JPH10211181A - Biological activity current source estimating device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To securely specify a current source without erroneously estimating a current source at a grid point far away from the other gird points. SOLUTION: A fine magnetic field from the biological activity current source of a subject M is measured by a multichannel SQUID sensor 1 and its magnetic field data are stored in a data collecting unit 5. A data analytic unit 8 estimates the group of grid points in the diagnostic object region of the subject M and obtains a current source on each grid point by a method of linear least squares using the condition of minimizing the sum of the error of squares of a calculated magnetic field and a measured magnetic field and the sum of squares with the weight of the power source. At the time, the square value of the current source is respectively weighted by a weighting factor decided based on an average distance between the grid point and all of the other gird points by each grid point. Then a current source corresponding to the case of minimizing the sum of these in the large area is estimated to be a real current source.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a biological activity current source estimating apparatus for estimating the position, orientation, and size of a biological activity current source, and more particularly, to more accurately estimate a biological activity current source. For technology.

[0002]

2. Description of the Related Art When a living body is stimulated, the polarization formed across a cell membrane is broken and a living activity current flows. This biological activity current appears in the brain and heart, and is recorded as an electroencephalogram and an electrocardiogram. The magnetic field generated by the biological activity current is recorded as a magnetoencephalogram and a magnetocardiogram.

Recently, as a device for measuring a minute magnetic field in a living body, SQUID (Superconducting Quantum Inter
face Device: A sensor using a superconducting quantum interferometer) has been developed. This sensor is placed outside the head, and the sensor can non-invasively measure a small magnetic field generated by a current dipole (hereinafter simply referred to as a current source), which is a biological activity current source, generated in the brain. From the measured magnetic field data, the position, direction, and size of the current source related to the lesion are estimated, and the estimated current source is displayed on a tomographic image obtained by an X-ray CT or MRI device, and the physics of the affected part is displayed. It is used to identify the target position.

Conventionally, as one of current source estimation methods, there is a method using a minimum norm method (for example, WHKullman
n, KDJandt, K. Rehm, HASchlitt, WJDallas and
WESmith, Advances in Biomagnetism, pp.571-574, P
lenum Pless, New York, 1989).

A conventional current source estimation method using the minimum norm method will be described below with reference to FIG. As shown in FIG. 7, a multi-channel SQUID sensor 1 is provided near the subject M. Multichannel SQUID sensor 1 is accommodated by immersing a number of magnetic sensors (pickup coil) S ₁ to S _m in the refrigerant such as liquid nitrogen in a container called a dewar.

On the other hand, a large number of grid points (1) to (n) are set in, for example, the brain, which is a diagnosis target area of the subject M, and an unknown current source (current dipole) is assumed at each grid point. Each current source is represented by a three-dimensional vector VP _j (j = 1 to n). Then, S
Magnetic field B _i .about.B _m detected by the magnetic sensor S ₁ to S _m of QUID sensor 1 is expressed by the following equation (1).

[0007]

(Equation 1)

In equation (1), VP _j = (P _jx , P _jy , P
_jz ) α _ij = (α _ijx, α _ijy, α _ijz ). Note that α _ij is the magnetic sensors S ₁ to S ₁ when a current source having a unit size in the X, Y, and Z directions is placed on a grid point.
Is a known coefficient which represents the strength of the magnetic field detected by each position of the S _m.

[B] = (B ₁ , B ₂ ,...,
B _m ) [P] = (P _1x , P _1y , P _1z , P _2x , P _2y , P _2z ,...,
(P _nx , P _ny , P _nz ), the expression (1) can be rewritten into a linear relational expression like the expression (2). [B] = A [P] (2) In the equation (2), A is a matrix having 3n × m elements represented by the following equation (3).

[0010]

(Equation 2)

Here, if the inverse matrix of A is represented by A ⁻ ,
[P] is represented by the following equation (4). (P) = A ^- (B) ......... (4), where the minimum norm method, the number of the formula m (magnetic sensor S ₁
~S number of _m) than, assuming the case where X of the current source, which is assumed the number of unknowns 3n (each lattice point, Y, unknown in the case of considering the size of the Z-direction) is large, the current source [ P]
Of the current source [P] by adding a condition that the norm | [P] | Note that a solution can be uniquely obtained by making the number m of the above equations equal to the number 3n of unknowns. In such a case, however, the solution becomes very unstable, so the minimum norm method is used. Is used.

