CN110163819B - Non-local mean value smoothing method for magnetic resonance diffusion weighted image - Google Patents
Non-local mean value smoothing method for magnetic resonance diffusion weighted image Download PDFInfo
- Publication number
- CN110163819B CN110163819B CN201910367288.9A CN201910367288A CN110163819B CN 110163819 B CN110163819 B CN 110163819B CN 201910367288 A CN201910367288 A CN 201910367288A CN 110163819 B CN110163819 B CN 110163819B
- Authority
- CN
- China
- Prior art keywords
- voxel
- diffusion
- magnetic resonance
- neighborhood
- equation
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000009792 diffusion process Methods 0.000 title claims abstract description 43
- 238000009499 grossing Methods 0.000 title claims abstract description 16
- 238000001914 filtration Methods 0.000 claims abstract description 4
- 238000010606 normalization Methods 0.000 claims description 2
- 239000000835 fiber Substances 0.000 abstract description 28
- 238000003384 imaging method Methods 0.000 description 6
- 238000002598 diffusion tensor imaging Methods 0.000 description 4
- 238000004088 simulation Methods 0.000 description 4
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 3
- 238000009826 distribution Methods 0.000 description 2
- 240000007594 Oryza sativa Species 0.000 description 1
- 235000007164 Oryza sativa Nutrition 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000002597 diffusion-weighted imaging Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012880 independent component analysis Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 235000009566 rice Nutrition 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
Classifications
-
- G06T5/70—
-
- 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]
- G06T2207/10092—Diffusion tensor magnetic resonance imaging [DTI]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
Abstract
The invention discloses a non-local mean smoothing method for a magnetic resonance diffusion weighted image, and relates to the technical field of magnetic resonance diffusion weighted image processing. The method comprises the following steps: s1, selecting a filtering parameter h and a neighborhood range R V ，R N For each voxel p i Is defined in the direction g of the diffusion-weighted gradient j Is a magnetic resonance diffusion weighting signal S (p i ，g j ) Steps S2 and S3 are carried out; s2, for voxel p i R of (2) V ×R V ×R V Neighborhood V pi Each voxel p of (3) k Calculating p k Is different from the diffusion weighted gradient direction g l Is a diffusion weighted signal pair S (p i ，g j ) Weights ω (p) i ，g j ，p k ，g l ) The method comprises the steps of carrying out a first treatment on the surface of the S3, pair S (p i ，g j ) Non-local mean smoothing is performed. The method can solve the problem that the accuracy of the estimation result is poor due to the fact that the existing SD method is susceptible to noise when being directly used for estimating the fiber structure in the voxel; the accuracy of the estimation result of the fiber structure in the voxels can be effectively improved.
Description
Technical Field
The invention relates to the technical field of magnetic resonance diffusion weighted image processing, in particular to a non-local mean value smoothing method for a magnetic resonance diffusion weighted image.
Background
Magnetic resonance diffusion imaging is currently the only non-invasive method capable of measuring the diffusion movement and imaging of water molecules in tissue on living bodies, and detects the microstructure of the tissue by measuring and quantifying the diffusion information of water molecules in tissue. The diffusion information of water molecules in different directions is contained in a set of diffusion weighted images (Diffusion Weighted Image, DWI) of different diffusion gradient directions, and the fiber structure information (mainly the fiber trend) in each voxel can be resolved by modeling the diffusion function. According to the trend of the fiber in each voxel, the whole three-dimensional structure of the tissue fiber bundle can be reconstructed by utilizing the fiber bundle tracking technology, and effective clinical statistical characteristics can be extracted from the structure, so that the structure is used for medical diagnosis, related research and the like.
Diffusion tensor imaging (Diffusion Tensor Imaging, DTI) is one of the earliest proposed diffusion models, and is now widely used in clinical and medical research. The modeling method is simple and stable, but the adopted second-order symmetrical tensor can only describe single average fiber trend in the voxels, but cannot be used for describing complex fiber structures in the voxels with fiber crossing, branching and the like.
Currently, there are methods that can estimate multiple fiber orientations within a voxel by more complex models, thus solving the limitations of diffusion tensor imaging methods. These methods mainly include: multi-tensor model, high order diffusion tensor, Q-Ball imaging, diffusion spectrum imaging, spherical deconvolution (Spherical Deconvolution, SD), independent component analysis, hybrid diffusion imaging, and the like. The SD-based method has the advantages of no need of specifying the number of fiber distributions, high calculation efficiency, capability of estimating the trend distribution of the fibers in the voxels under the imaging condition of low angular resolution, and the like, and is widely applied, but the SD-based method is very sensitive to noise.
