JP2009195571A - Noise removal method and biological information measuring apparatus using the method and brain magnetic field measuring apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve a problem wherein it is difficult to remove time-correlated noise from biometric data. <P>SOLUTION: A biological signal measuring part 10 measures a brain magnetic field signal to be generated from a living body. A noise generation time acquiring part acquires information concerning a heartbeat generation time when measuring the brain magnetic field signal. A biometric signal storage part 50 stores the measured brain magnetic field signal together with the information concerning the heartbeat generation time. A noise signal extracting part 20 synchronizes the brain magnetic field signal with the heartbeat generation time and performs averaging, thereby extracting a noise signal by cardiac magnetism. A noise removal processing part 40 acquires the brain magnetic field signal with the noise signal removed therefrom by subtracting the extracted noise signal from the brain magnetic field signal while matching with the heartbeat generation time. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a method for removing noise from biological measurement data, and a biological information measurement device and a cerebral magnetic field measurement device that can use the method.

  The biological measurement data always includes a signal to be acquired and a noise component. For example, when a human brain magnetic field is measured, the measurement data includes an electrocardiogram due to a biological phenomenon, hum noise generated from a commercial power supply, and Gaussian noise in addition to a brain magnetic field signal. Among these noises, electrocardiogram and hum noise are noises having time correlation, and Gaussian noise is noise having no time correlation. Noise removal is indispensable in order to obtain a desired signal from biological measurement data.

  As a conventional noise removal method, measurement is repeated and averaged to obtain a noise component, filtering, or independent component analysis (ICR). There is a method for separating a noise component independent of a signal.

  Biological phenomenon noises such as electrocardiogram and magnetocardiogram having a steep component are not sufficiently attenuated even if addition averaging or filtering is performed. In addition, a band elimination filter (BEF) is used to remove hum noise. However, since BEF attenuates all frequency components of commercial power, the original signal component is also attenuated. Moreover, in BEF, the harmonic component of a commercial power source cannot be attenuated and remains as noise.

  Biological phenomenon noise such as electrocardiogram and magnetocardiogram is independent of the brain magnetic field signal to be measured, so it can be separated in principle by independent component analysis. It is necessary to search repeatedly by a gradient method or the like, and the analysis takes a very long time. In addition, since the application conditions of independent component analysis are severe, it is difficult for actual measurement to satisfy all the conditions, which is not practical.

  The present invention has been made in view of these problems, and an object thereof is to provide a method for efficiently removing noise from biological measurement data, and a biological information measurement device and a cerebral magnetic field measurement device that can use the method. It is in.

  In order to solve the above-described problem, a noise removal method according to an aspect of the present invention includes a step of recording a biological measurement signal measured from a living body together with information related to the generation time of a noise signal having a time correlation, and the living body By referring to the information related to the generation time of the noise signal having the time correlation recorded in the memory with respect to the measurement signal, the average of the time correlation is obtained in synchronization with the generation time of the noise signal. Extracting a noise signal and referring to information on the time of occurrence of the time-correlated noise signal recorded in the memory, and adjusting the extracted noise signal to the living body in accordance with the time of occurrence of the noise signal. Obtaining a biological measurement signal from which the noise signal has been removed by subtracting it from the measurement signal and recording it in a memory.

  “Time-related noise signal generation time information” may be time stamp information of the noise signal generation time, and is data that serves as a basis for determining the noise signal generation time (trigger). Also good. For example, if there is rough waveform data of the noise signal, the noise signal generation time (trigger) can be determined by detecting the rising point when the noise signal value exceeds a predetermined threshold or by detecting the peak point of the noise waveform. Can be sought.

  Another aspect of the present invention is a biological information measuring device. The apparatus includes a measurement unit that measures a signal generated from a living body, an acquisition unit that acquires information on the generation time of a noise signal having a time correlation that is generated when the measurement unit measures the biological measurement signal, A storage unit that stores the biological measurement signal measured by the measurement unit together with information related to the generation time of the noise signal having the time correlation, and the generation time of the noise signal having the time correlation with respect to the biological measurement signal. A noise signal extraction unit that extracts the time-correlated noise signal by taking the averaging average in synchronization, and the extracted noise signal in accordance with the generation time of the time-correlated noise signal A noise removal processing unit that obtains a biological measurement signal from which the noise signal has been removed by subtracting from the biological measurement signal.

  Yet another embodiment of the present invention is a brain magnetic field measurement apparatus. The apparatus includes a measurement unit that measures a brain magnetic field signal generated from a living body, an acquisition unit that acquires information on a heartbeat generation time when the brain magnetic field signal is measured, and a brain magnetic field signal measured by the measurement unit. A storage unit that stores information related to the occurrence time of the heartbeat; and a noise signal extraction unit that extracts a noise signal due to the magnetocardiogram by taking an addition average in synchronization with the occurrence time of the heartbeat with respect to the brain magnetic field signal; And a noise removal processing unit that acquires the brain magnetic field signal from which the noise signal has been removed by subtracting the extracted noise signal from the brain magnetic field signal in accordance with the time of occurrence of the heartbeat.

  It should be noted that any combination of the above-described constituent elements and the expression of the present invention converted between a method, an apparatus, a system, a computer program, a data structure, a recording medium, and the like are also effective as an aspect of the present invention.

  According to the present invention, noise can be efficiently removed from biological measurement data, and the burden on the subject can be greatly reduced.

  An outline of an embodiment of the present invention will be described. In the embodiment, there is provided a method for removing noise having time correlation from biological measurement data. Extract and remove time-correlated noise from biological measurement data. Unlike Gaussian noise, time-correlated noise is often not removed only by averaging. Therefore, in the embodiment, time-correlated noise is synchronized with the noise signal itself, the noise signal is once extracted from the measurement data, and the extracted noise signal is subtracted from the measurement data.

  FIG. 6 shows the autocorrelation of time-correlated noise. The left figure shows the autocorrelation of hum noise with periodicity, and it can be seen that there is a period of 16 ms. The right figure shows the autocorrelation of the electrocardiogram waveform. There is a first peak around 700 ms, and the peak value is smaller than 1.0. The second peak is at about 1400 ms, but the peak value is further reduced than the first peak. This is because the electrocardiogram is not an accurate periodic noise such as hum noise, but fluctuates in time.

  In the biological measurement data, a noise component is mixed with a signal to be measured. As shown in FIG. 2, the biomedical signal is mixed with electrocardiogram due to a biological phenomenon and hum noise generated from a commercial power source as time-correlated noise and Gaussian noise as non-time-correlated noise.

  Biological phenomenon noise having a steep component such as electrocardiogram and magnetocardiogram is not sufficiently attenuated even when filtering is performed. Normally, a band elimination filter (BEF) is used to remove hum noise, but BEF can attenuate the frequency of the commercial power supply, but cannot attenuate its harmonics. As shown in FIG. 68, when BEF is used, all components of 60 Hz that is a power supply frequency are attenuated, and harmonics indicated by * are left.

  In general, when frequency filtering is applied, not only noise components but also frequency components of the original signal are discarded. As shown in FIG. 35, the magnetocardiographic noise includes a frequency component from 0.5 Hz to 40 Hz, which has a spectrum overlapping the brain magnetic field (several Hz to several tens Hz) to be measured. For this reason, it is impossible in principle to selectively remove only the magnetocardiographic component by the frequency filtering technique. Intuitively, removing the magnetocardiogram will also remove the brain magnetic field.

  Therefore, in this method, time information of noise having time correlation (hereinafter referred to as “trigger”) and measurement data are simultaneously recorded at the time of measurement. The measurement data includes a measurement signal and a trigger. When the measurement signal is divided by the width of the trigger period of the noise, and the divided measurement signals are synchronized at the trigger timing and the averaging is performed, only noise having a time correlation is extracted (left in FIG. 8). When the measurement signal is synchronized with the time information of the heartbeat and the addition average is taken, only the magnetocardiogram component is emphasized and extracted. Here, considering that the heartbeat fluctuates in time, a margin is provided before and after the addition average time, and a part of the time range for taking the addition average is partially overlapped. The width over which the time ranges for averaging are overlapped is determined according to the amount by which the heartbeat trigger fluctuates in the time axis direction. In this way, by providing a margin for the time range for taking the averaging, even if the heartbeat trigger timing slightly deviates back and forth, the heartbeat triggers match between the measurement signals divided by the heartbeat trigger cycle. Shifting in the time axis direction can be used for averaging.

  The timing at which the electrocardiogram signal changes sharply is detected as a trigger. As a trigger detection method, a rising point at which the value of the electrocardiogram signal exceeds the threshold may be detected, and this method is used in the embodiment. Another method of detecting the trigger is to detect the peak of the electrocardiogram signal. In any case, if a rough electrocardiogram waveform is obtained, heartbeat time information can be acquired by detecting a steep rising point from the electrocardiogram waveform as a trigger.

  Next, the noise component extracted by the averaging is removed from the measurement signal (right side in FIG. 8). Specifically, noise extracted in accordance with the time of heartbeats (heartbeat trigger) recorded simultaneously is subtracted from the original data. Thereby, the magnetocardiogram noise is removed from the measured signal. At this time, since there is a time margin before and after the averaging, it is possible to cope with noise having temporal fluctuations such as magnetocardiographic noise.

