JP7795396B2 - Magnetic field measurement device, magnetic field measurement method, and magnetic field measurement program - Google Patents

Description

The present invention relates to a magnetic field measurement device, a magnetic field measurement method, and a magnetic field measurement program.

US Pat. No. 6,299,499 describes a method for suppressing sensor noise in a spatially oversampling sensor array.
[Prior art documents]
[Patent Documents]
Patent Document 1: U.S. Patent Application Publication No. 2019/125268

A first aspect of the present invention provides a magnetic field measurement device. The magnetic field measurement device may include a magnetic sensor array composed of a plurality of magnetic sensor cells, each having a magnetic sensor. The magnetic field measurement device may include a magnetic field acquisition unit that acquires measurement data measured by the magnetic sensor array. The magnetic field measurement device may include a signal space separation unit that extracts first internal space data from the spatial distribution of the magnetic field indicated by measurement data from a portion of the magnetic sensor cells in the magnetic sensor array, and extracts second internal space data from the spatial distribution of the magnetic field indicated by measurement data from a plurality of magnetic sensor cells including another portion of the magnetic sensor cells in the magnetic sensor array. The magnetic field measurement device may include a calculation unit that calculates a projection matrix based on the first internal space data. The magnetic field measurement device may include a noise removal unit that removes noise from the second internal space data using the projection matrix.

The signal spatial separation unit may extract second internal spatial data from the spatial distribution of the magnetic field indicated by the measurement data from all the magnetic sensor cells of the magnetic sensor array.

The signal spatial separation unit may extract third internal spatial data from the spatial distribution of the magnetic field indicated by measurement data from a portion of the magnetic sensor cells of the magnetic sensor array that is different from the portion of the magnetic sensor cells from which the first internal spatial data is extracted. The calculation unit may calculate one or more projection matrices based on the first internal spatial data and the third internal spatial data.

The calculation unit may calculate an intersection between a time domain basis in the first internal space data and a time domain basis in the third internal space data, and calculate a projection matrix based on the intersection.

The calculation unit may calculate a first projection matrix based on the first internal space data, and calculate a second projection matrix based on the third internal space data. The noise removal unit may use the first projection matrix to remove a portion of the noise in the second internal space data, and use the second projection matrix to remove another portion of the noise in the second internal space data.

The system may include a subtraction unit that calculates the difference between the first internal space data and the third internal space data. The calculation unit may calculate a projection matrix based on the difference.

The calculation unit may calculate a basis in the time domain of the first internal space data by performing singular value decomposition on the first internal space data, and calculate a projection matrix based on the basis.

The magnetic field measurement device may further include an averaging processing unit that calculates the average value of multiple pieces of data for the first internal space data and the second internal space data. The calculation unit may calculate a projection matrix based on the average value of the first internal space data. The noise removal unit may use the projection matrix to remove noise from the average value of the second internal space data.

The noise removal unit may remove noise from the second internal space data by projecting the second internal space data into signal space using a projection matrix. The magnetic sensor may have a magnetoresistive element.

A second aspect of the present invention provides a magnetic field measurement method. The magnetic field measurement method may include a magnetic field acquisition step of acquiring measurement data measured by a magnetic sensor array composed of a plurality of magnetic sensor cells, each having a magnetic sensor. The magnetic field measurement method may include a signal space separation step of extracting first internal space data from the spatial distribution of the magnetic field indicated by the measurement data from a portion of the magnetic sensor cells in the magnetic sensor array, and extracting second internal space data from the spatial distribution of the magnetic field indicated by the measurement data from a plurality of magnetic sensor cells including another portion of the magnetic sensor cells in the magnetic sensor array. The magnetic field measurement method may include a calculation step of calculating a projection matrix based on the first internal space data. The magnetic field measurement method may include a noise removal step of removing noise from the second internal space data using the projection matrix.

A third aspect of the present invention provides a magnetic field measurement program. The magnetic field measurement program causes a computer to function as a magnetic field acquisition unit that acquires measurement data measured by a magnetic sensor array composed of a plurality of magnetic sensor cells, each having a magnetic sensor. The magnetic field measurement program causes the computer to function as a signal space separation unit that extracts first internal space data from the spatial distribution of the magnetic field indicated by measurement data from a portion of the magnetic sensor cells in the magnetic sensor array, and extracts second internal space data from the spatial distribution of the magnetic field indicated by measurement data from a plurality of magnetic sensor cells including another portion of the magnetic sensor cells in the magnetic sensor array. The magnetic field measurement program causes the computer to function as a calculation unit that calculates a projection matrix based on the first internal space data. The magnetic field measurement program causes the computer to function as a noise reduction unit that uses the projection matrix to reduce noise in the second internal space data.

Note that the above summary of the invention does not list all of the necessary features of the present invention. Subcombinations of these features may also constitute inventions.

1 shows the configuration of a magnetic field measuring device 10 according to this embodiment. 1 shows the configuration of a magnetic sensor unit 110 according to this embodiment. 2 shows the configuration of a magnetic sensor cell 220 in a magnetic sensor array 210 according to this embodiment. 3 shows an example of input/output characteristics of a magnetic sensor having a magnetoresistive element according to the present embodiment. 1 shows an example of the configuration of a sensor unit 300 according to this embodiment. 1 shows an example of input/output characteristics of the sensor unit 300 according to this embodiment. 1 shows an example of the configuration of a magnetic sensor 520 according to this embodiment. 10 shows a magnetic flux distribution when a feedback magnetic field is generated in the magnetic sensor 520 according to this embodiment. 2 shows an example of the arrangement of a plurality of magnetic sensor cells 220 in a magnetic sensor array 210 according to this embodiment. An example of the definition of space in the magnetic field measuring device 10 according to this embodiment will be shown. 1 shows the configurations of a magnetic sensor array 210, a sensor data collection unit 230, and a sensor data processing unit 1100 according to this embodiment. 1 shows a first configuration example of a part of a magnetic field measuring device 10 according to this embodiment. 10 shows a flow of signal separation of the spatial distribution of a magnetic field by the magnetic field measuring device 10 according to this embodiment. An example of a flow in which the magnetic field measuring device 10 according to this embodiment removes remaining noise from the internal space data ^Bin obtained by signal space separation will be shown. 10 shows a second configuration example of a part of the magnetic field measuring device 10 according to the present embodiment. 10 shows a third example of the configuration of a part of the magnetic field measuring device 10 according to the present embodiment. 10 shows a fourth example of the configuration of a part of the magnetic field measuring device 10 according to the present embodiment. 10 shows a fifth configuration example of a part of the magnetic field measuring device 10 according to the present embodiment. 10 shows a sixth configuration example of a part of the magnetic field measuring device 10 according to the present embodiment. 10 shows the measurement results of the internal space data ^Bin before noise is removed in the magnetic field measuring device 10 of the sixth configuration example according to this embodiment. 10 shows the measurement results of the internal space data ^Binproj after noise has been removed according to this embodiment. 99 illustrates an example computer 9900 in which aspects of the present invention may be embodied, in whole or in part.

The present invention will be described below through embodiments of the invention, but the following embodiments do not limit the scope of the invention as claimed. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention.

Figure 1 shows the configuration of a magnetic field measurement device 10 according to this embodiment. The magnetic field measurement device 10 according to this embodiment extracts multiple different internal space data from the spatial distribution of the magnetic field indicated by magnetic field measurement data measured using a magnetic sensor array. The magnetic field measurement device 10 then removes at least a portion of the noise in the internal space data using a projection matrix calculated from other internal space data. In this embodiment, an example is described in which the magnetic field measurement device 10 is a magnetocardiogram measurement device that measures magnetocardiograms, which are magnetic fields generated by the electrical activity of the human heart. However, this is not limited to this. The magnetic field measurement device 10 may be used to measure magnetocardiograms of living organisms other than humans, or to measure biomagnetic fields other than magnetocardiograms, such as brain magnetic fields. The magnetic field measurement device 10 may also be used for magnetic particle testing to detect surface and subsurface defects in steel materials and welds.

The magnetic field measurement device 10 comprises a main body 100 and an information processing unit 150. The main body 100 is a component for sensing the subject's cardiac magnetism, and includes a magnetic sensor unit 110, a head 120, a drive unit 125, a base unit 130, and a pole unit 140.

The magnetic sensor unit 110 is positioned on the subject's chest facing the heart during cardiac magnetic field measurement and senses the subject's cardiac magnetic fields. The head 120 supports the magnetic sensor unit 110 and positions it facing the subject. The drive unit 125 is located between the magnetic sensor unit 110 and the head 120 and changes the orientation of the magnetic sensor unit 110 relative to the head 120 when performing calibration. The drive unit 125 in this embodiment includes a first actuator that can rotate the magnetic sensor unit 110 360 degrees around the Z axis in the figure, and a second actuator that rotates the magnetic sensor unit 110 around an axis perpendicular to the Z axis (the X axis in the state shown in the figure). These are used to change the azimuth angle and zenith angle of the magnetic sensor unit 110. As shown as the drive unit 125 in the figure, the drive unit 125 has a Y-shape when viewed from the Y axis direction in the figure, and the second actuator can rotate the magnetic sensor unit 110 360 degrees around the X axis in the figure.

The base 130 is a base that supports the other components, and in this embodiment serves as a platform on which the subject stands when measuring magnetic fields. The pole 140 supports the head 120 at the height of the subject's chest. The pole 140 may be extendable in the vertical direction to adjust the height of the magnetic sensor unit 110 to the height of the subject's chest.

The information processing unit 150 is a component that processes measurement data from the main unit 100 and outputs it by display, printing, etc. The information processing unit 150 may be a computer such as a PC (personal computer), tablet computer, smartphone, workstation, server computer, or general-purpose computer, or may be a computer system in which multiple computers are connected. Alternatively, the information processing unit 150 may be a dedicated computer designed for processing information from magnetocardiography, or dedicated hardware realized by dedicated circuitry.

Figure 2 shows the configuration of the magnetic sensor unit 110 according to this embodiment. The magnetic sensor unit 110 has a magnetic sensor array 210 and a sensor data collection unit 230. The magnetic sensor array 210 has multiple magnetic sensor cells 220 and is capable of detecting input magnetic fields in three axial directions. This figure shows a case in which the magnetic sensor array 210 has multiple magnetic sensor cells 220 arranged in each of the X, Y, and Z directions (for example, 8 magnetic sensor cells 220 in the X direction, 8 magnetic sensor cells 220 in the Y direction, and 2 magnetic sensor cells 220 in the Z direction, for a total of 128 magnetic sensor cells 220).

The sensor data collection unit 230 is electrically connected to the multiple magnetic sensor cells 220 included in the magnetic sensor array 210 (not shown), collects sensor data (detection signals) from the multiple magnetic sensor cells 220, and supplies the data to the information processing unit 150.

Figure 3 shows the configuration of the magnetic sensor cells 220 in the magnetic sensor array 210 according to this embodiment. Each of the multiple magnetic sensor cells 220 has at least one sensor unit 300, each having a magnetoresistive element. This figure illustrates an example in which each of the multiple magnetic sensor cells 220 has three sensor units 300x-z (collectively referred to as "sensor units 300") and can detect an input magnetic field in three axial directions. However, each of the multiple magnetic sensor cells 220 is not limited to having three sensor units 300x-z; it is sufficient that at least a portion of the magnetic sensor array 210 can detect an input magnetic field in three axial directions. In this case, as described below, when spatially sampling each spherical harmonic function using the magnetic sensor array 210, it is necessary to detect the dependence on spatial frequencies related to the angular momentum of the magnetic field. Therefore, it is preferable that the positions of the sensor units 300 in the magnetic sensor array 210 are arranged as evenly as possible, at least in the azimuth and zenith directions. For similar reasons, it is desirable that the magnetic sensing axes of each sensor in the magnetic sensor array 210 be arranged as evenly as possible, at least in the azimuth and zenith directions. Sensor unit 300x is arranged along the X-axis and can detect magnetic fields in the X-axis direction. Sensor unit 300y is arranged along the Y-axis and can detect magnetic fields in the Y-axis direction. Sensor unit 300z is arranged along the Z-axis and can detect magnetic fields in the Z-axis direction. As shown in the enlarged view indicated by the dashed-dotted lines in this figure, in this embodiment, each sensor unit 300 has magnetic concentrators arranged on both ends of the magnetoresistive element. Therefore, each sensor unit 300 samples the spatial distribution of the magnetic field using a magnetoresistive element arranged in a narrow position between the magnetic concentrators, thereby clearly defining the sampling points in space in each axis direction. The configuration of each sensor unit 300 will be described in detail later.

