CN102749600B - Synthetic method of magnetic resonance multi-channel image - Google Patents
Synthetic method of magnetic resonance multi-channel image Download PDFInfo
- Publication number
- CN102749600B CN102749600B CN201210173379.7A CN201210173379A CN102749600B CN 102749600 B CN102749600 B CN 102749600B CN 201210173379 A CN201210173379 A CN 201210173379A CN 102749600 B CN102749600 B CN 102749600B
- Authority
- CN
- China
- Prior art keywords
- image
- channel
- data
- passage
- magnetic resonance
- 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
Landscapes
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention discloses a synthetic method of a magnetic resonance multi-channel image. The method includes the following steps: (1) obtaining multi-channel K space data; (2) comparing the maximum value of a module of K space data of each channel to obtain CHmin which is the number of a channel where the minimum maximum value is located; (3) conducting inverse Fourier transform on the K space data of each channel to obtain an original image of the channel; (4) enabling the image of each channel to an index phase of the image of the CHmin channel to obtain images with a phase portion only comprising coil sensitivity information; (5) conducting lowpass filtering processing on all images; (6) conducting principal component analysis on smoothened image data to obtain optimal matching filter vector; (7) conducting weighting and summation of the original image of each channel by using a matching filter to obtain a final synthetic image. The synthetic method can guarantee that the synthetic image is close to optimal signal-to-noise ratio, enables signal strength of the image to have unbiasedness, and can eliminate signal loss caused by phase oscillation.
Description
Technical field
The invention belongs to mr imaging technique field, be specifically related to a kind of synthetic method of magnetic resonance multi-channel image, for obtaining optimum signal-to-noise ratio image.
Background technology
Because magnetic resonance has the advantage such as high-resolution, "dead" injury, more and more extensive in application that is clinical and engineering field, along with phased array techniques is applied in magnetic resonance imaging, imaging time and signal noise ratio (snr) of image are obtained for optimization.The quality of multichannel image synthetic method is directly connected to the quality of final composograph.
The susceptibility of phased array surface coils all has locality, only has the tissue of close coil just to have high signal to noise ratio (S/N ratio), sharply declines away from coil region signal to noise ratio (S/N ratio), is obtained the image of even intensity by the surface coils synthesizing multiple diverse location.Optimum synthetic method is the sensitivity distribution figure obtaining each surface coils, is obtained the image of signal to noise ratio (S/N ratio) optimum by the image summation of its each passage of weighting.
The sensitivity distribution of coil to ask method to have a variety of, such as according to coil size positional information, utilize and directly obtain than Sa farr's law difficult to understand; The also method of useful magnetic resonance prescan: the full FOV image first gathering one group of low resolution, obtains the sensitivity distribution of corresponding coil divided by a benchmark image by the image of each passage.But in practical situations both, be difficult to Obtaining Accurate coil relative to tested accurate location information, and cause the sensitivity distribution figure tried to achieve not mate with image due to tested motion in scanning process.
Because coil sensitivities distribution plan not easily obtains, instead a kind of common method is quadratic sum (SOS) method, and the image of each passage square is sued for peace by again, finally extracts square root, and obtains the image of a signal to noise ratio (S/N ratio) suboptimum.Although it has the advantage being simple and easy to realize, there are some problems, as the general excess kurtosis of signal intensity, the noise of no signal part significantly strengthens.
43rd volume 682-690 page of the magnetic resonance magazines of people in medical science in 2000 such as David (David) has delivered a kind of synthetic method of adaptive matched filter, obtain signal to noise ratio (S/N ratio) close to optimum composograph, and change imaging time drawn game area image quality can be regulated by parameter, but their this method is for some special circumstances, noise as certain passage is very low, or the noise of each passage is all in varying level, the inhibition of noise will obviously reduce.The 47th volume 539-548 page of visiing the magnetic resonance magazine of the people such as moral in medical science in 2002 has delivered the hyperchannel synthetic method that a kind of coil sensitivities based on image area is estimated, by carrying out the sensitivity distribution image that low-pass filtering obtains Noise hardly to each channel image, the method of then carrying out similar square of summation is synthesized, they can obtain signal to noise ratio (S/N ratio) close to optimum composograph, and there is the unbiasedness of signal intensity, but ought testedly organize the magnetic susceptibility variation in some region acutely to cause image phase to be vibrated, low-pass filter can filter part useful signal, the signal intensity of the respective regions of final composograph can reduce.