  When this method is applied to hum noise removal, unlike BEF, as shown in FIG. 69, only hum noise including harmonic components can be removed without attenuating the signal component of the power supply frequency.

  As described above, the present method can cope with both the removal of periodic noise such as hum noise and the removal of temporally fluctuating noise (for example, heartbeat noise) due to biological phenomena.

  With this method, time-correlated noise can be removed from the biological measurement data in advance, and the remaining Gaussian noise can be removed by addition averaging, so the average number of additions required for normal noise removal is reduced from half to four minutes. It can be reduced to about 1. Therefore, the burden on the subject can be greatly reduced. For example, the experiment of measuring the magnetoencephalogram by letting the subject listen to the sound once a second has been repeated for about 1000 times in the past, and the experiment took nearly 20 minutes. Since this requires at least half to one-fourth averaging time, the experiment is shortened to about 5-10 minutes.

  It will be described in detail in an embodiment that this method can be applied to hum noise removal and magnetocardiogram noise removal in electroencephalogram measurement and MEG measurement. However, this is only an example, and the noise removal method of the present invention is not limited to brain waves and MEG measurement, and is generally applied to all technical fields aimed at removing time-correlated noise from biological measurement data. Can be applied.

  FIG. 1 is a configuration diagram of a brain magnetic field measurement apparatus 100 according to an embodiment of the present invention. The magnetoencephalogram measurement apparatus 100 is a magnetoencephalogram (MEG) measurement apparatus using a SQUID sensor, which will be described later. Here, a general configuration for measuring the magnetoencephalogram is omitted, and a configuration related to noise removal is described. To do.

  The biological signal measurement unit 10 measures a cerebral magnetic field signal generated from the living body and records it in the biological measurement signal storage unit 50. The noise generation time acquisition unit 30 acquires information on the heartbeat generation time based on the electrocardiogram waveform measured from the same living body and records it in the biological measurement signal storage unit 50. Recording of the cerebral magnetic field signal by the biological signal measurement unit 10 and recording of the heartbeat generation time by the noise generation time acquisition unit 30 are performed simultaneously.

  The noise signal extraction unit 20 extracts and extracts a noise signal due to the magnetocardiogram by taking an addition average in synchronism with the generation time of the heartbeat with respect to the cerebral magnetic field signal recorded in the biological measurement signal storage unit 50. The noise signal is stored in the noise signal storage unit 60.

  The noise removal processing unit 40 subtracts the noise signal stored in the noise signal storage unit 60 from the brain magnetic field signal recorded in the biological measurement signal storage unit 50 in accordance with the time of occurrence of the heartbeat, thereby removing the noise signal. The obtained cerebral magnetic field signal is acquired, and the cerebral magnetic field signal after removal of the noise signal is written back to the biological measurement signal storage unit 50.

  In the measured cerebral magnetic field signal, a magnetocardiographic component generated due to a heartbeat is mixed as noise. This method does not directly measure the magnetocardiogram signal, but uses only the heartbeat occurrence time information that can obtain an electrocardiogram waveform, and synchronizes the measured brain magnetic field signal with the heartbeat time to take the averaging The magnetocardiogram signal can be extracted from the measured brain magnetic field signal. If the extracted magnetocardiogram signal is subtracted from the measured brain magnetic field signal, the magnetocardiogram signal can be removed and only the brain magnetic field signal can be extracted.

  The biological signal to be measured is not limited to the cerebral magnetic field signal. The measurement target may be an electroencephalogram signal, and any signal generated from another living body can be the measurement target. Further, a signal mixed as noise in a biological signal to be measured is not limited to an electrocardiogram signal, and an arbitrary signal having a time correlation can be noise. For example, the same noise removal method can be applied to the case where periodic hum noise generated from an AC power supply of a measurement device is mixed in a biological measurement signal as a noise signal. That is, the hum noise signal can be extracted by acquiring the hum noise signal occurrence time information and averaging the biological measurement signals in synchronization with the occurrence time. By subtracting the extracted hum noise signal from the biological measurement signal, a biological measurement signal without hum noise can be obtained. Note that the generation time of the hum noise generated from the AC power supply can be obtained by detecting the timing at which the AC cuts zero from the minus direction to the plus direction. The generation time information may be recorded simultaneously with the measurement of the biological signal.

  In the following, embodiments of the present invention will be described in detail in line with the inventor's research regarding “effects of magnetocardiographic components on brain magnetic field measurement data”. Details of the operation of each component of the brain magnetic field measuring apparatus 100 will be understood with reference to the following research contents.

  In this study, we focused on the magnetocardiographic component of biological phenomenon noise in MEG measurement. It has been found that the influence of the magnetocardiographic component is different depending on the position of the focused ECD and that the magnetocardiographic component is difficult to attenuate only by the filter processing with a cutoff frequency of about 3 [Hz]. Therefore, we have developed a system that performs averaging and extracts magnetocardiographic components and removes them. As a result of hearing and visual MEG experiments and evaluation of this system, it was shown to have good performance.

The following explanation is presented.
In Chapter 1, the overview of various time-correlated noises measured during the MEG experiment was explained, and a problem was set regarding the removal of the magnetocardiographic component. In addition, the method of removing the magnetocardiogram component proposed in this method is explained.

  Chapter 2 explains the heart rate measurement system that can be used with MEG. The removal method of the magnetocardiographic component performed in this research must record the heartbeat simultaneously during MEG measurement. Therefore, development of a heart rate measuring device that does not cause a problem during MEG measurement and evaluation with an existing measuring device were performed.

  Chapter 3 explains the results of extracting magnetocardiographic components by MEG. The characteristics of the extracted magnetocardiographic components were clarified, and it was shown that it was difficult to attenuate by the conventional filter processing. In addition, using artificial data, it was suggested that the influence of the magnetocardiogram varies depending on the estimated site.

  Chapter 4 describes a program that can remove the magnetocardiogram using this method, which is not possible with existing magnetoencephalogram analysis software.

  In Chapter 5, an experiment that can measure the cerebral magnetic field responses of the primary auditory cortex and primary visual cortex was designed and evaluated. As a result, it was shown that the magnetocardiographic component is one factor of the fluctuation of the baseline by the addition average, and the effectiveness of this method was able to be shown.

1. Introduction MEG measurement is a fight against noise [1]. MEG (Magneto Encephalo Graphy) is a device that measures a magnetic field generated from a weak current generated in the cerebral cortex using a number of SQUID (Superconducting Quantum Interference Devices) sensors. Since the induced brain magnetic field is very weak, measurement is performed by attenuating external magnetic noise with a magnetic shield. However, this magnetic damping effect does not work against magnetic noise generated from a living body. Therefore, the measured brain magnetic field data includes many signals other than those intended (FIG. 2). Usually, the S / N ratio (Signal / Noise Ratio) is improved with respect to the measured brain magnetic field data by using a method such as addition averaging or signal processing.

  There are two types of measured noise: those caused by biological phenomena and those caused by the measurement environment (FIGS. 3 and 4).

  FIG. 2 is a diagram for explaining signals measured by the MEG. As a signal to be measured, a magnetic field signal in which a noise component, a magnetocardiogram component (noise due to a biological phenomenon), a hum noise (noise due to an environment), and the like are linearly mixed is measured.

  FIG. 3 is a diagram for explaining noise due to a biological phenomenon measured by MEG. The magnetic field waveform and isomagnetic field diagram which piled up the magnetic field signal measured from 160 SQUID sensors. In the isomagnetic diagram, green indicates the magnetic field sink (Sink), and red indicates the magnetic field source (Source). The isomagnetic field diagram is for the time of the yellow line. Since the eye movement of a) is close to the SQUID sensor, a large magnetic field change is observed. The respiration of e) is a very slow change. The isomagnetic field diagram is also very different.

  FIG. 4 is a diagram for explaining noise caused by the environment measured by the MEG. The magnetic field signal and power spectrum distribution which were measured from the SQUID sensor are shown. Peaks are seen at frequencies of multiples of 60 [Hz] of hum noise of 60 [Hz], 120 [Hz], and 180 [Hz]. This peak is thought to be generated from an AC power supply of 60 [Hz].

  There are noises that do not correlate with time and those that correlate with time (FIGS. 5 and 6). If signals other than the induced magnetic field fluctuate randomly with respect to time (if the signal is uncorrelated with time), the signal is attenuated to an amplitude of 1 / N by N times of averaging. On the other hand, since a noise signal correlated with time (noise having time correlation) does not fluctuate randomly, it is synchronized with the addition average and may remain without being canceled. In addition, since the biological phenomenon noise signal is usually a large signal that is about ten times as large as the brain magnetic field, even if the average number of additions is about 10, it has the same effect as the brain magnetic field. Therefore, time-correlated noise must be removed from the averaging as much as possible. On the other hand, there is a possibility that a noise component can be extracted by utilizing the time correlation.

  FIG. 5 is a diagram for explaining autocorrelation of random noise. Noise caused by measurement has no correlation with time. For this reason, if the averaging is performed, it is attenuated.

