Background technology
Echo-plane imaging (echo planar imaging, EPI) is the odd, even echo based on the contrary frequency readout gradient interleaved acquisition MR signal in direction.Due to the polarization effect that the magnetic susceptibility of imaging object, the factor such as chemical shift and magnetic field bump cause, K space data is caused to produce phase-modulation in phase encoding.The phase offset of signal receiving circuit and wave filter can cause occurring phase shifts between the odd, even echo of collection.In addition, the eddy current that the switch yard of high speed gradient causes also can affect the accuracy of echo sequential.All of these factors taken together causes the EPI image rebuild through Fourier transform to there is obvious artifact at phase-encoding direction, the havoc quality of image.See Fig. 1, anticipation image data is as shown in Fig. 1 (a), and the data that actual acquisition arrives are as shown in Fig. 1 (b).In Fig. 1 (b), thick line and fine rule divide into two groups, have a displacement each other.Artifact can be controlled by measures such as optimization sweep parameter, the performance improving wave filter, the homogeneity improving static magnetic field, reduction gradient eddy.But the raising due to system hardware performance is limited, low intensive artifact almost always appears in EPI image.
In order to overcome this shortcoming, there has been proposed the various artifact eliminating method based on aftertreatment.
Buonocore, at J.Magn Resorn Med2001, proposes the artifact eliminating method based on odd, even reconstruction in 45 (1): 96 ~ 108.But the method require artifact not exclusively overlapping with target image, and in real image target and artifact often overlap, therefore effect is not very good in actual applications for the method.In addition, the method also supposes that odd, even phase differential linearly changes, and this further restricts its application.K.J Lee is at J.Magn Reson Med2002, in 47 (4): 812 ~ 817, the method proposed based on convex set projection carrys out phase calibration, but his hypothetical target image and the difference of the image phase containing artifact linearly change, and suppose that the phase differential of every a line changes by same parameters simultaneously.In addition somebody proposes the method with second moment, but the method requires that image is symmetrical, requires that odd even phase differential linearly changes simultaneously, and the phase differential of all row presses same parameters change.
Method relatively more conventional at present adopts the method for phase recovery to eliminate artifact.When gathering K space data, gather two parts data altogether: answer image data and calibration data, by analyzing the calibration data collected, corresponding image data corrects, to eliminate artifact.But gather calibration data and add extra time loss, for the collection that high multiple accelerates, the time gathered spent by calibration data can limit image taking speed greatly.
Summary of the invention
The problem to be solved in the present invention is to provide a kind of K space reconstruction method and device, to avoid additionally gathering calibration data, utilizes calibration data to correct the K space data gathered.
For solving the problem, the method for the K space reconstruction that technical solution of the present invention provides, comprises the following steps:
A, repeatedly excitation plane echo imaging sequence, parallel acquisition K space data, the packet gathered is containing at least two data groups, and each data group is mutually certain angle on frequency coding direction, and has lap between each data group;
B, be the first group and the second group by the phase encoding line in described each data group according to the different Further Division in collection direction, it is contrary that the phase encoding line in described first group and the phase encoding line in described second group gather direction;
C, adds K space by each group and fixes, and the position that every group adds fashionable searching the strongest with all the other the group's correlativitys adding K space is fixed;
D, utilizes the K space data of above-mentioned steps c to carry out lack sampling data stuffing, realizes the reconstruction in K space.
Further, in described step a, described parallel acquisition K space data, the packet gathered is containing in the step of at least two data groups, and the value of the speedup factor that described each data group is corresponding is identical or different.
Further, in described step a, described each data group is mutually in the step of certain angle on frequency coding direction, the angle ranging from and is greater than 0 degree and is less than 180 degree.
Further, in described step a, described each data group is mutually in the step of certain angle on frequency coding direction, the angle ranging from 90 degree.
Further, in described step c, the position that described every group adds fashionable searching the strongest with all the other the group's correlativitys adding K space is fixed, further comprising the steps:
C1, puts into K space by a small group, transverse direction and/or vertically move this group, makes there is lap between this group and all the other groups having added K space;
C2, takes out the lap between this group and all the other groups having added K space respectively, and is unified to be interpolated into same public network grid space, calculates the similarity of described lap;
C3, repeats step c1 and c2, with obtain this group in K space corresponding to multiple position and the similarity of lap between all the other groups having added K space;
The multiple similarities obtained are contrasted by c4, determine this group to be added the position that this group is the strongest with all the other the group's correlativitys adding K space K space and fixes.
Further, in described step c2, the step of the similarity of the described lap of described calculating is specially: utilize variance evaluation method or relevancy estimating method to calculate the similarity of described lap.
Further, in described step c2, before calculating the similarity of described lap, further comprising the steps of: to the delivery value respectively of the lap between this group of described taking-up and all the other groups having added K space.
