Patent Application
20070035302
This is a Continuation Application of PCT Application No. PCT/JP2005/018822, filed Oct. 12, 2005, which was published under PCT Article 21(2) in Japanese.
This application is based upon and claims the benefit of priority from prior Japanese Patent Application No. 2004-326640, filed Nov. 10, 2004, the entire contents of which are incorporated herein by reference.
1. Field of the Invention
The present invention relates a phase unwrapping evolution method of subjecting phases of plural phase data to unwrapping evolution to be in a range of 2π, respectively, to decide a phase and relates to, for example, a technique that is applied at the time of creation of a phase map in data processing in an MRI (Magnetic Resonance Imaging) apparatus.
2. Description of the Related Art
As it is well known, in an MRI apparatus, in order to perform shimming processing for correcting a non-uniform component of a magnetic field, a phase map representing a distribution of fluctuation amounts of phases due to non-uniformity of the magnetic field is created. In this case, it is necessary to perform phase unwrapping evolution for calculating a true value φ+2nπ (n: integer) of a phase φ calculated by unwrapping a phase in a range of 2π.
As a conventional example of the phase unwrapping evolution, first, there is a method described in a literature “Radiology 1994; 192:555-561”. The method described in the literature is a method of repeating steps of deciding a phase of phase data at certain one point, referring to the phase of the phase data decided after that, and applying unwrapping evolution to any one of phase data for which a value is not decided yet to decide a phase. In other words, this method is characterized by performing, on the premise that a phase change is not steep between phase data in the neighborhood, estimation of a true phase of undecided phase data present near data for which a phase is decided.
However, actually, data having a steep phase change compared with data around the data such as phase data located at a boundary of tissues or phase data including noise may be present. When such phase data is processed according to the method described above, since the phase unwrapping evolution is performed with reference to decided data in the neighborhood, it is highly likely that a true phase is miss-estimated. In addition, since processing is advanced assuming that a phase of decided phase data is correct, if unwrapping evolution fails once to decide a phase while the true phase remains miss-estimated, miss-estimation of a true value occurs in series in the following unwrapping evolution that refers to the miss-estimated phase data.
As it is evident from this example, in the conventional method, since the phase unwrapping evolution is applied to phase data with reference to decided data in the neighborhood, it is highly likely that a true phase is miss-estimated. In addition, since processing is advanced assuming that a decided phase of phase data is correct, if unwrapping evolution fails once and a phase is decided while the true phase remains miss-estimated, miss-estimation of a true value occurs in series in the following unwrapping evolution that refers to the miss-estimated phase data. In this way, in the conventional method, since the phase unwrapping evolution is sequentially performed in order to match phases among respective data, there is a problem in that failure of the phase unwrapping evolution occurs in a wide range when there is only one data having a steep phase change compared with data around the data described above.
Thus, in order to finally reduce an influence exerted by such data, a method of performing phase unwrapping evolution between fragmented areas to match phases among plural phase data rather than applying phase unwrapping evolution to the data one by one is proposed in JP-A-2002-306445.
However, the method described in the literature is on the premise that the phase unwrapping evolution is performed in the respective areas before performing the phase unwrapping evolution among the areas and there is no reference to how unwrapping evolution inside the area is performed. Consequently, it is surmised that problems same as those in the method of the sequential phase unwrapping evolution described above are not solved and still present in the inside of the respective areas.
It is an object of the invention to provide a phase unwrapping evolution method that can suppress an influence of phase data, a true value of which tends to be miss-estimated, and perform robust and highly stable phase unwrapping evolution and a magnetic resonance imaging apparatus that can create an ideal phase map using the method and appropriately correct a non-uniform component of a magnetic field.
A phase unwrapping evolution method according to the invention is a phase unwrapping evolution method of subjecting phases of plural phase data in a target area to unwrapping evolution, respectively, to decide a phase, the phase unwrapping evolution method including: a first step of grouping the phase data under a predetermined condition; and a second step of matching phases among at least two groups grouped in the first step, wherein, in the second step, phases of all phase data in an arbitrary group subjected to the grouping are shifted by a predetermined amount and subjected to unwrapping evolution by a unit of group and, then, phases are matched between the group and the other groups.