Summary of the invention
The present invention seeks to: the synthetic method that a kind of magnetic resonance multi-channel image is provided, this synthetic method has more good robustness, the image close to optimum signal to noise ratio (S/N ratio) can be obtained, there is signal intensity unbiasedness simultaneously, the passage having special noise for some has same inhibition, can be good at the loss of signal eliminated because phase oscillation brings.
Technical scheme of the present invention is: a kind of synthetic method of magnetic resonance multi-channel image, it is characterized in that comprising the following steps:
1) the original frequency domain vector matrix data of each passage of magnetic resonance are obtained, i.e. K space data;
2) compare the maximal value of the mould of each passage K space data, obtain the port number CHmin at minimum maximal value place;
3) respectively inverse Fourier transform is done to the image of each passage, obtain original image matrix;
4) original image matrix of each passage is divided by the index PHASE DISTRIBUTION figure of CHmin channel image, obtains the image array that phase bit position only comprises Coil sensitivity information;
5) low-pass filtering treatment is carried out to this group image array, obtain the image array group after smoothing denoising;
6) principal component analysis (PCA) is carried out to the image array group after smoothing denoising, obtain the Optimum Matching filter vector of carrying out multichannel image synthesis;
7) each passage raw image data utilizes corresponding matched filter value to be weighted summation, and one group of wave filter can be applied to the synthesis of a point, also may be used for the synthesis of a zonule, thus obtains final composograph.
The concrete steps of the low-pass filtering treatment described in main technical schemes of the present invention are as follows:
1) by image domain data through discrete Fourier transformation to K spatial domain;
2) maximal value in more each passage K space, obtains the coordinate position (x, y) of maximum maximal value place channel C Hmax;
3) to the K space data of all passages all centered by (x, y) coordinate, cut out the data outside fixing square region, retain K space center data;
4) to above-mentioned steps 3) in block after the K space center data that obtain to carry out Kaiser window level and smooth, to remove gibbs artifact (gibbs ring);
5) zero filling is carried out to the K space center data of blocking after level and smooth, extend to original size, and coordinate position before ensureing K space center data maximums and not blocking is consistent;
6) to above-mentioned steps 5) in zero filling expand after the K space center data that obtain carry out inverse Fourier transform, obtain level and smooth image domain data.
The concrete steps of the principal component analysis (PCA) described in main technical schemes of the present invention are as follows:
1) phased array image domain data is first obtained, correlation matrix is asked to each pixel corresponding in each channel image, and the correlation matrix of all pixels in the region obtaining fixed size centered by this pixel, then expectation is asked to correlation matrixes all in this region;
2) successively to all correlation matrix eigs, obtain eigenvalue of maximum characteristic of correspondence vector, this proper vector is the principal component vector of correlation matrix, is also called the filter factor of respective center pixel, is also weighting coefficient.
Advantage of the present invention is:
The present invention obtains the method for susceptibility information and matched filter by combining image matrix, provide a kind of very stable composograph that can obtain close to optimum signal to noise ratio (S/N ratio), there is signal intensity unbiasedness simultaneously, the passage having special noise for some has same inhibition, eliminates the loss of signal because phase oscillation brings; And the present invention can need regulating parameter to strengthen region with the effect controlling composition algorithm time and application according to specific circumstances, there is the feature of flexible and changeable and good robustness.
Accompanying drawing explanation
Below in conjunction with drawings and Examples, the invention will be further described:
Fig. 1 is the synthetic method algorithm outline flowchart of a kind of magnetic resonance multi-channel image of the present invention.
Embodiment
Embodiment: shown in composition graphs 1, the synthetic method of a kind of magnetic resonance multi-channel image provided by the invention, its main technical step can brief summary be: 1) obtain multichannel K space data; 2) compare the maximal value of the mould of each passage K space data, obtain the port number CHmin at minimum maximal value place; 3) original image that inverse Fourier transform obtains this passage is carried out to the K space data of each passage; 4) image of each passage is divided by the index phase place of CHmin channel image, obtains the image that phase bit position only comprises Coil sensitivity information; 5) low-pass filtering treatment is carried out to all images; 6) principal component analysis (PCA) is carried out to the view data after level and smooth, obtain Optimum Matching filter vector; 7) each passage raw image data utilizes matched filter weighted sum, obtains final composograph.
Certainly, this enforcement synthesizes example with nuclear magnetic resonance four-way image, and its synthetic method is described in detail in detail, and here is the step of specific embodiments:
1. first obtain the original frequency domain vector matrix data of each passage of magnetic resonance, i.e. K space data matrix, the K space data of each passage, resolution is 256*256, and the signal noise ratio (snr) of image of each passage is lower, is convenient to like this compare.