The multiple magnetic sensor cells 220 are arranged at equal intervals of Δx along the X-axis, Δy along the Y-axis, and Δz along the Z-axis. The position of each magnetic sensor cell 220 in the magnetic sensor array 210 is represented by a set [i, j, k], where i is the X-axis position, j is the Y-axis position, and k is the Z-axis position. Here, i is an integer satisfying 1≦i≦Nx (Nx indicates the number of arranged magnetic sensor cells 220 in the X-axis), j is an integer satisfying 1≦j≦Ny (Ny indicates the number of arranged magnetic sensor cells 220 in the Y-axis), and k is an integer satisfying 1≦k≦Nz (Nz indicates the number of arranged magnetic sensor cells 220 in the Z-axis). Note that the above description illustrates an example in which the multiple magnetic sensor cells 220 are arranged at equal intervals along each axis. However, this is not limiting. The multiple magnetic sensor cells 220 may be arranged at different intervals in at least one of the X-axis direction, Y-axis direction, and Z-axis direction, for example.

In this diagram, the three-axis directions of the magnetic field detected by sensor units 300x, 300y, and 300z are the same as the three-dimensional direction in which the magnetic sensor cells 220 are arranged. This makes it easier to understand each component of the distribution of the measured magnetic field. Furthermore, sensor units 300x, 300y, and 300z are arranged within each magnetic sensor cell 220 so that they do not overlap each other when viewed from each of the three-dimensional directions in which the magnetic sensor cells 220 are arranged. Furthermore, in this diagram, sensor units 300x, 300y, and 300z are arranged with one end located on the gap side between multiple sensor units 300 and the other end extending in each of the three axial directions away from the gap. As an example, this figure shows an example in which a gap is provided in the lower left corner of the magnetic sensor cell 220 when viewed from the front, and the sensor units 300x, 300y, and 300z are arranged so that one end is in contact with the gap and the other end is away from the gap, extending in the X-axis, Y-axis, and Z-axis directions. In this figure, the sensor units 300x, 300y, and 300z are arranged along three mutually perpendicular sides from one corner of the cubic magnetic sensor cell 220, with the gap provided in the corner. It is also preferable that the coils or magnetic bodies of the sensor units 300x, 300y, and 300z, described below, are arranged so as not to overlap each other. This clarifies the measurement points and makes it easier to understand each component of the measured magnetic field. Furthermore, the other-axis sensitivities of the sensor units 300x, 300y, and 300z can be considered equivalent to each other. This other-axis sensitivity is generated by mutual interference between the coils or magnetic materials of the sensor units 300x, 300y, and 300z. However, the three-axis directions of the magnetic field to be detected may differ from the three-dimensional direction in which the magnetic sensor cells 220 are arranged. When the two directions are different, there are no restrictions on the placement of the sensor units 300 within the magnetic sensor cells 220 or the arrangement direction of the magnetic sensor cells 220, increasing the degree of freedom in the design of the magnetic sensor array 210.

Figure 4 shows an example of the input/output characteristics of a magnetic sensor having a magnetoresistive element according to this embodiment. In this figure, the horizontal axis represents the magnitude B of the input magnetic field input to the magnetic sensor, and the vertical axis represents the magnitude V_xMR0 of the detection signal from the magnetic sensor. The magnetic sensor may have, for example, a giant magnetoresistive (GMR) element or a tunnel magnetoresistive (TMR) element, and detects the magnitude of a magnetic field in a predetermined uniaxial direction.

Such magnetic sensors have high magnetic sensitivity, which is the slope of the detection signal V_xMR0 relative to the input magnetic field B, and can detect weak magnetic fields of around 10 pT. On the other hand, with magnetic sensors, the detection signal V_xMR0 saturates when the absolute value of the input magnetic field B is around 1 μT, for example, and the range over which the linearity of the input/output characteristics is good is narrow. Therefore, adding a closed loop that generates a feedback magnetic field to such a magnetic sensor can improve the linearity (also called linearity) of the magnetic sensor. Such magnetic sensors are described below.

Figure 5 shows an example configuration of the sensor unit 300 according to this embodiment. The sensor unit 300 is provided inside each of the multiple magnetic sensor cells 220, and includes a magnetic sensor 520, a magnetic field generation unit 530, and an output unit 540. Note that part of the sensor unit 300, such as the amplifier circuit 532 and the output unit 540, may be provided on the sensor data collection unit 230 side rather than on the magnetic sensor cell 220 side.

Like the magnetic sensor described in Figure 4, the magnetic sensor 520 has a magnetoresistive element such as a GMR element or a TMR element. The magnetic sensor 520 also has magnetic concentrators disposed on both ends of the magnetoresistive element. The magnetoresistive element of the magnetic sensor 520 may be configured so that, assuming the positive direction of the magnetic sensing axis is the +X direction, its resistance increases when a magnetic field is input in the +X direction and decreases when a magnetic field is input in the -X direction. In other words, by observing changes in the resistance of the magnetoresistive element of the magnetic sensor 520, the magnitude of the magnetic field B input to the magnetic sensor 520 can be detected. For example, if the magnetic sensitivity of the magnetic sensor 520 is S, the detection result of the magnetic sensor 520 in response to the input magnetic field B can be calculated as S x B. The magnetic sensor 520 is connected to a power source, for example, and outputs a voltage drop corresponding to changes in resistance as the detection result of the input magnetic field. The configuration of the magnetic sensor 520 will be described in detail later.

The magnetic field generation unit 530 applies a feedback magnetic field to the magnetic sensor 520 that reduces the input magnetic field detected by the magnetic sensor 520. For example, the magnetic field generation unit 530 generates a feedback magnetic field B_FB that is oriented in the opposite direction to the magnetic field B input to the magnetic sensor 520 and has approximately the same absolute value as the input magnetic field, thereby operating to cancel out the input magnetic field. The magnetic field generation unit 530 includes an amplifier circuit 532 and a coil 534.

The amplifier circuit 532 outputs a current corresponding to the detection result of the input magnetic field of the magnetic sensor 520 as the feedback current I_FB. If the magnetic resistance element of the magnetic sensor 520 is configured as a bridge circuit including at least one magnetic resistance element, the outputs of the bridge circuit are connected to the input terminal pairs of the amplifier circuit 532. The amplifier circuit 532 then outputs a current corresponding to the output of the bridge circuit as the feedback current I_FB. The amplifier circuit 532 includes, for example, a transconductance amplifier, and outputs the feedback current I_FB corresponding to the output voltage of the magnetic sensor 520. For example, if the voltage-current conversion coefficient of the amplifier circuit 532 is G, the feedback current I_FB can be calculated as G x S x B.

The coil 534 generates a feedback magnetic field B_FB corresponding to the feedback current I_FB. The coil 534 is wound so as to surround the magnetoresistive element of the magnetic sensor 520 and the magnetic concentrator plates disposed on both ends of the magnetoresistive element. It is desirable that the coil 534 generate a uniform feedback magnetic field B_FB throughout the entire magnetic sensor 520. For example, if the coil coefficient of the coil 534 is β, the feedback magnetic field B_FB can be calculated as β × I_FB. Here, the feedback magnetic field B_FB is generated in a direction that cancels out the input magnetic field B, so the magnetic field input to the magnetic sensor 520 is reduced to B - B_FB. Therefore, the feedback current I_FB is expressed by the following equation:

By solving equation (1) for the feedback current I_FB, it is possible to calculate the value of the feedback current I_FB in the steady state of the sensor unit 300. If the magnetic sensitivity S of the magnetic sensor 520 and the voltage-current conversion coefficient G of the amplifier circuit 532 are sufficiently large, the following equation can be calculated from equation (1):

The output unit 540 outputs an output signal V_xMR corresponding to the feedback current I_FB that the magnetic field generating unit 530 flows to generate the feedback magnetic field B_FB. The output unit 540 has, for example, a resistive element with a resistance value R, and outputs a voltage drop caused by the feedback current I_FB flowing through the resistive element as the output signal V_xMR. In this case, the output signal V_xMR is calculated from equation (2) as follows:

As described above, the sensor unit 300 generates a feedback magnetic field that reduces the magnetic field input from outside, thereby effectively reducing the magnetic field input to the magnetic sensor 520. As a result, the sensor unit 300 can prevent the detection signal V_xMR from saturating even if the absolute value of the input magnetic field B exceeds 1 μT, for example, by using a magnetoresistive element having the characteristics shown in Figure 4 as the magnetic sensor 520. The input/output characteristics of this sensor unit 300 are described below.

Figure 6 shows an example of the input/output characteristics of the sensor unit 300 according to this embodiment. In this figure, the horizontal axis represents the magnitude B of the input magnetic field input to the sensor unit 300, and the vertical axis represents the magnitude V_xMR of the detection signal from the sensor unit 300. The sensor unit 300 has high magnetic sensitivity and can detect a weak magnetic field of about 10 pT. Furthermore, the sensor unit 300 can maintain good linearity of the detection signal V_xMR even when the absolute value of the input magnetic field B exceeds 100 μT, for example.

In other words, the sensor unit 300 according to this embodiment is configured so that the detection results for the input magnetic field B are linear within a predetermined range of the input magnetic field B, for example, when the absolute value of the input magnetic field B is less than several hundred μT. By using such a sensor unit 300, it is possible to easily detect weak magnetic signals, such as magnetocardiogram signals.

Figure 7 shows an example configuration of the magnetic sensor 520 according to this embodiment. As an example, the magnetic sensor 520 according to this embodiment has a magnetoresistive element 702 and magnetic concentrators 704 and 706 arranged at one end and the other end of the magnetoresistive element 702. The magnetic concentrators 704 and 706 are arranged to sandwich the magnetoresistive element 702 between them. That is, the magnetic concentrators are arranged at both ends of the magnetoresistive element 702. In Figure 7, the magnetic concentrator 704, which is arranged at the right end of the magnetoresistive element 702 along the magnetic sensing axis when viewed from the front, is the magnetic concentrator arranged on the positive side of the magnetic sensing axis, and the magnetic concentrator 706, which is arranged at the left end of the magnetoresistive element 702, is the magnetic concentrator arranged on the negative side of the magnetic sensing axis. When a magnetic field is input to the magnetic concentrators 704 and 706 from the negative side to the positive side of the magnetic sensing axis, the resistance of the magnetoresistive element 702 may increase or decrease. The magnetic sensing axis may be along the direction of magnetization fixed by the magnetization fixed layer that forms the magnetoresistive element 702. The magnetic concentrators 704, 706 are made of a soft magnetic material such as iron. By placing the magnetic concentrators 704, 706, which are made of a soft magnetic material, at one end and the other end of the magnetoresistive element 702, the number of magnetic field lines passing through the magnetoresistive element 702 can be increased, thereby improving the sensitivity of the magnetic sensor 520.

