JP2010125113A - Matching device, magnetic resonance imaging apparatus, and matching method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a matching device for enabling a parameter to be close to optimization, and to provide a magnetic resonance imaging apparatus having the matching device, and also a matching method. <P>SOLUTION: Optimization is performed for angular parameters Rx, Ry, Rz. After the optimization of the angular parameters Rx, Ry, Rz, a correlation coefficient Cor is calculated, which expresses correlation between a sample brain 8b and a standard brain T. Then, it is determined whether the correlation coefficient Cor satisfies a prescribed correlation condition or not. When the correlation coefficient Cor does not satisfy the prescribed correlation condition, optimization of scaling parameters Sx, Sy, Sz is performed. Then, it is determined whether the correlation coefficient Cor satisfies the prescribed correlation condition or not. When the correlation coefficient Cor does not satisfy the prescribed correlation condition, the optimization is performed for the both of the angular parameters Rx, Ry, Rz and the scaling parameters Sx, Sy, Sz. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a matching device that matches a sample site to a standard site, a magnetic resonance imaging apparatus having the matching device, and a matching method that matches a sample site to a standard site.

Conventionally, magnetic resonance imaging (MRI)
A head of the subject is imaged using a (Resonance Imaging) apparatus to diagnose the subject's brain. When creating an image necessary for diagnosing the subject's brain, a portion of the brain is extracted from the MR image of the subject's head, and the extracted brain (sample brain) is used as a template called a standard brain (Template) (For example, refer patent document 1).
JP 2005-143764

  When the sample brain is matched with the standard brain, for example, affine transformation is used. In the affine transformation, for example, parameters such as a rotation parameter and a scaling parameter are optimized, and it is determined whether or not a sufficient correlation is obtained between the sample brain and the standard brain after the parameters are optimized. However, even if the parameters are optimized, the parameters may converge to the local solution instead of the optimal solution. In this case, sufficient correlation cannot be obtained between the sample brain and the standard brain, and matching is performed. May not work.

  In view of the above circumstances, the present invention provides a matching apparatus that can bring parameters close to optimization, a magnetic resonance imaging apparatus having the matching apparatus, and a matching method.

The matching device of the present invention that solves the above problems
A matching device that matches a sample part to a standard part,
Parameter optimization that optimizes one or more first parameters for matching the sample site to the standard site and one or more second parameters for matching the sample site to the standard site Means,
A correlation amount calculating means for calculating a correlation amount representing a correlation between the sample portion and the standard portion;
Control means for controlling the parameter optimization means and the correlation amount calculation means,
(A) the parameter optimization means optimizes the one or more first parameters based on a first norm between the sample site and the standard site;
(B) After the one or more first parameters are optimized, the correlation amount calculation means calculates a first correlation amount between the sample part and the standard part,
(C) If the first correlation amount does not satisfy the first correlation condition, the parameter optimization means optimizes the one or more second parameters based on the first norm;
(D) After the one or more second parameters are optimized, the correlation amount calculation means calculates a second correlation amount between the sample part and the standard part,
(E) When the second correlation amount does not satisfy the second correlation condition, the parameter optimization unit is configured to determine the one or more first parameters and the one or more based on the first norm. And a control means for optimizing both of the second parameter.

  The magnetic resonance imaging apparatus of the present invention includes the matching apparatus of the present invention.

Moreover, the matching method of the present invention includes:
A matching method for matching a sample part to a standard part,
Optimizing one or more first parameters for matching the sample site to the standard site based on a norm between the sample site and the standard site;
Calculating a first correlation amount representing a correlation between the sample site and the standard site after the one or more first parameters are optimized;
If the first correlation amount does not satisfy a first correlation condition, based on the norm, optimizing one or more second parameters for matching the sample site to the standard site;
Calculating a second correlation amount representing a correlation between the sample site and the standard site after the one or more second parameters have been optimized;
If the second correlation amount does not satisfy the second correlation condition, the step of optimizing both the one or more first parameters and the one or more second parameters based on the norm And have.

  In the present invention, the amount of correlation between the sample site and the standard site after optimizing one or more first parameters does not satisfy the first correlation condition, and one or more second parameters When the amount of correlation between the sample part and the standard part after optimizing the parameters does not satisfy the second correlation condition, both the first parameter and the second parameter are optimized. Therefore, the first parameter and the second parameter can be approximated to the optimal solution.

  The best mode for carrying out the invention will be described below in detail with reference to the drawings.

  FIG. 1 is a schematic diagram of a magnetic resonance imaging apparatus 1 according to an embodiment of the present invention.

Magnetic Resonance Imaging (MRI)
Resonance Imaging) 1) includes a coil assembly 2, a cradle 3, a receiving coil 4, a control device 5, and an input device 6.

  The coil assembly 2 includes a bore 21 in which the subject 8 is accommodated, a superconducting coil 22, a gradient coil 23, and a transmission coil 24. The superconducting coil 22 forms a static magnetic field B0, the gradient coil 23 applies a gradient pulse, and the transmission coil 24 transmits an RF pulse.