Disclosure of Invention
The invention provides a non-local mean value smoothing method for a magnetic resonance diffusion weighted image, which can solve the problem that the accuracy of an estimation result is poor due to the fact that the existing SD method is directly used for estimating a fiber structure in a voxel and is easily influenced by noise.
A non-local mean smoothing method for a magnetic resonance diffusion weighted image comprises the following steps:
s1, selecting a filtering parameter h and a neighborhood range R V ，R N For each voxel p i Is defined in the direction g of the diffusion-weighted gradient j Is a magnetic resonance diffusion weighting signal S (p i ，g j ) Performing smoothing operation of S2 and S3; h has a value of 1-20, R V The value range of R is 8-15 N The value range of (2) is 3-7;
s2, for voxel p i R of (2) V ×R V ×R V Neighborhood V pi Each voxel p of (3) k Calculating p according to equation (1) k Is different from the diffusion weighted gradient direction g l Is a diffusion weighted signal pair S (p i ，g j ) Weights ω (p) i ，g j ，p k ，g l )；
In equation (1), N pi And N pk Respectively voxel p i And p k R of (2) N ×R N ×R N Neighborhood, Z D For normalization constants, the distance Dis is calculated as in equation (2):
in equation (2), N D For neighborhood N pi The number of medium voxels, whose value is equal to the neighborhood N pk Number of medium voxels, q m And r m Respectively is neighborhood N pi And N pk The distance delta is calculated according to equation (3) for the mth voxel in (a):
s3, calculating the S (p) according to the equation (4) i ，g j ) Smoothing the non-local mean value;
in equation (4), G is the set of all diffusion-weighted gradient directions.
The invention provides a non-local mean smoothing method for a magnetic resonance diffusion weighted image, which can effectively solve the problem that the accuracy of an estimation result is poor due to the fact that the existing SD method is directly used for estimating a fiber structure in a voxel and is easily influenced by noise when being combined with the SD method; the accuracy of the estimation result of the fiber structure in the voxels can be effectively improved. The accuracy of the method is improved by about 200% compared with that of the method directly adopting the SD method by combining the method and the SD method and using the estimated intra-voxel fiber structure; compared with the method that the classical non-local mean method (NLM) is adopted to smooth the diffusion weighted image and then the SD method (NLM+SD) is adopted to estimate and obtain the intra-voxel fiber structure, the accuracy is improved by about 36%.
Drawings
FIG. 1 is a schematic view of an actual intra-voxel fiber structure of simulation data provided by an embodiment of the present invention;
FIG. 2 is a schematic diagram of an intra-voxel fiber structure obtained directly by SD method;
FIG. 3 is a schematic view of an intra-voxel fiber structure obtained by NLM+SD method;
fig. 4 is a schematic representation of the intra-voxel fiber structure obtained by the method of the present invention in combination with the SD method.
Detailed Description
One embodiment of the present invention will be described in detail below with reference to the attached drawings, but it should be understood that the scope of the present invention is not limited by the embodiment.
In this embodiment, a gaussian mixture model method is adopted to simulate and survive a set of magnetic resonance diffusion weighted image simulation data with resolution of 16×16, the set of data comprises 30 magnetic resonance diffusion weighted images with different diffusion weighted gradient directions, the diffusion sensitivity factor b value is 1000, the actual fiber trend of each voxel is shown in fig. 1, and each line segment in the figure represents one trend of the fiber in the voxel where the line segment is located. And adding rice noise with certain intensity into the generated simulation data to obtain a group of simulation magnetic resonance diffusion weighted image data with a signal-to-noise ratio of 25. The non-local mean smoothing method for the magnetic resonance diffusion weighted image is used for processing the group of noisy data, and comprises the following steps of:
s1, selecting a filtering parameter h and a neighborhood range R V ，R N For each voxel p i Is defined in the direction g of the diffusion-weighted gradient j Is a magnetic resonance diffusion weighting signal S (p i ，g j ) Performing the smoothing operation of the step S2 and the step S3; in this embodiment, h has a value of 7, R V The value is 10, R N The value is 5.