  FIG. 6 is a diagram for explaining time-correlated noise. The left figure shows the autocorrelation of periodic hum noise. As indicated by red circles, peaks occur periodically at regular intervals. The right figure shows electrocardiogram autocorrelation. The portion of the RR interval is indicated by a red line. Heart beats occur at regular intervals, but fluctuate in time.

  When the MEG experiment is performed, the movement of the subject is suppressed as shown in FIG. Suppresses biological phenomenon noise caused by movement of the head. Also, devise experimental tasks that make it difficult for eye movements to occur when they are watched. Nevertheless, biological phenomenon noise is generated from breathing and the heart. However, the noise of biological phenomena such as respiration and eye movements can be excluded from the addition averaging because the waveform changes several times as much as the magnetic field data measured normally. Further, the hum noise generated from the power source can be greatly attenuated by the band elimination filter (BEF) at the time of measurement and the filter processing after the measurement. However, the magnetocardiographic component continues to occur at regular intervals and is difficult to exclude.

  FIG. 7 is a diagram for explaining the scenery during the MEG experiment. The dewar and the subject's head are filled with an elastic material. Thereby, the lateral movement of the head is suppressed. In addition, the chin rest [4] produced by Nomura is used to suppress the head's vertical movement. There is a report that the marker coil deviation before using the chin rest is 9.72 [mm] at the maximum, and the marker coil deviation after using the chin rest is 0.40 [mm] at the maximum. is there.

  Noise removal by signal processing techniques such as PCA and ICA has also been proposed [5] [6]. Independent component analysis (ICA) is a method of separating a plurality of observed signals into source signals based on statistical independence. Since noise and induced magnetic field are uncorrelated, they can be separated from each other. However, since MEG is measured using a large number of SQUID sensors, it becomes multidimensional data, and analysis takes a lot of time [7].

  The purpose of this study is to evaluate the performance of this system by focusing on the magnetocardiographic component of the biological phenomenon noise, and developing the system to remove the magnetocardiographic component and investigating the effect on the equivalent current dipole estimation. That is.

  The removal method of the magnetocardiographic component used in this research is a method of extracting the magnetocardiographic component by using heartbeat as a trigger and extracting the magnetocardiographic component using the addition average. Since the magnetocardiogram component is generated at a constant rhythm, we thought that recording the heartbeat at the same time would be effective in removing only the magnetocardiogram component.

1.1. Subtraction method by addition averaging The removal method of the magnetocardiographic component proposed in this study records the movement of the heart at the same time as the brain magnetic field measurement. Using this as a trigger, addition averaging is performed, a magnetocardiographic component generated by the heartbeat is extracted, and subtraction is performed off-line from measurement data (FIG. 8).

  Generally, when exercise load is applied, the QRS time and QT interval of the electrocardiogram waveform are shortened [8]. It is said that when a load such as copying or a calculation task is given, the RR interval changes and the mental load is reflected [9]. In this study, we thought that the extracted magnetocardiogram waveform was almost similar because it was almost resting during the experiment. The fluctuation of the RR interval can be dealt with by obtaining a magnetocardiogram waveform for one period and removing it from the heartbeat trigger.

  FIG. 8 is a diagram for explaining a technique for removing the magnetocardiographic component proposed by this research. First, based on the heartbeat time information, addition averaging is performed on the signal measured by the SQUID. The signal obtained by averaging is considered to be a magnetic field component derived from the magnetocardiogram. Thereafter, removal is performed with the extracted magnetocardiographic component using the time of the heartbeat used for the averaging.

  In order to realize this method, a heartbeat measuring device that does not affect MEG measurement and software for performing this method are required.

1.2. Baseline correction Baseline correction is applied to the data that has been averaged. Baseline correction removes any sustained neural activity or baseline fluctuations. FIG. 9 shows an operation performed by the baseline correction.

  FIG. 9 is a diagram for explaining baseline correction. The average value of each CH in the specified time interval is obtained, and the original data is subtracted for each sample. 100 [ms] from before the stimulation trigger is performed in the entire interval. This operation directly changes the measurement data.

  In order to change the measurement data directly, there is a possibility that the signal that is originally obtained may be erased depending on the interval for obtaining the average. Also, the criteria for which part to use are not clearly determined.

1.3. About electrocardiogram and magnetocardiogram The heart is an organ that functions as a pump that circulates blood throughout the body by contracting the heart muscle. The heart has a ventricle and an atrium separated by a septum and a valve membrane on the left and right. In addition to this, it provides a tissue that issues and transmits electrical commands for controlling contraction of the myocardium constituting the atria and ventricles (FIG. 10).

  FIG. 10 is a diagram for explaining the names and features of each part of the heart. The sinus node is the site that creates a constant rhythm of myocardial contraction. Propagation due to depolarization of the left and right atrial myocardium stops at the tricuspid valve and the mitral valve, and propagation to the left and right ventricle is transmitted only to the atrioventricular node. The stimulation conduction system of the ventricle is Purkinje fiber deletion from the His bundle to the left and right ventricular myocardium. It is transmitted very quickly during this time.

  Cardiomyocytes remain negative at rest and are depolarized by excitation. At this time, Na + flows into the cell. The cells actively expel this inflowed Na +. When 3 Na + are discharged, 2 K + flows. At this time, an electric current is generated outward. Ca2 + is required when the heart muscle contracts. Immediately after the depolarization of the myocardium, since there is much Na +, Na + is discharged and Ca2 + is taken in (an outward current is generated). When Ca 2+ increases, Ca 2+ is discharged and Na + is taken in (inward current is generated). After the myocardium contracts, repolarization occurs in preparation for the next depolarization. At this time, the intracellular K + is excreted to return to the original resting membrane potential [12].

  Thus, cardiomyocytes are depolarized by Na +, contracted by Ca2 +, and repolarized by K +. Explain myocardial excitation conduction (propagation of depolarization). At rest, the cell remains negative, but depolarization is propagated by adjacent cardiomyocytes. At this time, myocardial cells become positive. A current dipole is generated in the wave front of excitatory conduction [11].

  A mechanism for generating a potential difference on the body surface will be described. When there is an electric current associated with myocardial excitement and contraction, a feedback current is generated in the surrounding organ having resistance. For this purpose, the potential difference is measured at an arbitrary point (FIG. 2 of reference [10]).

  The electrocardiogram waveform and the contraction timing of cardiomyocytes will be described. The measured electrocardiographic waveform represents the contraction timing of the myocardium. As shown in FIG. 3 of the reference [10], the electrocardiogram shows the electrical state of the heart and appears in the order of P wave, Q wave, R wave, S wave, T wave, and U wave. During PQ, left and right atrial contraction occurs. Excitatory conduction from the His bundle to Purkinje fiber extinction occurs during QRS. In other words, it represents the beginning and end of depolarization of the ventricular muscle. ST is repolarized due to subsequent contraction.

  The magnetocardiogram current source is thought to be a current dipole accompanying excitatory conduction. It is thought that the propagation of myocardial excitement is measured as a magnetocardiogram.

2. About heart rate measurement system that can be used in MEG The heart is an organ that carries blood to the whole body by contraction of the myocardium. This change can be measured by a volume change (pulse wave) in the blood vessel and a potential change (electrocardiogram) on the body surface caused by muscle contraction of the myocardium. In order to remove the magnetocardiogram in this study, a signal synchronized with the heartbeat must be recorded simultaneously during MEG measurement. Therefore, we examined a heart rate measurement device that does not cause problems in noise and experimental tasks even if it is measured simultaneously with MEG measurement.

2.1. Evaluation of system for measuring heart rate In MEG experiment, there is an electrical stimulation device as a stimulation device for somatosensory sensation. Depending on the experimental task, the subject may perform a button press operation. It is necessary to prevent the system that measures the heart rate developed in this case from being affected by such stimulation and movement (Fig. 11).

  FIG. 11 is a diagram illustrating an example of an experimental task. Some somatic stimulation devices use electrical stimulation equipment. Depending on the experimental task, a button press operation may be performed. During these experiments, the heart rate measurement system should not be affected.

2.2. Heartbeat measurement using pulse waves Pulse waves are the transmission of arterial pressure waves generated by heartbeat [13]. As a method for measuring this transmission, there are a method of measuring fluctuations using an air dynamic pressure sensor and a method of using the fact that the absorption amount of hemoglobin in infrared light is different.

  As a device for measuring a pulse wave using an air dynamic pressure sensor, a pulse wave observation device (FIG. 12) manufactured by MI Lab Inc. was used. As the pulse wave observation device using infrared rays, a heart rate measuring device constituted by infrared LEDs and optical sensors (photodiodes) was manufactured (FIG. 13 is a circuit diagram, and FIG. 14 is a manufactured device).

  FIG. 12 is a diagram for explaining a pulse wave observation device (captured by a change in air pressure). The pulse wave observation device captures slight air pressure fluctuations in the air bag, converts it into voltage, and outputs it. When in use, use the air bag in close contact with the subject's finger with surgical tape. At this time, it is necessary to consider the delay in the tube's air transmission and the delay in blood flow from the heart to the finger.

  FIG. 13 is a circuit diagram of a pulse wave observation device (device using infrared rays). Changes in blood flow can be measured by placing a finger between the infrared LED and the photodiode.

  FIG. 14 is a diagram for explaining the manufactured pulse wave observation device. Infrared light cannot be seen by human eyes. In the left figure, the infrared LED portion is shining. This is recorded because the optical sensor of the digital camera is sensitive to infrared light. On the right is a pulse wave measurement scene. Measure with your finger to hide the infrared LED and photodiode.

2.3. Manufacture of an electrocardiograph The electric field generated by the excitement of myocardial cells almost instantaneously reaches the body surface via surrounding organs. An electrocardiograph is a device that captures this electric field change. For electrocardiograms for clinical purposes, electrodes called standard 12 leads (limb lead and chest lead) are applied [15] (FIG. 15). In this study, electrocardiograms are not used for clinical purposes, so it is only necessary to capture and amplify QRS electric field changes caused by myocardial contraction (FIG. 16). 17 and 18 show the circuit diagram of the produced electrocardiograph and an overview of the apparatus.

  FIG. 15 is a diagram for explaining standard 12-lead electrode attaching positions. The attachment positions of the electrodes for limb guidance and chest guidance are shown. This electrode attachment position is used for clinical purposes.

  FIG. 16 is a diagram for explaining a typical electrocardiogram waveform. When myocardial cells depolarize, the myocardium contracts. P waves represent atrial contraction. The QRS wave indicates the beginning and end of ventricular depolarization, that is, the contraction of the ventricle, and the T wave indicates the end of the contraction of the myocardium. Details are explained in 1.3.

  FIG. 17 is a circuit diagram of the produced electrocardiograph. The AD 620 is an instrumentation amplifier. This is an operational amplifier dedicated to differential amplification with high input impedance. With VR, the gain can be amplified up to 1 to 10,000 times. After applying a 0.5 [Hz] high-pass filter and a 40 [Hz] low-pass filter composed of RC, non-inverting amplification is performed and output.

  FIG. 18 is a diagram for explaining the produced electrocardiograph. The produced electrocardiograph is installed outside the shield room. Extend the cable into the shield room and attach the electrode to the subject.

  As an electrode to be attached to the body of a subject, a disposable electrode Vitrode F-150S manufactured by Nihon Kohden was used. It was confirmed that the electrode was not a magnetic material using a magnet before the experiment. It was found that the commercially available product cannot be used for MEG experiments because stainless steel is used for the electrode lead connecting the disposable electrode and the electrocardiograph. Therefore, I made my own electrode lead without using magnetic material. Copper was used for the conducting wire, and a rubber foot used for electronic work was processed and used for the connecting portion with Vitrode (FIG. 19).

  FIG. 19 is a diagram for explaining the manufactured electrode lead (left) and the Nihon Kohden electrode lead (right). The self-made electrode lead is made so that the copper wire part contacts the disposable electrode. Nihon Kohden's electrode leads (disposable electrode lead wires BC-IXX: XX have different model numbers depending on the color) have a disposable electrode in close contact with a spring structure.

  Apply digital filter to the measured ECG data. This filter processing is for removing baseline fluctuation (˜2 [Hz]) and hum noise (60 [Hz] component) due to the movement of the subject under measurement.

2.4. Heartbeat measurement results of the pulse wave observation device and the manufactured electrocardiograph The results of measuring the heart rate using the pulse wave observation device of FIGS. 12 and 14 are shown (FIGS. 20 and 21). Both subjects of this measurement had an amplitude peak about 3 times in 2 seconds. Since the interval between heartbeats at rest is 0.8 to 1.7 times per second [14], the normal interval is considered to be measured normally.

  FIG. 20 is a diagram for explaining a result of pulse wave measurement using the air dynamic pressure sensor. When you move your finger, the baseline fluctuates greatly. Moreover, if the air bag is touched, it will be a big change and a pulse wave will not be observed normally.

  FIG. 21 is a diagram for explaining the result of pulse wave measurement using an optical sensor. Basically, you cannot measure unless you keep your finger between the light sensor and the infrared LED. For this reason, the operation of pressing a button during the MEG experiment cannot be performed.

  A 3-50 [Hz] FIR type bandpass filter was applied to the measured electrocardiographic waveform. Thereby, the fluctuation of the baseline of the electrocardiogram waveform and the hum noise caused by the movement of the subject can be removed (FIG. 22). At this time, a filter with linear phase characteristics was used to prevent phase distortion [16].

  FIG. 22 is a diagram for explaining an electrocardiogram waveform before applying a filter and an electrocardiogram waveform after a band-pass filter. The hum noise appears in a fine and regular cycle. Since the frequency of the AC power supply is 60 [Hz], a band pass filter of 3-50 [Hz] was applied so as to attenuate the band.

  The produced electrocardiograph can obtain and amplify the potential difference between the two points (in the preliminary experiment, the insensitive electrode is connected to the electrocardiograph GND for noise reduction). . Therefore, the waveform varies depending on which part the electrode is attached to. Therefore, the electrode was regarded as the apex, and measurement was performed by attaching the electrode so that the heart fits in the triangle (FIGS. 23 and 24).

  FIG. 23 is a diagram for explaining a difference in waveform due to a difference in electrode attachment position in chest guidance. Red is a positive input, blue is a negative input, and black is connected to a dead electrode (electrocardiograph GND).

  FIG. 24 is a diagram for explaining a difference in waveform due to a difference in electrode attachment position in limb guidance. Red is a positive input, blue is a negative input, and black is connected to a dead electrode (electrocardiograph GND). Overall, the amplitude was lower than the chest lead. This is thought to be due to organs and the like becoming resistance components due to the distance from the heart.

  FIG. 25 shows the result when the button is operated with the electrical stimulator used in the MEG experiment. There was no change in the ECG waveform.

  FIG. 25 is a diagram for explaining an electrocardiogram waveform when an electrical stimulation or button pressing operation is performed during electrocardiogram measurement. The red line represents the start timing of each stimulus. Electrical stimulation outputs 0.5 [mA] to the wrist. The button was pushed with the index finger. In both cases, there was no change in the ECG waveform.

  FIG. 26 shows the result of measuring the heart rate at the same time using the manufactured pulse wave observation device (FIG. 14) and the electrocardiograph. The peak values of the electrocardiographic R wave and pulse wave had a delay of approximately 200 [ms]. This delay changes because the distance from the heart to the pulse wave measurement site varies depending on the subject.

  FIG. 26 is a diagram for explaining heart rate measurement of a pulse wave observation device using an optical sensor and a manufactured electrocardiograph. The pulse wave was delayed compared to the electrocardiogram. This is because a change in blood flow is captured.

2.5. Discussion In this study, MEG measurements and heartbeats are recorded simultaneously, and only the magnetocardiogram component is extracted by averaging. For this reason, a device for measuring heartbeats stably even during an experimental task is required.
There are problems with pulse wave observation devices that use an air dynamic pressure sensor and those that use infrared light. Depending on the experimental task, the button may be pressed. From FIG. 20, the pulse wave observation device using the air dynamic pressure sensor cannot be used. A pulse wave observation device using an optical sensor cannot bring the sensor into a shield room. This is because the change in the current of the optical sensor is mixed into the SQUID sensor. In addition, the use of an optical fiber was also considered, but the attenuation of the infrared wavelength was large, and the pulse wave component could not be measured.

  The produced electrocardiograph was able to measure heart rate stably even after electrical stimulation and button pressing. In addition, unlike the pulse wave, since the change in the electric field is measured, there is no delay. Therefore, in this study, we decided to extract the magnetocardiographic component using an electrocardiograph as a heart rate measuring device.

3. Extraction of magnetocardiographic component by MEG There are many devices that measure the function using magnetism generated from a living body. A magnetoencephalogram is literally a measurement device that is designed to target brain function. In addition, there are magnetocardiogram measuring devices for examining the function of the heart. In this research, since there was no report of the magnetocardiogram using the MEG measuring device, it was investigated how the magnetocardiogram is measured by the SQUID sensor.

  The magnetocardiogram measuring device shown in reference [17] is used. A SQUID sensor array is placed directly above the heart to perform measurement. The MEG measuring device arranges the SQUID sensor so as to cover the head.

  The magnetocardiographic component measured using the magnetocardiogram measuring device is the result of superposing the SQUID sensors for 8 ch as shown in the reference [17].

3.1. Experimental Method An electrode was attached to the subject as shown in FIG. Moreover, the output of the electrocardiograph was connected to the trigger channel of the MEG measuring device, and the electrocardiogram was recorded simultaneously. Subjects were 4 males aged 22-23.

  FIG. 27 is a diagram for explaining an electrode attachment position. Inverted input, non-inverted input, and insensitive electrode were pasted so that the heart could be captured in a triangle with the electrode at the apex.

  For the extraction of the magnetocardiographic component, the R wave portion of the electrocardiogram recorded at the same time was added and averaged using a trigger. The software used for analysis is MEG Laboratory (Yokogawa Electric, Eagle Technology) for magnetoencephalography. The addition average was performed 200 times or more. Filter processing and baseline correction processing used for normal brain magnetic field analysis were not performed.