  The cradle 3 is configured to move in the z direction and the −z direction. As the cradle 3 moves in the z direction, the subject 8 is transported to the bore 21. When the cradle 3 moves in the −z direction, the subject 8 transported to the bore 21 is unloaded from the bore 21.

  The receiving coil 4 is attached to the head 8 a of the subject 8. An MR (Magnetic Resonance) signal received by the receiving coil 4 is transmitted to the control device 5.

  The control device 5 includes coil control means 51 to dividing means 60.

  The coil control unit 51 controls the gradient coil 23 and the transmission coil 24 so that a pulse sequence for imaging the subject 8 is executed.

  The data processing means 52 processes the data collected by the receiving coil 4 from the head 8a of the subject 8, and reconstructs the volume data DV (see FIG. 5) of the head 8a of the subject 8.

  The brain extraction means 53 extracts the brain 8b from the volume data DV.

  The center-of-gravity matching means 54 matches the center of gravity Gs of the sample brain 8b described later with the center of gravity Gt of the standard brain T. (See FIG. 8).

  The brain surface point extracting means 55 extracts the surface point Xs (i) of the sample brain 8b.

  The parameter optimizing means 56 optimizes one or both of angle parameters Rx, Ry, and Rz and scaling parameters Sx, Sy, and Sz, which will be described later.

  The correlation coefficient calculation means 57 calculates correlation coefficients Cor and Cor ′ (see equations (3) and (11)), which will be described later.

  The correlation coefficient determination unit 58 determines whether or not the correlation coefficients Cor and Cor ′ calculated by the correlation coefficient calculation unit 57 satisfy a correlation condition (see equations (4) and (12)) described later.

  The control unit 59 controls the brain surface point extraction unit 55, the parameter optimization unit 56, and the correlation coefficient calculation unit 57 based on the determination result of the correlation coefficient determination unit 58.

  The dividing unit 60 divides the sample brain 8b into a matching region Rm and a non-matching region Rout (see FIG. 14A).

  The input device 6 inputs a command input by the operator 61 to the control device 5.

  The MRI apparatus 1 is configured as described above. Next, the operation of the MRI apparatus 1 will be described.

  2 to 4 are diagrams illustrating an example of a processing flow of the MRI apparatus 1.

  In step S <b> 1, the operator 9 operates the input device 6 to set a plurality of slices on the head 8 a of the subject 8. Thereafter, an imaging execution command for imaging the head 8a of the subject 8 is input. When an imaging execution command is input, the coil control means 51 (see FIG. 1) controls the gradient coil 23 and the transmission coil 24 so that a scan for imaging the head 8a of the subject 8 is executed. . By executing the scan, the receiving coil 4 receives the MR signal from the head 8a of the subject 8. The MR signal received by the receiving coil 4 is transmitted to the data processing means 52 of the control device 5. After scanning the head 8a of the subject 8, the process proceeds to step S2.

  In step S2, the data processing means 52 (see FIG. 1) reconstructs the volume data of the head 8a of the subject 8 from the slice data collected from the head 8a of the subject 8.

  FIG. 5 is a diagram schematically showing the reconstructed volume data DV, and FIG. 6 is a diagram schematically showing an example of a sagittal section of the volume data DV.

  As shown in FIG. 5, the reconstructed volume data DV represents the head 8 a of the subject 8. The volume data DV includes data such as the scalp 8c in addition to the data of the brain 8b of the subject 8. After reconstructing the volume data DV, the process proceeds to step S3.

  In step S3, the brain extracting means 53 (see FIG. 1) extracts the brain 8b from the volume data DV. In the present embodiment, in the matching process (see step S4 in FIG. 2) to be described later, the process of matching the brain 8b of the subject 8 with the standard brain is performed. The process of extracting the brain 8b from the above is performed. The brain extracting means 53 uses the gap G1 between the brain 8b and the scalp 8c, the gap G2 between the brain 8b and the nasal cavity 8d (see FIGS. 5 and 6), and the like from the volume data DV to the brain 8b. To extract. As an extraction method, erosion, dilation, or the like can be applied.

  FIG. 7 is a schematic diagram showing the extracted brain 8b. After extracting the brain 8b, the process proceeds to step S4.

  In steps S4 to S23, a process of matching the extracted brain (hereinafter referred to as “sample brain”) 8b with the standard brain is performed. This matching process is performed by affine transformation. In order to perform affine transformation, first, in step S4, the center-of-gravity matching means 54 (see FIG. 1) matches the center of gravity of the sample brain 8b with the center of gravity of the standard brain.

  FIG. 8 is a diagram showing a state in which the center of gravity of the sample brain 8b is matched with the center of gravity of the standard brain. The sample brain 8b is indicated by a thick line, and the standard brain T is indicated by a thin line.