In order to solve the problem, present invention also offers a kind of K space imaging device, comprising:
Collecting unit, be suitable for by repeatedly excitation plane echo imaging sequence, parallel acquisition K space data, the packet gathered is containing at least two data groups, and each data group is mutually certain angle on frequency coding direction, and has lap between each data group;
Grouped element, being suitable for the phase encoding line in described each data group according to the different Further Division in collection direction is the first group and the second group, and it is contrary that the phase encoding line in described first group and the phase encoding line in described second group gather direction;
Correcting unit, is suitable for that each group is added K space and fixes, and the position that every group adds fashionable searching the strongest with all the other the group's correlativitys adding K space is fixed;
Reconstruction unit, is suitable for utilizing the K space data corrected through correcting unit to carry out lack sampling data stuffing, realizes the reconstruction in K space.
Technical solution of the present invention passes through repeatedly excitation plane echo imaging sequence, and gather and organize K space data more, each data group is mutually certain angle on frequency coding direction, lap is had between group and group, utilize the data of lap, the K space data gathered is corrected, realize artifact and eliminate.Technical solution of the present invention does not need additionally to gather the artifact that calibration data can realize in Echo-plane imaging and corrects, and saves the time gathering calibration data and need to expend, improves the image taking speed of magnetic resonance.
Embodiment:
Below in conjunction with drawings and Examples, the present invention is further illustrated.
Fig. 2 is the schematic flow sheet of K space reconstruction method of the present invention.Please refer to Fig. 1, described K space reconstruction method comprises the following steps:
A, repeatedly excitation plane echo imaging sequence, parallel acquisition K space data, the packet gathered is containing at least two data groups, and each data group is mutually certain angle on frequency coding direction, and has lap between each data group;
B, be the first group and the second group by the phase encoding line in described each data group according to the different Further Division in collection direction, it is contrary that the phase encoding line in described first group and the phase encoding line in described second group gather direction;
C, adds K space by each group and fixes, and the position that every group adds fashionable searching the strongest with all the other the group's correlativitys adding K space is fixed;
D, utilizes the K space data of step c to carry out lack sampling data stuffing, realizes the reconstruction in K space.
Below in conjunction with specific embodiment, technical solution of the present invention is described in further detail.
Embodiment one
Repeatedly excite EPI (multishot EPI, MS-EPI), parallel acquisition K space data.Wherein, utilize readout gradient field to switch continuously after one time radio-frequency pulse excites and gather multiple gtadient echo, fill the many phase encoding lines in K space, need repeatedly radio-frequency pulse to excite the filling that just can complete whole K space with the EPI collection of corresponding number of times and the roundabout filling of data.That carries out required for MS-EPI excites number of times, depends on K phase encode step level and echo train length (ETL).If K phase encode step level is 128, ETL=16, then needs to carry out 8 times and excite.
Before actual acquisition K space data, the K space data that will gather is divided into multiple data group, and described multiple data group is mutually certain angle between any two on frequency coding direction, and has partial data overlapping.For convenience of explanation, the K space data that the present embodiment will gather is divided into two groups, and the angle of two data groups on frequency coding direction is 90 degree, and each data group is according to certain speedup factor, certain track gathers, and is specifically 2 to be described with parallel acquisition speedup factor.
Fig. 3 is gradient schematic diagram when gathering K space data in the embodiment of the present invention on phase encoding and frequency coding direction.Please refer to Fig. 3, excite EPI pulse train, be equivalent to gradient with image data group a and data group b, image data group a and image data group b sequence used and exchange, be i.e. G
pE'=G
rO, G
rO'=G
pE, G here
rO=GX, G
pE=GY, namely in X direction, phase encoding is along Y-direction in the frequency coding direction of a group.It should be noted that, in figure 3, shown sequence is ignored layer and is selected the details such as gradient and radio-frequency pulse.
Fig. 4 is the acquisition trajectories schematic diagram of point two groups of collection K space data in the embodiment of the present invention.Please refer to Fig. 4, in figure, solid line represents the data of actual acquisition, dotted line represents deficient image data, the phase-encoding direction of a group is along Y-direction, the phase-encoding direction of b group in X direction, frequency coding direction perpendicular to phase-encoding direction, namely for the sampling shown in Fig. 4, when image data group a and data group b, the phase encoding of pulse train and frequency encoding gradient exchange.Only show the partial data in two groups of data in the figure.As shown in the figure, data group a lack sampling in the Y direction, data group b lack sampling in the X direction, data group a becomes 90 degree with data group b on frequency coding direction, and has partial data overlapping between the two.
Next, need to be optimized process to the data group a collected and data group b, be specially: be the first group and the second group by the phase encoding line in described each data group according to the different Further Division in collection direction, it is contrary that the phase encoding line in described first group and the phase encoding line in described second group gather direction; Each group is added K space fix, the position that every group adds fashionable searching the strongest with all the other the group's correlativitys adding K space is fixed.