A magnetic resonance imaging apparatus according to the invention is an apparatus that performs shimming processing with an object of correcting a non-uniform component of a magnetic field, the magnetic resonance imaging apparatus including: grouping means that groups phase data of a phase map representing a distribution of fluctuation amounts of phases due to non-uniformity of the magnetic field under a predetermined condition; and phase matching means that matches phases among at least two groups grouped by the grouping means, wherein the phase matching means shifts phases of all phase data in an arbitrary group subjected to the grouping by a predetermined amount and performs unwrapping evolution by a unit of group and, then, matches phases between the group and the other groups.
An embodiment of the invention will be hereinafter explained with reference to the drawings.
A phase unwrapping evolution method proposed in the invention roughly consists of three steps as shown in
A processing order of the method will be explained. First, grouping of phase data is performed. Subsequently, steps S1 to S3 of performing the phase unwrapping evolution targeting the groups as described above and, then, performing merging of the target groups are repeatedly applied until an ending condition is satisfied in judgment step S4.
The roughly divided three steps will be explained, respectively.
First, a step of grouping phase data will be explained. Rules in determining whether phase data, a belonging group of which is not determined yet, should be included in a certain group will be explained.
(1) If a phase difference between phase data belonging to a certain group and phase data, a belonging group of which is not determined yet, is within a certain threshold value, the un-belonging data is included in the same group. If the phase difference exceeds the threshold value, the un-belonging data is included in another group.
(1-1) As a minimum condition, the threshold value is set to a value smaller than 180 [deg] that is generally used as an index indicating that phase unwrapping occurs if a difference between data is larger than the value. As the threshold value is smaller, the grouping is performed more finely. As an example, it is desirable to set the threshold value to 30 to 90 [deg]. However, it is needless to mention that, if the minimum condition described above is satisfied, the threshold value is not limited to this. Belonging-decided phase data, for which a phase difference from un-belonging phase data is calculated, is desirably data located in the neighborhood of the un-belonging data. For example, in two dimensions and three dimensions, data adjacent to the un-belonging data in the four directions or obliquely adjacent to the un-belonging data is set as data for which a difference is calculated.
(1-2) In (1-1) above, a threshold value for determining whether groups are the same is set in advance and data for which a phase difference is calculated is limited to data in the neighborhood of the un-belonging data. An example of grouping different from this is described. As data for which a difference from un-belonging phase data is calculated, data in an arbitrary position not only in the neighborhood such as phase data located in the center of gravity of groups is set as a pair for a phase difference calculation together with the un-belonging phase data. In this case, a threshold value may be fixed to a certain value in the same manner as (1-1) and used. However, it is more effective to use a value weighted according to a distance between both the data as a threshold value. There is also a method not using a phase difference as an index. As an example of the method, it is also possible that an approximate straight line fitted according to a value of a phase is drawn between belonging group data and un-belonging group data and a condition that, if fluctuation in inclination of an approximate formula is equal to or smaller than a certain value before and after including the un-belonging group data at an end of the approximate straight line, the data are included in the same group and, if the fluctuation is not equal to or smaller than the value, the data are included in different groups is adopted.
(2) For grouping of phase data, a belonging group of which is not determined yet, one belonging-decided phase data is associated with the phase data in (1). However, as a method other than (1), a condition that, if a difference between an average value of phase data of a group and a phase of un-belonging phase data is within a certain threshold value, the data are included in the same group and, if the difference is not within the threshold value, the data are included in different groups may be adopted. The threshold value in this case is naturally set to a value smaller than 180 8 deg] as in (1). However, it is more desirable to set the threshold value smaller than the threshold value regarded as desirable in (1), for example, set the threshold value to 30 [deg] or less. As the average value of phase data of a group, an average of all phase data belonging to the group may be calculated. Alternatively, an average of phase data of a part of a group may be calculated by, for example, limiting the phase data to data located in the neighborhood of un-belonging phase data.
(3) As another method, if un-belonging phase data is included in a group, a condition that, if fluctuation in phase variance of phase data in the group is within a certain range, for example, 10% before and after the un-belonging phase data, the un-belonging phase data is included in the group may be adopted. In this case, as in (2), a range in which variance is calculated may be set in the entire group or variance may be calculated with phase data of a part of the group by, for example, limiting the phase data to data in the neighborhood of the un-belonging phase data.