Note that while this figure shows an example in which magnetic concentrators are disposed at both one end and the other end of the magnetoresistive element 702, magnetic concentrators may be disposed at only one end or the other end of the magnetoresistive element 702. However, to further increase the sensitivity of the magnetic sensor 520, it is preferable to provide magnetic concentrators at both one end and the other end of the magnetoresistive element 702. Furthermore, if magnetic concentrators are provided at both one end and the other end of the magnetoresistive element 702, the position of the magnetoresistive element 702, which is located in a narrow position sandwiched between the two magnetic concentrators 704 and 706, becomes the magnetic sensing area, i.e., the spatial sampling point. This makes the magnetic sensing area clearer, further enhancing compatibility with the signal space separation (SSS) technology described below. In this way, by using a magnetic sensor 520 in each sensor unit 300, in which magnetic concentrators 704 and 706 are arranged on both ends of a magnetoresistive element 702, the magnetic field measuring device 10 of this embodiment can sample the spatial distribution of the magnetic field at an extremely narrow position (for example, 100 μm or less) sandwiched between the magnetic concentrators on both ends in each axial direction, as shown in FIG. 3. This results in higher sampling accuracy (positional accuracy) than when sampling the spatial distribution of the magnetic field using a SQUID coil (up to 2 cm) that measures biomagnetic fields.

Figure 8 shows the magnetic flux distribution when a feedback magnetic field is generated in the magnetic sensor 520 according to this embodiment. In Figure 8, components having the same functions and configurations as those in Figure 7 are designated by the same reference numerals, and descriptions are omitted except for differences. In the magnetic sensor 520 according to this embodiment, when a feedback current is supplied to the feedback coil 534, the feedback coil 534 generates a feedback magnetic field, generating a magnetic flux distribution as shown by the dashed line in this figure. The magnetic flux generated by this feedback magnetic field is spatially distributed so as to cancel the spatial distribution of the magnetic field input to the magnetoresistive element 702 and magnetically amplified by the magnetic concentrators 704 and 706. Therefore, when the magnetic concentrators 704 and 706 are arranged at both ends of the magnetoresistive element 702 as shown in this figure, the magnetic field distribution at the position of the magnetoresistive element 702 can be accurately canceled by the feedback magnetic field, thereby achieving a sensor with high linearity between the input magnetic field and the output voltage.

Figure 9 shows an example of the arrangement of multiple magnetic sensor cells 220 in a magnetic sensor array 210 according to this embodiment. For ease of explanation, in Figures 2 and 3, the magnetic sensor array 210 is shown as being planar. However, in reality, the magnetic sensor array 210 may have a curved surface that is curved in at least one direction, as shown in this figure. The multiple magnetic sensor cells 220 may then be arranged three-dimensionally so as to be located at lattice points included in the curved surface. As an example, the magnetic sensor array 210 may be arranged three-dimensionally so as to be arranged in an arc-shaped cross section.

That is, the multiple magnetic sensor cells 220 may be arranged in a cross-sectional arc shape along the chest of the measured object, with the center of gravity of the measured object at the center. In this case, each magnetic sensor cell 220 is arranged at a lattice point included in the curved surface shape in a three-dimensional lattice space. Note that here, the lattice points are lattice-like points arranged at equal intervals at predetermined intervals in the X, Y, and Z directions. As an example, when viewed from any one of the X, Y, and Z directions, each magnetic sensor cell 220 is arranged so as to follow a curved surface that has a convex shape in a direction perpendicular to the one direction. This figure shows an example in which each magnetic sensor cell 220 is arranged so as to follow a curved surface that has a convex shape in the positive direction of the Z axis when viewed from the Y direction. The magnetic sensor array 210 may form a curved surface shape that is convex in the positive direction of the Z axis by arranging each magnetic sensor cell 220 at a lattice point in a three-dimensional lattice space so that each vertex of each magnetic sensor cell 220 is positioned as far in the negative direction of the Z axis as possible, without exceeding a predetermined curved surface that is convex in the positive direction of the Z axis.

More specifically, in the cross-sectional view of this figure, the multiple magnetic sensor cells 220 on the inner side (negative Z-axis side), i.e., magnetic sensor cells 220[1,j,1] to 220[8,j,1], are arranged outside the arc indicated by the dashed-dotted line 915 so as to be positioned outside the inscribed circle of the magnetic sensor array 210 indicated by the reference numeral 910. Furthermore, the multiple magnetic sensor cells 220 on the outer side (positive Z-axis side), i.e., magnetic sensor cells 220[1,j,2] to 220[8,j,2], are arranged inside the arc indicated by the dashed-double-dot line 925 so as to be positioned inside the circumscribed circle of the magnetic sensor array 210 indicated by the reference numeral 920. The centers of these inscribed and circumscribed circles are common and coincide with the coordinate origin in the signal separation calculations described below.

In this manner, the magnetic sensor array 210 may be configured by arranging multiple magnetic sensor cells 220, each having a magnetic sensor 520, to form a surface that covers at least a portion of the object to be measured. In particular, in the magnetic sensor array 210, the multiple magnetic sensor cells 220 may be arranged three-dimensionally so as to be positioned at lattice points between two curved surfaces that are curved in one direction. These curved surfaces may be formed in a substantially parabolic shape. This allows the magnetic sensor array 210 to position the sensor unit 300 not only in one direction facing the heart but also in multiple directions, thereby enabling sensing of cardiac magnetism from multiple directions. Furthermore, since the magnetic sensor array 210 according to this embodiment has the magnetic sensor cells 220 formed in a rectangular parallelepiped shape, as an example, the shape of the magnetic sensor array 210 can be easily changed. In other words, the magnetic sensor array 210 according to this embodiment can adopt various shapes that can be configured by arranging the magnetic sensor cells 220 at lattice points, providing a high degree of design freedom. Therefore, as shown in this figure, the magnetic sensor array 210 can easily form a curved surface shape in three-dimensional space by arranging multiple magnetic sensor cells 220 at lattice points included in the curved surface shape in three-dimensional space. The magnetic field measurement device 10 measures the magnetic field by arranging the magnetic sensor array 210 so that the chest of the subject is located toward the center of the curved surface, i.e., so that the heart, which is the source of the magnetic field to be measured, is located toward the center of the curved surface. This allows the magnetic field measurement device 10 to perform signal spatial separation (described below) using measurement data measured at a position close to the heart, which is the source of the magnetic field to be measured, thereby separating the magnetic field to be measured and the disturbance magnetic field with high accuracy. In this case, it is preferable that the curvature of the curved surface of the magnetic sensor array 210 is approximately the same as the curvature around the chest of the subject, as this allows the magnetic field to be measured closer to the heart, which is the source of the magnetic field to be measured.

Figure 10 shows an example of the definition of space in the magnetic field measurement device 10 according to this embodiment. The area inside the circle inscribed in the magnetic sensor array 210, indicated by the reference numeral 910, is defined as the internal space, and the area outside the inscribed circle is defined as the external space. When measuring cardiac magnetism, the magnetic field measurement device 10 positions the magnetic sensor array 210 so that the object to be measured (i.e., the heart of the subject) is positioned opposite the surface formed by the magnetic sensor array 210 in the internal space. In this case, the cardiac magnetism to be measured is defined as the magnetic field to be measured Bin generated in the internal space.

Here, if the magnetic sensor array 210 measures the magnetic field of the heart and performs signal separation (signal separation here can also be called signal space separation) on the spatial distribution of the magnetic field indicated by the measurement data, and the magnetic field Bin can be accurately separated from other disturbance magnetic fields, the electrical activity of the heart can be more accurately understood. However, when the magnetic field measurement device 10 measures the magnetic field Bin, various environmental magnetic fields may interfere as disturbance magnetic fields. Such disturbance magnetic fields may include, for example, the external space magnetic field Bout generated in external space. Therefore, in reality, the signal-space separated estimate of the magnetic field Bin to be measured, i.e., the internal space data ^Bin (where "^" means "estimated value"), may contain, in addition to the magnetic field Bin to be measured generated by the electrical activity of the heart, interference components that could not be accurately separated as disturbance magnetic fields and remain. Such interference components may include components caused by the external space magnetic field Bout that should have been separated as external space data ^Bout.

The component resulting from the external space magnetic field Bout increases in proportion to the magnitude of the sensor error. If the magnetic sensor 520 includes a magnetoresistive element 702, errors in the sensor's main axis sensitivity and other axis sensitivity may increase due to manufacturing errors in the magnetoresistive element 702, etc. Furthermore, if the magnetic sensor 520 includes magnetic concentrators 704 and 706 positioned at both ends of the magnetoresistive element 702, errors in the sensor's main axis sensitivity and other axis sensitivity may increase due to misalignment between the magnetic concentrators 704 and 706 and the magnetoresistive element 702. For these reasons, as the sensor error of the magnetic sensor 520 increases, noise resulting from the external space magnetic field Bout becomes significant. The magnetic field measuring device 10 according to this embodiment removes at least a portion of the remaining noise from the internal space data ^Bin obtained by signal space separation.

Figure 11 shows the configuration of the magnetic sensor array 210, sensor data collection unit 230, and sensor data processing unit 1100 according to this embodiment. The sensor data processing unit 1100 may be included in the information processing unit 150 of Figure 1.

The magnetic sensor array 210 is configured by arranging multiple magnetic sensor cells 220, each having a magnetic sensor 520, to form a surface that covers at least a portion of the object to be measured. As an example, each of the multiple magnetic sensor cells 220 has multiple sensor units 300x-z as described above. This figure shows the portions of the multiple magnetic sensor cells 220 that the magnetic sensor array 210 has in each dimensional direction, relating to positions [i, j, k], [i+1, j, k], [i, j+1, k], and [i, j, k+1].

The sensor data collection unit 230 has multiple AD converters 1110 and a clock generator 1112. The multiple AD converters 1110 are provided corresponding to each of the multiple sensor units 300x-z of the magnetic sensor cell 220, and convert the analog detection signal (sensor output signal V_xMR in Figure 6) output by the corresponding sensor unit 300 into digital measurement data (Vx, Vy, Vz). Here, Vx, Vy, and Vz are measurement values (e.g., digital voltage values) obtained by converting the detection signals from sensor units 300x, 300y, and 300z into digital form.

The clock generator 1112 generates a sampling clock and supplies a common sampling clock to each of the multiple AD converters 1110. Each of the multiple AD converters 1110 then performs AD conversion in accordance with the common sampling clock supplied from the clock generator 1112. Therefore, all of the multiple AD converters 1110 that perform AD conversion on the outputs of the three-axis sensor units 300x-z located in different positions operate synchronously. This allows the multiple AD converters 1110 to simultaneously sample the detection results of the three-axis sensor units 300x-z located in different spaces.

The sensor data processing unit 1100 has a plurality of magnetic field acquisition units 1120, each corresponding to a respective one of the plurality of magnetic sensor cells 220, a plurality of calibration calculation units 1130, a plurality of data output units 1140, a basis vector storage unit 1150, a data storage unit 1155, a signal space separation unit 1160, and a calculation processing unit 1170.

The magnetic field acquisition unit 1120 is connected to three AD converters 1110, each connected to a corresponding magnetic sensor cell 220, and acquires measurement data measured by the sensor units 300x-z in the multiple magnetic sensor cells 220 that make up the magnetic sensor array 210. Specifically, the magnetic field acquisition unit 1120 may be configured using a flip-flop or the like that latches and acquires the digital measurement data (Vx, Vy, Vz) converted by the AD converter 1110 at a predetermined timing T.