  The center-of-gravity matching means 54 determines the center of gravity Gs of the sample brain 8b and the center of gravity Gt of the standard brain T (FIG. 8A). After determining the centroids Gs and Gt, the values of the movement parameters tx, ty, and tz for matching the centroid Gs of the sample brain 8b with the centroid Gt of the standard brain T are determined. The center-of-gravity matching means 54 moves the center of gravity Gs of the sample brain 8b by the determined movement parameters tx, ty, and tz (FIG. 8B). As a result, the center of gravity Gs of the sample brain 8b can be matched with the center of gravity Gt of the standard brain T. After performing the center of gravity adjustment, the process proceeds to step S5.

  In step S5, parameter n is initialized. The parameter n is a parameter that represents the number of times that the determination in step S21 (see FIG. 4) described later has been performed. After the parameter n is initialized, the process proceeds to step S6.

  In step S6, the brain surface point extraction means 55 (see FIG. 1) extracts the surface point Xs (i) of the sample brain 8b. i is a variable representing an arbitrary position on the surface of the sample brain 8b, and here, i is an integer of 1 to m.

  FIG. 9 is a diagram schematically showing the surface point Xs (i) extracted from the sample brain 8b.

  In FIG. 9, only three surface points Xs (1), Xs (2), and Xs (m) are extracted as the surface points Xs (i) extracted from the sample brain 8b in consideration of the visibility of the drawing. It is shown. However, actually, the surface point Xs (i) is extracted over the entire surface of the sample brain 8b. After extracting the surface point Xs (i) of the sample brain 8b, the brain surface point extracting means 55 determines the surface point X (i) of the standard brain T corresponding to the surface point Xs (i) of the sample brain 8b ( (See FIG. 10).

  FIG. 10 is a diagram schematically showing the surface point X (i) of the standard brain T. As shown in FIG.

  Also in FIG. 10, as the surface point X (i) of the standard brain T corresponding to the surface point Xs (i) of the sample brain 8b, representatively three surface points X (1), X (2), and X (m Only) is shown. However, in practice, the surface point X (i) is determined over the entire surface of the standard brain T. Then, it progresses to step S7.

  In step S7, the parameter optimization means 56 (see FIG. 1) optimizes the angle parameters Rx, Ry, and Rz for rotating the sample brain 8b with respect to the standard brain T (see FIG. 11).

  FIG. 11 is a diagram for explaining the concept of the angle parameters Rx, Ry, and Rz.

  The angle parameter Rx is a parameter that defines a rotation angle when the sample brain 8b is rotated around the rotation axis Ax that passes through the gravity centers Gs and Gt and extends in the X-axis direction. The angle parameter Ry is a parameter that defines a rotation angle when the sample brain 8b is rotated about the rotation axis Ay that passes through the gravity centers Gs and Gt and extends in the Y-axis direction. The angle parameter Rz is a parameter that defines a rotation angle when the sample brain 8b is rotated about the rotation axis Az that passes through the gravity centers Gs and Gt and extends in the Z-axis direction.

The parameter optimization means 56 changes the angular parameters Rx, Ry, and Rz, and the norm (Norm) between the surface point Xs (i) of the sample brain 8b and the surface point X (i) of the standard brain T And the angle parameters Rx, Ry, and Rz are obtained when the norm is minimized. The norm is, for example, an L2 norm or an Euclidean distance, and as a method for calculating the norm, a method of Chamfer Distance can be used. The parameter optimization means 56 determines the angle parameters Rx, Ry, and Rz when the norm is minimized as the optimized angle parameters Rx1, Ry1, and Rz1. Therefore, the optimized angle parameters Rx1, Ry1, and Rz1 are expressed by the following equations (1) and (2).

  As an optimization method, various methods such as a steepest descent method, a quasi-Newton method, a simplex method, and a conjugate gradient method can be applied. After the angle parameters Rx, Ry, and Rz are optimized, the process proceeds to step S8.

In step S8, the correlation coefficient calculating means 57 (see FIG. 1) calculates the correlation coefficient Cor between the sample brain 8b and the standard brain T after the angle parameters Rx, Ry, and Rz are optimized. . The correlation coefficient Cor is calculated using the following equation (3).

  After the correlation coefficient Cor is calculated, the process proceeds to step S9.

In step S9, the correlation coefficient determination unit 58 (see FIG. 1) determines whether or not the correlation coefficient Cor calculated by the correlation coefficient calculation unit 57 satisfies a predetermined correlation condition. The correlation condition is expressed by the following formula (4), for example.
k1 ≦ | Cor | ≦ k2 (4)
Where | Cor |: absolute value of correlation coefficient Cor,
k1, k2: constants

  k1 and k2 are, for example, 0.7 and 1, respectively. The values of k1 and k2 may be predetermined values as default values, or may be values that can be changed by the operator 61.

  When the correlation coefficient determination unit 58 determines that the correlation coefficient Cor satisfies the expression (4), the control unit 59 (see FIG. 1) generates a command for ending the matching process, and as a result, the matching The process ends. However, when the correlation coefficient determination unit 58 determines that the correlation coefficient Cor does not satisfy the expression (4), the control unit 59 determines that the brain surface point extraction unit 55 and the parameter are set so that the matching process is continued. The optimization unit 56 and the correlation coefficient calculation unit 57 are controlled. When the matching process is continued, the process proceeds to step S10.