As an example of an actual grouping procedure, at first, since there is no phase data, a belonging group of which is determined, the procedure is started by setting phase data in an arbitrary position as a start point and giving a first group ID to the phase data at the start point. Subsequently, concerning un-belonging data adjacent to the start point in the four directions or obliquely adjacent to the phase data in two dimensions and three dimensions, it is determined whether the un-belonging phase data belongs to the same group at the start point using any one of the conditions (1) to (3). Thereafter, grouping of un-belonging phase data adjacent to any one of phase data for which belonging groups are determined in the same manner is performed using any one of the conditions (1) to (3). The work described above is repeated until groups of all the phase data are determined.
Note that, a place where a start point is set and an order of progress of grouping are not specifically designated. For example, phase data in the center may be set as a start point to trace phase data spirally or phase data at an end may be set as a start point to trace phase data in an axial direction row by row. However, an order of performing grouping has to be set to prevent omission of phase data in the grouping.
A condition for grouping may be always fixed to the same condition among the conditions (1) to (3) or may be freely changed in the middle of processing. However, it is desirable to use the condition (1-1) in (1) while the number of phase data belonging to a group is small.
The method of grouping using phases of phase data has been explained. As another example of grouping, for example, the same grouping as described above may be performed on the basis of an absolute value rather than phases of phase data. Besides, as an example of completely different grouping not using information on phase data, phase data may be classified into an arbitrary shape, for example, a rectangular in the case of two dimensions and a rectangular parallelepiped in the case of three dimensions.
In performing grouping of phase data, it is not always necessary to perform the grouping targeting phase data itself obtained first. The grouping may be performed targeting phase data subjected to some processing. As a content of the processing, for example, it is conceivable to perform arbitrary filter processing such as differential filtering or smoothing filtering, perform an arbitrary arithmetic operation, or excluding phase data with small signal value such as a background from phase data in advance according to threshold value processing.
Merging of groups will be explained with reference to an example. First, it is assumed that five groups A, B, C, D, and E are formed according to grouping of phase data. Two or more arbitrary groups, for example, A and B are selected as targets of unwrapping evolution to perform unwrapping evolution in order to match phases between both the groups.
After that, both the groups A and B are merged to form a new group F. At this stage, remaining groups are the four groups C, D, E, and F. In the same manner, unwrapping evolution may be performed targeting two or more arbitrary groups, for example, C, D, and E in order to match phases among the groups to merge the groups. In this case, as the merging of the groups, the three groups C, D, and E may be merged at a time or only a part of the groups such as C and E may be merged.
Assuming that the two groups D and E are merged to form a new group G, groups remaining at this point are C, F, and G. Subsequently, unwrapping evolution is performed targeting the groups C, F, and G and the three groups are merged to form a group H. At this point, a group of phase data is only H. The phase unwrapping evolution processing proposed in this method ends.
Note that, in this example, a condition for ending repetition of unwrapping evolution and merging is that one group is left. However, the condition is not limited to this. The ending condition may be set arbitrarily. For example, the number of times of repetition of unwrapping evolution and merging, the number of groups, the number of boundaries of groups, variance of phases at the time of end of merging, or the like is set as an index. A condition for determining groups selected as targets of the phase unwrapping evolution processing may also be set arbitrarily. As an example of the condition for selecting target groups, priorities are determined according to group IDs (in the example described above, A, B, etc.), the number of belonging phase data, the number of adjacent groups, or the like and a group with a highest priority and groups located in the neighborhood of the group are selected as targets or a combination of all target groups or groups with a minimum phase difference and variance among the groups is selected as a target.
Lastly, the phase unwrapping evolution processing targeting groups and merging of the groups will be explained. In the following description, for simplicity of explanation, it is assumed that there are only four phase data “a” to “d” on a two-dimensional plane as shown in
The phase unwrapping evolution processing according to this method is performed in accordance with an evaluation value that is calculated on the basis of a phase of a target group. A condition satisfied by the evaluation value calculated on the basis of a phase is not specifically limited. In the following description, phase unwrapping evolution will be explained with reference to several examples of the condition.