S2, for voxel p i R of (2) V ×R V ×R V Neighborhood V pi Each voxel p of (3) k Calculating p according to equation (1) k Is different from the diffusion weighted gradient direction g l Is a diffusion weighted signal pair S (p i ，g j ) Weights ω (p) i ，g j ，p k ，g l )；
In equation (1), N pi And N pk Respectively voxel p i And p k R of (2) N ×R N ×R N Neighborhood, Z D To normalize the constant, the distance Dis is calculated according to equation (2),
in equation (2), N D For neighborhood N pi The number of medium voxels, whose value is equal to the neighborhood N pk Number of medium voxels, q m And r m Respectively is neighborhood N pi And N pk The m-th voxel of (2) and the distance delta is calculated according to the equation (3);
s3, calculating the S (p) according to the equation (4) i ，g j ) Smoothing the non-local mean value;
in equation (4), G is the set of all diffusion-weighted gradient directions.
And S4, estimating the fiber structure in the voxels by using an SD method according to the magnetic resonance diffusion weighted data smoothed by the steps.
The intra-voxel fiber structure obtained by directly adopting the SD method is shown in fig. 2, the intra-voxel fiber structure obtained by adopting the SD method (NLM+SD) after the diffusion weighted image is smoothed by adopting the classical non-local mean method (NLM) is shown in fig. 3, and the intra-voxel fiber structure obtained by adopting the method of the invention and the SD method is shown in fig. 4. The average angle errors for the results shown in fig. 2, 3 and 4 were 14.8 degrees, 7.5 degrees and 4.8 degrees, respectively. The combined use of the SD method according to the present invention improves the accuracy by about 200% and 36% respectively compared to the direct use of the SD method and the NLM+SD method.
The foregoing disclosure is merely illustrative of some embodiments of the invention, but the embodiments are not limited thereto and variations within the scope of the invention will be apparent to those skilled in the art.
Claims (1)
1. The non-local mean smoothing method for the magnetic resonance diffusion weighted image is characterized by comprising the following steps of:
s1, selecting a filtering parameter h and a neighborhood range R V ，R N For each voxel p i Is defined in the direction g of the diffusion-weighted gradient j Is a magnetic resonance diffusion weighting signal S (p i ，g j ) Performing smoothing operation of S2 and S3; h has a value of 1-20, R V The value range of R is 8-15 N The value range of (2) is 3-7;
s2, for voxel p i R of (2) V ×R V ×R V Neighborhood V pi Each voxel p of (3) k Calculating p according to equation (1) k Is different from the diffusion weighted gradient direction g l Is a diffusion weighted signal pair S (p i ，g j ) Weights ω (p) i ，g j ，p k ，g l )；
In equation (1), N pi And N pk Respectively voxel p i And p k R of (2) N ×R N ×R N Neighborhood, Z D For normalization constants, the distance Dis is calculated as in equation (2):
in equation (2), N D For neighborhood N pi The number of medium voxels, whose value is equal to the neighborhood N pk Number of medium voxels, q m And r m Respectively is neighborhood N pi And N pk The distance delta is calculated according to equation (3) for the mth voxel in (a):
s3, calculating the S (p) according to the equation (4) i ，g j ) Smoothing the non-local mean value;
in equation (4), G is the set of all diffusion-weighted gradient directions.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910367288.9A CN110163819B (en) | 2019-04-21 | 2019-04-21 | Non-local mean value smoothing method for magnetic resonance diffusion weighted image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910367288.9A CN110163819B (en) | 2019-04-21 | 2019-04-21 | Non-local mean value smoothing method for magnetic resonance diffusion weighted image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110163819A CN110163819A (en) | 2019-08-23 |
| CN110163819B true CN110163819B (en) | 2023-08-01 |
Family
ID=67633463
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910367288.9A Active CN110163819B (en) | 2019-04-21 | 2019-04-21 | Non-local mean value smoothing method for magnetic resonance diffusion weighted image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110163819B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104616270A (en) * | 2015-03-03 | 2015-05-13 | 哈尔滨工业大学 | Multi-tensor-based magnetic resonance diffusion weighted image structure adaptive smoothing method |