  In step S10, the brain surface point extraction means 55 extracts the surface point Xs (i) of the sample brain 8b. This extraction method is the same as step S5. After extracting the surface points Xs (i) and X (i), the process proceeds to step S11.

  In step S11, the parameter optimization means 56 optimizes the scaling parameters Sx, Sy, and Sz for changing the size of the sample brain 8b (see FIG. 12).

  FIG. 12 is a diagram for explaining the concept of the scaling parameters Sx, Sy, and Sz.

  The scaling parameter Sx is a parameter that defines an enlargement ratio or a reduction ratio when the sample brain 8b is enlarged or reduced in the X-axis direction. The scaling parameter Sy is a parameter that defines an enlargement rate or a reduction rate when the sample brain 8b is enlarged or reduced in the Y-axis direction. The scaling parameter Sz is a parameter that defines an enlargement rate or a reduction rate when the sample brain 8b is enlarged or reduced in the Z-axis direction.

The parameter optimization means 56 calculates the norm between the surface point Xs (i) of the sample brain 8b and the surface point X (i) of the standard brain T while changing the scaling parameters Sx, Sy, and Sz. When the norm is minimized, the scaling parameters Sx, Sy, and Sz are obtained. The parameter optimizing unit 56 determines the scaling parameters Sx, Sy, and Sz when the norm is minimized as the optimized scaling parameters Sx1, Sy1, and Sz1. Therefore, the optimized scaling parameters Sx1, Sy1, and Sz1 are expressed by the following equations (5) and (6).

  After optimizing the scaling parameters Sx, Sy, and Sz, the process proceeds to step S12.

  In step S12, the correlation coefficient calculating means 57 calculates the correlation coefficient Cor between the sample brain 8b and the standard brain T after the scaling parameters Sx, Sy, and Sz are optimized. The correlation coefficient Cor is calculated using the above equation (3). After the correlation coefficient Cor is calculated, the process proceeds to step S13.

  In step S13, the correlation coefficient determination unit 58 determines whether or not the correlation coefficient Cor calculated by the correlation coefficient calculation unit 57 in step S12 satisfies a predetermined correlation condition. As the correlation condition, for example, Equation (4) can be used.

  When the correlation coefficient determination unit 58 determines that the correlation coefficient Cor satisfies the expression (4), the control unit 59 generates a command for ending the matching process, and as a result, the matching process ends. However, if the correlation coefficient determination unit 58 determines that the correlation coefficient Cor does not satisfy the expression (4), the control unit 59 causes the brain surface point extraction unit 55, The parameter optimization unit 56 and the correlation coefficient calculation unit 57 are controlled. When the matching process is further continued, the process proceeds to step S14.

  In step S14, the brain surface point extraction means 55 extracts the surface point Xs (i) of the sample brain 8b. This extraction method is the same as step S5. After extracting the surface point Xs (i), the process proceeds to step S15.

  In step S15, the parameter optimization means 56 optimizes both the angle parameters Rx, Ry, and Rz and the scaling parameters Sx, Sy, and Sz.

The parameter optimization unit 56 changes the angle parameters Rx, Ry, and Rz and the scaling parameters Sx, Sy, and Sz while changing the surface point Xs (i) of the sample brain 8b and the surface point X ( The norm between i) and the angle parameters Rx, Ry, and Rz and the scaling parameters Sx, Sy, and Sz are obtained when the norm is minimized. The parameter optimization means 56 determines the angle parameters Rx, Ry, and Rz when the norm is minimized as the optimized angle parameters Rx1, Ry1, and Rz1, and the scaling parameter Sx when the norm is minimized. , Sy, and Sz are determined as optimized scaling parameters Sx1, Sy1, and Sz1. Therefore, the optimized angle parameters Rx1, Ry1, and Rz1 and the optimized scaling parameters Sx1, Sy1, and Sz1 are expressed by the following equations (7) and (8).

  After the angle parameters Rx, Ry, and Rz and the scaling parameters Sx, Sy, and Sz are optimized, the process proceeds to step S16.

  In step S16, the correlation coefficient calculating unit 57 determines whether the angle parameters Rx, Ry, and Rz and the scaling parameters Sx, Sy, and Sz are optimized, and the sample brain 8b and the standard brain T after the optimization. The correlation coefficient Cor of is calculated. The correlation coefficient Cor is calculated using the above equation (3). After the correlation coefficient Cor is calculated, the process proceeds to step S17.

  In step S17, the correlation coefficient determination means 58 determines whether or not the correlation coefficient Cor calculated by the correlation coefficient calculation means 57 satisfies a predetermined correlation condition. As the correlation condition, for example, Equation (4) can be used.

  When the correlation coefficient determination unit 58 determines that the correlation coefficient Cor satisfies the expression (4), the control unit 59 generates a command for ending the matching process, and as a result, the matching process ends. However, if the correlation coefficient determination unit 58 determines that the correlation coefficient Cor does not satisfy the expression (4), the control unit 59 causes the brain surface point extraction unit 55, The parameter optimization unit 56 and the correlation coefficient calculation unit 57 are controlled. When the matching process is further continued, the process proceeds to step S18.

  In step S18, the brain surface point extraction means 55 extracts the surface point Xs (i) of the sample brain 8b. However, in this embodiment, by the time step S18 is reached, the optimization of the angle parameter (step S7), the optimization of the scaling parameter (step S11), and the optimization of both the angle parameter and the scaling parameter ( Step S15) is executed. Even though these optimizations are performed, the fact that the correlation coefficient Cor does not satisfy the expression (4) is considered that the sample brain 8b could not be successfully extracted in step S3. . Hereinafter, an example in which the sample brain 8b cannot be successfully extracted will be described with reference to FIG.

  FIG. 13A is an example of a sagittal section of the volume data DV shown in FIG. 5 when the subject 8 has sinusitis, and is a diagram showing the brain extracted from FIG. 13A. .

  As described with reference to FIGS. 5 and 6, when extracting the brain 8 b from the volume data DV, the brain extracting means 53 includes the gap G 1 between the brain 8 b and the scalp 8 c, the brain 8 b and the nasal cavity 8 d, and A gap G2 between the two is used. However, when the subject 8 has sinusitis, as shown in FIG. 13, inflammation 8e due to sinusitis develops between the brain 8b and the nasal cavity 8d. Therefore, the gap G2 between the brain 8b and the nasal cavity 8d is filled with the inflammation 8e due to sinusitis, and the brain extracting means 53 cannot recognize the gap G2 between the brain 8b and the nasal cavity 8d. As a result, as shown in FIG. 13B, the inflammation 8e due to sinusitis is also extracted together with the brain 8b. In this case, the brain surface point extraction means 55 cannot distinguish between the brain 8b and the inflammation 8e due to sinusitis, and also extracts the surface point Xs (i) from the region of the inflammation 8e. Therefore, the surface point Xs (i) extracted from the region of the inflammation 8e becomes an obstacle, and it becomes difficult to match the sample brain 8b with the standard brain T well.

  Therefore, in step S18, in order to prevent the brain surface point extraction means 55 from extracting the surface point Xs (i) from the region of inflammation 8e, the dividing means 60 (see FIG. 1) applies the sample brain 8b to the matching target. Is divided into a matching area used as a matching area and a non-matching area excluded from matching (see FIG. 14).

  FIG. 14 is a diagram for explaining how the sample brain 8b is divided into a matching target and a non-matching region.

  14A is a sagittal section of a sample brain 8b of a subject in which sinusitis inflammation 8e occurs, and FIG. 14B is a diagram showing a sagittal section of the standard brain T. FIG.

  As shown in FIG. 14A, the inflammation 8e due to sinusitis occurs on the lower side of the frontal lobe FL of the brain 8b. Therefore, if the lower side of the sample brain 8b is a non-matching region Rout to be excluded from the matching target, and the upper side of the sample brain 8b is the matching region Rm used as a matching target, the brain surface point extracting means 55 is The number of surface points Xs (i) extracted from the region of the inflammation 8e can be reduced to zero or sufficiently small. Therefore, the dividing unit 60 sets a slice plane Sb for dividing the sample brain 8b into the matching region Rm and the non-matching region Rout. By setting this slice plane Sb to the sample brain 8b, the sample brain 8b can be divided into a matching region Rm and a non-matching region Rout.

  However, in order to divide the sample brain 8b into the matching region Rm and the non-matching region Rout, it is necessary to set the slice surface 8 between the frontal lobe FL and the sinusitis inflammation 8e. Hereinafter, how the slice surface 8 is set between the frontal lobe FL and the sinusitis inflammation 8e in the present embodiment will be described.

  As described above, the inflammation 8e due to sinusitis occurs on the lower side of the frontal lobe FL of the brain 8b. Therefore, if a rough range of the frontal lobe FL of the brain 8b can be predicted, the slice plane 8b can be set between the frontal lobe FL and the sinusitis inflammation 8e. Therefore, in the present embodiment, data of the standard brain T is used in order to predict a rough range of the frontal lobe FL of the brain 8b. As shown in FIG. 14B, since the standard brain T does not include the sinusitis inflammation 8e, the range of the frontal lobe FL can be known for the standard brain T. Therefore, with respect to the standard brain T, a slice plane ST is defined below the frontal lobe FL, a region above the slice plane ST is defined as a matching region Tm, and a region below the slice plane ST is defined as a non-matching region Tout. . The dividing means 60 stores the position of the slice plane ST with respect to the standard brain T. The dividing unit 60 calculates a position corresponding to the slice plane ST of the standard brain T from the range of the sample brain 8b, and determines the calculated position as the position of the slice plane Sb of the sample brain 8b. Therefore, the slice plane 8 can be set between the frontal lobe FL and the sinusitis inflammation 8e.

  FIG. 15 is a diagram schematically showing the surface point Xs (i) extracted from the matching region Rm by the brain surface point calculating means.

  FIG. 15 representatively shows only three surface points Xs (1), Xs (2), and Xs (m) as surface points Xs (i) extracted from the matching region Rm of the sample brain 8b. ing.

  The surface point Xs (i) is not extracted from the non-matching region Rout of the sample brain 8b. Therefore, it can be seen that the brain surface point extracting means 55 does not extract the surface point Xs (i) from the region of the sinusitis inflammation 8e included in the non-matching region Rout. The brain surface point extraction means 55 extracts the surface point Xs (i) from the matching region Rm of the sample brain 8b, and then the surface point of the standard brain T corresponding to the surface point Xs (i) of the matching region Rm of the sample brain 8b. X (i) is determined. However, the brain surface point extraction means 55 extracts the surface point X (i) from the matching region Tm of the standard brain T.

  FIG. 16 is a diagram schematically showing the surface point X (i) of the standard brain T. As shown in FIG.

  Also in FIG. 16, only three surface points X (1), X (2), and X (m) are representatively shown as surface points X (i) of the standard brain T. However, in practice, the surface point X (i) is determined over the entire surface of the matching region Tm of the standard brain T. Thereafter, the process proceeds to step S19.

  In step S19, the parameter optimization means 56 optimizes both the angle parameters Rx, Ry and Rz and the scaling parameters Sx, Sy and Sz.

The parameter optimization means 56 changes the angle parameters Rx, Ry, and Rz and the scaling parameters Sx, Sy, and Sz while changing the surface point Xs (i) of the matching region Rm of the sample brain 8b and the standard brain T. A norm between the matching region Tm and the surface point X (i) is calculated, and angle parameters Rx, Ry, and Rz and scaling parameters Sx, Sy, and Sz are obtained when the norm is minimized. The parameter optimization means 56 determines the angle parameters Rx, Ry, and Rz when the norm is minimized as the optimized angle parameters Rx1, Ry1, and Rz1, and the scaling parameter Sx when the norm is minimized. , Sy, and Sz are determined as optimized scaling parameters Sx1, Sy1, and Sz1. Therefore, the optimized angle parameters Rx1, Ry1, and Rz1 and the optimized scaling parameters Sx1, Sy1, and Sz1 are expressed by the following equations (9) and (10).

  After the angle parameters Rx, Ry, and Rz and the scaling parameters Sx, Sy, and Sz are optimized, the process proceeds to step S20.

In step S20, the correlation coefficient calculation means 57 calculates the correlation coefficient Cor ′ between the matching region Rm of the sample brain 8b and the matching region Tm of the standard brain T after the parameters are optimized in step S19. . The correlation coefficient Cor ′ is calculated using, for example, the following equation (11).

  After the correlation coefficient Cor 'is calculated, the process proceeds to step S21.

In step S21, the correlation coefficient determination unit 58 determines whether or not the correlation coefficient Cor ′ calculated by the correlation coefficient calculation unit 57 satisfies a predetermined correlation condition. The correlation condition is expressed by the following formula (12), for example.
k1 ≦ | Cor ′ | ≦ k2 (12)
Where | Cor '|: absolute value of correlation coefficient Cor',
k1, k2: constants

  When the correlation coefficient determination unit 58 determines that the correlation coefficient Cor ′ satisfies the expression (12), the control unit 59 generates a command for ending the matching process, and as a result, the matching process ends. . However, if the correlation coefficient determination unit 58 determines that the correlation coefficient Cor 'does not satisfy the expression (12), the process proceeds to step S22.

  In step S22, the parameter n is incremented and updated to n = 1. Updating to n = 1 means that the determination in step S21 has been made once (see step S5). When the parameter n is updated, the process proceeds to step S23.

  In step S23, the control means 59 determines whether or not the parameter n is n <N0. If n <N0, the control unit 59 returns to step S6 (see FIG. 2) or step S18 (see FIG. 3), and the control unit 59 continues the brain surface point extraction unit 55, parameter optimization unit so that the matching process is further continued. 56 and the correlation coefficient calculating means 57 are controlled. Therefore, the matching process is repeatedly executed until the parameter n reaches N0. However, depending on the shape and size of the sample brain 8b, the correlation coefficient may not satisfy a predetermined correlation condition, no matter how many times the matching process is executed. Therefore, when the matching process is repeatedly executed and the parameter n reaches n = N0, the control unit 59 generates an instruction for ending the matching process. As a result, the matching process ends.

  In the present embodiment, after the angle parameter optimization (step S7) is executed, if the correlation coefficient Cor does not satisfy a predetermined correlation condition, the scaling parameter is further optimized (step S11), and the angle parameter and the scaling are performed. Both parameter optimization (step S15) is performed. Therefore, when the parameter is optimized, even if the parameter converges to the local solution instead of the optimal solution, the parameter can be converged to the optimal solution or can be converged to the local solution closer to the optimal solution.