3.2. About MEG Measuring Device A full-head 160-channel magnetoencephalograph (manufactured by Eagle Technology) was used for cerebral magnetic field measurement and magnetocardiographic component measurement. The SQUID sensor was a Gradio-meter and the Baseline was 50.0 [mm]. The S / N ratio calculated using the noise and signal peak values was 0.03 [dB]. Subjects are asked to enter a Magnetically Shielded Room (MSR). For cerebral magnetic field measurement, a chin rest was used, and the space between the Dewar and the head was filled as shown in FIG.

  FIG. 28 is a diagram for explaining a full-head 160-channel magnetoencephalograph. The subject sits in a chair and puts his head in the dewar and performs brain magnetic field measurements.

Table 1 shows the parameters common to all measurements.
(Table 1)
MEG measurement parameters Sampling frequency [Hz] 1000
Low Pass Filter cutoff frequency [Hz] 200
Center frequency [Hz] of Band Elimination Filter 60

3.3. Extraction result of magnetocardiographic component by MEG The result of addition averaging using an electrocardiographic waveform as a trigger is shown. FIG. 29 shows the waveform obtained by the averaging, and FIG. 30 shows the waveform change of the SQUID channel.

  FIG. 29 is a diagram for explaining a magnetocardiogram waveform and an isomagnetic field diagram extracted by addition averaging. In the isomagnetic diagram, green indicates the magnetic field sink (Sink), and red indicates the magnetic field source (Source). An isomagnetic field diagram at a time when the change is intense (red line, blue line) is shown.

  FIG. 30 is a diagram for explaining a magnetocardiographic component included in each SQUID sensor. When each SQUID sensor was roughly divided into left and right, the channel that changed most on the left side was 108 ch, and the channel that changed most on the right side was 71 ch.

  Since an electrocardiogram is used as a trigger and a waveform as shown in FIG. 29 is obtained, it is considered that the magnetic field is induced by a heartbeat. FIG. 31 shows a comparison between the electrocardiogram waveform recorded at the same time and the measured magnetic field data of the channel having a remarkable change found from FIG.

  FIG. 31 is a diagram showing a comparison between an electrocardiogram waveform and measured magnetic field data. A change synchronized with 71 ch and 108 ch is observed over the QRS wave of the electrocardiogram waveform. Subtracting 108ch from 71ch is easier to understand.

  FIG. 32 is a diagram showing a comparison between an electrocardiogram waveform and a measured magnetocardiogram. The magnetocardiogram component corresponding to the P wave, QRS wave, and T wave is measured. A portion corresponding to the QRS wave was greatly measured.

  FIG. 33 shows the results of extracting the magnetocardiographic components of three subjects. Magnetocardial waveforms corresponding to P, QRS, and T were confirmed in all subjects.

  FIG. 33 is a diagram for explaining the magnetocardiographic component extraction results of three subjects. The isomagnetic field diagram shows the same position of the magnetocardiographic component as the blue line in FIG. The waveform of the magnetocardiographic component corresponding to the QRS wave differs depending on the subject. The peak value of the amplitude also varied depending on the subject. When four subjects were measured and the absolute value of the amplitude was taken, the maximum amplitude was 820 [fT], the minimum was 333 [fT], and the average was 485 ± 229 [fT].

  It was investigated whether the magnetocardiogram waveform changed with time (FIG. 34). The frequency component of the magnetocardiographic component is shown in FIG.

  FIG. 34 is a diagram for explaining the similarity of the magnetocardiogram waveforms. A, B, and C in the figure each represent a time interval. An average of 300 times was performed at about 200 [s]. There was almost no change in the magnetocardiogram waveform.

  FIG. 35 is a diagram for explaining the power spectrum distribution (only 108ch) of the magnetocardiogram waveform. Since the magnetocardiographic component of QRS is a steep change, the component is included in a wide frequency range. According to the figure, components of approximately 1 to 50 [Hz] are included. The component of 1 to 50 [Hz] is a band included in the brain magnetic field analysis.

3.4. FIG. 36 shows the result of the estimation of the equivalent current dipole for the extracted magnetocardiogram component.

  FIG. 36 is a diagram for explaining the result of performing the equivalent current dipole estimation on the extracted component. The estimated position color corresponds to the GOF and Intensity. Intensity of 420 [nA · m] was measured at the time when the amplitude of the magnetocardiogram waveform was the highest. The estimated position was estimated in the middle of the brain. The estimation algorithm used Sarvas method.

  In order to examine how the magnetocardiographic component affects the induced magnetic field response, magnetic field data as shown in FIG. 37 was artificially created. MEG Laboratory was used for creation. Equivalent current dipoles were manually placed near the right primary auditory cortex and primary visual cortex on the right (FIG. 38). The magnetocardiographic component was mixed with the magnetic field data generated by each equivalent current dipole.

  FIG. 37 is a diagram for explaining a mixed waveform of a magnetocardiographic component and an equivalent current dipole at a specific portion. The Intensity of the equivalent current dipole was set to 10 [nA · m], and an induced magnetic field was created. Since the amplitude of the magnetocardiographic component is larger than the induced magnetic field, it was multiplied by 0.1 and mixed.

  FIG. 38 is a diagram for explaining isomagnetic field diagrams when a current dipole is present in the primary auditory cortex and the primary visual cortex. The isomagnetic field diagram produced by each current dipole is shown.

  Equivalent current dipole estimation was performed using the created magnetic field data. The channels used for estimation are shown in FIG.

  FIG. 39 is a diagram for explaining channels used for analysis. The blue frame is used to estimate the equivalent current dipole near the primary auditory cortex. The red frame is used to estimate the equivalent current dipole near the primary visual cortex.

  40 and 41 show the results of equivalent current dipole estimation for the mixed magnetic field data.

  FIG. 40 is a diagram for explaining the influence of the magnetocardiographic component on the equivalent current dipole. In the figure above, the colored vertical bars represent Intensity and the dots represent GOF. The positions and colors of the equivalent current dipoles shown on the lower MRI tomogram correspond to each other. The position of the current dipole near the primary auditory cortex is moved downward by the magnetocardiographic component. It also affects Intensity.

  FIG. 41 is a diagram for explaining the influence of the magnetocardiographic component on the equivalent current dipole. In the figure above, the colored vertical bars represent Intensity and the dots represent GOF. The positions and colors of the equivalent current dipoles shown on the lower MRI tomogram correspond to each other. The position of the equivalent current dipole in the vicinity of the primary visual cortex is greatly moved by the magnetocardiographic component. Intensity has also changed significantly. It is greatly influenced by the vicinity of the primary auditory area in FIG.

  The influence of the magnetocardiographic component was different depending on the part. The position of the equivalent current dipole in FIG. 40 mainly moved up and down, whereas in FIG. 41, it moved to a place other than the brainstem or brain.

3.5. Discussion With this method, it was possible to extract magnetocardiographic components from four subjects. The difference in the measured magnetocardiographic waveform is considered to be due to the difference in how the chair sits, the subject's heart position, and the distance from the SQUID to the heart.

  It has been reported that exercise load appears as shortening of QRS time and QT interval of electrocardiogram, and mental load appears as fluctuation of RR interval [8] [9]. During the MEG experiment, since no work was performed, it was thought that the magnetocardiogram waveform would be similar. Since the magnetocardiogram waveform was hardly changed from the results of the magnetocardiogram obtained by dividing the time (FIG. 34), it is considered that this method can be used even in the MEG experiment.

  When performing brain magnetic field analysis, filter processing is performed offline. However, when the frequency component of the magnetocardiogram was analyzed, the magnetocardiogram component was widely distributed. When removing the magnetocardiogram component, it was found that the filter processing normally used attenuates the component of the induced magnetic field response and is difficult to remove. Therefore, it is considered that there is a sufficient value for removing the magnetocardiographic component.

  Equivalent current dipoles were performed on the extracted magnetocardiographic components. Although this result may differ depending on the estimation algorithm, the magnetocardiogram waveform corresponding to QRS was estimated to have an equivalent current dipole at the center of the brain (near the brainstem).

  When the magnetic field data mixed with the magnetocardiogram was artificially created and the equivalent current dipole was estimated, the equivalent current dipole position shifted greatly. The equivalent current dipole near the primary auditory cortex was estimated mainly in the downward direction, and the equivalent current dipole near the primary visual cortex was estimated at the position between the brainstem and the brainstem. The equivalent current dipole is estimated at a position where the boundary of the isomagnetic field diagram is formed. FIG. 42 shows a summary of the isomagnetic field diagram generated by the cardiac magnetic field component.

  FIG. 42 is an isomagnetic field diagram given by the magnetocardiogram component in the vicinity of the primary auditory cortex and in the vicinity of the primary visual cortex. Equivalent current dipoles are estimated at the positions of the magnetic field blowing and sucking (yellow arrows). The effect of the magnetocardiographic isomagnetic field diagram on the isomagnetic field diagram created by the primary auditory cortex is influenced by blowing and suctioning. The boundary between the discharge and suction of magnetocardiograms affects the isomagnetic field diagram created by one visual cortex.