2. compare the maximal value of the mould of each passage K space data, obtain the port number CHmin at minimum maximal value place;
3. respectively inverse Fourier transform is done to the image of each passage, obtain original image matrix;
4. the original image matrix of each passage is divided by the index PHASE DISTRIBUTION figure of CHmin channel image, obtains the image array that phase bit position only comprises Coil sensitivity information:
Ci represents the view data of i-th passage, and CCHmin represents the view data of CHmin passage, || represent the amplitude portion of data.
5. image domain data step 4 obtained
through discrete Fourier transformation to K spatial domain, the maximal value of more each passage K space data, obtains the coordinate position (x, y) of maximum maximal value place channel C Hmax; The K space data of all passages is all with (x, y) centered by coordinate, cut out the data outside fixed length square region, retain K space center data, and it is level and smooth to carry out Kaiser window to the K space data after blocking, parameter gets 4, to remove gibbs artifact (gibbs ring); Zero filling is carried out to the K space data of blocking after level and smooth, extends to original size, and coordinate position before ensureing k-space maximal value and not blocking is consistent.Finally carry out inverse Fourier transform, obtain level and smooth image domain data
<> represents and carries out low-pass filtering to data;
6. the image array after pair smoothing denoising
carry out principal component analysis (PCA): correlation matrix is asked to each pixel corresponding in each channel image, and the correlation matrix of all pixels in the region obtaining fixed size centered by this pixel, then ask its average, be expressed as follows:
The jth row of the <R> of <R (j, k) > correlation matrix, kth column element.E represents and asks expectation to data, and (x, y) represents the coordinate figure of each pixel of image, and * represents the complex conjugate asking data, and SROI represents the area-of-interest of a fixed size.
Then to correlation matrix <R> eig, eigenvalue of maximum characteristic of correspondence vector is obtained: the transposition of W=(w1 w2 w3 w4) T, T representing matrix.This characteristic vector W is the principal component vector of correlation matrix, is also called the filter factor of respective center pixel, is also weighting coefficient;
7. each passage raw image data utilizes corresponding matched filter value to be weighted summation, and one group of wave filter can be applied to the synthesis of a point, also may be used for the synthesis of a zonule, thus obtains final composograph, is expressed as follows:
I(x,y)=W(x,y)
H<C
CHmin(x,y)>
I (x, y) represents the image pixel value of composograph at position (x, y) place, and H represents and carries out conjugate transposition operation to data.
Above embodiment is only the present invention's a kind of embodiment wherein, and it describes comparatively concrete and detailed, but therefore can not be interpreted as the restriction to the scope of the claims of the present invention.It should be pointed out that for the person of ordinary skill of the art, without departing from the inventive concept of the premise, can also make some distortion and improvement, these all belong to protection scope of the present invention.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.
Claims (3)
1. a synthetic method for magnetic resonance multi-channel image, is characterized in that comprising the following steps:
1) the original frequency domain vector matrix data of each passage of magnetic resonance are obtained, i.e. K space data;
2) compare the maximal value of the mould of each passage K space data, obtain the port number CHmin at minimum maximal value place;
3) respectively inverse Fourier transform is done to the image of each passage, obtain original image matrix;
4) original image matrix of each passage is divided by the index PHASE DISTRIBUTION figure of CHmin channel image, obtains the image array that phase bit position only comprises Coil sensitivity information;
5) low-pass filtering treatment is carried out to this group image array, obtain the image array group after smoothing denoising;
6) principal component analysis (PCA) is carried out to the image array group after smoothing denoising, obtain the Optimum Matching filter vector of carrying out multichannel image synthesis;
7) each passage raw image data utilizes corresponding matched filter value to be weighted summation, and one group of wave filter can be applied to the synthesis of a point, also may be used for the synthesis of a zonule, thus obtains final composograph.
2. the synthetic method of a kind of magnetic resonance multi-channel image according to claim 1, is characterized in that: the concrete steps of described low-pass filtering treatment are as follows:
1) by image domain data through discrete Fourier transformation to K spatial domain;
2) maximal value in more each passage K space, obtains the coordinate position (x, y) of maximum maximal value place channel C Hmax;
3) to the K space data of all passages all centered by (x, y) coordinate, cut out the data outside fixing square region, retain K space center data;
4) to above-mentioned steps 3) in block after the K space center data that obtain to carry out Kaiser window level and smooth, to remove gibbs artifact;
5) zero filling is carried out to the K space center data of blocking after level and smooth, extend to original size, and coordinate position before ensureing K space center data maximums and not blocking is consistent;
6) to above-mentioned steps 5) in zero filling expand after the K space center data that obtain carry out inverse Fourier transform, obtain level and smooth image domain data.