By adding the condition that the norm | [P] | of the current source [P] is minimized, the above equation (4) becomes the following equation (5).
It is represented as [P] = A ⁺ [B] (5) where A ⁺ is a generalized inverse matrix expressed by the following equation (6). ^{A + = A t (AA t} ) -1 ......... (6) However, A ^t is the transpose matrix of A.

By solving the above equation (5), the current sources VP _j on each grid point
Are estimated, and the one having the largest value among them is assumed to be close to the true current source. This is the principle of the current source estimation method using the minimum norm method.

Further, there has been proposed a method of repeatedly obtaining a minimum norm solution while subdividing a grid point division in order to improve the position resolution of the minimum norm method (for example, Y.Oka).
da, J.Huang and C.Xu, 8th International Conference
on Biomagnetism, Munster, August 1991). Hereinafter, FIG.
This method will be described briefly with reference to FIG.

FIG. 8 is an enlarged view of a part of the lattice point group N shown in FIG. 7, and a symbol J in the figure represents a true current source estimated using the minimum norm method described above. Is a grid point where a current source close to Around this grid point J, a subdivided grid point group M (indicated by small black dots in FIG. 8) is additionally set. Then, a current source closer to the true current source is estimated by using the same method as described above, with the newly set grid point group M included in the initially set grid point group N.

However, the conventional example having such a configuration has the following problems. According to the conventional method shown in FIG. 8, the subdivided grid point group M is newly set in addition to the initially set grid point group N, so that the number of grid points increases. Therefore, there is a problem that the number of elements of the vector [P] in the equation (5) increases and the calculation accuracy of the minimum norm solution decreases.

Therefore, the present applicant has previously disclosed in Japanese Patent Laid-Open No. 6-34.
No. 3613 and JP-A-6-343614 disclose a grid point moving minimum norm method for moving another group of grid points near a grid point where a current source having a large value exists among current sources estimated by the minimum norm method. is suggesting. This method attempts to estimate the current source accurately while maintaining the accuracy of the minimum norm solution by estimating the current source with the number of grid points being the same as the previous time and narrowing only the grid point interval. Is what you do.

However, in the current source estimating method using the minimum norm method, the size of the current source in the X, Y, and Z directions assumed on each grid point is larger than the number m of magnetic sensors (the number of equations). Since it is assumed that the number of unknowns 3n (n is the number of lattice points) considered is large (3n> m), a coefficient matrix indicating the relationship between the unknown current source on each lattice point and the measured magnetic field is obtained. The rank may drop and the solution may become unstable. Further, in the process of specifying the optimal current source, whether or not the minimum grid point interval has become equal to or less than a preset value (convergence determination value) is used as a criterion. May be different.

Therefore, the present applicant further proposes Japanese Patent Application No. 6-47.
According to No. 220, grid points are set with a smaller number than the number of magnetic sensors, and a condition that a square error between a magnetic field exerted by an unknown current source on each of these grid points and a magnetic field measured by the magnetic sensor is minimized. To determine the unknown current source, and determine whether or not the square error between the magnetic field calculated from the obtained current source and the magnetic field measured by the magnetic sensor is globally minimized. In this case, a method of moving another group of grid points to a grid point near a grid point having a large current source among the obtained current sources is proposed. Here, the square error is globally minimum when the above described movement of the grid points and the calculation processing of the current source are repeated a plurality of times. It means that the value is the minimum among the errors.

According to this method, the number of unknown current sources on each set grid point is made smaller than the number of magnetic sensors, and the current source on each grid point is set as a condition for obtaining a current source from a measured magnetic field. Since the condition that the square error between the magnetic field exerted by the source and the actual measured magnetic field is minimized is employed, the current source can be estimated more accurately. Furthermore, since the current source when the square error is globally minimized is estimated as a true current source, it is not necessary to set a convergence determination value in the process of specifying the final current source, and the final current source becomes unnecessary. The source can be uniquely identified.

However, this method still has the following points to be improved. That is, according to the above-described method of estimating the current source using the linear least squares method, the moments (directions and magnitudes) of the current sources on each grid point are varied, and the magnetic fields of these current sources are canceled out. Current sources on each grid point are estimated such that the resulting magnetic field is the same as the measured magnetic field. That is, the estimated moment of each current source is likely to vary. Further, when noise is discretely mixed into the measurement magnetic field, even the noise component may be calculated as a solution (current source).

[0022]

Therefore, the present applicant has
Japanese Patent Application No. 6-152680 (JP-A-7-327943)
), The above-mentioned drawbacks are eliminated, and in particular, the variation in the physical quantity of the estimated current source is suppressed, and even if noise is mixed in the measured magnetic field, the biological activity current source estimation device is hardly affected by the noise. Suggested. This biological activity current source estimating apparatus is configured as follows. That is, the minute magnetic field measured by each magnetic sensor is converted into digital data, and then collected and stored. In addition, a plurality of grid points are virtually set in the diagnosis target area with a number smaller than the number of magnetic cesans. Then, a square error between a magnetic field exerted by an unknown current source on each grid point and collected and stored magnetic field data, and a weighted sum of squares of the current source (hereinafter, referred to as
A current source on each grid point can be obtained by calculation, with the condition that the sum of the current source and the current source is also minimized.

Subsequent to the calculation of the current source, a function including a square error between the magnetic field calculated from the obtained current source and the magnetic field data actually measured and collected and stored by the magnetic sensor is globally minimized. Is determined. When it is determined that the current value is not the minimum, another group of grid points is moved to a position near a grid point where a current source having a large value exists among the current sources obtained by the calculation, and the group of grid points is rearranged. The calculation, determination, and grid point group rearrangement of the current source are repeated, and when the function including the square error is globally determined to be the minimum, the current source corresponding to the magnetic field is determined. Estimated as a true current source.

In the biological activity current source estimating apparatus according to the above prior application, the penalty term which is a weighted sum of squares of the current source is:
Since the value of the current source becomes smaller as the current source is solidified (converged), the variation in the physical quantity of the current source calculated is suppressed, and the discretely existing noise is erroneously detected. It is expected that the possibility of being estimated as a solution is reduced. However, even in the biological activity current source estimating apparatus of the prior application, a current source may be estimated at a grid point far from another grid point, and accurate identification of the current source may be hindered. That is,
For example, it is considered that the inconvenience that discretely existing noise is erroneously estimated as a solution still remains without being sufficiently eliminated.

In view of the above circumstances, the present invention provides an apparatus for estimating a biological activity current source that can accurately specify a current source that is not erroneously estimated at a grid point far from other grid points. The task is to provide.

[0026]

In order to solve the above-mentioned problems, the inventor has studied from various angles, and obtained a weighted sum of squares of the current sources used in calculating the current sources on each grid point. Pays attention to the penalty term. That is, it was presumed that the penalty term might not function properly when calculating the current source. This penalty term (which will be described in detail later) is [λΣ (wi
.Pi ² )], where i: 1 to 3
n and λ are weighting coefficients for a penalty term, and wi is a coefficient for canceling the influence of the distance from the magnetic sensor to the magnetic field measurement surface. In short, the effect of this penalty term is that a current source is not found at a grid point where the penalty term has a large value, and a current source is found at a grid point where the penalty term has a small value. . Based on this, in order to prevent an inappropriate current source caused by measurement magnetic field noise from being estimated at a grid point (distant grid point) far from other grid points, It was inferred that the penalty term should have a large value if there is a remote grid point.

Therefore, when the current penalty term is examined again, the presence / absence of a remote grid point is hardly related to the weight coefficient of the penalty term, and the penalty term is large due to the presence of the remote grid point. It did not become a value, and it turned out that the situation where the current source was estimated even at the remote grid point was caused. That is, conventionally, an appropriate constant is used as the weighting coefficient λ of the penalty term, or a simple average value of the shortest distance between each grid point is used. It was concluded that the uniform value irrespective of the presence or absence of an existing remote grid point resulted in the estimation of an inappropriate current source due to noise at the remote grid point.

Based on the above inference results, we continued to study measures to make the presence or absence of a remote grid point appear as a difference in the weighting factor of the penalty term. The weighting (weighting factor) of the squared value is not uniform, and a weighting factor is set for each grid point assuming that it corresponds to the average value of the distance between each grid point and all other grid points. For example, if there is a remote grid point, the penalty term becomes a large value, which leads to the finding that it is possible to prevent a situation in which an improper current source caused by noise is estimated at the remote grid point. The invention was completed.

Therefore, in order to solve the above-mentioned problems, the biological activity current source estimating apparatus of the present invention is an apparatus for estimating a physical quantity such as a position, a size, and a direction of the biological activity current source. A plurality of magnetic sensors arranged in close proximity to the diagnosis target area for measuring a minute magnetic field generated by a biological activity current source in the diagnosis target area; and (b) converting magnetic field data measured by each of the magnetic sensors into digital data. (C) data collection means for collecting and storing the magnetic field data converted to the digital data; and (d) the number of the magnetic field sensors in the diagnosis target area is smaller than the number of the magnetic sensors. (E) the square of the magnetic field exerted by the unknown current source on each of the grid points and the magnetic field data stored in the data collection means. Current source calculating means for obtaining an unknown current source by adding a condition that minimizes the sum of the weighted sum of squares of the current source; and (f) a magnetic field calculated from the obtained current source. Determining means for determining whether a function including a square error from magnetic field data actually measured by the magnetic sensor and stored in the data collecting means is globally minimum; and (g) the square When it is determined that the function including the error is not globally minimum, among the current sources on each grid point obtained by the current source calculation unit, the other is moved to the vicinity of the grid point where the current source having a large value exists. (H) repeating the processing by the grid point group rearrangement means for moving the grid point group and rearranging the grid point group; and (h) the current source calculation means, the determination means, and the grid point group rearrangement means. The function including the square error is global by the judgment means. Current source specifying means for estimating the current source corresponding to the magnetic field when it is determined to be the minimum as a true current source; and (i) determining the current source estimated by the current source specifying means as a diagnosis target area of the subject. A biological activity current source estimating device comprising: display means for displaying a superimposed superimposed image on the tomographic image of the tomographic image. The weighting factors are determined based on the weighting factors, and the square values of the corresponding current sources are weighted by the respective weighting factors to obtain a weighted sum of squares of the current sources.

[Operation] The operation at the time of current source estimation in the life activity current source estimation device of the present invention is as follows.
The minute magnetic field measured by each magnetic sensor is converted into digital data by the data conversion unit, and then stored in the data collection unit. Then, a plurality of grid points are virtually set in the diagnosis target area by the grid point setting means. The number of the lattice points is set smaller than the number of the magnetic cesans. A condition is added that minimizes the sum of the square error between the magnetic field exerted by the unknown current source on each grid point and the magnetic field data stored in the data collection means, and the weighted sum of squares of the current source. Then, the current source on each grid point is obtained by the current source calculating means. Then, at this time, the current source calculating means calculates the weighted sum of squares of the current source, for each grid point,
A weighting factor is determined based on the average distance between the grid point and all other grid points, and a corresponding current source is weighted by each weighting factor.

Here, the penalty term, which is the weighted sum of squares of the current sources, has the property that the value decreases as the current sources are more solidified (closer together). Of the physical quantity of the current source is suppressed, and the presence of a remote grid point appears as a difference in the value of each weight coefficient in the weighted sum of squares of the current source. Since the weighted sum of squares has a large value, discretely existing noise is hardly found as a solution.

Subsequently, a function including a square error between the magnetic field calculated from the current source calculated by the current source calculating means and the magnetic field data actually measured by the magnetic sensor and stored in the data collecting means is globally determined. It is determined by the determination means whether or not the minimum has been reached. If it is determined that it is not the minimum, the grid point group rearrangement means moves another grid point group to a vicinity of a grid point where a current source having a large value exists among the current sources obtained by the current source calculation means. Then, the grid point group is rearranged. Each process by the current source calculation unit, the determination unit, and the grid point group rearrangement unit is repeated. When the determination unit determines that the function including the square error is globally minimum, the current source identification unit determines the magnetic field. Is estimated as a true current source. The current source estimated by the current source specifying unit is displayed by the display unit so as to be superimposed on the tomographic image of the diagnosis target region of the subject.

[0033]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of a biological activity current source estimating apparatus according to the present invention. <First Embodiment> FIG. 1 is a block diagram showing a schematic configuration of a living activity current source estimating apparatus according to an embodiment. In the drawing, reference numeral 2 denotes a magnetically shielded room.
A bed 3 in which the subject M is supine, a multi-channel SQUID sensor 1 which is disposed in close proximity to, for example, the brain of the subject M, and non-invasively measures a small magnetic field generated by a biological activity current source generated in the brain. Is provided. As described above, the multi-channel SQUID sensor 1 has a large number of magnetic sensors housed in a dewar in a state of being immersed in the refrigerant. In the present embodiment, when the brain of the subject M is a sphere, each magnetic sensor is constituted by a pair of coils for detecting a magnetic field component in the radial direction.

The magnetic field data detected by the multi-channel SQUID sensor 1 is supplied to a data conversion unit 4 and converted into digital data, and then collected in a data collection unit 5. The stimulating device 6 is for giving an electrical stimulus (or a sound, a light stimulus, or the like) to the subject M. The positioning unit 7 is a multi-channel S
This is a device for grasping the positional relationship of the subject M with respect to a three-dimensional coordinate system based on the QUID sensor 1. For example, small coils are attached to a plurality of locations of the subject M, and power is supplied to these small coils from the positioning unit 7.
Then, the magnetic field generated from each coil is detected by the multi-channel SQUID sensor 1 to grasp the positional relationship of the subject M with respect to the multi-channel SQUID sensor 1. In addition, other methods for grasping the positional relationship of the subject M with respect to the SQUID sensor 1 include a method of attaching a projector to a dewar and irradiating the subject M with a light beam to grasp the positional relationship between the two. Alternatively, various methods disclosed in Japanese Patent Application Laid-Open Nos. 5-237065 and 6-788925 are used.

The data analysis unit 8 is for estimating a current source in the diagnosis target area of the subject M based on the magnetic field data collected by the data collection unit 5.
The data analysis unit 8 has functions as a grid point setting unit, a current source calculation unit, a determination unit, a grid point group rearrangement unit, and a current source identification unit according to the present invention, as will be apparent from the following description. . The magneto-optical disk 9 provided in association with the data analysis unit 8 stores, for example, a tomographic image obtained by an X-ray CT apparatus or an MRI apparatus, and a current source estimated by the data analysis unit 8 is The image is superimposed on these tomographic images (TV) and displayed on the color monitor 10 or the color printer 1
1 is printed out. That is, in the present invention, the display of the estimated current source is not only displayed on the screen of the color monitor 10 but also printed on the surface of a sheet such as a film. In addition,
A tomographic image obtained by an X-ray CT apparatus or MRI apparatus is shown in FIG.
May be directly transmitted to the data analysis unit 8 via the communication line 12 shown in FIG.

As described above, the multi-channel SKU
The positional relationship of the subject M with respect to the three-dimensional coordinate system based on the ID sensor 1 is measured and stored. After collecting the magnetic field data in the data collection unit 5, the data analysis unit 8 executes a current source estimation process. Hereinafter, description will be given with reference to the flowchart shown in FIG.

Similar to the conventional example shown in FIG. 7, a three-dimensional grid point group N is assumed in a diagnosis target area, for example, the brain (step S1). Here, the unknown number 3n (the number of current sources assumed in the X, Y, and Z directions for each grid point) for the grid point group N is the number of magnetic sensors S _{1 to} S _m constituting the multi-channel SQUID sensor 1. The number of grid points of the grid point group N is set so as to be smaller than the number m.

Then, each coefficient of the matrix A represented by the equation (3) is calculated using Biot-Savart's law (each coefficient of the matrix A is calculated for each movement of a lattice point described later). Thereafter, a current source on each grid point is obtained by a linear least squares method with a penalty term (step S2). Hereinafter, the process of step S2 will be described in detail.

Using the above equation (4), the magnetic sensor S ₁
Measured by to S _m, determine the magnetic field data stored in the data collecting unit 5 current sources on each grid point from [Bd] [P]. Equation (4) is shown below again. (P) = A ^- [Bd] ......... (4)

The matrix A is expressed by the above equation (3).
It is a matrix having × m elements. In equation (4), no solution is obtained because the number m of equations (the number of magnetic sensors S _{1 to} S _m ) is greater than the number of unknowns 3n. Therefore, the square error between the magnetic field (B) which assumed current source and the measured field [Bd] on each grid point (P) is on the magnetic sensor S ₁ to S _m | [Bd] - [B] | A current source [P] is calculated using the linear least squares method by adding a condition of minimizing the evaluation function f expressed by the sum of the following and a penalty term. This evaluation function is shown in equation (7).

[0041]

(Equation 3)

The first term of the evaluation function is the square error of the magnetic field,
The term is a penalty term. Λi in the penalty term is a weighting factor of the penalty term, and wi is a coefficient for canceling the influence of the distance from the magnetic sensor to the magnetic field measurement surface. Since the penalty term has a property that its value becomes smaller as the current source is solidified, the current source is obtained under the condition that only the square error of the magnetic field of the first term in the evaluation function is minimized. In this case, the tendency that the current source is likely to vary is suppressed, and the adoption of discrete noise components as solutions is reduced.