The calibration calculation unit 1130 is connected to the magnetic field acquisition unit 1120 and calibrates the measurement data acquired by the magnetic field acquisition unit 1120 using calibration parameters. An overview of the calibration of measurement data by the calibration calculation unit 1130 is as follows. Let B(Bx, By, Bz) be the magnetic field input to the magnetic sensor cell 220 at position [i, j, k], and let V(Vx, Vy, Vz) be the detection result of the triaxial magnetic sensor by the sensor units 300x, 300y, and 300z. In this case, if the magnetic sensor characteristics of the triaxial magnetic sensor are represented by a matrix S, the detection result V of the triaxial magnetic sensor can be expressed as follows:

Here, Sxx, Syy, and Szz represent the sensitivity in the main axis direction (main axis sensitivity) of sensor units 300x, 300y, and 300z, respectively, and Sxy, Sxz, Syx, Syz, Szx, and Szy represent the sensitivity in the other axis directions (other axis sensitivity). Also, Vos,x, Vos,y, and Vos,z represent the offsets of sensor units 300x, 300y, and 300z, respectively. Here, the main axis direction refers to the direction that sensor units 300x, 300y, and 300z primarily measure, and the other axis direction refers to the direction that they primarily do not measure. In measuring a magnetic field, the main axis direction is the direction in which the magnetic sensor shows maximum sensitivity when a magnetic field is input (input axis direction, sensitivity axis direction). The other axis direction is the axis perpendicular to the main axis direction. For example, when sensor unit 300x measures the X-axis direction, the main axis direction is the X-axis, and the other axes are the Y-axis and Z-axis directions. Ideally, magnetic sensor 520 only has main axis sensitivity, but it may also have other axis sensitivity depending on the process, etc. Magnetic sensor 520 also has other axis sensitivity generated by the mutual interference described above. A vertical vector represented by the three components of sensor unit 300's main axis sensitivity and other axis sensitivity is called a sensitivity vector. For example, the sensitivity vector nx of sensor unit 300x is represented by the three components (Sxx, Sxy, Sxz). In this case, the output of sensor unit 300x is the dot product of the input magnetic field to the sensor and the sensitivity vector nx. Similarly, the sensitivity vector ny of sensor unit 300y is represented by the three components (Syx, Syy, Syz), and the sensitivity vector nz of sensor unit 300z is represented by the three components (Szx, Szy, Szz).

Since each sensor unit 300 has linearity in its detection results for the input magnetic field within the range of the input magnetic field to be detected, each element of matrix S has a substantially constant coefficient that is unrelated to the magnitude of the input magnetic field B. Furthermore, even if the sensor unit 300 has other-axis sensitivity, as long as the detection results of the sensor unit 300 are linear, each element of matrix S will have a substantially constant coefficient that is unrelated to the magnitude of the input magnetic field B.

Therefore, by using the inverse matrix S-1 of the matrix S and the offset (Vos, x, Vos, y, Vos, z), the calibration calculation unit 1130 can convert the measurement data V (Vx, Vy, Vz) into magnetic field measurement data B (Bx, By, Bz) indicating the original input magnetic field, as shown in the following equation. In other words, the calibration calculation unit 1130 calibrates the digital measurement data V from the magnetic field acquisition unit 1120 using the main axis sensitivity, the other axis sensitivity, and the offset. As a result, the calibration calculation unit 1130 corrects the offset, the sensitivity in the main axis direction, and the sensitivity in the other axis direction. This conversion also holds true when the sensor units 300x to 300z are equipped with the magnetic flux concentrator plates described above. This is because the magnetic sensor cell 220 is configured as a triaxial magnetic sensor using the sensor units 300x to 300z, which enables conversion using linear algebra. Note that offset calibration may be omitted if a high-pass filter or the like is provided between the output of the sensor unit 300 and the calibration calculation unit 1130 to convert the measurement data V into AC components. That is, the calibration calculation unit 1130 may calibrate the digital measurement data V from the magnetic field acquisition unit 1120 using at least one of the main axis sensitivity, the other axis sensitivity, and the offset. Note that these calibration parameters may be calculated in advance by measuring a known DC or AC magnetic field. Furthermore, the calibration calculation unit 1130 in this embodiment is only required to calibrate the output from each magnetic sensor cell 220 as components expressed in a coordinate system formed by three independent vectors, and does not necessarily need to correct the output to three-axis components expressed in a coordinate system of three orthogonal vectors (a so-called Cartesian coordinate system). That is, when all the magnetic sensor cells 220 are measuring the same magnetic field, each calibration calculation unit 1130 corresponding to each magnetic sensor cell 220 may calibrate the digital measurement data V from the corresponding magnetic field acquisition unit 1120 into the same magnetic field measurement data B expressed in independent three-axis components.

The calibration calculation unit 1130 uses the environmental magnetic field measurement data to calculate the inverse matrix S-1 of the matrix S and the offset (Vos, x, Vos, y, Vos, z), and converts the magnetic field measurement data acquired by the magnetic field acquisition unit 1120 into magnetic field measurement data B using these calibration parameters, and supplies it to the data output unit 1140.

As described above, because each sensor unit 300 has linearity, the calibration calculation unit 1130 can convert the measurement data into magnetic field measurement data B using approximately constant coefficients. In other words, the approximately constant coefficients used by the calibration calculation unit 1130 can be determined as a set of calibration parameters using environmental magnetic field data.

The data output unit 1140 supplies the magnetic field measurement data B calibrated by the calibration calculation unit 1130 to the data storage unit 1155.

The data storage unit 1155 stores the magnetic field measurement data B supplied from the data output unit 1140 and supplies the magnetic field measurement data B to the signal space separation unit 1160.

The basis vector storage unit 1150 pre-stores the basis vectors required for the signal space separation unit 1160 to perform signal separation of the magnetic field measurement data B, and supplies these to the signal space separation unit 1160. Such basis vectors may be calculated from the position and magnetic sensitivity of each magnetic sensor 520.

The signal space separation unit 1160 acquires measurement data B from multiple different combinations of magnetic sensor cells 220 of the magnetic sensor array 210 from the magnetic field measurement data B stored in the data storage unit 1155 for each combination. The signal space separation unit 1160 extracts first internal space data from the spatial distribution of the magnetic field indicated by the measurement data B from some of the magnetic sensor cells 220 of the magnetic sensor array 210, and extracts second internal space data from the spatial distribution of the magnetic field indicated by the measurement data from multiple magnetic sensor cells 220 including another portion of the magnetic sensor cells 220 of the magnetic sensor array 210. The signal space separation unit 1160 extracts third internal space data from the spatial distribution of the magnetic field indicated by the measurement data B from another portion of the magnetic sensor cells 220 different from the portion of the magnetic sensor cells 220 of the magnetic sensor array 210 from which the first internal space data was extracted. The signal space separating unit 1160 may extract second internal space data from the spatial distribution of the magnetic field indicated by the measurement data B from all the magnetic sensor cells 220 of the magnetic sensor array as the target for noise removal. The signal space separating unit 1160 supplies the extracted internal space data to the calculation processing unit 1170.

The calculation processing unit 1170 calculates a projection matrix based on the first internal space data and uses the projection matrix to remove noise from the second internal space data. The calculation processing unit 1170 outputs the internal space data from which noise has been removed. Details of this calculation processing will be described later.

Figure 12 shows a first example configuration of a portion of the magnetic field measurement device 10. The arithmetic processing unit 1170 has a noise removal unit 1172 and a calculation unit 1174. The signal space separation unit 1160 acquires, from the data storage unit 1155, measurement data B from all magnetic sensor cells 220 in the magnetic sensor array 210, measurement data B_A from only the magnetic sensor cells 220 in the even channels (even-numbered) of the magnetic sensor array 210, and measurement data B_B from only the magnetic sensor cells 220 in the odd channels (odd-numbered) of the magnetic sensor array 210. This figure shows a portion of the magnetic sensor array 210 on the X-Y plane, and the magnetic sensor cells 220 in the magnetic sensor array 210 for which corresponding measurement data has been acquired by the signal space separation unit 1160 are indicated by hatching.

The signal space separation unit 1160 performs signal separation of the measurement data B. Based on the position and magnetic sensitivity of each magnetic sensor 520, the signal space separation unit 1160 separates the spatial distribution of the magnetic field indicated by the measurement data, i.e., the spatial distribution of the magnetic field indicated by the magnetic field measurement data B obtained by calibrating the digital measurement data V, into internal space data ^Bin and external space data ^Bout. For example, the signal space separation unit 1160 performs signal separation of the spatial distribution of the magnetic field based on basis vectors calculated from orthonormal functions, the positions of each magnetic sensor 520, and the magnetic sensitivity. In this case, the signal space separation unit 1160 may perform series expansion of the basis vectors to separate the spatial distribution of the magnetic field. More specifically, the signal space separation unit 1160 may calculate the expansion coefficients of the basis vectors using the least squares method. Note that such orthonormal functions may be expressed as spherical harmonic functions. Details of signal separation will be described later. In this way, the signal space separation unit 1160 separates the spatial distribution of the magnetic field indicated by the magnetic field measurement data B into internal space data ^Bin and external space data ^Bout.

The signal space separation unit 1160 may similarly perform signal separation on the measurement data B_A and B_B using information from the corresponding magnetic sensor cells 220, and extract internal space data ^Bin_A and ^Bin_B, respectively. The signal space separation unit 1160 supplies the internal space data ^Bin, which is the data after signal separation, to the noise removal unit 1172, and supplies the internal space data Bin_A and ^Bin_B to the calculation unit 1174.

The calculation unit 1174 calculates a projection matrix based on the internal space data ^Bin_A and the internal space data ^Bin_B using a method such as principal component analysis. For example, the calculation unit 1174 may calculate bases Cin_A and Cin_B from the internal space data ^Bin_A and the internal space data ^Bin_B, respectively, and calculate the projection matrix based on the common part of the bases Cin_A and Cin_B. For example, the calculation unit 1174 calculates the time domain base Cin_A for the internal space data Bin_A and the time domain base Cin_B for the internal space data Bin_A by performing singular value decomposition on the internal space data Bin_A and ^Bin_B, respectively. The internal spatial data Bin_A includes the signal components of the measurement target and the sensor noise of the even channels, and the internal spatial data ^Bin_B includes the signal components of the measurement target and the sensor noise of the odd channels. Therefore, the common part of the basis Cin_A and the basis Cin_B corresponds to the signal components of the measurement target with the basis corresponding to the sensor noise removed.

More specifically, the calculation unit 1174 calculates the common fluctuation component as the common part based on the correlation between the bases Cin_A and Cin_B. In this case, the calculation unit 1174 may, for example, perform singular value decomposition on the covariance matrix (or correlation matrix) of the bases Cin_A and Cin_B, determine the correlation coefficient between the bases Cin_A and Cin_B from the magnitude of the singular values, and extract the bases whose correlation coefficient exceeds a predetermined threshold as the common fluctuation component. The calculation unit 1174 then calculates a projection matrix P from the common part of the bases Cin_A and Cin_B, and supplies this projection matrix P to the noise removal unit 1172.

The noise removal unit 1172 removes noise from the internal space data ^Bin by projecting the internal space data ^Bin onto signal space using the projection matrix P from the calculation unit 1174. The noise removal unit 1172 may output the internal space data ^Binproj from which the noise has been removed. Details of this calculation process will be described later.

Figure 13 shows the flow of signal separation of the spatial distribution of the magnetic field by the magnetic field measurement device 10 according to this embodiment. In this figure, the magnetic field measurement device 10 separates the spatial distribution of the magnetic field indicated by the magnetic field measurement data B into internal space data ^Bin and external space data ^Bout.