3. the synthetic method of a kind of magnetic resonance multi-channel image according to claim 1, is characterized in that: the concrete steps of described principal component analysis (PCA) are as follows:
1) phased array image domain data is first obtained, correlation matrix is asked to each pixel corresponding in each channel image, and the correlation matrix of all pixels in the region obtaining fixed size centered by this pixel, then expectation is asked to correlation matrixes all in this region;
2) successively to all correlation matrix eigs, obtain eigenvalue of maximum characteristic of correspondence vector, this proper vector is the principal component vector of correlation matrix, is also called the filter factor of respective center pixel, is also weighting coefficient.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210173379.7A CN102749600B (en) | 2012-05-30 | 2012-05-30 | Synthetic method of magnetic resonance multi-channel image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210173379.7A CN102749600B (en) | 2012-05-30 | 2012-05-30 | Synthetic method of magnetic resonance multi-channel image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102749600A CN102749600A (en) | 2012-10-24 |
CN102749600B true CN102749600B (en) | 2014-12-24 |
Family
ID=47029949
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210173379.7A Active CN102749600B (en) | 2012-05-30 | 2012-05-30 | Synthetic method of magnetic resonance multi-channel image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102749600B (en) |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104778341B (en) * | 2014-01-09 | 2017-08-22 | 上海联影医疗科技有限公司 | Magnetic resonance coil merges coefficient calculation method, MR imaging method and its device |
CN105629187B (en) * | 2015-12-31 | 2018-12-18 | 上海联影医疗科技有限公司 | The method and system of field uniformity are received for optimizing magnetic resonance radio frequency coil |
CN106199473B (en) * | 2016-06-24 | 2019-04-12 | 天津大学 | A kind of more b value diffusion magnetic resonance imaging optimization methods based on noise Ratio Weighted |
CN107783067B (en) * | 2016-08-30 | 2020-02-07 | 上海联影医疗科技有限公司 | Magnetic resonance imaging method |
CN106842089B (en) * | 2017-01-23 | 2019-08-23 | 上海联影医疗科技有限公司 | A kind of MR imaging method and system |
CN109581254B (en) | 2017-09-29 | 2021-07-30 | 西门子(深圳)磁共振有限公司 | Phase deviation obtaining method and system and phase calibration method and system |
CN108090937B (en) * | 2018-02-09 | 2021-03-19 | 奥泰医疗***有限责任公司 | Self-adaptive correction method for magnetic resonance diffusion weighted imaging |
CN109938733B (en) | 2019-05-14 | 2020-06-30 | 浙江大学 | High-quality magnetic resonance image synthesis method |
CN110638433B (en) * | 2019-09-08 | 2022-09-30 | 佛山协智领航科技有限公司 | Three-dimensional pulse wave image denoising method and system based on channel weighted robustness principal component analysis |
CN112001856A (en) * | 2020-07-29 | 2020-11-27 | 东软医疗***股份有限公司 | Training method of denoising model, image noise removing method and related device |
CN113985334A (en) * | 2021-11-08 | 2022-01-28 | 电子科技大学 | Method for evaluating signal-to-noise ratio of magnetic resonance scanning image |
CN115035210B (en) * | 2022-08-10 | 2022-11-11 | 天津恒宇医疗科技有限公司 | PS-OCT visibility improving method and system based on polarization multi-parameter fusion |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7141973B2 (en) * | 2003-08-13 | 2006-11-28 | National Research Council Of Canada | Magnetic resonance experiments by spatial encoding using the RF transmission coil |
CN101185571A (en) * | 2006-11-22 | 2008-05-28 | 株式会社东芝 | Magnetic resonance imaging apparatus |
JP2008307228A (en) * | 2007-06-14 | 2008-12-25 | Ge Medical Systems Global Technology Co Llc | Image formation method, location compensation method, and magnetic resonance imaging apparatus |
CN101627910A (en) * | 2008-07-17 | 2010-01-20 | 株式会社东芝 | Magnetic resonance imaging apparatus and magnetic resonance imaging method |
DE102010033329A1 (en) * | 2010-08-04 | 2012-02-09 | Siemens Aktiengesellschaft | Method for determining magnet resonance system-controlling sequence, involves carrying out optimization of multi-channel pulse train or k-area-gradient trajectory in high frequency-pulse-optimization process |
-
2012