As a characteristic configuration of the present invention, the weighting factor λi of the penalty term in equation (7) is used.
Is calculated for each grid point, and a weight coefficient λi for each grid point is determined based on the average distance between the grid point and the other grid points. The shorter the average distance is, the smaller the weight coefficient λi is. As a result, when the lattice point is far from other lattice points, the weight coefficient λi increases, and as a result, the effect of the penalty term in equation (7) increases, and at lattice points far from the other lattice points, Estimation of the current source is no longer required, and variations in the estimated current source due to the influence of noise components can be suppressed. On the other hand, when the lattice points are locally gathered, the weighting factor λi becomes smaller, and as a result, the influence of the penalty term in equation (7) becomes smaller, so that unnecessary concentration of the current source can be avoided. Also.

By adding the condition of minimizing the evaluation function f, equation (4) is represented by [P] = A ⁺ [Bd] (8). Here, A ⁺ is obtained by the following equation (9). A ⁺ = (A ^t A + Γ · W) ⁻¹ · A ^t (9) (9) Γ in the expression (9) is a matrix represented by the following expression (10), and (9)
W in the equation is a matrix represented by the following equation (11).

[0045]

(Equation 4)

[0046]

(Equation 5)

Here, in general, as the current source Pi is closer to the magnetic field measurement surface, a larger magnetic field is measured.
By performing as in (12), for the current source Pi to be obtained,
The effect of the distance to the magnetic field measurement surface (ie, the current source Pi
Tend to be estimated close to the magnetic field measurement plane).

[0048]

(Equation 6)

[0049] When the above-mentioned (8) and (9) current source by formula (P) is obtained, the process proceeds to step S3, the magnetic field estimated current source (P) is on the magnetic sensor S ₁ to S _m [ It is determined whether or not the sum of the square error between B] and the measured magnetic field [Bd] and the above-described penalty term is globally minimum. Here, the fact that the sum of the square error of the magnetic field and the penalty term is the global minimum is determined in step S4 described later.
When the grid point group is moved multiple times in the process of repeating
The sum of the square error of the magnetic field and the penalty term obtained by the above-described method at the position of each lattice point has the smallest value. The determination as to whether or not the sum of the square error of the magnetic field and the penalty term is globally minimum is made in step S
In the process of repeating Step 4, for the rearranged grid points,
By the above-described processing of steps S2 and S3, the sum of the square error of the magnetic field and the penalty term obtained respectively is stored, and the respective values are compared to determine the minimum value as the global minimum value. Good.

If it is determined that it is not the minimum, step S
Proceeding to 4, the current source on another grid point is moved around the grid point where the current source having a large value exists among the current sources [P] obtained in step S2. As a result, a new lattice point group having the same number of lattice points as the original lattice point group N and a close interval is obtained. It should be noted that the current source having a large value is not necessarily one, and there are usually a plurality of current sources.

A method for bringing another group of grid points closer to a grid point where a current source having a large value exists is not particularly limited.
For example, the following method is exemplified. The magnitude of the current source at each grid point obtained in step S2 is considered as a mass, and it is assumed that an attractive force acts between each grid point due to gravity. Then, each lattice point approaches a lattice point having a large mass, and a lattice point group having a higher density is obtained as the lattice point is closer to a lattice point having a large mass. The moving distance of each grid point is set appropriately.
Further, as another grid point moving method, for example, eight grid points each having a small estimated current are moved to a vertex of a cube centered on a grid point having a large estimated current. At this time, the size of the cube is, for example, half the distance between a grid point having a large estimated current and the closest grid point.

After moving the grid points, the process returns to step S2, and the condition that the above-mentioned evaluation function is minimized is added to the moved grid point group, and the current source [P] is calculated by the linear least square method. Ask for a new one. Then, in step S3, it is determined again whether or not the sum of the square error and the penalty term is globally minimum. If it is not, the above-described steps S2 to S4 are repeated.

If it is determined in step S3 that the current is the minimum, the current source [P] for the globally minimum magnetic field [B] in step S5 is estimated as a true current source.
The current source [P] estimated as a true current source is superimposed on the X-ray CT image or MR tomographic image stored in the magneto-optical disk 9 shown in FIG. .