(A) An average of phase differences between phase data located in neighborhoods of a boundary of groups is set as an evaluation value. Phases of phase data in the groups are shifted by 2nπ to minimize the evaluation value.
The number of combinations of data for which a phase difference is calculated varies according to how neighborhoods of the boundary are set. However, the number of combinations is not specifically limited either. In
However, even if “n” minimizing an average of phase differences is determined, if one group is shifted by 2nπ+C (C is arbitrary) and the other group is shifted by C, a shift amount minimizing the average of phase differences has values equivalent to infinite combinations depending on a way of setting C. Thus, it is assumed that the shift amount is C=0, that is, phase shift of one group is not performed. A sign of “n” is different and a value of “n” changes depending on which of the groups A and B is subjected to phase shift. Therefore, a group for which a phase is shifted only has to be determined in accordance with, for example, a level of a priority of the group described in the explanation of merging.
In the following description, the explanation is continued assuming that a phase of phase data of the group B is shifted to minimize an average phase difference from A and combinations for calculating a phase difference are set to the 4-neighbors.
When a phase of phase data p is φp, an evaluation value αBA between the groups B and A is represented as follows.
αBA=( (φb−φa)+(φc−φa) )/2
In this case, “n” is calculated according to the following expression.
n=−int((int(αBA/π)+sign(αBA))/2) (1)
Note that int(x) means an integer part obtained by omitting the figures below a decimal point of x. A value of sign(x) is −1 when x is negative and is 1 in other cases. When “n” is calculated, 2nπ is added to phases (φb, φc, and φd) of all phase data belonging to the group B to update values of φb, φc, and φd. The group B is merged with A to form a new group C having four phase data “a” to “d” as shown in
(B) A difference of average values of phases of a part of or all of phase data belonging to a group is set as an evaluation value. The phases of the phase data in the group are shifted by 2nπ to minimize the evaluation value.
This is the same as (A) except that αBA=(φb+φc+φd)/3-φa in the example in
(C) Variance of a part of or all of phase data of a target group is set as an evaluation value. Phases of phase data in the group are shifted by 2nπ to minimize the evaluation value.
This exactly means that, instead of shifting phases to reduce a difference of averages in (B), phases are shifted to minimize variance. In the example in
Unwrapping evolution concerning one-to-one merging of groups has been explained as shown in
A method of applying merging to three or more groups by limiting targets to a part of the target groups will be explained. For simplicity of explanation, as shown in
It is considered to shift phase data of the group A to merge the phase data with B or C. In this case, plural evaluation values are calculated for each combination of merging. When it is assumed that the method (A) explained above is used as a method of calculating evaluation values and boundaries for calculating a phase difference are 8-neighbors, evaluation values αAB and αAC are calculated as αAB=φa-φb and αAC=((φa−φc)+(φa−φd))/2. “n” is calculated for a group having an optimum value among evaluation values obtained for each candidate of a merging destination. A phase of data of the group A is shifted and merged.
A condition concerning an evaluation value for determining an optimum group is not specifically limited. For example, a group having a minimum absolute value of an evaluation value may be set as a merging destination. A method of calculating an evaluation value may be any method as described above. For example, merging may be performed by unwrapping evolution for a combination in which an average or variance of phases is minimized.
(Specific Examples)
1. A Result of Acquisition of Phase Data
For simplicity of explanation, it is assumed that nine phase data in total consisting of three phase data x three phase data on a two-dimensional plane are obtained as shown in
2. Grouping of Phase Data
Before applying unwrapping evolution to obtained phase data, first, grouping of the phase data is performed. In grouping the phase data, it is assumed that, for example, a rule that, when phases of data adjacent to each other in the 4-neighbors in the four directions of the data are compared, if a difference of the phases is within 90 [deg], the data belong to the same group and, if the difference exceeds 90 [deg], the data belong to different groups (the rule written in (1-1) above) is applied. As a result of performing grouping under the rule, as shown in
3. Phase Unwrapping Evolution by a Unit of Group
Phase unwrapping evolution by a unit of group that is performed after the grouping of the phase data ends will be explained assuming that, for example, the following rule is applied.