- 2012-05-30 CN CN201210173379.7A patent/CN102749600B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7141973B2 (en) * | 2003-08-13 | 2006-11-28 | National Research Council Of Canada | Magnetic resonance experiments by spatial encoding using the RF transmission coil |
CN101185571A (en) * | 2006-11-22 | 2008-05-28 | 株式会社东芝 | Magnetic resonance imaging apparatus |
JP2008307228A (en) * | 2007-06-14 | 2008-12-25 | Ge Medical Systems Global Technology Co Llc | Image formation method, location compensation method, and magnetic resonance imaging apparatus |
CN101627910A (en) * | 2008-07-17 | 2010-01-20 | 株式会社东芝 | Magnetic resonance imaging apparatus and magnetic resonance imaging method |
DE102010033329A1 (en) * | 2010-08-04 | 2012-02-09 | Siemens Aktiengesellschaft | Method for determining magnet resonance system-controlling sequence, involves carrying out optimization of multi-channel pulse train or k-area-gradient trajectory in high frequency-pulse-optimization process |
Non-Patent Citations (2)
Title |
---|
一种基于互信息量的并行磁共振图像重建新算法;黄鑫等;《中国生物医学工程学报》;20071220;第26卷(第6期);全文 * |
磁共振部分K空间数据图像重构;方青;《中国优秀硕士学位论文全文数据库医药卫生科技辑》;20070615(第6期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN102749600A (en) | 2012-10-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102749600B (en) | Synthetic method of magnetic resonance multi-channel image | |
US6160398A (en) | Adaptive reconstruction of phased array NMR imagery | |
US11079456B2 (en) | Method of reconstructing magnetic resonance image data | |
US20140126796A1 (en) | Mri reconstruction with motion-dependent regularization | |
RU2626184C2 (en) | Method, device and system for reconstructing magnetic resonance image | |
WO2009112987A1 (en) | Coil selection for parallel magnetic resonance imaging | |
CN104781685A (en) | Image reconstruction for dynamic MRI with incoherent sampling and redundant HAAR wavelets | |
US20090115413A1 (en) | Device and method for parallel magnetic resonance imaging | |
US20120286783A1 (en) | Accelerated MRI with Nonlinear Spatial Encoding Gradients | |
CN111754416A (en) | System and method for background noise reduction in magnetic resonance images | |
WO2003050761A2 (en) | Method for compensating for motion artifacts in magnetic resonance imaging by phase line prediction | |
US9208587B2 (en) | Systems and methods for compressed sensing for multi-shell magnetic resonance imaging | |
Haldar et al. | Linear Predictability in MRI Reconstruction: Leveraging Shift-Invariant Fourier Structure for Faster and Better Imaging. | |
US20030038632A1 (en) | Method and apparatus for enhanced multiple coil imaging | |
US7956611B2 (en) | Magnetic resonance imaging apparatus and method | |
US11262424B2 (en) | Method and apparatus for hardware RF receiver channel reduction | |
Patel et al. | Medical image fusion based on multi-scaling (DRT) and multi-resolution (DWT) technique | |
RU2525747C2 (en) | Noise matching in coupled antenna arrays | |
WO2013181517A2 (en) | System and method for magnetic resonance imaging using highly accelerated projection imaging | |
CN108333545B (en) | Magnetic resonance image reconstruction method based on high-pass filtering | |
JP4481591B2 (en) | Method, system and computer product for gradient nonlinearity k-space correction | |
CN111537930B (en) | Magnetic resonance parameter imaging method and equipment based on gradient waveform adjustment | |
US9360543B2 (en) | System and method for parallel magnetic resonance image reconstruction using digital beamforming | |
US11885863B2 (en) | Simultaneous use of joint and individual information terms in multicontrast compressive sensing reconstruction | |
US20230162326A1 (en) | Method for training post-processing device for denoising mri image and computing device for the same |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C53 | Correction of patent for invention or patent application | ||
CB02 | Change of applicant information |
Address after: Xing Pu Road Suzhou City Industrial Park in Jiangsu province 215126 building K No. 128 Applicant after: SUZHOU LONWIN MEDICAL SYSTEMS CO., LTD. Address before: Xing Pu Road Suzhou City Industrial Park in Jiangsu province 215126 building K No. 128 Applicant before: Suzhou Anke Medical System Co., Ltd. |
|
COR | Change of bibliographic data |
Free format text: CORRECT: APPLICANT; FROM: SUZHOU ANKE MEDICAL SYSTEM CO., LTD. TO: SUZHOU LANGRUN MEDICAL SYSTEM CO., LTD. |
|
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |