CN103584864A - Magnetic resonance imaging method and device - Google Patents

Abstract

The invention discloses a magnetic resonance imaging method and device. The magnetic resonance imaging method comprises collecting through a PROPELLER (Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction) algorithm to obtain a plurality of K-space bars; performing Fourier transformation on the K-space bars to obtain temporary reconstructed images which are corresponding to the K-space bars respectively; enabling the temporary reconstructed image which is corresponding to one K-space bar to be served as a reference image, enabling the temporary reconstructed images which are corresponding to other temporary reconstructed images to be served as images to be performed registration and working out optimal motion parameters of the images to be performed registration relative to the reference image through image registration; correcting the K-space bars which are obtained before the temporary reconstructed images according to obtained optimal motion parameters; rearranging the corrected K-space bars, performing the Fourier transformation on a rearranged result and obtaining imaging images. The magnetic resonance imaging method is less influenced by acquisition parameters such as the echo train length and provides possibilities for eliminating motion artifacts due to the fact that estimation of motion parameters is performed based on registration of image fields.

Description

A kind of MR imaging method and device

Technical field

The present invention relates to mr techniques field, relate in particular to a kind of MR imaging method and device.

Background technology

Nuclear magnetic resonance (MRI, Magnetic Resonance Imaging), because the advantages such as not damaged and multiparameter imaging have obtained clinical practice widely, is one of important detection methods of current clinical radiology.Yet nuclear magnetic resonance is because the time of data acquisition is longer, imaging process is subject to motion artifacts, thereby motion artifacts appears in image, and then may affect doctor's diagnosis.Therefore, how effectively to overcome the impact of motion on imaging, be one of the focus of MRI investigation and technical barrier always.

Conventionally use the enhancing Reconstructed cycle to rotate overlapping parallel lines (PROPELLER, Periodically Rotated Overlapping Parallel Lines with Enhanced Reconstruction) method at present.PROPELLER method is by utilizing the data in K space center overlap sampling region to estimate the movable information of examinate in gatherer process, thereby realizes the motion compensation to K space bar; Yet, traditional PROPELLER algorithm for reconstructing is because echo train can not be long, and the line number that single K space bar comprises is limited, causes the data volume in overlap sampling region too little, inevitably reduce precision and the robustness of kinematic parameter estimation, finally affected the eradicating efficacy of motion artifacts.

Summary of the invention

The main technical problem to be solved in the present invention is that a kind of MR imaging method and device are provided.

According to the embodiment of the present invention aspect, a kind of MR imaging method is provided, comprising: data acquisition step, utilize PROPELLER algorithm to gather MR data, obtain a plurality of K space bar; Interim reconstruction procedures, converts each K space bar, obtains corresponding with each K space bar respectively interim reconstruction image; Calculation of parameter step, usings interim reconstruction image corresponding to certain K space bar as with reference to image, and interim reconstruction image corresponding to other K space bar, as image subject to registration, calculates image subject to registration with respect to the optimal movement parameter of reference picture by image registration; Revise step, the K space bar obtaining before to interim reconstruction procedures according to the optimal movement parameter obtaining is revised; Reestablishment imaging step, resets revised K space bar, and carries out Fourier transformation to resetting result, obtains image.

Another aspect according to the embodiment of the present invention, provides a kind of MR imaging apparatus, comprising: data acquisition module, for utilizing PROPELLER algorithm to gather MR data, obtains a plurality of K space bar; The interim module of rebuilding, for each K space bar is converted, obtains corresponding with each K space bar respectively interim reconstruction image; Parameter calculating module, for usining interim reconstruction image corresponding to certain K space bar as with reference to image, interim reconstruction image corresponding to other K space bar, as image subject to registration, calculates image subject to registration with respect to the optimal movement parameter of reference picture by image registration; Correcting module, revises for the K space bar interim reconstruction module being received according to the optimal movement parameter obtaining; Reestablishment imaging module, for revised K space bar is reset, and carries out Fourier transformation to resetting result, obtains image.

The invention has the beneficial effects as follows: by the registration based on image area, carry out the estimation of kinematic parameter, it is subject to the impact of the acquisition parameters such as echo train legth less, for effectively eliminating motion artifacts, provide probability.

Accompanying drawing explanation

Fig. 1 is the schematic flow sheet of the MR imaging method of the embodiment of the present invention 1;

Fig. 2 is the sample track schematic diagram in the K space of an embodiment of the present invention;

Fig. 3 is the processing procedure schematic diagram of the phasing of an embodiment of the present invention;

Fig. 4 is the schematic flow sheet of the MR imaging method of the embodiment of the present invention 2;

Fig. 5 is the structural representation of the MR imaging apparatus of the embodiment of the present invention 3.