When phase unwrapping evolution is performed targeting a certain group, in a group adjacent to the target group in 4-neighbors in the four directions, unwrapping evolution for shifting phases of all data of the target group by 2nπ to minimize a phase difference of phase data located at a boundary is performed and, then, the target group and the adjacent group are merged.
One-to-one merging of groups is performed every time unwrapping evolution is performed once and the processing is continued until the groups are merged into one group.
When there are plural groups adjacent to a target group of phase unwrapping evolution in 4-neighbors in the four directions, the target group is merged with a group having a smallest average of a phase difference of phase data located at boundaries by carrying out the phase unwrapping evolution.
4. Designation of a Target Group of Unwrapping Evolution
First, a target group of unwrapping evolution is designated. An order of designation of a target group is not specifically designated. For example, as shown in
5. Groups of Target Groups
As groups to be candidates of a merging destination adjacent to the target group D in 4-neighbors, there are two groups B and C as shown in
6. Calculation of an Evaluation Value for Each Group of Target Groups
First, a phase shift amount at the time when the target group is subjected to unwrapping evolution is calculated for each group of target groups.
When calculation of a phase difference is performed in the case of a group of target groups 1 consisting of D and B shown in
In the rule determined above, with an average phase difference of boundary data after carrying out unwrapping evolution for each group of target groups as an evaluation value, one-to-one merging is performed with a pair having a smaller evaluation value. Thus, an evaluation value is calculated as follows for each of the groups of target groups.
In the case of the group of target groups 1:
|−220−(−150)|/1=70
In the case of the group of target groups 2:
|−220−(−140)|/1=80
Therefore, the group B having a smaller evaluation value is selected as a merging destination of the target group D.
7. Merging of Groups
Groups of phase data present at the point when the group D is subjected to phase unwrapping evolution and merged with the group B are three groups (A, B, and C) as shown in
8. Repeated Application of Phase Unwrapping Evolution by a Unit of Group
As shown in FIGS. 6(h) and 6(i), there are two groups of target groups for the target group C. Since an average phase difference of boundary data in a group of target groups 3 consisting of C and A is ((−140)−140)+((−80)−160))/2=−260, “n” is calculated as 1 from Expression (1) above. Therefore, in the group of target groups 3, a shift amount of data by unwrapping evolution of the target group C is +360 [deg]. On the other hand, since an average phase difference in a group of target groups 4 consisting of C and B is ((−140)−(−220))/1=80, “n” is calculated as 0. Therefore, in the group of target groups 4, a phase shift amount by unwrapping evolution of the target group C is 0. In other words, data of the target group C does not fluctuate. When calculation of an evaluation value is performed in each group of target groups after phase unwrapping evolution, in the case of the group of target groups 3, |(220−140)+(280−160)|/2=100 and, in the case of the group of target groups 4, |(−140)−(−220)|/1=80. Thus, the group of target groups 4 is optimum. Therefore, the group C is merged with the group B this time.
9. A Final Result by Application of the Proposed Method
At this point, the two groups (A and B) are still present as shown in
Note that, in the above explanation, the condition of grouping, the way of forming a pair of phase data for calculating a phase difference, the ending condition of the phase unwrapping evolution processing by a unit of group, and the like are always the same during the processing. However, the conditions are only an example. Conditions are not limited to the conditions and it is possible to adopt arbitrary conditions. It is also possible that conditions are changed in the middle of the processing.
The above explanation is an explanation of the procedure in the invention. As it is evident from the explanation, in the invention, as the unwrapping evolution processing proceeds, groups increase and information on a phase difference among the group increases. Thus, the influence of phase data that tends to be a cause of failure of unwrapping evolution gradually decreases, robustness becomes higher than that in the conventional method, and a more correct result is obtained. In addition, a group once subjected to unwrapping evolution continues to be subjected to unwrapping evolution as a new group created by merging. Thus, even if a phase is decided once, the group is subjected to unwrapping evolution many times in order to match the phase with phases of other groups. As a result, it is possible to prevent a series of occurrence of failure of unwrapping evolution. Failure of unwrapping evolution is reduced compared with the conventional method and stability of unwrapping evolution is improved.