In step 1210, the basis vector storage unit 1150 stores basis vectors. As an example, before measuring the magnetic field to be measured, the basis vector storage unit 1150 stores, as basis vectors, signal vectors output by each of the multiple magnetic sensors 520 when the magnetic sensor array 210 detects a magnetic field having a spatial distribution of spherical harmonic functions. That is, the basis vectors may be calculated in advance from the positions and sensitivity vectors of each magnetic sensor 520 in the magnetic sensor array 210 and stored in the basis vector storage unit 1150. That is, the basis vector storage unit 1150 stores, as basis vectors, magnetic field signal vectors obtained by spatially sampling spherical harmonic functions when a predetermined point in space is designated as the coordinate origin. In other words, the basis vector storage unit 1150 calculates in advance, using the positions and sensitivity vectors of each magnetic sensor 520, magnetic field signal vectors that represent the spatial magnetic field in two subspaces (an internal space including the signal source to be measured and an external space) based on a series expansion of a spherical harmonic function, and stores the calculated vectors as basis vectors. Here, a spherical harmonic function is a function obtained by restricting a homogeneous polynomial that is a solution to an n-dimensional Laplace equation to a unit sphere, and has orthonormality on the sphere. Note that, as an example, this figure shows a case where step 1210, in which the basis vector storage unit 1150 stores the basis vectors, is the first step in the flow for signal separation of the spatial distribution of the magnetic field by the magnetic field measurement device 10. However, the basis vector storage unit 1150 may store the basis vectors in advance before the flow for signal separation of the spatial distribution of the magnetic field by the magnetic field measurement device 10. The basis vector storage unit 1150 may also store signal vectors that are predetermined based on simulation results or the like as basis vectors.

Next, in step 1220, the signal space separation unit 1160 acquires the magnetic field measurement data B measured by the magnetic sensor array 210 and calibrated by the calibration calculation unit 1130 from the data storage unit 1155.

Furthermore, in step 1230, the signal space separation unit 1160 obtains from the basis vector storage unit 1150 the signal vectors that were stored as basis vectors by the basis vector storage unit 1150 in step 1210. Note that in this flow, either step 1220 or step 1230 may be performed first.

In step 1240, the signal space separation unit 1160 performs a series expansion on the spatial distribution of the magnetic field indicated by the magnetic field measurement data B acquired in step 1220, using the signal vector acquired in step 1230 as a basis vector. Then, from the vector obtained by the series expansion, the signal space separation unit 1160 performs signal separation on the spatial distribution of the magnetic field into internal space data ^Bin (an estimated value of the magnetic field to be measured) and external space data ^Bout (an estimated value of the disturbance magnetic field). Note that the orthonormal function may be a spherical harmonic function. Furthermore, when performing signal separation, the signal space separation unit 1160 calculates the series expansion coefficients of the basis vectors using the least squares method. This is explained in detail below.

The static magnetic field B(r) can be calculated as the spatial gradient of the potential V(r) using the potential V(r) that satisfies the Laplace equation ΔV(r) = 0, as shown in the following equation: where r is a position vector representing the position from the coordinate origin, Δ is the Laplacian, μ is magnetic permeability, and ∇ is an operator representing a vector differential operation.

The solution of the Laplace equation is generally given by the spherical harmonic functions Yl,m, which are orthonormal functions.
Since the solution is in the form of a series expansion using (θ, φ), the potential V(r) can be expressed as follows: where |r| is the absolute value of the position vector r (distance from the coordinate origin), θ and φ are two argument angles in spherical coordinates, l is the azimuthal quantum number, m is the magnetic quantum number, α and β are multipole moments, and L in and L out are the series numbers for the space in front of and behind the magnetic sensor array 210, respectively, as seen from the subject. The azimuthal quantum number l is a positive integer, and the magnetic quantum number m is an integer between -l and +l. That is, for example, when l is 1, m is -1, 0, and 1, and when l is 2, m is -2, -1, 0, 1, and 2. Note that because there are no single magnetic poles in a magnetic field, the azimuthal quantum number l in (Equation 7) starts from 1, not 0. The first term in (Equation 7) is a term inversely proportional to the distance from the coordinate origin, and represents the potential existing in the space in front of the magnetic sensor array 210 as seen from the subject. The second term in (Equation 7) is a term proportional to the distance from the coordinate origin, and represents the potential existing in the space behind the magnetic sensor array 210 as seen from the object being measured.

Therefore, according to (Equation 6) and (Equation 7), the static magnetic field B(r) can be expressed by the following equation: Here, the first term in (Equation 8) represents a magnetic field source existing in the space in front of the magnetic sensor array 210 as viewed from the subject, i.e., the cardiac magnetic field (magnetic field to be measured) created by the electrical activity of the heart. Also, the second term in (Equation 8) represents a disturbance magnetic field created by a magnetic field source existing in the space behind the magnetic sensor array 210 as viewed from the subject.

When the solution to Laplace's equation is expressed in the form of a series expansion using spherical harmonics, the general solution is an infinite series. However, as long as an SNR (signal-to-noise ratio, i.e., the ratio of the magnetic field signal to be measured to the disturbance magnetic field and sensor noise) sufficient for measuring biomagnetic fields is obtained, it is said that a series of around 10 terms is sufficient in practice. Furthermore, for signal spatial separation in magnetoencephalography, Lin = 8 and Lout = 3 are generally used. Therefore, in this embodiment, Lin and Lout can be values similar to those above. However, the values of Lin and Lout are not limited to these and can be any numerical values sufficient to sufficiently suppress the disturbance magnetic field and calculate only the magnetic field to be measured.

Here, for a total of N sensors used in the magnetic sensor array 210, a1,m and b1,m are defined as follows: where n1, n2, ... nN are sensitivity vectors of each sensor unit. Note that a1,m and b1,m are vectors with dimensions equal to three times the number of magnetic sensor cells 220 (because there are sensor units 300x, y, and z). In other words, they are vectors with dimensions equal to the total number of sensors. As an example, this vector (a1,m, b1,m) is calculated from the output of each magnetic sensor cell 220 using data corrected by the calibration calculation unit 1130. In this way, the values of a1,m and b1,m calculated including the sensitivity in the main axis directions of the sensor unit 300x, y, and z and the sensitivity corrections in the other axis directions (corrected sensitivity vectors) are stored in the basis vector storage unit 1150. In the magnetic field measurement device 10 according to the present embodiment, in which the basis vector storage unit 1150 stores the values of a l,m and b l,m calculated including correction of the magnetic sensitivity (main axis sensitivity, other axis sensitivity), the calibration calculation unit 1130 performs correction on the data acquired by the magnetic field acquisition unit 1120 during operation, thereby making it possible to correct the magnetic sensitivity (main axis sensitivity, other axis sensitivity) of each magnetic sensor cell 220. As another example, when the basis vector storage unit 1150 stores default values of a l,m and b l,m that have not been corrected for the magnetic sensitivity (main axis sensitivity, other axis sensitivity) (using uncorrected sensitivity vectors), the calibration calculation unit 1130 converts the output of each magnetic sensor cell 220 into data in which the magnetic sensitivity has been corrected so as to match the default sensitivity vector of each sensor unit, as shown in (Equation 5), and outputs the data to the data output unit 1140. Thereafter, the signal space separation unit 1160 performs calculations.

Since al,m and bl,m are defined in this way, the sensor output vector Φ output from each magnetic sensor cell 220 at a certain time can be expressed by the following equation.

Furthermore, Sin, Sout, Xin, and Xout are defined as follows. That is, Sin is defined as a Lin x (Lin + 2) column vector in total, where each vector a is arranged in a column when m = -l to l for each l, from l = 1 to l = Lin. Also, Sout is defined as an Lout x (Lout + 2) column vector in total, where each vector b is arranged in a column when m = -l to l for each l, from l = 1 to l = Lout. Also, Xin is defined as a Lin x (Lin + 2) row vector in total, where the transposed vectors are each arranged in a column when m = -l to l for each l _, from l = 1 to l = Lin. Also, Xout is defined as a vector with a total of Lout × (Lout + 2) rows, obtained by transposing the vectors in which each multipole moment β _l,m is arranged in order in a column from l = 1 to l = Lin, where m = an integer from -1 to l for each l.

Then, the sensor output vector Φ can be expressed in the form of an inner product of a matrix S and a column vector X, as shown in the following equation. Here, the matrix S indicates a basis vector, and is obtained, for example, by the signal space separation unit 1160 from the basis vector storage unit 1150 in step 1230. Furthermore, the column vector X indicates a coefficient related to the basis vector.

In step 1240, the signal space separation unit 1160 according to this embodiment determines a column vector X that satisfies Φ = SX by least squares approximation using the following equation based on the model equation of the sensor output vector Φ obtained in (Equation 12). This allows the signal space separation unit 1160 to solve the spatial distribution of the magnetic field in step 1240. That is, the signal space separation unit 1160 becomes able to estimate the spatial distribution of the magnetic field. Specifically, the signal space separation unit 1160 uses the sensor output vector Φ in Equation 12 as magnetic field measurement data B, and can estimate the magnetic field to be measured Bin as internal space data ^Bin = SinXin, and the disturbance magnetic field as external space data ^Bout = SoutXout. In this case, if the magnitude of the external space data ^Bout exceeds a predetermined range, the signal space separation unit 1160 may issue a warning that the magnetic field to be measured Bin cannot be estimated with high accuracy. This allows the magnetic field measuring device 10 to prevent in advance measurement of the magnetic field Bin to be measured in situations such as when the device is broken or when there is a disturbance magnetic field so large that the magnetic field Bin to be measured cannot be estimated with high accuracy. In this case, the signal space separating unit 1160 may determine that the magnitude of the disturbance magnetic field exceeds a predetermined range when the magnitude of any of the components of SoutXout exceeds a predetermined threshold, or may determine that the magnitude of the disturbance magnetic field exceeds a predetermined range when the sum or average of the magnitudes of the components of SoutXout exceeds a predetermined threshold.

The signal space separation unit 1160 may also perform signal separation on the measurement data B_A and B_B in the same manner as in steps 1210-1240, extracting the internal spatial data ^Bin_A and ^Bin_B, respectively. The signal space separation unit 1160 may perform signal separation on each of the measurement data B, B_A, and B_B in parallel, or may perform signal separation sequentially. Here, ^Bin_A and ^Bin_B may be composed only of data corresponding to the sensors of the even and odd channels, respectively, or may be created by complementing the data of excluded channels. For example, for internal spatial data ^Bin_A using even channels, the coefficients Xin and Xout multiplied by the basis vectors are calculated based on data from only the even channels, but by adding the basis vectors of all channels, including the odd channels, to Sin, it is possible to estimate the internal spatial data of all channels.

Then, the signal space separation unit 1160 supplies the internal space data ^Bin, ^Bin_A, and ^Bin_B, which are the data after signal separation in step 1240, to the calculation processing unit 1170.

As a result, the magnetic field measurement device 10 according to this embodiment can separate the spatial distribution of the magnetic field indicated by the magnetic field measurement data measured using the magnetic sensor array 210, which is configured by arranging multiple magnetic sensor cells 220, each having a magnetic sensor 520, to form a surface that covers at least a portion of the object being measured, into internal space data ^Bin, ^Bin_A, and ^Bin_B and external space data ^Bout. Here, since each of the multiple sensor units 300 has a magnetic concentrator, the magnetic sensitivity of the sensor unit 300 can be increased and the spatial sampling points can be clarified, further enhancing compatibility with signal spatial separation technology.

Figure 14 shows an example of a flow in which the magnetic field measuring device 10 according to this embodiment removes residual noise from the internal space data ^Bin after signal space separation.

In step 1310, the calculation processing unit 1170 acquires the internal space data ^Bin, ^Bin_A, and ^Bin_B as time-series data from the signal space separation unit 1160. In other words, the calculation processing unit 1170 acquires the internal space data ^Bin, ^Bin_A, and ^Bin_B extracted by the signal space separation unit 1160 through signal separation from the magnetic field measurement data B at each time t1, t2, ... tn. Here, the internal space data ^Bin, ^Bin_A, and ^Bin_B are each a matrix whose vertical direction corresponds to each magnetic sensor cell 220 from which measurement data was acquired and whose horizontal direction corresponds to each time.