  Comparing the isomagnetic field diagram generated by the equivalent current dipole near the primary auditory cortex and the isomagnetic field diagram generated by the magnetocardiogram component, the magnetocardiogram component mainly affects the suction and the blowout. In the case of the primary visual cortex, the boundary is similar to the isomagnetic field diagram of the magnetocardiographic component. Therefore, it can be considered that the equivalent current dipole of the primary visual cortex is strongly influenced by the magnetocardiogram and is estimated near the brain stem.

  These results are conditions in which the magnetocardiogram is stronger than the induced magnetic field. It can be considered that the greater the number of addition averages, the smaller the influence because the amplitude of the magnetocardiogram becomes smaller. Therefore, it is different from the brain magnetic field data of the actual MEG experiment. However, it was shown that the influence of the magnetocardiographic component differs depending on the position of the equivalent current dipole.

4). Program to remove the magnetocardiographic component The existing analysis software can extract the magnetocardiographic component from the simultaneously recorded electrocardiographic waveform and the addition average. However, the magnetocardiographic component cannot be removed from the measured brain magnetic field data. In order to confirm the knowledge and effectiveness obtained in this study, a program for extracting and removing the magnetocardiographic component was developed (FIG. 43). The appendix explains how to use the program. Here, the processing of the program and the result of removing the magnetocardiogram are shown.

  FIG. 43 is a diagram showing a screen shot (as of January 2008) of a program for removing the magnetocardiogram component. Microsoft Visual C ++ 2005 (Microsoft Corporation) was used as the development environment. MFC (Microsoft Foundation Class) was used for GUI development. The name of the program is “SubMCG” with the meaning of drawing the magnetocardiogram.

4.1. FIG. 44 shows a flowchart of the manufactured program. This program sets the time when the threshold value is exceeded as a trigger from the simultaneously measured ECG waveforms, creates an addition average waveform, and extracts the magnetocardiographic component. The magnetocardiographic component is subtracted from the SQUID sensor in accordance with the trigger timing from the obtained magnetocardiographic component.

  FIG. 44 is a flowchart of the program. The program is roughly divided into four tasks. (1) Pre-processing of the electrocardiogram waveform, (2) Trigger extraction from the electrocardiogram waveform, (3) Addition average for extraction of the magnetocardiogram component, (4) Remove the magnetocardiogram component from the measurement data of the SQUID sensor.

(1) Pre-processing of ECG waveform Filter to ECG waveform. An inter-channel average value is obtained for all SQUID channels. Find the maximum ECG waveform. Until all the samples have been searched, the average value between channels is obtained, and the process of obtaining the maximum value of the electrocardiogram waveform is repeated.

(2) Trigger extraction from electrocardiogram waveform A threshold is obtained from the maximum value of the electrocardiogram waveform. A magnetocardiographic component extraction trigger is detected from the electrocardiogram waveform and the threshold value. Until all the samples have been searched, the process of detecting the magnetocardiographic component extraction trigger from the electrocardiogram waveform and the threshold value is repeated.

(3) Addition averaging for extraction of magnetocardiographic components If there is a large change before and after the trigger time, the addition averaging step is skipped, and if there is no significant change before and after the trigger time, addition averaging is performed. Repeat this until all samples have been searched.

(4) Subtract the magnetocardiographic component from the SQUID sensor measurement data Subtract the magnetocardiographic component from each SQUID measurement data from the magnetocardiographic component extraction trigger. Repeat this until all samples have been searched.

4.2. Results of removing the magnetocardiographic component using the program The result of removing the magnetocardiographic component with the manufactured program is shown (FIG. 45).

  FIG. 45 is a diagram for explaining the result of removing the magnetocardiographic component using this program (screen shot of the program). The magnetocardiographic component synchronized with the electrocardiographic waveform is measured. The lower window shows the result of removing the magnetocardiogram component and the extracted magnetocardiogram component.

5. Evaluation of the magnetocardiographic component removal system A system to remove the magnetocardiographic component was fabricated. An MEG experiment was conducted to evaluate this system.

5.1. Experimental conditions We designed an experiment that can measure the induced cerebral magnetic field response in the primary auditory cortex and primary visual cortex.
The auditory experiment used the experimental system of reference [4] and the acoustic stimulator [19]. The sound stimulus was presented to the subject's left ear at the timing shown in FIG. The presented sound stimulus is a sine wave of 1000 [Hz], and the sound pressure is 60 [dB (A)]. The sound pressure was measured using Ono Sokki's LA-5570 (calibrated). A characteristic (JIS C1505 Type1) was used for the measurement. The background noise was 17-25 dB (A).

  FIG. 46 is a diagram for explaining the presented sound stimulus. The sound stimulus was 200 [ms], and the rise / fall time was 10 [ms]. The presentation interval varies from 700 [ms] to 1000 [ms].

  The visual experiment used the experimental system of reference [20]. For visual stimulation, Gabor approximated by the Cos function was presented to the subject (FIG. 47). The luminance was measured using a spectral radiance meter CS-1000A of Minolta Co., Ltd. (currently Konica Minolta) (FIG. 47).

  FIG. 47 is a diagram for explaining the presented visual stimulus. The background luminance was 15.1 [cd / m 2], and the average luminance of Gabor stimulation was 25.2 [cd / m 2].

5.2. Analysis Method Meg Laboratory (Yokogawa Electric, Eagle Technology) was used for the analysis of brain magnetic field data. Using the stimulus presented to the subject as a trigger, the averaging was performed. During addition averaging, eye movements and extremely large waveforms (2 [pT] or more) were excluded. A band-pass filter with a cutoff frequency of 2-50 [Hz] was applied to the data obtained by averaging. Equivalent current dipole estimation was performed on the data. The estimation algorithm is Sarvas, and the current dipole model is Moving Dipole (a model that assumes that the position moves). The X, Y, and Z axes of the current dipole position used in the results are each an MRI coordinate system (FIG. 48).

  FIG. 48 is a diagram for explaining the MRI coordinate system.

5.3. Evaluation method of this method In order to show the removal effect of the magnetocardiogram component, the equivalent current dipole estimation was performed using the magnetoencephalogram data from which the magnetocardiogram component was removed and the brain magnetic field data from which the magnetocardiogram component was not removed (FIG. 49). . In addition, comparison was made by changing the number of average additions. The average number of additions is 20 times, 50 times, 100 times, 300 times or more.

  FIG. 49 is a diagram for explaining an evaluation method for removing the magnetocardiographic component. The difference between the equivalent current dipole positions of the removed magnetocardiographic component data and the unremoved data was obtained with the same average number of additions.

5.4. Experimental results The results of performing an equivalent current dipole estimation on the brain magnetic field data presenting 300 or more stimuli are shown (FIGS. 50 and 51). The estimation results of the primary auditory cortex and the primary visual cortex are consistent with known knowledge of both position and latency. Using the localized time, the effects of removing the magnetocardiographic component were compared.

  FIG. 50 is a diagram for explaining the magnetic field response and the equivalent current dipole position induced by the auditory stimulus. Between 70 and 130 [ms] from the start of stimulation, an equivalent current dipole was localized in the primary auditory cortex.

  FIG. 51 is a diagram for explaining the magnetic field response induced by the visual stimulus and the equivalent current dipole position. Between 115 and 135 [ms] from the start of stimulation, an equivalent current dipole was localized in the primary visual cortex.

  The result of having compared with the brain magnetic field data which removed the magnetocardiogram component is shown. The comparison results of the magnetocardiographic component removal in the primary auditory cortex are shown in FIGS.

  FIG. 52 is a diagram for explaining the difference in equivalent current dipole position when the average number of additions is 20 in the primary auditory cortex. The Z axis has changed significantly.

  FIG. 53 is a diagram for explaining a difference in equivalent current dipole position in the case of 50 addition averages in the primary auditory cortex. The difference from the original data is getting smaller. There is a difference on the Z axis.

  FIG. 54 is a diagram for explaining a difference in equivalent current dipole position when the average number of additions is 100 in the primary auditory cortex. The difference between the X and Y axes is the same as in FIG. The Z axis changed with time.

  FIG. 55 is a diagram for explaining a difference in equivalent current dipole position in the primary auditory cortex. The averaging was performed the same number of times as the data to be compared. Compared with FIGS. 52 to 54, the X, Y, and Z axes were flat.

  Comparison results of the removal of the magnetocardiographic component in the primary visual cortex are shown in FIGS. Deviations of 40 [mm] or more were omitted from the estimation results.

  FIG. 56 is a diagram for explaining a difference in equivalent current dipole position when the number of average additions is 20 in the primary visual cortex. The X, Y, and Z axes were displaced by 10 [mm] or more.

  FIG. 57 is a diagram for explaining the difference in equivalent current dipole position when the number of average additions is 50 in the primary visual cortex. Compared to FIG. 56, the position does not change with time. The X, Y, and Z axes deviate between 10 and 20 [mm].

  FIG. 58 is a diagram for explaining the difference in equivalent current dipole position when the average number of additions is 100 in the primary visual cortex. The Z axis has changed significantly. The X and Y axes approached the original data.