Be Liang Ge group by each the data group Further Division in the data group a collected and data group b.For data group a, the process being Liang Ge group by a data group Further Division is described.Fig. 5 is by schematic diagram that the data group a Further Division collected is group a1 and group a2 in the embodiment of the present invention.In figure, arrow represents along a collection direction during phase encoding line image data, different according to data acquisition direction, be that group a1 and group a2, group a1 are contrary with the collection direction of the data in group a2 by data group a Further Division, the collection direction with data in a small group is identical.
The K space data collected is added the fixing schematic diagram in K space by Fig. 6 in the embodiment of the present invention.Please refer to Fig. 6, group a1, a2 and group b1, b2 are added in K space, when each group adds K space, find the position the strongest with other group's correlativitys adding K space and fix.First, choosing arbitrary group is benchmark group, as group a1, this group is added in K space fixing.Please refer to Fig. 7, then group a1 and group b1 is taken out, put into same K space, transverse direction and/or vertically move group b1, make there is lap between group a1 and group b1.The data of group a1 and group b1 lap are extracted out, obtains two data group D1 and D2, D1 and D2 is interpolated into same public network grid space, to calculate the similarity of this lap.The impact of translation motion in gatherer process, before the similarity calculating described lap, carries out delivery value to D1 and D2.Calculate the step of the similarity of described lap herein, be specially: calculated by methods such as variance evaluation or correlation estimations.The similarity of the described lap calculated is recorded.Then, by repeatedly transverse direction and/or vertically move group b1, multiple laps between acquisition group a1 and group b1, according to the step of the similarity of lap between aforementioned calculating group, calculate the similarity of the multiple laps between group a1 and group b1 obtained.Finally, similarity corresponding for the multiple laps calculated is compared, to determine optimum value, the relative position of the group b1 that described optimum value is corresponding and group a1 is exactly the strongest position of group b1 and group's a1 correlativity, according to this position, group b1 is added K space and fixes.Determine that the step of optimum value is specifically as follows herein and find variance minimum value.According to identical method, find group a2 and fix the strongest position of group's b1 correlativity, group a2 being added K space and fixes.Searching group b2 and fixed the position that between group a1, a2, correlativity is the strongest, adds K space by group b2 and fixes.
When group a2 is joined K space, find and organize the strongest position of correlativity with fixing all the other, namely group a2 and position that between group a1, b1, correlativity is the strongest is found, but because group a1 is parallel with group's a2 data acquisition direction, lap is not had between two groups, therefore when a2 group is added K space, without the need to calculating the correlativity of two groups.In like manner, when group b2 is joined K space, without the need to calculating the correlativity between group b1 and group b2.
It should be noted that, when group a1, a2, b1, b2 are added K space, unrestricted on time order and function, group a1 first can be added K space as benchmark group, also group a2, b1 or b2 can be added K space at first as benchmark group.
After the K space data collected is reentered into K space, the data be filled on K phase encode line obtain optimization.Utilize these through the K space data optimized, carry out follow-up lack sampling data filling and image reconstruction, the image artifacts because phase offset causes can be eliminated.In addition, utilize the K space reconstruction method of the present embodiment, without the need to additionally gathering calibration data, having saved acquisition time, having improve magnetic resonance imaging speed.
Corresponding above-mentioned K space reconstruction method, the present embodiment additionally provides a kind of K space reconstruction device, as shown in Figure 8, comprises with lower unit:
Collecting unit 81, be suitable for by repeatedly excitation plane echo imaging sequence, parallel acquisition K space data, the packet gathered is containing at least two data groups, and each data group is mutually certain angle on frequency coding direction, and has lap between each data group;
Grouped element 82, being suitable for the phase encoding line in described each data group according to the different Further Division in collection direction is the first group and the second group, and it is contrary that the phase encoding line in described first group and the phase encoding line in described second group gather direction;
Correcting unit 83, is suitable for that each group is added K space and fixes, and the position that every group adds fashionable searching the strongest with all the other the group's correlativitys adding K space is fixed;
Reconstruction unit 84, is suitable for utilizing the K space data corrected through correcting unit to carry out lack sampling data stuffing, realizes the reconstruction in K space.
In the present embodiment, the concrete enforcement of described K space reconstruction device can with reference to described in the present embodiment for the enforcement of the method for artifact correction in echo-planar imaging, do not repeat them here.
To sum up, K space reconstruction method provided by the invention and device, repeatedly excitation plane echo imaging sequence, gather and organize K space data more, utilize the data of lap, calculate the similarity between each data group, to determine the relative position between each data group, the data collected are added K space and fixes.Technical solution of the present invention carries out phase recovery to the data that phase offset occurs in gatherer process, realizes the correction of artifact, does not need additionally to gather calibration data simultaneously, shortens acquisition time, improve the image taking speed of magnetic resonance.
Although the present invention discloses as above with preferred embodiment; so itself and be not used to limit the present invention, any those skilled in the art, without departing from the spirit and scope of the present invention; when doing a little amendment and perfect, therefore protection scope of the present invention is when being as the criterion of defining with claims.