In step 1320, the magnetic field measuring apparatus 10 calculates time-domain bases Cin_A and Cin_B for the internal space data ^Bin_A and ^Bin_B, respectively. For example, the calculation unit 1174 calculates the time-domain bases Cin_A and Cin_B by performing singular value decomposition on the internal space data ^Bin_A and ^Bin_B, respectively. More specifically, the calculation unit 1174 calculates the time-domain bases Cin_A and Cin_B, respectively, using the following equation. That is, the calculation unit 1174 normalizes the transposed matrix of the internal space data ^Bin_A in a certain time domain by dividing it by the Frobenius norm, and then applies singular value decomposition. Then, the calculation unit 1174 calculates the left singular vector Uin_A after singular value decomposition as the time-domain base Cin_A for the internal space data ^Bin_A. Similarly, the calculation unit 1174 divides the transposed matrix of the internal space data ^Bin_B in a certain time domain by the Frobenius norm to normalize it, and then performs singular value decomposition.The calculation unit 1174 then calculates the left singular vector Uin_B after the singular value decomposition as the basis Cin_B in the time domain for the external space data ^Bin_B.

In step 1330, the magnetic field measuring device 10 calculates a basis with a strong correlation between the time domain basis Cin_A in the internal space data ^Bin_A and the time domain basis Cin_B in ^Bin_B, based on techniques such as principal component analysis. For example, the calculation unit 1174 performs singular value decomposition on the covariance matrix (or correlation matrix) of the bases Cin_A and Cin_B, and calculates the coefficient of correlation between the bases Cin_A and Cin_B based on the magnitude of the singular value. Such a correlation coefficient is close to 1 (strong correlation) between bases corresponding to time changes common to the bases Cin_A and Cin_B of magnetocardiogram signals, etc., and close to 0 (uncorrelated) between bases corresponding to random noise, etc.

In step 1340, the magnetic field measuring device 10 extracts, as a common part L, a fluctuation component common to the time domain basis Cin_A in the internal space data ^Bin_A and the time domain basis Cin_B in the internal space data ^Bin_B. For example, the calculation unit 1174 extracts a part where the correlation coefficient exceeds a predetermined threshold as the common part. In this way, the calculation unit 1174 extracts a basis with a strong correlation as a common fluctuation component based on the strength of correlation between the time domain basis Cin_A in the internal space data ^Bin_A and the time domain basis Cin_B in the external space data ^Bin_B. Note that if the predetermined threshold is set too high, the basis of the signal component may be removed, resulting in attenuation of the magnetocardiogram signal, whereas if it is set too low, the basis of noise may also be included. Therefore, it is preferable to set the predetermined threshold to an appropriate value.

In this way, the calculation processing unit 1170 extracts the common part L by performing principal component analysis on the time domain basis Cin_A in the internal space data ^Bin_A and the time domain basis Cin_B in the internal space data ^Bin_B. However, this is not limited to this. Instead of principal component analysis, the calculation processing unit 1170 may extract the common fluctuation component as the common part L using a different method, such as independent component analysis.

In step 1350, the magnetic field measuring apparatus 10 calculates a projection matrix P from the common part L of the bases Cin_A and Cin_B. More specifically, the calculation unit 1174 calculates the projection matrix P for the signal space projection (SSP) operation by the following equation: In the following equation, P represents the projection matrix, and L represents the extracted common part.

In step 1360, the magnetic field measuring apparatus 10 removes noise from the internal space data ^Bin. For example, the arithmetic processing unit 1170 projects the internal space data ^Bin onto the signal space of the measurement target by multiplying the internal space data ^Bin by a projection matrix P. More specifically, the arithmetic processing unit 1170 removes noise using the following equation. In the following equation, ^Binproj represents the internal space data after noise has been removed by SSP.

In the magnetic field measuring device 10 according to this embodiment, the projection matrix P is calculated from the basis L of the signal components from which noise has been removed, so only the signal components of the internal space data ^Bin are projected, enabling the sensor noise components to be efficiently removed.

Figure 15 shows a second configuration example of a portion of the magnetic field measurement device 10 according to this embodiment. This figure shows a portion of the magnetic sensor array 210 in the X-Y plane, with magnetic sensor cells 220 of the magnetic sensor array 210 for which corresponding measurement data has been acquired by the signal space separation unit 1160 indicated by hatching. This figure shows the signal space separation unit 1160 and the calculation processing unit 1170 of the magnetic field measurement device 10 of the second configuration example, and the other configuration may be similar to that of the magnetic field measurement device 10 of Figures 1 to 11. The magnetic field measurement device 10 of the second configuration example does not extract common parts, but calculates two projection matrices and performs two projection calculations. The magnetic field measurement device 10 of the second configuration example may be similar in other configuration and operation to the magnetic field measurement device 10 of the first configuration example. Below, the configuration and operation that differ from the magnetic field measurement device 10 of the first configuration example will be mainly described.

The calculation unit 1174 calculates a projection matrix P1 based on the internal space data ^Bin_A, and calculates a projection matrix P2 based on the internal space data ^Bin_B. For example, similar to step 1320, the calculation unit 1174 calculates a time-domain basis Cin_A for the internal space data ^Bin_A by singular value decomposition of the internal space data ^Bin_A. Then, the calculation unit 1174 calculates a projection matrix P1 from the basis Cin_A. In the following equation, P1 represents the projection matrix, and L1 represents the basis Cin_A.

Furthermore, the calculation unit 1174 calculates a time domain basis Cin_B of the internal space data ^Bin_B by singular value decomposition of the internal space data ^Bin_B, as in step 1320. Then, the calculation unit 1174 calculates a projection matrix P2 from the basis Cin_B. In the following equation, P2 represents the projection matrix, and L2 represents the basis Cin_B.

The calculation unit 1174 supplies the calculated projection matrices P1 and P2 to the noise removal unit 1172.

The noise removal unit 1172 removes a portion of the noise in the internal spatial data ^Bin using the projection matrix P1, and removes another portion of the noise in the internal spatial data ^Bin using the projection matrix P2. The noise removal unit 1172 multiplies the internal spatial data ^Bin by the projection matrix P1 and projects the internal spatial data ^Bin onto the signal space of the even-numbered channel measurement target, thereby removing the odd-numbered channel sensor noise contained in the internal spatial data ^Bin. Furthermore, the noise removal unit 1172 multiplies the internal spatial data ^Bin by the projection matrix P2 and projects the internal spatial data ^Bin onto the signal space of the odd-numbered channel measurement target, thereby removing the even-numbered channel sensor noise contained in the internal spatial data ^Bin. Note that the noise removal unit 1172 may first apply either the projection matrix P1 or P2 to the internal spatial data ^Bin. In this way, the noise removal unit 1172 can perform noise removal in multiple stages. The noise removal unit 1172 may output the internal spatial data ^Binproj with noise removed. Furthermore, L1 and L2 may use all of the bases Cin_A and Cin_B, or may be calculated by retaining bases corresponding to signals with large singular values when singular value decomposition is performed on the internal spatial data ^Bin_A and the internal spatial data ^Bin_B, and excluding bases corresponding to noise with small singular values. In this case, too, removing bases with small singular values is advantageous for noise reduction, but removing too many will result in attenuation of signal components, so it is desirable to set an appropriate threshold value when performing calculations.

Figure 16 shows a third configuration example of a portion of the magnetic field measurement device 10 according to this embodiment. This figure shows a portion of the magnetic sensor array 210 in the X-Y plane, with magnetic sensor cells 220 of the magnetic sensor array 210 for which corresponding measurement data has been acquired by the signal space separation unit 1160 indicated by hatching. This figure shows the signal space separation unit 1160 and the calculation processing unit 1170 of the magnetic field measurement device 10 of the third configuration example, and other configurations of the third configuration example may be similar to those of the magnetic field measurement device 10 of Figures 1 to 11. The magnetic field measurement device 10 of the third configuration example does not extract common parts, but calculates three projection matrices and performs three projection calculations. The magnetic field measurement device 10 of the third configuration example may be similar in other configurations and operations to the magnetic field measurement device 10 of the first configuration example. Below, the configurations and operations that differ from those of the magnetic field measurement device 10 of the first configuration example will be mainly described.

The signal space separation unit 1160 acquires from the data storage unit 1155 four different combinations of magnetic field measurement data measured by the magnetic sensor array 210 and calibrated by the calibration calculation unit 1130. For example, the signal space separation unit 1160 acquires from the data storage unit 1155 measurement data B from all magnetic sensor cells 220 in the magnetic sensor array 210, and measurement data B_A, B_B, and B_C from magnetic sensor cells 220 in the magnetic sensor array 210, skipping one different column every three columns. Specifically, the signal space separation unit 1160 may acquire measurement data B_A from the magnetic sensor cells 220 in the second, third, fifth, sixth, eighth, and so on columns among the magnetic sensor cells 220 in the magnetic sensor array 210, acquire measurement data B_B from the magnetic sensor cells 220 in the first, third, fourth, sixth, seventh, and so on columns, and acquire measurement data B_C from the magnetic sensor cells 220 in the first, second, fourth, fifth, seventh, eighth, and so on columns. The signal space separation unit 1160 performs signal separation on the measurement data B, B_A, B_B, and B_C, respectively, in the same manner as in steps 1210 to 1240 of FIG. 13 , and extracts internal space data ^Bin, ^Bin_A, ^Bin_B, and ^Bin_C. The signal space separation unit 1160 supplies the internal spatial data ^Bin to the noise removal unit 1172, and supplies the internal spatial data ^Bin_A, ^Bin_B, and ^Bin_C to the calculation unit 1174.

As in step 1320 and the magnetic field measuring device 10 of the second configuration example, the calculation unit 1174 calculates a projection matrix P1 based on the internal space data ^Bin_A, and calculates a projection matrix P2 based on the internal space data ^Bin_B. Furthermore, as in step 1320, the calculation unit 1174 calculates a time domain basis Cin_C for the internal space data ^Bin_C by singular value decomposition of the internal space data ^Bin_C. Then, the calculation unit 1174 calculates a projection matrix P3 from the basis Cin_C. Note that in the following equation, P3 represents the projection matrix, and L3 represents the basis Cin_C.

The calculation unit 1174 supplies the calculated projection matrices P1, P2, and P3 to the noise removal unit 1172.

The noise removal unit 1172 removes noise from the internal space data ^Bin using the projection matrices P1, P2, and P3. For example, the noise removal unit 1172 removes sensor noise from the magnetic sensor cells 220 not used in calculating the projection matrix P1 (i.e., noise from the magnetic sensor cells 220 in the first, fourth, seventh, etc. columns included in the measurement data ^Bin) using a projection calculation of the projection matrix P1. The noise removal unit 1172 removes sensor noise from the magnetic sensor cells 220 not used in calculating the projection matrix P2 (i.e., noise from the magnetic sensor cells 220 in the second, fifth, eighth, etc. columns included in the measurement data ^Bin) using a projection calculation of the projection matrix P2. The noise removal unit 1172 removes sensor noise from the magnetic sensor cells 220 not used in calculating the projection matrix P3 from the internal space data ^Bin (i.e., noise from the magnetic sensor cells 220 in the third, sixth, eighth, etc. columns included in the measurement data ^Bin) through a projection calculation of the projection matrix P3. The noise removal unit 1172 can remove noise by multiplying the internal space data ^Bin by the projection matrices P1, P2, and P3, respectively. Note that the order of the projection calculation of the projection matrices P1, P2, and P3 on the internal space data ^Bin is not limited. The noise removal unit 1172 may output the internal space data ^Binproj from which noise has been removed.