  FIG. 59 is a diagram for explaining a difference in equivalent current dipole position using data obtained by removing the magnetocardiographic component in the primary visual cortex. The averaging was performed as many times as the data to be compared. The X and Z axes are no longer different from the original data. The Y axis was greatly displaced.

  The average value of the standard deviations of the XYZ axes of the MRI coordinate system was determined, and the relationship of the number of addition averages was summarized (FIG. 60).

  FIG. 60 is a diagram showing the number of averaging times and the standard deviation of the MRI coordinate system XYZ axes. As the average number of additions increased, the change in standard deviation was constant. In the primary auditory cortex, there was no significant change after 50 addition averages. In the primary visual cortex, there was no significant change after 100 addition averages.

5.5. Discussion This method can remove only the magnetocardiographic component. For this reason, the averaging is necessary to some extent. Since the response of the primary visual cortex is sufficiently smaller than the response of the primary auditory cortex (FIGS. 50 and 51), the result of the primary visual cortex with a small average number of additions is shifted by 20 to 30 [mm]. I think that it became.

  As for the result of the primary auditory cortex, the Z axis changed greatly. This is the same as the result suggested in Chapter 3.4. Even in the primary visual cortex, there were results that affected the Z and Y axes (FIGS. 58 and 59). In summary, it was found that the magnetocardiographic component affects the estimated position as shown in FIG.

  FIG. 61 is a diagram for explaining the influence of the isomagnetic field diagram of the magnetocardiographic component on the equivalent current dipole estimation. Red represents a magnetic field blowout, and green represents a suction. When the equivalent current dipole estimation is performed using the blue frame, it is particularly affected in the Z-axis direction. The channel using the red frame is greatly affected by the magnetocardiographic component due to the boundary between blowing and sucking.

  When the average number of additions in the visual and auditory comparison results increases, the difference does not change with time, and only a certain difference appears (FIGS. 55 and 59). In addition, it can be seen from FIG. 60 that only a certain difference occurs even when the average number of additions is increased. This constant difference is considered to be a magnetocardiographic component.

  Since the analysis data is subjected to a band pass filter of 2-50 [Hz], components such as direct current are attenuated. From this, it can be said that the magnetocardiographic component is one factor that distorts the baseline. Since the baseline correction that is usually used directly changes the measurement data, the measurement data is greatly distorted depending on where the reference data is subtracted. Also, it is not effective for data whose baseline is fluctuating. By using this method, it is the same as removing the magnetocardiogram as one factor distorting the base line and applying the base line correction to the whole section.

By using this method, it is possible to exclude the baseline component created by the magnetocardiogram and the influence on the estimated position. In this experiment, the influence of the degree shown in Table 2 could be excluded by using 100 times of brain magnetic field addition average data.
(Table 2)
Type of average position experiment of equivalent current dipole by removing magnetocardiogram component MRI X axis [mm] MRI Y axis [mm] MRI Z axis [mm]
Hearing 0.1 ± 1.6 [mm] 2.4 ± 2.4 [mm] 1.5 ± 3.2 [mm]
Visual 4.6 ± 2.8 [mm] 2.0 ± 2.0 [mm] 16.4 ± 5.6 [mm]

6). Summary The following shows what has been made possible by this study.
Development of an electrocardiograph that can measure heart rate even when performing an experimental task in a shielded room,
Development of a program that extracts the magnetocardiogram component by averaging with ECG,
Development of a program that removes magnetocardiographic components from measured SQUID channel data.

In addition, what has become clear is shown below.
The magnetocardiogram component is also measured by MEG and cannot be removed by filtering.
It was shown that the influence of the magnetocardiographic component varies depending on the measurement site.
When the head is viewed from above, the left and right positions affect the MRI coordinate system Z axis, and the middle position affects the MRI coordinate system Z and Y axes.
When the magnetocardiographic component is averaged, it becomes a factor in the fluctuation of the base line.
By using this method, the magnetocardiogram component was removed and the effect could be shown.

7). References
[1] Masaguchi Higuchi, Magnetometer support system, Med.Imag.Tech.Vol.1l No.2 June 1993.
[2] Hiroshi Irino, Tadao Hori, Use of Event Related Potential (ERP) in Psychological Research, Bulletin of Faculty of Science, Hiroshima University IV, Vol. 26, pp15-31.
[3] Yukio Kosugi, Toshimitsu Takeshi (2000), Electronic Information Communication Engineering Series, Bioinformatics, Morikita Publishing Co., Ltd., 1st edition, pp13-14.
[4] Ryosuke Nomura (2006), Brain magnetic field response during listening to Shepard Tone, 2006 Kanazawa Institute of Technology, Graduate School of Human Information Engineering, pp7-25.
[5] Tetsuo Kobayashi, Shinya Kurishiro, Study on reduction of spontaneous brain magnetic field components in multichannel induced magnetoencephalography, IEICE Technical Report 1996-06, pp7-12.
[6] Shiro Ikeda, Noboru Murata, Analysis of MEG data using Independent Component Analysis, IEICE technical report, Vol. 98, No. 129, pp. 29-36.
[7] Takeshi Kaijima, Yuko Mizuno (Matsumoto), Susumu Date, Kazuhiro Shinozaki, Shinji Shimojo, Clustering analysis of signal components using independent component analysis (ICA) of MEG data, Special Issue of Journal of Biomagnetic Society of Japan, Vol. 16, No. 1, pp. 256-257.
[8] Michiyo Hirano, Relationship between recovery time of post-exercise QRS time and QT interval and ECG abnormalities, Bulletin of Iwate Women's College of Nursing, Vol.2 (19950831), pp15-21.
[9] Shigetaka Yamagishi, Ryoichi Tsuda, Frequency analysis of electrocardiograms of mental workload, Proceedings of the 19th Kumamoto Prefectural Industry-Academia-Government Technology Exchange Meeting, January 2005, pp206-207
[10] Transistor Technology Editorial Department, Tomoaki Ueda (2006), Transistor Technology January 2006, CQ Publishing Co., Ltd., 43 (1), January 2006, pp223-233.
[11] Supervised by Satoshi Yamaguchi, edited by Keiji Tsukada, Reading of magnetocardiogram, Corona Co., Ltd., 2006, 1st edition, 1st edition, pp10-12.
[12] Takeshi Yamaguchi, Bedside Basic Cardiology Cardiomyocyte Electrophysiology-From Ion Channel to ECG, Arrhythmia-Medical Science International, 2006, 1st Edition, 5th Edition, pp23-90.
[13] Masaharu Yoshimura (1970), Interpretation of pulse wave interpretation, Chugai Medical Co., Ltd., 4th edition, pp1-31.
[14] DALE DUBIN, MD, Illustrated electrocardiogram text, Bunkodo, Inc., 2007, 6th edition, 3rd print, p66.
[15] DALE DUBIN, MD, Illustrated electrocardiogram text, Bunkodo Co., Ltd., 2007, 6th edition, 3rd print, p36-54.
[16] Eve Thomas, Shogo Nakamura, Practice Digital Signal Processing, Tokyo Denki University Press, 2005, 1st Edition, 4th Edition, pp132-154.
[17] Masahiro Murakami, Hiroyuki Suzuki, Shigeaki Naito (2006), Hitachi Heart Magnetic Measurement System MC-6400, Lab.Clin.Pract., 24 (2), pp132-139.
[18] Transistor Technology Editorial Department, Akihiro Yamaguchi (2002), Transistor Technology October 2002, CQ Publishing, 39 (10), October 2002, pp173-186.
[19] Noriyuki Kajita (2005), Brain magnetic field response related to musical scale perception, 2005 Kanazawa Institute of Technology, Graduate School of Human Information Engineering, pp41-56.
[20] Kazuo Yamaguchi (2003), Measurement of brain magnetic field response related to Cognitive Timer, 2003 Kanazawa Institute of Technology, Graduate School of Human Information Engineering, pp40-50.
[21] Mitsuya Yamauchi, Statistical Method for Psychology and Education <Second Edition>, Science Co., Ltd., Second Edition, Eighth Printing, pp68-78.

8). Appendix 8.1. About the produced program Describe how to use the program. The default values for the number of SQUID channels and the total number of channels are matched to the MEG measuring device of the Human Information Systems Laboratory.

FIG. 62 is an explanatory diagram of a screen of the magnetocardiogram component removal program.
The magnetocardiogram component is removed by pressing the Open Bin button, Search button, SUB Pre button, and SUB Out button in the yellow frame. The red frame sets the threshold value and the ECG channel to be read.

Usage 1. Select ECG CH and filter.
2. Read the file (Open Bin button).
3. Pressing the Search button updates the ECG window and trigger list.
4). Select the detected trigger or select the position of the waveform. If all of the R wave components of the ECG waveform are normally red, go to 5.
If the number of trigger lists is extremely small or extremely large, try the following.
I. If the ECG waveform is inverted, check the inversion and press the Search button.
II. If the ECG threshold (red line) is too high or too low, set ECG Coeff. The threshold value of this program is the Peak Threshold Ratio of the ECG waveform. Increasing the ECG Coef value increases the threshold.
III. If a trigger is set where the ECG waveform is not an R wave, select the trigger and press the Delete button. Then the trigger is deleted.
IV. If the read waveform is not an electrocardiogram, start over from 1.
5. When the Sub Pre button is pressed, the result of removing the magnetocardiographic component from the pre-processing for removing the magnetocardiographic component and the data selected by the MEG CH can be displayed.
6). The file from which the magnetocardiogram component has been removed is output using the Sub Out button.
If the name of the read file is “abc.bin”, it is output to the same folder with the file name “abc_MCG.con”.