The specific embodiment

Below by the specific embodiment, by reference to the accompanying drawings the present invention is described in further detail.

Embodiment 1:

As shown in Figure 1, the MR imaging method of the present embodiment comprises the steps S101 ~ S111:

Step S101, data acquisition step, adopts PROPELLER algorithm to gather MR data, obtains a plurality of K space bar;

K space refers to the frequency data space consisting of the MR data collecting after phase code and frequency coding, and the frequency space that magnetic resonance samples data form, is also frequency domain space corresponding to magnetic resonance image (MRI).

Can to MR data, sample by conventional PROPELLER algorithm, for example, first use usual manner as FSE(fast spin echo) sequence, single-shot echo planar imaging sequence etc. gather near one group of K space line K space center, be assumed to be L bar, obtain a K space bar, then the center fixed angle of every rotation around K space gathers next K space bar data in the same way, until complete the sampling in whole K space.The sample track in K space as shown in Figure 2.

Step S103, phasing step, carries out phasing to a plurality of K space bar obtaining, so that the center of each K space bar and the center superposition in K space;

Because gradient magnetic is not completely linear, and be subject to the impact of eddy current in imaging process, there is skew in the center of each K space bar and the center of image data, thereby cause Tiao center, K space not overlap with the center of image data, therefore need to proofread and correct it, by the center alignment of all K space bars.Bearing calibration can adopt conventional correcting algorithm.Consider the characteristic of Fourier transformation, skew occurs in K space center can cause phase place slowly changing of image area generation, therefore can reach by removing the phase place of image area low frequency space variation the object of correction.Embodiment obtains a phase place for image area low frequency space by initial data being applied to a two-dimentional quarter window, with the data F (x of certain K space bar _k, y _k) be example, as shown in Figure 3, be specially:

(1) by the data F (x of K space bar _k, y _k) copy portion, to the F (x copying _k, y _k) add two-dimentional quarter window function, get its result and carry out two-dimensional Fourier transform, obtain I ₁(x, y);

(2) to original F (x _k, y _k) do equally two-dimensional Fourier transform, obtain I ₂(x, y);

(3) by I ₁(x, y) and I ₂(x, y) does phase place and subtracts each other, and removes after phase center side-play amount, does two-dimentional inverse Fourier transform and can obtain the data after phasing.

After phasing, each Tiao center, K space is aligned in the central point in K space.

Certainly, if the center of each K space bar and the center of image data do not exist skew, can omit this step S103.

Step S105, interim reconstruction procedures, converts each K space bar after proofreading and correct, and obtains corresponding with each K space bar respectively interim reconstruction image;

Each K space bar comprises K space center area data, by upper and lower zero filling, the data matrix of L * N can be become to the matrix of N * N, then by two-dimensional Fourier transform, can obtain the interim reconstruction image that single space bar is corresponding; Here, L represents the line number of the phase code line that each K space bar gathers, and N represents the number of every phase code line image data.K space bar is transformed to the conversion process of interim reconstruction image, the present embodiment adopts two-dimensional Fourier transform; Certainly, if there is other mapping modes, adopt other alternative approach to be also fine.

Each K space bar can reconstruct a width intermediate images, if there is motion in the object being checked, these images are not identical, between them, there is a motion, and these kinematic parameters will be estimated just, once in motion parameter estimation out, just can be mapped to it in the bar of K space, and it is proofreaied and correct.

Step S107, calculation of parameter step, using interim reconstruction image corresponding to certain K space bar as with reference to image, and interim reconstruction image corresponding to other K space bar, as image subject to registration, calculates image subject to registration with respect to the optimal movement parameter of reference picture by image registration;

The kinematic parameter of embodiment obtains based on image area, can adopt method for registering images to calculate, the intermediate images reconstructing for K space bar, usings certain intermediate images as with reference to image, and the method by image registration calculates other intermediate images with respect to the kinematic parameter of reference picture.Conventionally select intermediate images that first K space bar reconstructs as with reference to image, this selection is also not limited to this certainly.Here, method for registering images can adopt conventional method for registering images, as the method based on template matching, method based on characteristic matching etc.; The method based on image interpolation mutual information that also can adopt the present embodiment to provide.