Note that, in the above explanation, for simplicity of explanation, phase unwrapping evolution is limited to the case in which phase data is present only on two dimensions. However, the same holds true in the case in which phase data is present on three dimensions. In that case, phase data seen as neighborhoods of a boundary is simply evolved in three-dimensional directions as well.
The method may be applied to all phase data at a time. Alternatively, the method may be applied stepwise while an application area of the method is extended in such a manner that phase data is divided into certain areas and the method is applied in the respective areas independently and, then, applied in a large area. For example, it is also possible that, first, the method is independently applied by a unit of two-dimensional plane and, after phases are matched in a two-dimensional plane shape, phase data of the respective two-dimensional planes are consolidated as one group and the method is applied to three dimensions this time. In that case, an evaluation value used in phase unwrapping evolution for each stage may be continuously the same or may be changed according to a step.
Phases among phase data are matched by the application of the method explained above. However, it is likely that several offsets are still added to values of the phases as a whole. In some cases, it is conceivable that only matching among phases is not an object, a phase has some meaning, and it is desired to adjust an offset by shifting overall phase data by a certain amount in order to use the phase as an appropriate value depending on processing. In such a case, for example, it is conceivable that some evaluation values such as average values of phases are compared before and after application of the method and overall phases are shifted to minimize the evaluation values.
As another example, it is conceivable that, when the invention is applied to data processing such as shimming in an MRI apparatus, a center frequency is calculated separately and a phase is shifted by the center frequency after application of the method.
An embodiment of an MRI apparatus to which the invention is applied will be hereinafter explained.
The RF coil 3 is a coil for transmitting an RF pulse (also referred to as high-frequency magnetic field or rotating magnetic field) to a patient and receiving a magnetic resonance signal from the patient. Instead of using the RF coil 3 for both transmission and reception in this way, a transmission coil and a reception coil may be provided separately. A transmitter 5 is a device for supplying a high-frequency pulse corresponding to a Larmor frequency peculiar to a target atomic nucleus to the RF coil 3 to bring spin of the target atomic nucleus into an excited state. A receiver 6 has a function for receiving a high-frequency magnetic resonance signal, which is emitted in a process in which the excited spin relaxes, via the RF coil 3, amplifying the magnetic resonance signal, subjecting the magnetic resonance signal to quadrature phase detection, and further subjecting the magnetic resonance signal to analog/digital conversion.
A computer system 11 captures the magnetic resonance signal digitized by the receiver 6 and subjects the magnetic resonance signal to two-dimensional Fourier transformation (2DFT) to thereby reconstitute a magnetic resonance image. This image is displayed on a display unit 12. A sequencer 10 controls operation timings for the transmitter 5, the receiver 6, and the gradient magnetic field power supplies 7, 8, and 9 for the X, Y, and Z axis and executes a pulse sequence for determining a shimming value and a pulse sequence for imaging.
In this embodiment, both shimming by the shim coil and shimming called an FUC method (Field Uniformity Correction method) performed by giving an offset to a gradient magnetic field are used. The FUC method is a method of directly correcting a first-order non-uniform component of a static magnetic field by superimposing offsets on gradient magnetic fields Gx, Gy, and Gz. In this embodiment, the FUC method is used to make it possible to indirectly correct a higher order, that is, second-order non-uniform component. A principle of realizing the indirect correction is in a method of determining a shimming value. Details of the principle will be described later.
The multi-channel type shim coil 15 includes plural shim coils that correct a high-order non-uniform magnetic field component that are not directly and indirectly corrected by the FUC method and have different non-uniform magnetic field components of static magnetic fields to be corrected. In general, shim coils for thirteen channels or eighteen channels are prepared. In this embodiment, as in the FUC method, it is made possible to indirectly correct components other than a non-uniform component indirectly corrected by each of the shim coils, respectively. A shim coil power supply 16 is constituted to be capable of independently supplying electric currents (shim currents) to each of the plural shim coils of the multi-channel type shim coil 15.
A shim controller 14 captures the magnetic resonance signal digitized by the receiver 6, calculates a spatial magnetic field distribution on the basis of the magnetic resonance signal, and calculates a shimming value for each component on the basis of the magnetic field distribution. Then, the shim controller 14 supplies to the shim coil power supply 16 a shimming value of a non-uniform component to be corrected by the multi-channel type shim coil 15 while changing the shimming value according to movement of a partial area where data is collected. The shim coil power supply 16 supplies a shim current corresponding to the shimming value to a shim coil corresponding thereto.