| CN105574849A (en) * | 2015-11-25 | 2016-05-11 | 天津大学 | Diffusion kurtosis tensor based white matter microstructure feature visualization method |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2141506B1 (en) * | 2008-07-01 | 2019-04-03 | The Regents of The University of California | Identifying fiber tracts using magnetic resonance imaging (MRI) |
| WO2010073251A2 (en) * | 2008-12-25 | 2010-07-01 | Medic Vision - Brain Technologies Ltd. | Denoising medical images |
| EP2458397B1 (en) * | 2010-11-24 | 2016-11-16 | Universite de Rennes 1 | Diffusion MRI for detecting a direction of at least one fibre in a body |
| CN104520897B (en) * | 2012-08-06 | 2018-05-15 | 皇家飞利浦有限公司 | Picture noise reduces and/or image resolution ratio improves |
| EP2933743B1 (en) * | 2014-04-18 | 2019-05-08 | Université de Rennes 1 | Method of characterizing molecular diffusion within a body from a set of diffusion-weighted magnetic resonance signals and apparatus for carrying out such a method |
| CN103985099B (en) * | 2014-05-30 | 2017-01-25 | 成都信息工程学院 | Dispersion tensor magnetic resonance image tensor domain non-local mean denoising method |
| CN104200442B (en) * | 2014-09-19 | 2017-11-21 | 西安电子科技大学 | Non-local mean MRI image denoising method based on improved canny rim detections |
| CN104574298B (en) * | 2014-12-25 | 2018-09-28 | 天津大学 | A kind of noise-reduction method of more b values diffusion weightings images based on mutual information |
-
2019
- 2019-04-21 CN CN201910367288.9A patent/CN110163819B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104616270A (en) * | 2015-03-03 | 2015-05-13 | 哈尔滨工业大学 | Multi-tensor-based magnetic resonance diffusion weighted image structure adaptive smoothing method |
| CN105574849A (en) * | 2015-11-25 | 2016-05-11 | 天津大学 | Diffusion kurtosis tensor based white matter microstructure feature visualization method |
Non-Patent Citations (2)
| Title |
|---|
| 基于SVD的快速组匹配"磁共振指纹"新方法;黄敏等;生物医学工程研究(第02期);全文 * |
| 灵敏度编码磁共振谱成像(SENSESI)技术及图像重建方法;黄敏等;北京生物医学工程(第03期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110163819A (en) | 2019-08-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Descoteaux et al. | Deterministic and probabilistic tractography based on complex fibre orientation distributions | |
| St-Onge et al. | Surface-enhanced tractography (SET) | |
| Garyfallidis et al. | Dipy, a library for the analysis of diffusion MRI data | |
| JP5123191B2 (en) | Diffusion tensor imaging using highly constrained image reconstruction methods | |
| Reisert et al. | Fiber continuity: an anisotropic prior for ODF estimation | |
| CN104574298B (en) | A kind of noise-reduction method of more b values diffusion weightings images based on mutual information | |
| US20030053669A1 (en) | Magnetic resonance angiography method and apparatus | |
| US7034531B1 (en) | Diffusion MRI using spherical shell sampling | |
| US8417005B1 (en) | Method for automatic three-dimensional segmentation of magnetic resonance images | |
| CN111445546A (en) | Image reconstruction method and device, electronic equipment and storage medium | |
| Ehricke et al. | Regularization of bending and crossing white matter fibers in MRI Q-ball fields | |
| EP2791907B1 (en) | Method of reconstructing a signal in medical imaging on the basis of perturbed experimental measurements, and medical imaging device implementing this method | |
| Aja-Fernández et al. | Validation of deep learning techniques for quality augmentation in diffusion MRI for clinical studies | |
| CN111899213B (en) | Cerebrovascular reconstruction method based on high-angle-resolution diffusion imaging | |
| CN110163819B (en) | Non-local mean value smoothing method for magnetic resonance diffusion weighted image | |
| Pai et al. | Evaluation of fiber bundles across subjects through brain mapping and registration of diffusion tensor data | |
| Descoteaux et al. | Diffusion tensor sharpening improves white matter tractography | |
| Nedjati-Gilani et al. | Regularized super-resolution for diffusion MRI | |
| Shi et al. | Diffusion coefficient orientation distribution function for diffusion magnetic resonance imaging | |
| Kop et al. | Kidney segmentation from ultrasound images using gradient vector force | |
| CN112581385A (en) | Diffusion kurtosis imaging tensor estimation method, medium and equipment based on multiple prior constraints | |
| CN108961171B (en) | Mammary gland DTI image denoising method | |
| Gonzalez et al. | Modeling diffusion‐weighted MRI as a spatially variant Gaussian mixture: Application to image denoising | |
| CN110197498B (en) | Non-convex regularized deconvolution-based intra-voxel multi-fiber trend estimation method | |
| Stamm et al. | Diffusion directions imaging (DDI) |
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 |