The method based on image interpolation mutual information in the present embodiment specifically, adopts normalized mutual information as the similarity measure of image registration.If reference picture is A, image to be calibrated is B, and in image A, the gray value of pixel P is a=A (p), and in image B, the gray value of pixel P is b=B (p), and normalized mutual information can be expressed as:

$\mathrm{MI}(A,B)=\frac{H\left(A\right)+H\left(B\right)}{H(A,B)}$

Wherein, the entropy that H (A) is image A, the entropy that H (B) is image B, H (A, B) is combination entropy, its computing formula is:

$H(A,B)=-\underset{I,J}{\mathrm{\Σ}}{\mathrm{\ρ}}_{\mathrm{ab}}(i,j)\log {\mathrm{\ρ}}_{\mathrm{ab}}(i,j)$

$H\left(A\right)=-\underset{i}{\mathrm{\Σ}}{\mathrm{\ρ}}_a\left(i\right)\log {\mathrm{\ρ}}_a\left(i\right)=-\underset{i}{\mathrm{\Σ}}\left(\underset{j}{\mathrm{\Σ}}{\mathrm{\ρ}}_{\mathrm{ab}}(i,j)\right)\log \left(\underset{j}{\mathrm{\Σ}}{\mathrm{\ρ}}_{\mathrm{ab}}(i,j)\right)$

$H\left(B\right)=-\underset{j}{\mathrm{\Σ}}{\mathrm{\ρ}}_b\left(j\right)\log {\mathrm{\ρ}}_b\left(j\right)=-\underset{j}{\mathrm{\Σ}}\left(\underset{i}{\mathrm{\Σ}}{\mathrm{\ρ}}_{\mathrm{ab}}(i,j)\right)\log \left(\underset{i}{\mathrm{\Σ}}{\mathrm{\ρ}}_{\mathrm{ab}}(i,j)\right)$

Wherein, ρ _ab(i, j) is joint probability density, can be calculated by joint histogram HIST (i, j), and its computing formula is:

$\mathrm{\ρ}(i,j)=\frac{\mathrm{HIST}(i,j)}{\underset{i,j}{\mathrm{\Σ}}\mathrm{HIST}(i,j)}$

Because image A and image B obtain through Fourier transformation, its pixel value is floating number, therefore, can not directly according to traditional algorithm, obtain joint histogram.

The present embodiment utilizes the thought of bilinear interpolation to obtain joint histogram, and its circular is: define a joint histogram matrix H IST, matrix size is N _a* N _b, be respectively the number of greyscale levels (in a kind of realization, number of greyscale levels is 256) of image A and image B, and this matrix all elements initial value be made as to 0.Travel through each pixel P, obtain pixel value a and b that two width images are ordered at P, get the maximum integer that [a] represents to be not more than a, order

$\left\{\begin{array}{c}\mathrm{HIST}\left(\right[a],[b\left]\right)=\mathrm{HIST}\left(\right[a],[b\left]\right)+\left(\right[a]+1-a)*\left(\right[b]+1-b)\\ \mathrm{HIST}\left(\right[a],[b]+1)=\mathrm{HIST}\left(\right[a],[b]+1)+\left(\right[a]+1-a)*(b-[b\left]\right)\\ \mathrm{HIST}\left(\right[a]+1,[b\left]\right)=\mathrm{HIST}\left(\right[a]+1,[b\left]\right)+(a-[a\left]\right)*\left(\right[b]+1-b)\\ \mathrm{HIST}\left(\right[a]+1,[b]+1)=\mathrm{HIST}\left(\right[a]+1,[b]+1)+(a-[a\left]\right)*(b-[b\left]\right)\end{array}\right.$

Process of image registration based on normalized mutual information can be described as: find a conversion T ₀make image F subject to registration reach maximum with the mutual information of reference picture after this conversion; And the process of registration is found optimal transformation T exactly ₀process, or be called the search procedure of optimal movement parameter.

Suppose that the motion in imaging process is the rigid motion in layer, kinematic parameter has three, is respectively translation parameters (Δ x, Δ y), anglec of rotation θ.Search optimal movement parameter is by changing this three parameter values, treat registering images and carry out spatial alternation, and according to the computational methods of aforementioned normalized mutual information, value corresponding to kinematic parameter while finding normalized mutual information to obtain maximum, this value is optimal movement parameter.Here spatial alternation can adopt conventional spatial alternation algorithm to realize, and has adopted common bilinear interpolation method to carry out the spatial alternation of image in a kind of embodiment, and concrete formula is with conventional bilinear interpolation method, and the present invention does not limit this.