<Simulation> A simulation was performed to visually confirm the operation of the above-described method.
A current source in which a current of 10 nA was flowing through a cube having a side of 20 mm was estimated. FIG. 3 shows a simulation result estimated according to the first embodiment by setting 32 grid points in a magnetic field (signal-to-noise ratio S / N = 20) measured by a current source using 129 magnetic sensors. In addition, the simulation was performed in the same manner as in the first embodiment except that the weighting factor λ of the penalty term in the equation (7) was made to be the same as the simple average value of the distance between all the lattice points as in the prior application. FIG. 6 shows the results. In FIG. 6, large current sources 1d and d2 are estimated at remote grid points, which are error current sources that do not exist due to measurement noise. In FIG. 3 according to the first embodiment, no error current source is recognized at all, and it can be seen that the current source is accurately estimated in the present invention even when there is measurement noise.

<Second Embodiment> The method described in the first embodiment can perform a good estimation when the magnetic field source is assumed to be a current dipole. In step S3), a penalty term is added, and the estimated current source tends to solidify (locally concentrate). Therefore, when the current sources are distributed with a certain degree of spread, there is a problem that it is difficult to correctly estimate these current sources. In the present embodiment, in view of this point, a current source having a spread can be correctly estimated. Since the schematic configuration of the apparatus is the same as that of the first embodiment shown in FIG. 1, the description is omitted here.

Hereinafter, referring to the flowchart of FIG.
The current source estimating process in the data analysis unit 8 will be described. In step S11, as in the first embodiment, a grid point group is set evenly in the diagnosis target area. In step S12, Upon current source from the magnetic field measured by the magnetic sensor S ₁ to S _m [Bd] a [P] obtained by linear least squares method, as in step S2 of the first embodiment, the measured magnetic field [ Bd] the square error between the magnetic field assumed current sources (P) is on the magnetic sensor S ₁ to S _m on each lattice point [B] | [Bd] - [B] | and the sum of the penalty term The condition that the evaluation function f represented by is minimized is added.

Of course, also here, the weighting factor λi of the penalty term in the equation (7) is calculated for each grid point, and the average distance between the grid point and the other is calculated. Weighting factor .lambda.i of The shorter the average distance is, the smaller the weight coefficient λi is. As a result, when the lattice point is far from other lattice points, the weight coefficient λi increases, and as a result, the effect of the penalty term in equation (7) increases, and at lattice points far from the other lattice points, No more estimating the current source,
It goes without saying that the variation of the estimated current source due to the influence of the noise component can be suppressed. Also, when the lattice points are locally gathered, the weighting factor λi becomes smaller, and as a result, the influence of the penalty term in Equation (7) becomes smaller, so that unnecessary concentration of the current source can be avoided. The same is true.

When the current source [P] is obtained, step S13 is performed.
The willing, the estimated current source (P) is the magnetic sensors S ₁ ~
Square error between the magnetic field measured with the magnetic field (B) on S _m [Bd] determines whether it is globally minimized. That is, in this step S13, the penalty term is removed from the specific reference of the optimum current source, so that it is possible to prevent the unnecessary solidified current source from being adopted as the optimum current source.

If it is determined that the current source is not the minimum, the process proceeds to step S14 as in the first embodiment, and the current sources on the other grid points are moved around the grid point where the maximum current source exists. Returning to step S12, a current source [P] is newly obtained for the moved lattice point group by the linear least squares method to which a penalty term is similarly added. Then, it is determined again in step S13 whether or not the square error is globally minimum. If it is determined that the square error is not minimum, step S1 described above is performed.
Steps 2 to S14 are repeated. If it is determined in step S13 that the current source is the minimum, the current source [P] for the globally minimum magnetic field [B] in step S15 is estimated as a true current source, and the current source is regarded as a tomographic region in the region of interest. The display is overlaid on the screen and the processing is completed.

<Simulation> A simulation was performed in the same manner as in the first embodiment. A current source in which a current of 10 nA was flowing through a cube having a side of 20 mm was estimated. FIG. 5 shows the simulation results. No error current source is observed, and the estimated current source spread is wider than in the case of FIG. 3 of the first embodiment. It can be seen that the second embodiment is suitable when the current source spread is large. .

The present invention is not limited to the above embodiment, but can be modified as follows. (1) In the above embodiment, the number of magnetic sensors is 129 and the number of grid points set is 32. However, the number of magnetic sensors and the number of grid points set are larger than those in the embodiment. Or less.