The shim controller 14 sets a shimming value of a first-order non-uniform component as an offset value and supplies the offset value to the sequencer 10 while changing the offset value according to movement of a partial area where data is collected. The sequencer 10 adds the offset value to a criteria value and supplies the added value to the gradient magnetic field power supplies 7, 8, and 9. The gradient magnetic field power supplies 7, 8, and 9 supply a gradient magnetic field current corresponding to the added value to the XYZ axis gradient magnetic field coil 2. Consequently, a first-order component is shimmed. The shim controller 14 adjusts a reference frequency of quadrature phase detection in the receiver 6 according to a shimming value corresponding to a zero-th component, that is, shift of a resonant frequency to thereby shim the zero-th component.
An example of a method of determining a shimming value will be explained.
Shimming means correcting a non-uniform magnetic field component in order to improve uniformity of static magnetic fields in a target area as much as possible. As a method of calculating a shimming value, there are the following methods.
(1) Acquire a magnetic resonance signal from a target area without superimposing a gradient magnetic field thereon and calculate a shim current value with a longest signal attenuation time constant of the magnetic resonance signal.
(2) Acquire a magnetic resonance signal from a target area without superimposing a gradient magnetic field thereon, subject the magnetic resonance signal to Fourier transformation, and calculate a shim current value minimizing a frequency band of conversion data of the magnetic resonance signal.
(3) Spatially calculate a magnetic field distribution as a phase map, evolve (resolve) the magnetic field distribution for each magnetic field component set as a target of shimming, and calculate a shim current value required for obtaining magnetic field intensity stabilizing a magnetic field distribution for each magnetic field component.
A best one among the methods is the method of (3). The method of (3) is adopted in this embodiment. In this method, for example, considering that an extremely thin slice area with thickness of 3 mm is an object of concern and a slice direction is a most general Z direction, fall in accuracy is apprehended when components z1, z3, z5, and the like representing non-uniformity in the Z direction are calculated from a magnetic field distribution only in the thin one slice area. In this embodiment, the apprehension is solved by calculating the components from a magnetic field distribution of a wide area, that is, entire plural slice areas.
In this embodiment, in correcting a non-uniform component in a static magnetic field, a shimming value is calculated for each spatially different partial area and the non-uniform component of the static magnetic field is corrected with the shimming value different for each partial area. For example, in a multi-slice method, a magnetic field distribution for a wide area over plural slice areas, that is, a magnetic field distribution (a second-order magnetic field in the FUC method) higher in order than a non-uniform component of an n-th order term (n=1 in the FUC method) of a static magnetic field to be corrected is calculated, a shape of the magnetic field distribution is approximated by an n-th order equation (a linear equation in the FUC method), and a shimming value of each magnetic field component on the basis of the n-th order equation is obtained. Note that the spatially different partial area means, for example, an arbitrary sectional area shown in
A method of determining shimming values of a zero-th order component and a first-order component will be specifically explained with the case in which the multi-slice method is also used as an example.
A magnetic field distribution in the Z axis direction is shown in
Slice area #1; C1=c11(Z)+c10+b0 (1)
Slice area #2; C2=c21(Z)+c20+b0 (2)
Slice area #3; C3=c31(Z)+c30+b0 (3)
Therefore, a shimming value of a zero-th order component for the slice area #1 is given as −c10, a shimming value of a first-order component for the slice area #1 is given as −c11, a shimming value of a zero-th component for the slice area #2 is given as −c20, a shimming value of a first-order component for the slice area #2 is given as −c21, a shimming value of a zero-th order component for the slice area #3 is given as −c30, and a shimming value of a first-order component for the slice area #3 is given as −c31.
Shimming for a first-order component is realized by giving an offset to a gradient magnetic field according to a shimming value of the first-order component. Shimming for a zero-th component, that is, shift of a resonance frequency is realized by adjusting a reference frequency of quadrature phase detection in the receiver 6 according to a shimming value of the zero-th component.