Because kinematic parameter has three, while therefore finding mutual information value maximum, corresponding parameter value is exactly a problem of asking multiparameter extreme value.Can adopt the conventional algorithm of asking multiparameter extreme-value problem to realize, such as POWELL method, genetic algorithm, simulated annealing etc., also can adopt the POWELL method that the present embodiment provides to carry out searching moving parameter in conjunction with one dimension Fibonacci method, and specific algorithm is as follows:

(1) making initial motion parameter is X ⁰=(Δ x, Δ y, θ), selects three directions of search, is respectively d ^(1,1)=(1,0,0), d ^(1,2)=(0,1,0), d ^(1,3)=(0,0,1), sets the error ε allowing, with seasonal k=1.Here initial motion parameter can be made as 0 entirely.

(2) from X ^{(k, 0)}=X ^k-1set out, successively along three direction d ^{(k, 1)}, d ^{(k, 2)}, d ^{(k, 2)}search for,

$\left\{\begin{array}{cc}{X}^{(k,j)}=X^{(k,j-1)}+{\mathrm{\λ}}_j{d}^{(k,j)}& (j=\mathrm{1,2,3})\\ {\mathrm{\λ}}_j:f(X^{(k,j-1)}+{\mathrm{\λ}}_j{d}^{(k,j)})=\underset{\mathrm{\λ}}{\min }& f(X^{(k,j-1)}+{\mathrm{\λ}}_j{d}^{(k,j)})\end{array}\right.$

In above formula, ask λ _jprocess be one dimension extreme-value problem, adopt Fibonacci method to try to achieve.Make d ^{(k, 4)}=X ^{(k, 3)}-X ^{(k, 0)}, from X ^{(k, 3)}set out along d ^{(k, 4)}carry out acceleration search and obtain X ^k.

(3) as ‖ X ^k-X ^k-1‖ < ε searches for and stops, and the parameter searching is X ^k; Otherwise, make k=k+1, return to step (2).

By above-mentioned process of image registration, can obtain optimal movement parameter.Visible, the acquisition process of this optimal movement parameter only occurs in image area, does not relate to the acquisition parameter in frequency domain, that is to say, the kinematic parameter obtaining is like this subject to the impact of the acquisition parameters such as echo train legth less, for effectively eliminating motion artifacts, provides probability.

Step S109, revises step and also claims motion compensation step, according to the kinematic parameter obtaining, the K space bar obtaining through phasing step is revised;

According to kinematic parameter, each K space bar data can be revised, just can make the image after rebuilding eliminate the artifact causing due to motion.From the character of Fourier transformation, rotatablely moving of image area is equivalent to the rotation of K spatial domain, and therefore proofreading and correct rotatablely moves only needs to counter-rotate according to the anglec of rotation in the kinematic parameter obtaining the sampling angle of each K space bar.

And the translational motion of image area corresponding be that the phase place of K spatial domain changes, therefore revising translational motion need to be multiplied by former K spatial data a phase factor, is specifically calculated as follows:

Wherein, f ₀ ^(j)(u, v) j K space bar data for gathering,

be the sampling angle of j K space bar, (Δ x, Δ y) is the translation parameters in the aforementioned kinematic parameter obtaining, and (M, N) is image size dimension, f ^(j) _corr(u, v) is revised K space bar data.

Step S111, reestablishment imaging step, resets revised K space bar, and carries out Fourier transformation to resetting result, obtains image.

After need to combining, the image data of revising just can reconstruct final image.This is actually the Problems of Reconstruction of a non-Cartesian sampled data, can adopt conventional method for reconstructing, such as backprojection reconstruction, conjugate phase reconstruction, gridding reconstruction etc., and the present embodiment does not limit this.For example, can adopt common gridding method for reconstructing in a kind of embodiment, the K spatial data convolution that is about to non-Cartesian sampling is interpolated on equally distributed cartesian grid, and then rebuilds and obtain final image by Fourier transformation.

The present embodiment is the registration based on image area to the estimation of kinematic parameter, usings normalized mutual information as similarity measure, is subject to the impact of the acquisition parameters such as echo train legth less, for effectively eliminating motion artifacts, provides probability.In addition, embodiment is not subject to the impact of field intensity height, can be applicable to the magnetic resonance imaging system of any field intensity.

Embodiment 2:

As shown in Figure 4, the MR imaging method of the present embodiment comprises step S401 ~ S411, wherein step S401, S403, S405, S407, S409, S411 are similar with step S101, S103, S105, S107, S109, the S111 of embodiment 1 respectively, repeat no more.The difference of the present embodiment and embodiment 1 is, increases step S406 after step S405 and before step S407, in order to interim reconstruction image filtering.