  In the present embodiment, even if both the angle parameter and the scaling parameter are optimized (step S15), if the correlation coefficient Cor does not satisfy a predetermined correlation condition, the sample brain 8b and the standard brain T The matching target is limited to the matching regions Rm and Tm. Therefore, even if the sample brain 8b cannot be extracted successfully in step S3 (see FIG. 13), the parameters can be converged to the optimum solution or converged to the local solution closer to the optimum solution.

  In the present embodiment, the angle parameters Rx, Ry, and Rz are first optimized (step S7). If the correlation coefficient Cor does not satisfy a predetermined correlation condition, the scaling parameters Sx, Sy, and Sz are optimized. (Step S11). However, the scaling parameters Sx, Sy, and Sz may be optimized first, and then the angle parameters Rx, Ry, and Rz may be optimized.

  In the present embodiment, the angle parameters Rx, Ry, and Rz and the scaling parameters Sx, Sy, and Sz are optimized. However, if the parameter is for matching the sample brain 8b with the standard brain T, the angle parameter Parameters other than Rx, Ry, and Rz and scaling parameters Sx, Sy, and Sz may be optimized.

  In the present embodiment, the optimized parameters are calculated using equations (1), (5), (7), and (9). However, an optimized parameter may be calculated based on another formula.

  In the present embodiment, the correlation coefficient Cor (see Equation (3)) and the correlation coefficient Cor ′ (see Equation (11)) are used as the correlation amounts representing the correlation between the sample brain and the standard brain. However, the correlation between the sample brain and the standard brain may be expressed by a correlation amount different from the correlation coefficients Cor and Cor ′.

  In the present embodiment, when determining whether or not the correlation coefficients Cor and Cor ′ satisfy a predetermined correlation condition, the correlation conditions of Expressions (4) and (12) are used as an example. The correlation condition may be used.

  In the present embodiment, the slice plane ST for dividing the standard brain T into the matching area Tm and the non-matching area Tout is defined below the frontal lobe FL, but the position of the slice plane ST is changed as necessary. You may be able to do it. In this embodiment, the slice plane Sb of the sample brain 8b is determined based on the slice ST set for the standard brain T. However, the slice plane Sb is determined based on the data of the sample brain 8b. May be.

  In this embodiment, if n <N0 (see step S23), the process returns to step S6 (see FIG. 2) or step S18 (see FIG. 3), but returns to another step (for example, step S10). May be.

  Furthermore, in the present embodiment, a method for matching the sample brain 8b with the standard brain T has been described. However, the present invention can also be applied when matching another organ other than the brain.

1 is a schematic view of a magnetic resonance imaging apparatus 1 according to an embodiment of the present invention. It is a figure which shows an example of the processing flow of the MRI apparatus. It is a figure which shows an example of the processing flow of the MRI apparatus. It is a figure which shows an example of the processing flow of the MRI apparatus. It is a figure which shows roughly the reconfigure | reconstructed volume data DV. It is a figure which shows roughly an example of the sagittal cross section of the volume data DV. It is the schematic which shows the extracted brain 8b. It is a figure which shows a mode when matching the gravity center of the sample brain 8b with the gravity center of a standard brain. It is a figure which shows roughly the surface point Xs (i) extracted from the sample brain 8b. 2 is a diagram schematically showing a surface point X (i) of a standard brain T. FIG. It is a figure explaining the concept of angle parameters Rx, Ry, and Rz. It is a figure explaining the concept of scaling parameters Sx, Sy, and Sz. FIG. 16 is a sagittal section (a) of the volume data DV shown in FIG. 5 when the subject 8 has sinusitis, and a diagram (b) showing the brain extracted from FIG. 13 (a). It is a figure explaining how the sample brain 8b is divided | segmented into the object of matching and a non-matching area | region. It is a figure which shows roughly the surface point Xs (i) extracted from the matching area | region Rm by the brain surface point calculation means. 2 is a diagram schematically showing a surface point X (i) of a standard brain T. FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 MRI apparatus 2 Coil assembly 3 Cradle 4 Receiving coil 5 Control apparatus 6 Input apparatus 8 Subject 8a Head 21 Bore 22 Superconducting coil 23 Gradient coil 24 Transmitting coil 51 Coil control means 52 Data processing means 53 Brain extraction means 54 Center of gravity alignment Means 55 Brain surface point extraction means 56 Parameter optimization means 57 Correlation coefficient calculation means 58 Control means 59 Dividing means

Claims (18)