Figure 17 shows a fourth configuration example of a portion of the magnetic field measurement device 10 according to this embodiment. This figure shows a portion of the magnetic sensor array 210 in the X-Y plane, with the magnetic sensor cells 220 of the magnetic sensor array 210 for which corresponding measurement data has been acquired by the signal space separation unit 1160 indicated by hatching. This figure shows the signal space separation unit 1160 and the calculation processing unit 1170 of the magnetic field measurement device 10 of the fourth configuration example, and other configurations of the fourth configuration example may be similar to those of the magnetic field measurement device 10 of Figures 1 to 11. The magnetic field measurement device 10 of the fourth configuration example does not extract common parts, but removes noise in the internal space data extracted from the measurement data of some of the magnetic sensor cells 220 of the magnetic sensor array 210 using a projection matrix calculated from the measurement data of other magnetic sensor cells 220 of the magnetic sensor array 210. The magnetic field measurement device 10 of the fourth configuration example may be similar in other configuration and operation to the magnetic field measurement device 10 of the first configuration example. Below, we will mainly describe the configuration and operation that differ from the magnetic field measurement device 10 of the first configuration example.

The signal space separation unit 1160 acquires measurement data B_A from only the magnetic sensor cells 220 in the even channels (even-numbered) of the magnetic sensor array 210, and measurement data B_B from only the magnetic sensor cells 220 in the odd channels (odd-numbered) of the magnetic sensor array 210, from the data storage unit 1155. The signal space separation unit 1160 may extract internal spatial data ^Bin_A and ^Bin_B from the measurement data B_A and B_B, respectively, in the same manner as steps 1210-1240 of FIG. 13. The signal space separation unit 1160 supplies the internal spatial data ^Bin_A to the noise removal unit 1172 and supplies the internal spatial data ^Bin_B to the calculation unit 1174.

The calculation unit 1174 calculates the projection matrix P2 based on the internal space data ^Bin_B, similar to step 1320 and the magnetic field measuring device 10 of the second configuration example. The calculation unit 1174 supplies the projection matrix P2 to the noise removal unit 1172.

The noise removal unit 1172 removes noise from the internal space data ^Bin_A using the projection matrix P2. For example, the noise removal unit 1172 removes sensor noise from the magnetic sensor cells 220 of the even channels that are not used in calculating the projection matrix P2 from the internal space data ^Bin_A through a projection operation using the projection matrix P2. The noise removal unit 1172 can remove noise by multiplying the internal space data ^Bin_A by the projection matrix P2. The noise removal unit 1172 may output the internal space data ^Binproj from which the noise has been removed.

Figure 18 shows a fifth configuration example of a portion of the magnetic field measurement device 10 according to this embodiment. This figure shows a portion of the magnetic sensor array 210 in the X-Y plane, with magnetic sensor cells 220 of the magnetic sensor array 210 for which corresponding measurement data has been acquired by the signal space separation unit 1160 indicated by hatching. This figure shows the signal space separation unit 1160 and the calculation processing unit 1170 of the magnetic field measurement device 10 of the fifth configuration example, and other configurations of the fifth configuration example may be similar to those of the magnetic field measurement device 10 of Figures 1 to 11. The magnetic field measurement device 10 of the fifth configuration example includes an averaging processing unit 1165 between the signal space separation unit 1160 and the noise removal unit 1172 and calculation unit 1174, and performs noise removal using the average value of the internal spatial data. Other than the averaging processing unit 1165, the magnetic field measurement device 10 of the fifth configuration example may be similar in configuration and operation to the magnetic field measurement device 10 of the first configuration example. Below, configurations and operations that differ from the magnetic field measurement device 10 of the first configuration example will be mainly described.

The averaging processing unit 1165 calculates the average value of multiple pieces of data for the internal spatial data ^Bin, ^Bin_A, and ^Bin_B. The averaging processing unit 1165 may calculate the average value by taking a moving average of multiple pieces of internal spatial data that are continuous on the time axis and supplied from the signal spatial separation unit 1160, and supply the average value to the noise reduction unit 1172. Alternatively, the averaging processing unit 1165 may divide the internal spatial data ^Bin, ^Bin_A, and ^B that are continuous on the time axis into specific intervals and perform an arithmetic average of each piece of data. For example, the data may be divided into intervals of 1,000 samples each, such as time t1 to t1000, t1001 to t2000, etc., and averaged as the average of the data between times t1 and t1001, the average of the data between t2 and t1002, ..., the average of the data between t1000 and t2000, and supplied to the noise reduction unit 1172. The averaging processing unit 1165 can further suppress noise by calculating the average value of multiple pieces of data for each of the internal space data ^Bin, ^Bin_A, and ^Bin_B.

Similar to steps 1310-1360 in FIG. 14, the calculation unit 1174 calculates a projection matrix P based on the internal space data average values ^Bin_A and ^Bin_B, and the noise removal unit 1172 removes noise from the internal space data average value ^Bin using the projection matrix P. The noise removal unit 1172 may output the internal space data ^Binproj from which the noise has been removed. The averaging processing unit 1165 may also be provided subsequent to the noise removal unit 1172.

Figure 19 shows a sixth example configuration of a portion of the magnetic field measurement device 10 according to this embodiment. This figure shows a portion of the magnetic sensor array 210 in the X-Y plane, with magnetic sensor cells 220 of the magnetic sensor array 210 for which corresponding measurement data has been acquired by the signal space separation unit 1160 indicated by hatching. This figure shows the signal space separation unit 1160 and the calculation processing unit 1170 of the magnetic field measurement device 10 of the sixth example configuration, and other configurations of the sixth example configuration may be similar to those of the magnetic field measurement device 10 of Figures 1 to 11. The magnetic field measurement device 10 of the sixth example configuration includes a subtraction unit 1176 between the signal space separation unit 1160 and the calculation unit 1174, and calculates the projection matrix P using the subtraction results from the signal space separation unit 1160. The magnetic field measurement device 10 of the sixth example configuration may be similar in other configuration and operation to the magnetic field measurement device 10 of the first example configuration. Below, the configuration and operation that differ from those of the magnetic field measurement device 10 of the first example configuration will be mainly described.

The subtraction unit 1176 calculates the difference ^Bin_A-B between the internal spatial data ^Bin_A and internal spatial data ^Bin_B supplied from the signal space separation unit 1160. The internal spatial data ^Bin_A includes the signal component of the measurement target and sensor noise of the even channel, while the internal spatial data ^Bin_B includes the signal component of the measurement target and sensor noise of the odd channel. Therefore, the difference ^Bin_A-B is data indicating sensor noise, with the signal component of the measurement target removed. The subtraction unit 1176 supplies this difference ^Bin_A-B to the calculation unit 1174.

The calculation unit 1174 calculates the basis Cin_A based on the difference ^Bin_A-B from the subtraction unit 1176. Similar to steps 1310-1320 in FIG. 14, the calculation unit 1174 calculates the time-domain basis Cin_A-B for the difference ^Bin_A-B by singular value decomposition of the difference ^Bin_A-B. Here, the basis Cin_A-B is a basis vector corresponding to the noise component.

The calculation unit 1174 calculates the projection matrix P from the basis Cin_A-B, similar to step 1350 in FIG. 14. In (Equation 16), the projection matrix P can be calculated as L = Cin_A-B. The calculation unit 1174 supplies the projection matrix P to the noise removal unit 1172.

The noise removal unit 1172 removes noise from the internal space data ^Bin using the projection matrix P. Here, since the projection matrix P is calculated from noise component data, a projection operation is performed using the projection matrix P as shown in the following equation to project the internal space data ^Bin onto the signal space of the measurement target, which is orthogonal to the noise space. In the following equation, I represents a unit matrix. The noise removal unit 1172 may output the internal space data ^Binproj from which noise has been removed.

Figure 20 shows the measurement results of the internal spatial data ^Bin before noise removal in the magnetic field measurement device 10 of the sixth configuration example according to this embodiment. The measurement was performed using 300 sensors, sampling 4,000 points at 1-millisecond intervals for a simulated magnetic field signal from a phantom coil, and calculating the internal spatial data ^Bin through signal spatial separation. In this figure, the horizontal axis represents time, and the vertical axis represents the magnitude of the internal spatial data ^Bin before noise removal, showing data extracted from some sensors. As shown in this figure, it can be seen that the internal spatial data ^Bin contains large sensor-derived noise (indicated by arrows in this figure) superimposed on the magnetic field signal (R-wave). This large noise is called Barkhausen jump noise, and occurs when domain walls are trapped by impurities in magnetoresistive elements such as GMR and TMR.

Figure 21 shows the measurement results of the internal space data ^Binproj after noise removal in the magnetic field measurement device 10 of the sixth configuration example according to this embodiment. The internal space data ^Bin, ^Bin_A, and ^Bin_B were calculated based on the data of the even and odd channels, respectively. Furthermore, the data of the excluded channels of ^Bin_A and ^Bin_B were complemented and output by adding the basis vectors of the excluded channels during calculation, as described above. The time-domain basis Cin_A-B of ^Bin_A-B was calculated by extracting the top five (300 total) singular value-decomposition bases of ^Bin_A-B in order of largest singular value, and then calculating the projection matrix P. In this figure, the horizontal axis represents time, and the vertical axis represents the magnitude of the internal space data ^Binproj after noise removal. As shown in this figure, it can be seen that noise due to Barkhausen jumps was successfully removed by the projection operation. Similar noise removal effects were also confirmed for magnetic field measurement devices 10 with other configuration examples related to this embodiment.

As mentioned above, when performing signal space separation processing, noise originating from the sensor remains in the internal space data ^Bin extracted by signal space separation. To remove such noise, one possible method is to calculate a projection matrix from the internal space data ^Bin extracted by signal space separation using singular value decomposition, and then apply this projection matrix to the internal space data ^Bin. However, because the sensor noise remaining in the internal space data ^Bin is in the same signal space as the internal space data ^Bin, it is not possible to effectively separate the noise components using a projection operation.

Another possible method is to calculate a projection matrix from measurement data B before signal space separation using singular value decomposition, apply this projection matrix to data B, and then perform signal space separation on measurement data B. However, the data before signal space separation contains a much larger environmental magnetic field (microtesla) while the biomagnetic signal and sensor noise are tiny (picotesla). Therefore, even if you try to separate the bases of the signal and sensor noise using singular value decomposition, it is not possible to accurately calculate anything other than the bases of the environmental magnetic field. Therefore, it is difficult to separate and calculate the bases of the signal and sensor noise and then remove the sensor noise using a projection operation.

In contrast, according to the magnetic field measuring device 10 of this embodiment, multiple data sets are acquired from different combinations of magnetic sensor cells 220 in the magnetic sensor array 210, and signal space separation is performed. Then, a projection matrix is calculated from one data set, and other data sets are subjected to noise removal. This makes it possible to separate noise contained in the signal from noise contained in the projection matrix and perform noise removal, thereby further improving sensor noise removal performance. The magnetic field measuring device 10 may use the internal space data ^Binproj from which noise has been removed to display various displays related to magnetocardiography on a display or the like.

Note that the magnetic field measurement device 10 of the first to fifth configuration examples, which calculates a projection matrix from signal components, is more suitable for removing noise smaller than the 1/f noise of the sensor, while the magnetic field measurement device 10 of the sixth configuration example, which calculates a projection matrix from noise components, is more suitable for removing larger noise such as Barkhausen jumps. The magnetic field measurement device 10 may also include multiple of the arithmetic processing units 1170 of the first to fifth configuration examples, and may simultaneously perform the calculations of the first to fifth configuration examples, calculate the average value of the multiple types of internal space data ^Binproj obtained, and output the average value as the internal space data ^Binproj. The magnetic sensor cells 220 that acquire the internal space data used by the arithmetic processing unit 1170 may be selected from all of the magnetic sensor cells 220 in the magnetic sensor array 210, or may be selected from a portion of the magnetic sensor cells 220 in the magnetic sensor array 210.