Owing to lacking a lot of high-frequency signals, it is generally fuzzyyer that step S405 rebuilds the image obtaining, and because phase-encoding direction exists, block, image has obvious gibbs artifact, so also need further image to be carried out to Filtering Processing (being step S406).The present embodiment adopts dimensional Gaussian window function to carry out filtering, and formula is as follows:

$G(k_x,k_y)=\frac{-(k_x^2+k_y^2)}{2d^2}$

Wherein, k _xand k _ybe the coordinate to certain pixel after frequency domain by image transformation, d is constant, is Gauss's window width parameter.Then, filtered gradation of image value is normalized to [0255] interval.

The present embodiment carries out filtering to interim reconstruction image, makes subsequent calculations kinematic parameter more accurate, further for effectively eliminating motion artifacts, provides probability.In addition, embodiment is not subject to the impact of field intensity height, can be applicable to the magnetic resonance imaging system of any field intensity.

Embodiment 3:

As shown in Figure 3, the present embodiment has proposed the MR imaging apparatus corresponding with the MR imaging method of embodiment 1 or embodiment 2, comprising:

Data acquisition module, for utilizing PROPELLER algorithm to gather MR data, obtains a plurality of K space bar;

Phase correction module, for a plurality of K space bar obtaining is carried out to phasing, so that the center of each K space bar and the center superposition in K space;

The interim module of rebuilding, for each K space bar after proofreading and correct is carried out to Fourier transformation, obtains corresponding with each K space bar respectively interim reconstruction image; Or, the interim module of rebuilding is except for carrying out Fourier transformation by each K space bar after proofreading and correct, obtain corresponding with each K space bar respectively interim reconstruction image, also for after obtaining the interim reconstruction image corresponding with each K space bar, interim reconstruction image is carried out to filtering, and filtered gradation of image value is normalized;

Parameter calculating module, for usining interim reconstruction image corresponding to certain K space bar as with reference to image, interim reconstruction image corresponding to other K space bar, as image subject to registration, calculates image subject to registration with respect to the optimal movement parameter of reference picture by image registration;

Correcting module, revises for the K space bar interim reconstruction module being received according to the kinematic parameter obtaining;

Reestablishment imaging module, for revised K space bar is reset, and carries out Fourier transformation to resetting result, obtains image.

The related processing procedure of above modules is identical with the process of mentioning in aforementioned the inventive method embodiment, repeats no more.

To sum up, the K space bar that the embodiment of the present invention first collects single-shot transforms to respectively image area, by after Filtering Processing again the new method based on image interpolation mutual information carry out registration, in conjunction with the extremum search algorithm of optimizing, finally can obtain rapidly and accurately kinematic parameter, reconstruct again the image without motion artifacts after then utilizing kinematic parameter to proofread and correct image data.The embodiment of the present invention is the registration based on image area to the estimation of kinematic parameter, using the mutual information of interpolation as similarity measure, be subject to the impact of the acquisition parameters such as echo train legth less, the precision of registration is higher compared with prior art, robustness is better, can effectively eliminate motion artifacts.The present invention is not subject to the impact of field intensity height, can be applicable to the MRI imaging system of any field intensity.

It will be appreciated by those skilled in the art that, in above-mentioned embodiment, all or part of step of the whole bag of tricks can come instruction related hardware to complete by program, this program can be stored in a computer-readable recording medium, and storage medium can comprise: read only memory, random access memory, disk or CD etc.

Above content is in conjunction with concrete embodiment further description made for the present invention, can not assert that specific embodiment of the invention is confined to these explanations.For general technical staff of the technical field of the invention, without departing from the inventive concept of the premise, can also make some simple deduction or replace.

Claims (10)

1. a MR imaging method, is characterized in that, comprising:

Data acquisition step, utilizes PROPELLER algorithm to gather MR data, obtains a plurality of K space bar;

Interim reconstruction procedures, converts each K space bar, obtains corresponding with each K space bar respectively interim reconstruction image;

Calculation of parameter step, usings interim reconstruction image corresponding to certain K space bar as with reference to image, and interim reconstruction image corresponding to other K space bar, as image subject to registration, calculates image subject to registration with respect to the optimal movement parameter of reference picture by image registration;

Revise step, the K space bar obtaining before to interim reconstruction procedures according to the optimal movement parameter obtaining is revised;

Reestablishment imaging step, resets revised K space bar, and carries out Fourier transformation to resetting result, obtains image.

