ultrasonic image
CN106991355A (en) * 2015-09-10 2017-07-28 天津中科智能识别产业技术研究院有限公司 The face identification method of the analytical type dictionary learning model kept based on topology
CN107274462A (en) * 2017-06-27 2017-10-20 哈尔滨理工大学 The many dictionary learning MR image reconstruction methods of classification based on entropy and geometric direction
CN107274461A (en) * 2017-06-14 2017-10-20 合肥工业大学 A kind of hyperpolarized gas lung MR image reconstruction method
CN107301630A (en) * 2017-06-28 2017-10-27 重庆大学 A kind of CS MRI image reconstructing methods constrained based on ordering structure group non-convex

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2757775A2 (en) * 2013-01-16 2014-07-23 Inview Technology Corporation Generating modulation patterns for the acquisition of multiscale information in received signals
CN106991355A (en) * 2015-09-10 2017-07-28 天津中科智能识别产业技术研究院有限公司 The face identification method of the analytical type dictionary learning model kept based on topology
CN106023293A (en) * 2016-05-26 2016-10-12 北京爱科声科技有限公司 Three-dimensional reconstruction method based on C-scan ultrasonic image
CN107274461A (en) * 2017-06-14 2017-10-20 合肥工业大学 A kind of hyperpolarized gas lung MR image reconstruction method
CN107274462A (en) * 2017-06-27 2017-10-20 哈尔滨理工大学 The many dictionary learning MR image reconstruction methods of classification based on entropy and geometric direction
CN107301630A (en) * 2017-06-28 2017-10-27 重庆大学 A kind of CS MRI image reconstructing methods constrained based on ordering structure group non-convex

Non-Patent Citations (7)

* Cited by examiner, † Cited by third party
Title
Dynamic MRI reconstruction using low-rank and 3D sparsifying transform with separation of background and dynamic components;Changfeng Xi等;《2017 IEEE 2nd Advanced Information Technology, Electronic and Automation Control Conference (IAEAC)》;20171002;第2563-2567页 *
MRI reconstruction using a joint constraint in patch-based total variational framework;Shujun Liu等;《J. Vis. Commun. Image R.》;20170320;第150–164页 *
MRI reconstruction via enhanced group sparsity and nonconvex regularization;Shujun Liu等;《Neurocomputing》;20170701;第108–121页 *
Synergistic PET and SENSE MR Image Reconstruction Using Joint Sparsity Regularization;Abolfazl Mehranian等;《IEEE Transactions on Medical Imaging》;20170418;第20-34页 *
基于分析模型的CS-MR图像重建;高向珍;《中国优秀硕士学位论文全文数据库 信息科技辑》;20150715;I138-864 *
基于群稀疏系数估计的图像重构算法;刘书君等;《仪器仪表学报》;20151231;第36卷(第12期);第2756-2764页 *
联合低秩与稀疏先验的高光谱图像压缩感知重建;孙玉宝等;《电子学报》;20141130;第42卷(第11期);第2219-2224页 *

Also Published As

Publication number Publication date
CN107993204A (en) 2018-05-04

Similar Documents

Publication Publication Date Title
CN107993204B (en) MRI image reconstruction method based on image block enhanced sparse representation
CN107610194B (en) Magnetic resonance image super-resolution reconstruction method based on multi-scale fusion CNN
CN104063886B (en) Nuclear magnetic resonance image reconstruction method based on sparse representation and non-local similarity
CN104156994B (en) Compressed sensing magnetic resonance imaging reconstruction method
WO2018099321A1 (en) Generalized tree sparse-based weighted nuclear norm magnetic resonance imaging reconstruction method
CN107301630B (en) CS-MRI image reconstruction method based on ordering structure group non-convex constraint
CN109522971B (en) CS-MRI image reconstruction method based on classified image block sparse representation
CN107274462B (en) Classified multi-dictionary learning magnetic resonance image reconstruction method based on entropy and geometric direction
CN103400402A (en) Low-rank structure-based sparse compressive sensing MRI (Magnetic Resonance Imaging) image reconstruction method
CN109712077B (en) Depth dictionary learning-based HARDI (hybrid automatic repeat-based) compressed sensing super-resolution reconstruction method
CN111754598B (en) Local space neighborhood parallel magnetic resonance imaging reconstruction method based on transformation learning
CN102332153A (en) Kernel regression-based image compression sensing reconstruction method
CN112991483B (en) Non-local low-rank constraint self-calibration parallel magnetic resonance imaging reconstruction method
CN110942496A (en) Propeller sampling and neural network-based magnetic resonance image reconstruction method and system
CN106981046B (en) Single image super resolution ratio reconstruction method based on multi-gradient constrained regression
CN109920017B (en) Parallel magnetic resonance imaging reconstruction method of joint total variation Lp pseudo norm based on self-consistency of feature vector
CN107993205A (en) A kind of MRI image reconstructing method based on study dictionary with the constraint of non-convex norm minimum
CN107240131A (en) A kind of breast image method for registering based on iteration texture Deformation Field
CN114758028B (en) High-performance reconstruction method of undersampled magnetic resonance image based on optimized regularization and CNN
CN114004764B (en) Improved sensitivity coding reconstruction method based on sparse transform learning
CN108346167B (en) MRI image reconstruction method based on simultaneous sparse coding under orthogonal dictionary
CN115511996A (en) Compressed sensing magnetic resonance image fast reconstruction method based on deep learning
CN114140404A (en) Lung multi-core MRI (magnetic resonance imaging) double-domain super-resolution reconstruction method based on artificial intelligence
CN113487507A (en) Dual-domain recursive network MR reconstruction method based on multi-module feature aggregation
Cheng et al. MRI Reconstruction Based on Transfer Learning Dynamic Dictionary Algorithm

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
TA01 Transfer of patent application right
TA01 Transfer of patent application right

Effective date of registration: 20220419

Address after: No. 78, building 13, No. 287, shuangfengwu Road, Wuhou District, Chengdu, Sichuan 610000

Applicant after: Chengdu Guoyi Technology Co.,Ltd.

Address before: 400044 No. 174 Sha Jie street, Shapingba District, Chongqing

Applicant before: Chongqing University

GR01 Patent grant
GR01 Patent grant