A matching device that matches a sample part to a standard part,
Parameter optimization that optimizes one or more first parameters for matching the sample site to the standard site and one or more second parameters for matching the sample site to the standard site Means,
A correlation amount calculating means for calculating a correlation amount representing a correlation between the sample portion and the standard portion;
Control means for controlling the parameter optimization means and the correlation amount calculation means,
(A) the parameter optimization means optimizes the one or more first parameters based on a first norm between the sample site and the standard site;
(B) After the one or more first parameters are optimized, the correlation amount calculation means calculates a first correlation amount between the sample part and the standard part,
(C) When the first correlation amount does not satisfy the first correlation condition, the parameter optimization unit optimizes the one or more second parameters based on the first norm;
(D) After the one or more second parameters are optimized, the correlation amount calculation means calculates a second correlation amount between the sample part and the standard part,
(E) When the second correlation amount does not satisfy the second correlation condition, the parameter optimization unit is configured to determine the one or more first parameters and the one or more based on the first norm. And a control means for optimizing both of the second parameter of the matching device. Correlation amount determining means for determining whether or not the first correlation amount satisfies the first correlation condition and determining whether or not the second correlation amount satisfies the second correlation condition is provided. And
The matching device according to claim 1, wherein the control unit controls the parameter optimization unit and the correlation amount calculation unit based on a determination result of the correlation amount determination unit. The control means includes
After both the one or more first parameters and the one or more second parameters are optimized, a third correlation amount between the sample site and the standard site is calculated. The matching device according to claim 2, wherein the correlation amount calculation unit is controlled. The correlation amount determining means includes
The matching device according to claim 3, wherein it is determined whether or not the third correlation amount satisfies the third correlation condition.   5. The matching device according to claim 3, further comprising a sample dividing unit that divides the sample portion into a matching region used as a matching target and a non-matching region excluded from the matching target. The dividing means includes
The matching device according to claim 5, wherein the standard part is divided into a matching region that is a matching target and a non-matching region that is excluded from the matching target. The control means includes
When the third correlation amount does not satisfy a third correlation condition, the one or more first ones are based on a second norm between the matching region of the sample region and the matching region of the standard region. The matching device according to claim 6, wherein the parameter optimization unit is controlled such that both a parameter and the one or more second parameters are optimized. The control means includes
A fourth correlation between the matching region of the sample region and the matching region of the standard region after both the one or more first parameters and the one or more second parameters are optimized The matching device according to claim 7, wherein the correlation amount calculation unit is controlled so that the amount is calculated. The correlation amount determining means includes
The matching device according to claim 8, wherein the fourth correlation amount determines whether or not the fourth correlation condition is satisfied. The control means includes
If the fourth correlation amount does not satisfy a fourth correlation condition, both the one or more first parameters and the one or more second parameters are optimized based on the second norm The matching device according to claim 8 or 9, wherein the parameter optimization unit is controlled so that the processing to be performed is repeatedly executed. The control means includes
When the fourth correlation amount does not satisfy the fourth correlation condition,
(A) the parameter optimization means optimizes the one or more first parameters based on a first norm between the sample site and the standard site;
(B) After the one or more first parameters are optimized, the correlation amount calculation means calculates a first correlation amount between the sample part and the standard part,
(C) When the first correlation amount does not satisfy the first correlation condition, the parameter optimization unit optimizes the one or more second parameters based on the first norm;
(D) After the one or more second parameters are optimized, the correlation amount calculation means calculates a second correlation amount between the sample part and the standard part,
(E) When the second correlation amount does not satisfy the second correlation condition, the parameter optimization unit is configured to determine the one or more first parameters and the one or more based on the first norm. Both the second parameter of and
(F) After both of the one or more first parameters and the one or more second parameters are optimized, the correlation amount calculating means may determine whether the correlation between the sample part and the standard part Calculate the third correlation amount,
(G) When the third correlation amount does not satisfy the third correlation condition, the parameter optimization unit is based on a second norm between the matching region of the sample region and the matching region of the standard region. Optimizing both the one or more first parameters and the one or more second parameters;
(H) After both of the one or more first parameters and the one or more second parameters are optimized, the correlation amount calculating means includes a matching region of the sample region and the standard region. Calculating a fourth correlation amount with the matching region;

The matching device according to claim 8 or 9, wherein the parameter optimization unit and the correlation amount calculation unit are controlled so that the processes (a) to (h) are repeatedly executed.   The matching device according to claim 1, further comprising a center-of-gravity matching unit that matches the center of gravity of the sample part with the center of gravity of the standard part.   13. The one or more first parameters are three angle parameters, and the one or more second parameters are three scale parameters. Matching device.   The matching device according to claim 1, wherein the first norm is a norm between a surface point of the sample part and a surface point of the standard part.   The matching according to any one of claims 7 to 11, wherein the second norm is a norm between a surface point of the matching region of the sample region and a surface point of the matching region of the standard region. apparatus.   The matching device according to claim 1, wherein the sample site is a sample brain, and the standard site is a standard brain.   A magnetic resonance imaging apparatus comprising the matching apparatus according to claim 1. A matching method for matching a sample part to a standard part,
Optimizing one or more first parameters for matching the sample site to the standard site based on a norm between the sample site and the standard site;
Calculating a first correlation amount representing a correlation between the sample site and the standard site after the one or more first parameters have been optimized;
Optimizing one or more second parameters for matching the sample site to the standard site based on the norm if the first correlation quantity does not satisfy a first correlation condition;
Calculating a second correlation amount representing a correlation between the sample site and the standard site after the one or more second parameters have been optimized;
If the second correlation amount does not satisfy a second correlation condition, the step of optimizing both the one or more first parameters and the one or more second parameters based on the norm When,
A matching method.

Images