(2) In the above embodiment, the diagnosis target area is the head, but it goes without saying that the diagnosis target area may be a part other than the head.

[0063]

As is apparent from the above description, according to the present invention, the square error between the magnetic field exerted by the unknown current source on each grid point and the measured magnetic field data, and the weighting of the current source When a current source on each grid point is obtained by adding a condition of minimizing the sum of the square sum, weighting of the square value of the current source in the penalty term is performed on each grid point and all other grid points. Since it is assumed that it corresponds to the average value of the distance of each grid point, the variation of the physical quantity of the current source obtained is suppressed, and the noise caused by the grid point estimated alone at a distant position The current source is not erroneously determined as a solution, and the spread of the current source can be accurately specified. At the same time, a decrease in the number of grid points for which the current source is erroneously estimated means an increase in the number of grid points for which the current source is correctly estimated, which leads to an improvement in the accuracy of current source analysis.

In the present invention, as in the case of the prior application, in order to improve the accuracy of current source estimation, there is a current source having a large value among the current sources previously obtained without increasing the number of grid points. Another grid point group is moved to the vicinity of the grid point to be changed, and the current source is estimated again. That is, since the number of unknowns is constant, not only the calculation accuracy of the current source calculation can be maintained, but also the sum of the square error of the magnetic field and the weighted sum of squares of the current source is globally minimized. Is estimated as the true current source, the setting of the convergence determination value is unnecessary in the process of specifying the final current source, and the final current source can be uniquely specified. .

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of an embodiment of a life activity current source estimating apparatus according to the present invention.

FIG. 2 is a flowchart showing a flow of a current source estimating process by the device of the first embodiment.

FIG. 3 is an explanatory diagram showing a simulation result of current source estimation according to the first embodiment.

FIG. 4 is a flowchart illustrating a flow of a current source estimation process performed by the device according to the second embodiment;

FIG. 5 is an explanatory diagram showing a simulation result of current source estimation according to the second embodiment.

FIG. 6 is an explanatory diagram showing a simulation result of current source estimation by a conventional device.

FIG. 7 is an explanatory diagram of an estimation process according to a conventional example.

FIG. 8 is an explanatory diagram of an estimation process according to another conventional example.

[Explanation of symbols]

1 ... multichannel SQUID sensor 2 ... magnetic shield room 4 ... data conversion unit 5 ... data collection unit 8 ... data analysis unit 10 ... color monitor M ... subject S ₁ to S _m ... magnetic sensor .lambda.i ... weighting factor

Claims (1)

[Claims] 1. An apparatus for estimating a physical quantity such as a position, a size, and a direction of a biological activity current source, comprising: (a) a biological activity current source disposed close to a diagnostic target area of a subject; A plurality of magnetic sensors for measuring a minute magnetic field by a source; and (b) data conversion means for converting magnetic field data measured by each of the magnetic sensors into digital data;
(C) data collection means for collecting and storing the magnetic field data converted to the digital data; and (d) a plurality of grids in the diagnosis target area so that the number thereof is smaller than the number of the magnetic sensors. (E) a square error between a magnetic field exerted by an unknown current source on each of the grid points and magnetic field data stored in the data collection unit, and a weighted 2 of the current source. Current source calculating means for obtaining an unknown current source by adding a condition of minimizing the sum of the power sum, and (f) a magnetic field calculated from the obtained current source and actually measured by the magnetic sensor. Determining means for determining whether a function including a square error with the magnetic field data stored in the data collection means is globally minimum; and (g) determining that the function including the square error is globally minimum. If not determined Then, among the current sources on each grid point obtained by the current source calculation means, another grid point group is moved to a position near a grid point where a current source having a large value exists, and the grid point group is rearranged. (H) repeating each processing by the current source calculating means, the determining means, and the grid point group relocating means, and the determining means globally minimizes a function including a square error. Current source identification means for estimating the current source corresponding to the magnetic field when determined to be a true current source,
(I) a biological activity current source estimating apparatus, comprising: display means for displaying the current source estimated by the current source specifying means so as to be superimposed on a tomographic image of the diagnosis target area of the subject.
In the current source calculating means, a weighting factor is determined for each grid point based on an average distance between the grid point and all other grid points, and a square value of a corresponding current source is determined by each weighting factor. , And a weighted sum of squares of the current source is calculated.