Various embodiments of the present invention may be described with reference to flowcharts and block diagrams, where the blocks may represent (1) stages of a process in which operations are performed or (2) sections of apparatus responsible for performing the operations. Particular stages and sections may be implemented by dedicated circuitry, programmable circuitry supplied with computer-readable instructions stored on a computer-readable medium, and/or a processor supplied with computer-readable instructions stored on a computer-readable medium. Dedicated circuitry may include digital and/or analog hardware circuitry and may include integrated circuits (ICs) and/or discrete circuits. Programmable circuitry may include reconfigurable hardware circuitry including logical AND, logical OR, logical XOR, logical NAND, logical NOR, and other logical operations, flip-flops, registers, memory elements such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), etc.

A computer-readable medium may include any tangible device capable of storing instructions that are executed by a suitable device, such that the computer-readable medium having instructions stored thereon comprises an article of manufacture, including instructions that can be executed to create means for performing the operations specified in the flowchart or block diagram. Examples of computer-readable media may include electronic storage media, magnetic storage media, optical storage media, electromagnetic storage media, semiconductor storage media, etc. More specific examples of computer-readable media may include floppy disks, diskettes, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), electrically erasable programmable read-only memory (EEPROM), static random access memory (SRAM), compact disc read-only memory (CD-ROM), digital versatile disc (DVD), Blu-ray (RTM) disc, memory stick, integrated circuit card, etc.

The computer-readable instructions may include either assembler instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state-setting data, or source or object code written in any combination of one or more programming languages, including object-oriented programming languages such as Smalltalk®, JAVA®, C++, etc., and conventional procedural programming languages such as the "C" programming language or similar programming languages.

The computer-readable instructions may be provided to a processor or programmable circuitry of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, either locally or over a local area network (LAN), a wide area network (WAN) such as the Internet, etc., which executes the computer-readable instructions to create means for performing the operations specified in the flowcharts or block diagrams. Examples of processors include computer processors, processing units, microprocessors, digital signal processors, controllers, microcontrollers, etc.

Figure 22 shows an example of a computer 9900 in which aspects of the present invention may be embodied, in whole or in part. Programs installed on the computer 9900 may cause the computer 9900 to function as or perform operations associated with an apparatus or one or more sections of the apparatus according to embodiments of the present invention, and/or to perform a process or steps of a process according to embodiments of the present invention. Such programs may be executed by the CPU 9912 to cause the computer 9900 to perform specific operations associated with some or all of the blocks in the flowcharts and block diagrams described herein.

The computer 9900 according to this embodiment includes a CPU 9912, RAM 9914, a graphics controller 9916, and a display device 9918, which are interconnected by a host controller 9910. The computer 9900 also includes input/output units such as a communications interface 9922, a hard disk drive 9924, a DVD-ROM drive 9926, and an IC card drive, which are connected to the host controller 9910 via an input/output controller 9920. The computer also includes legacy input/output units such as a ROM 9930 and a keyboard 9942, which are connected to the input/output controller 9920 via an input/output chip 9940.

The CPU 9912 operates according to programs stored in the ROM 9930 and RAM 9914, thereby controlling each unit. The graphics controller 9916 retrieves image data generated by the CPU 9912 into a frame buffer or the like provided in the RAM 9914 or into itself, and causes the image data to be displayed on the display device 9918.

The communication interface 9922 communicates with other electronic devices via a network. The hard disk drive 9924 stores programs and data used by the CPU 9912 in the computer 9900. The DVD-ROM drive 9926 reads programs or data from the DVD-ROM 9901 and provides the programs or data to the hard disk drive 9924 via the RAM 9914. The IC card drive reads programs and data from an IC card and/or writes programs and data to an IC card.

The ROM 9930 stores therein boot programs and the like that are executed by the computer 9900 upon activation, and/or programs that depend on the hardware of the computer 9900. The input/output chip 9940 may also connect various input/output units to the input/output controller 9920 via parallel ports, serial ports, keyboard ports, mouse ports, etc.

The programs are provided on a computer-readable medium such as a DVD-ROM 9901 or an IC card. The programs are read from the computer-readable medium, installed on the hard disk drive 9924, RAM 9914, or ROM 9930, which are also examples of computer-readable media, and executed by the CPU 9912. The information processing described in these programs is read by the computer 9900, resulting in cooperation between the programs and the various types of hardware resources described above. An apparatus or method may be configured by implementing information manipulation or processing in accordance with the use of the computer 9900.

For example, when communication is performed between the computer 9900 and an external device, the CPU 9912 may execute a communication program loaded into the RAM 9914 and instruct the communication interface 9922 to perform communication processing based on the processing described in the communication program. Under the control of the CPU 9912, the communication interface 9922 reads transmission data stored in a transmission buffer processing area provided in the RAM 9914, hard disk drive 9924, DVD-ROM 9901, or a recording medium such as an IC card, and transmits the read transmission data to the network, or writes received data received from the network to a reception buffer processing area or the like provided on the recording medium.

The CPU 9912 may also cause all or necessary portions of a file or database stored on an external recording medium such as a hard disk drive 9924, a DVD-ROM drive 9926 (DVD-ROM 9901), an IC card, etc. to be read into the RAM 9914, and perform various types of processing on the data in the RAM 9914. The CPU 9912 then writes the processed data back to the external recording medium.

Various types of information, such as various types of programs, data, tables, and databases, may be stored on the recording medium and may undergo information processing. The CPU 9912 may perform various types of processing on data read from RAM 9914, including various types of operations, information processing, conditional judgment, conditional branching, unconditional branching, information search/replacement, etc., as described throughout this disclosure and specified by the program's instruction sequence, and write the results back to RAM 9914. The CPU 9912 may also search for information in files, databases, etc. on the recording medium. For example, if multiple entries, each having an attribute value of a first attribute associated with an attribute value of a second attribute, are stored on the recording medium, the CPU 9912 may search for an entry that matches a condition specified by the attribute value of the first attribute from among the multiple entries, read the attribute value of the second attribute stored in the entry, and thereby obtain the attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

The programs or software modules described above may be stored on computer-readable media on or near the computer 9900. Recording media such as a hard disk or RAM provided within a server system connected to a dedicated communications network or the Internet can also be used as computer-readable media, thereby providing the programs to the computer 9900 via the network.

The present invention has been described above using embodiments, but the technical scope of the present invention is not limited to the scope described in the above embodiments. It will be clear to those skilled in the art that various modifications and improvements can be made to the above embodiments. It is clear from the claims that such modifications and improvements can also be included within the technical scope of the present invention.

The order of execution of each process, such as operations, procedures, steps, and stages, in the devices, systems, programs, and methods shown in the claims, specifications, and drawings is not specifically stated as "before," "prior to," or the like, and it should be noted that processes can be performed in any order, unless the output of a previous process is used in a subsequent process. Even if the operational flow in the claims, specifications, and drawings is described using "first," "next," etc. for convenience, this does not mean that it is necessary to perform the processes in that order.

10 Magnetic field measurement device 100 Main body 110 Magnetic sensor unit 120 Head 125 Driving unit 130 Base unit 140 Pole unit 150 Information processing unit 210 Magnetic sensor array 220 Magnetic sensor cell 230 Sensor data collection unit 300 Sensor unit 520 Magnetic sensor 530 Magnetic field generation unit 532 Amplification circuit 534 Coil 540 Output unit 702 Magnetic resistance element 704, 706 Magnetic flux concentrator 1100 Sensor data processing unit 1110 AD converter 1112 Clock generator 1120 Magnetic field acquisition unit 1130 Calibration calculation unit 1140 Data output unit 1150 Basis vector storage unit 1155 Data storage unit 1160 Signal space separation unit 1165 Averaging processing unit 1170 Arithmetic processing unit 1172 Noise removal unit 1174 Calculation unit 1176 Subtraction unit 9900 Computer 9901 DVD-ROM
9910 Host controller 9912 CPU
9914 RAM
9916 Graphics controller 9918 Display device 9920 Input/output controller 9922 Communication interface 9924 Hard disk drive 9926 DVD-ROM drive 9930 ROM
9940 Input/Output Chip 9942 Keyboard

Claims (12)

a magnetic sensor array including a plurality of magnetic sensor cells each having a magnetic sensor;
a magnetic field acquisition unit that acquires measurement data measured by the magnetic sensor array;
a signal space separation unit that extracts first internal spatial data from a spatial distribution of a magnetic field indicated by measurement data from a portion of magnetic sensor cells of the magnetic sensor array, and extracts second internal spatial data from a spatial distribution of a magnetic field indicated by measurement data from a plurality of magnetic sensor cells including another portion of magnetic sensor cells of the magnetic sensor array;
a calculation unit that calculates a projection matrix based on the first internal space data;
a noise removal unit that removes noise from the second internal space data using the projection matrix. The magnetic field measuring device according to claim 1 , wherein the signal spatial separation unit extracts the second internal spatial data from a spatial distribution of the magnetic field indicated by measurement data from all magnetic sensor cells of the magnetic sensor array. the signal space separation unit extracts third internal space data from a spatial distribution of a magnetic field indicated by measurement data from a portion of magnetic sensor cells different from the portion of magnetic sensor cells of the magnetic sensor array from which the first internal space data is extracted;
The magnetic field measuring device according to claim 1 , wherein the calculation unit calculates one or more of the projection matrices based on the first internal space data and the third internal space data. The magnetic field measuring device according to claim 3 , wherein the calculation unit calculates a common part between a basis of the time domain in the first internal space data and a basis of the time domain in the third internal space data, and calculates the projection matrix based on the common part. the calculation unit calculates a first projection matrix based on the first internal space data and calculates a second projection matrix based on the third internal space data;
4. The magnetic field measuring device according to claim 3, wherein the noise removal unit removes a part of the noise in the second internal space data using the first projection matrix and removes another part of the noise in the second internal space data using the second projection matrix. a subtraction unit that calculates a difference between the first internal space data and the third internal space data,
The magnetic field measuring device according to claim 3 , wherein the calculation unit calculates the projection matrix based on the difference. 7. The magnetic field measuring device according to claim 1, wherein the calculation unit calculates a basis of a time domain of the first internal space data by performing singular value decomposition on the first internal space data, and calculates a projection matrix based on the basis. an averaging processing unit that calculates an average value of a plurality of data for the first internal space data and the second internal space data,
the calculation unit calculates the projection matrix based on an average value of the first internal space data;
The magnetic field measuring device according to claim 1 , wherein the noise removal unit removes noise from an average value of the second internal space data by using the projection matrix. The magnetic field measuring device according to claim 1 , wherein the noise removal unit removes noise from the second internal space data by projecting the second internal space data onto a signal space using the projection matrix. The magnetic field measuring device according to claim 1 , wherein the magnetic sensor includes a magnetoresistive element. a magnetic field acquisition stage for acquiring measurement data measured by a magnetic sensor array configured with a plurality of magnetic sensor cells each having a magnetic sensor;
a signal spatial separation step of extracting first internal spatial data from a spatial distribution of a magnetic field indicated by measurement data from a portion of magnetic sensor cells of the magnetic sensor array, and extracting second internal spatial data from a spatial distribution of a magnetic field indicated by measurement data from a plurality of magnetic sensor cells including another portion of magnetic sensor cells of the magnetic sensor array;
a calculation step of calculating a projection matrix based on the first internal space data;
a noise reduction step of reducing noise in the second internal space data using the projection matrix. Computer,
a magnetic field acquisition unit that acquires measurement data measured by a magnetic sensor array configured with a plurality of magnetic sensor cells each having a magnetic sensor;
a signal space separation unit that extracts first internal spatial data from a spatial distribution of a magnetic field indicated by measurement data from a portion of magnetic sensor cells of the magnetic sensor array, and extracts second internal spatial data from a spatial distribution of a magnetic field indicated by measurement data from a plurality of magnetic sensor cells including another portion of magnetic sensor cells of the magnetic sensor array;
a calculation unit that calculates a projection matrix based on the first internal space data;
a magnetic field measurement program that causes the magnetic field measurement program to function as a noise removal unit that removes noise from the second internal space data by using the projection matrix.