2. MR imaging method as claimed in claim 1, it is characterized in that, before described interim reconstruction procedures, also comprise: phasing step, a plurality of K space bar obtaining is carried out to phasing, so that the center of each K space bar and the center superposition in K space;

In described calculation of parameter step, adopt normalized mutual information as the similarity measure of image registration, kinematic parameter is optimized to search.

3. MR imaging method as claimed in claim 2, is characterized in that, the calculating of described normalized mutual information comprises:

Histogram calculation step, the joint histogram of employing bilinear interpolation mode computing reference image and image subject to registration;

Mutual information calculation procedure, calculates normalized mutual information according to the joint histogram obtaining.

4. MR imaging method as claimed in claim 3, is characterized in that, the computing formula of described joint histogram is

$\left\{\begin{array}{c}\mathrm{HIST}\left(\right[a],[b\left]\right)=\mathrm{HIST}\left(\right[a],[b\left]\right)+\left(\right[a]+1-a)*\left(\right[b]+1-b)\\ \mathrm{HIST}\left(\right[a],[b]+1)=\mathrm{HIST}\left(\right[a],[b]+1)+\left(\right[a]+1-a)*(b-[b\left]\right)\\ \mathrm{HIST}\left(\right[a]+1,[b\left]\right)=\mathrm{HIST}\left(\right[a]+1,[b\left]\right)+(a-[a\left]\right)*\left(\right[b]+1-b)\\ \mathrm{HIST}\left(\right[a]+1,[b]+1)=\mathrm{HIST}\left(\right[a]+1,[b]+1)+(a-[a\left]\right)*(b-[b\left]\right)\end{array}\right.$

Wherein, a is the pixel value of pixel P on reference picture, and b is the pixel value of pixel P on image subject to registration, and [a] represents to be not more than the maximum integer of a, and [b] represents to be not more than the maximum integer of a, and HIST is the joint histogram of reference picture and image subject to registration.

5. the MR imaging method as described in claim 2-4 any one, is characterized in that, describedly kinematic parameter is optimized to search comprises: adopt POWELL algorithm in conjunction with the maximum of one dimension Fibonacci method search normalized mutual information; When normalized mutual information reaches maximum, the value that kinematic parameter is corresponding is that image subject to registration is with respect to the optimal movement parameter of reference picture.

6. the MR imaging method as described in claim 1-5 any one, it is characterized in that, described interim reconstruction procedures also comprises: obtain, after the interim reconstruction image corresponding with each K space bar, interim reconstruction image being carried out to filtering, and filtered gradation of image value being normalized.

7. a MR imaging apparatus, is characterized in that, comprising:

Data acquisition module, for utilizing PROPELLER algorithm to gather MR data, obtains a plurality of K space bar;

The interim module of rebuilding, for each K space bar is converted, obtains corresponding with each K space bar respectively interim reconstruction image;

Parameter calculating module, for usining interim reconstruction image corresponding to certain K space bar as with reference to image, interim reconstruction image corresponding to other K space bar, as image subject to registration, calculates image subject to registration with respect to the optimal movement parameter of reference picture by image registration;

Correcting module, revises for the K space bar interim reconstruction module being received according to the optimal movement parameter obtaining;

Reestablishment imaging module, for revised K space bar is reset, and carries out Fourier transformation to resetting result, obtains image.

8. MR imaging apparatus as claimed in claim 7, is characterized in that, also comprises: phase correction module, for a plurality of K space bar that data acquisition module is obtained, carry out phasing, so that the center of each K space bar and the center superposition in K space; Described interim reconstruction module receives each K space bar after phase correction module is proofreaied and correct; Described parameter calculating module adopts normalized mutual information as the similarity measure of image registration, and kinematic parameter is optimized to search.

9. MR imaging apparatus as claimed in claim 8, is characterized in that, described parameter calculating module is optimized search to kinematic parameter and comprises: adopt POWELL algorithm in conjunction with the maximum of one dimension Fibonacci method search normalized mutual information; When normalized mutual information reaches maximum, the value that kinematic parameter is corresponding is that image subject to registration is with respect to the optimal movement parameter of reference picture.

10. the MR imaging apparatus as described in claim 7-9 any one, it is characterized in that, described interim reconstruction module also, for after obtaining the interim reconstruction image corresponding with each K space bar, is carried out filtering to interim reconstruction image, and filtered gradation of image value is normalized.

Images