Consequently, as shown in
A pulse sequence in the case in which the multi-slice method is used together with a field echo method is shown in
A second-order magnetic distribution before correction is shown in
In the MRI apparatus according to the constitution described above, when a non-uniform component of a static magnetic field is corrected, a shimming value is calculated for each spatially different partial area and a non-uniform component of a static magnetic field is corrected for each partial area on the basis of respective shimming values. According to the method, a shimming value is calculated for each spatially different partial area like a slice in the multi-slice method and a non-uniform component of a static magnetic field is corrected for each partial area on the basis of respective shimming values. Thus, uniformity of the static magnetic field is improved and an effect of the correction extends to a component higher in order than a non-uniform component directly corrected by a shimming value calculated. It is possible to perform correction of non-uniformity of the static magnetic field at higher accuracy. This makes it possible to obtain a satisfactory image with less artifact even in imaging in which high magnetic field uniformity is required such as echo planar imaging (EPI), spectroscopy (MRS), spectroscopy imaging (MRSI), water fat separation, and fat control.
When a first-order component of a non-uniform static magnetic field is corrected by offset of a gradient magnetic field, an offset value is calculated for spatially different partial area and the first-order component of the non-uniform static magnetic field is corrected for each partial area on the basis of respective offset values. According to the method, since an offset value is calculated for each spatially different partial area like a slice in the multi-slice method and a first-order component of a non-uniform static magnetic field is corrected for each partial area on the basis of respective offset values. Thus, uniformity of the static magnetic field is improved and an effect of the correction extends to a component in an order equal to or higher than a second order. It is possible to perform correction of non-uniformity of the static magnetic field at higher accuracy. This makes it possible to obtain a satisfactory image with less artifact even in imaging in which high magnetic field uniformity is required such as echo planar imaging (EPI), spectroscopy (MRS), spectroscopy imaging (MRSI), water fat separation, and fat control. Moreover, it is possible to cope with the effect only through a software change of calculating an offset value for each partial area.
In such an MRI apparatus, when a phase map is created in shimming processing for correcting a non-uniform component of a magnetic field, phase unwrapping evolution is performed.
In the phase unwrapping evolution, in the conventional method, as shown in
When the phase unwrapping evolution method according to the invention is applied, grouping for grouping phase data obtained by shimming under a predetermined condition is performed to match phases among at least two groups. In this case, after phases of all phase data in an arbitrary group subjected to the grouping are shifted by a predetermined amount and subjected to unwrapping evolution by a unit of group, phases are matched with phases of the other groups. In order to match phases, a center frequency is calculated separately for each group in advance and phases are shifted by the center frequency.
By executing the processing described above, even when data having a steep phase change compared with data around the data is present at the time of phase map creation in the shimming processing, it is possible to surely perform unwrapping evolution and remarkably reduce failure due to an influence of noise.
Since groups increase as the unwrapping evolution processing advances and information on a phase difference among the groups increases, it is possible to remarkably reduce an influence of phase data that tends to cause failure of unwrapping evolution. Thus, robustness is higher than that in the conventional method. In addition, a group once subjected to unwrapping evolution continues to be subjected to unwrapping evolution as a new group created by merging. Thus, even if a phase is decided once, the group is subjected to wrapping evolution many times in order to match the phase with phases of other groups. As a result, it is possible to prevent a series of occurrence of failure of unwrapping evolution. Failure of unwrapping evolution is reduced compared with the conventional method and stability of unwrapping evolution is improved.
Note that it is assumed that the unwrapping evolution method according to the invention includes performance of additional processing for deciding a phase after execution of the unwrapping evolution method.
The invention is not limited to the embodiment itself. It is possible to embody the invention by modifying the components without departing from the spirit of the invention at a stage of carrying out the invention. It is possible to form various inventions according to appropriate combinations of the plural components disclosed in the embodiment. For example, several components may be deleted from all the components described in the embodiment. Moreover, components used in different embodiments may be appropriately combined.
The unwrapping evolution method according to the invention is applied in, for example, data processing in an MRI (Magnetic Resonance Imaging) apparatus at the time of creation of a phase map.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2004-326640 | Nov 2004 | JP | national |
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/JP05/18822 | Oct 2005 | US |
| Child | 11404840 | Apr 2006 | US |