Trigger list import / export The trigger list Import / Export button supports import / export of triggers in MEG Laboratory. You can also import and export triggers using the MEG Laboratory.

  FIG. 63 shows brain magnetic field data that can be read by this program. MEG Laboratory can export a binary file having a structure as shown in FIG.

  FIG. 63 is a diagram for explaining a binary file structure that can be read by a program. The MEG measuring device of the Human Information System Laboratory records 192 ch data per sample. This program was adapted to this specification. The data of 1ch and 1 sample is double type 8 bytes.

  In this program, the measured ECG waveform is digitally filtered to eliminate baseline fluctuations and hum noise. When a band pass filter is performed with an IIR filter, the waveform is distorted due to phase distortion (FIG. 64). For this purpose, this program implements two types of filters: an FIR type band pass filter and an IIR type band elimination filter. DSPLinks of Digital Filter Co., Ltd. was used for calculating the filter coefficient of the IIR filter. The tap coefficient of the FIR filter was calculated with reference to the reference [14].

  FIG. 64 is a diagram for explaining waveform distortion due to phase distortion. A band pass filter of 3-50 [Hz] is applied to Raw Data. The IIR filter has been distorted (* in the figure). The FIR filter does not cause distortion.

  This program can remove the magnetocardiographic component by recording simultaneously with the electrocardiogram. Therefore, it cannot be applied to data that does not measure ECG simultaneously. Therefore, we tried to extract ECG triggers from the channel, which clearly shows the change. From the channel (108CH) in which the waveform changes remarkably and a typical magnetocardiogram, the Pierce product moment correlation coefficient [21] was obtained, and trigger extraction was attempted.

  As a typical magnetocardiogram, an average value for four subjects was used (FIG. 65). The part corresponding to the QRS with remarkable change was used.

  FIG. 65 is a diagram showing a typical magnetocardiogram waveform. An average value of the measured magnetocardiogram waveforms for four subjects was obtained, and only the portion where the change was remarkable was used.

  FIG. 66 is a diagram for explaining a result of an attempt to extract a trigger using correlation. The blue line represents the ECG trigger time. The red line represents the correlation between 108ch and the waveform of FIG. The time when r = 0.85 or more was plotted.

  From the results shown in FIG. 66, there are portions where the trigger time obtained from the electrocardiogram and the time obtained from the correlation coincide with each other, but the portions that do not coincide are also output. Trigger extraction could not be performed simply by obtaining the correlation. I think that it is necessary to search using some constraint condition. For example, the human electrocardiographic interval (RR interval) may be excluded by utilizing a characteristic that becomes substantially constant at rest. If a plurality of other channels whose changes are clearly known are used, it may be possible to extract an electrocardiographic trigger only from data measured from the SQUID sensor.

8.2. Removal of hum using this method The cause of hum is the AC power supply. Therefore, if we obtain the period of the AC power supply, we should be able to remove only the hum noise mixed in the SQUID sensor by this method. FIG. 67 shows the frequency analysis result of the magnetic field data that is not subjected to the band elimination filter of 60 [Hz] during measurement.

  FIG. 67 is a diagram showing a frequency distribution of magnetic field data without applying a band elimination filter of 60 [Hz] at the time of measurement. An averaged frequency distribution for 4 ch SQUID sensors used in the analysis is shown. Peaks are observed in the component of 60 [Hz] and its harmonics, 120 [Hz] and 180 [Hz] (* in the figure).

  FIG. 68 shows the result of removing hum noise by applying a band elimination filter of 60 [Hz] offline, and FIG. 69 shows the result of removing hum noise using this method.

  FIG. 68 is a diagram for explaining the result of hum noise removal using a band elimination filter. Only the 60 [Hz] component is attenuated (red arrow). However, harmonic components are not removed.

  FIG. 69 is a diagram for explaining the result of hum noise removal using this method. The peaks of 60 [Hz] and harmonic components 120 [Hz] and 180 [Hz] disappeared. Compared with the result using the band elimination filter, only the component of 60 [Hz] is not attenuated.

  Since the AC power source and the phase that cause the hum noise are completely fixed and removed, the brain magnetic field component of 60 [Hz] can be removed without being attenuated. For this reason, it is considered that this method is effective for removing hum noise.

  Microsoft Visual C ++ 2005 from Microsoft Corporation was used for program development. The program of reference [19] was used as the sound stimulus creation program for the auditory experiment.

  The present invention has been described based on the embodiments. The embodiments are exemplifications, and it will be understood by those skilled in the art that various modifications can be made to combinations of the respective constituent elements and processing processes, and such modifications are within the scope of the present invention. .

It is a block diagram of the brain magnetic field measuring apparatus which concerns on this Embodiment. It is a figure explaining the signal measured by MEG. It is a figure explaining the noise by the biological phenomenon measured by MEG. It is a figure explaining the noise resulting from the environment measured by MEG. It is a figure explaining the autocorrelation of random noise. It is a figure explaining the noise with a time correlation. It is a figure explaining the scenery in MEG experiment. It is a figure explaining the technique which removes the magnetocardiogram component which this research proposes. It is a figure explaining baseline correction. It is a figure explaining the name and characteristic of each site | part of the heart. It is a figure explaining the example of an experiment task. It is a figure explaining a pulse wave observation device. It is a circuit diagram of a pulse wave observation device. It is a figure explaining the manufactured pulse wave observation apparatus. It is a figure explaining the electrode pasting position of standard 12 induction. It is a figure explaining a typical electrocardiogram waveform. It is a circuit diagram of the produced electrocardiograph. It is a figure explaining the produced electrocardiograph. It is a figure explaining the manufactured electrode lead and the electrode lead made from Nihon Kohden. It is a figure explaining the result of pulse wave measurement using an air dynamic pressure sensor. It is a figure explaining the result of pulse wave measurement using an optical sensor. It is a figure explaining the electrocardiogram waveform before applying a filter and the electrocardiogram waveform after a band pass filter. It is a figure explaining the difference in the waveform by the difference in an electrode sticking position by chest guidance. It is a figure explaining the difference in the waveform by the difference in an electrode sticking position by limb guidance. It is a figure explaining the electrocardiogram waveform when carrying out electrical stimulation and a button pushing operation | movement during electrocardiogram measurement. It is a figure explaining the heart rate measurement of the pulse wave observation apparatus using an optical sensor, and the produced electrocardiograph. It is a figure explaining the sticking position of an electrode. It is a figure explaining a full head type 160 channel magnetoencephalograph. It is a figure explaining the magnetocardiogram extracted by the addition average and an isomagnetic field diagram. It is a figure explaining the magnetocardiogram component contained in each SQUID sensor. It is a figure which shows the comparison with an electrocardiogram waveform and measurement magnetic field data. It is a figure which shows the comparison of an electrocardiogram waveform and the measured magnetocardiogram. It is a figure explaining the magnetocardiographic component extraction result of three test subjects. It is a figure explaining the similarity of a magnetocardiogram waveform. It is a figure explaining the power spectrum distribution of a magnetocardiogram waveform. It is a figure explaining the result of having performed equivalent current dipole estimation with respect to the extraction component. It is a figure explaining the mixed waveform of a magnetocardiogram component and the equivalent current dipole of a specific part. It is a figure explaining an isomagnetic field diagram when there is a current dipole in the primary auditory cortex and the primary visual cortex. It is a figure explaining the channel used for the analysis. It is a figure explaining the influence which a magnetocardiogram component has on an equivalent current dipole. It is a figure explaining the influence which a magnetocardiogram component has on an equivalent current dipole. It is an isomagnetic field diagram which a magnetocardiographic component gives to the primary auditory field vicinity and the primary visual field vicinity. It is a figure which shows the screenshot of the program which removes a magnetocardiogram component. It is a flowchart of a program. It is a figure explaining the removal result (screen shot of a program) of the magnetocardiographic component using this program. It is a figure explaining the presented sound stimulus. It is a figure explaining the presented visual stimulus. It is a figure explaining an MRI coordinate system. It is a figure explaining the evaluation method of a magnetocardiogram component removal. It is a figure explaining the magnetic field response induced by auditory stimulation, and an equivalent current dipole position. It is a figure explaining the magnetic field response induced by visual stimulation, and an equivalent electric current dipole position. It is a figure explaining the difference of the equivalent electric current dipole position in the case of the addition average frequency | count of 20 times in a primary auditory cortex. It is a figure explaining the difference of the equivalent electric current dipole position in the case of 50 addition average frequency | counts in a primary auditory cortex. It is a figure explaining the difference of the equivalent electric current dipole position in the case of the addition average frequency | count of 100 times in a primary auditory cortex. It is a figure explaining the difference of the equivalent electric current dipole position in a primary auditory cortex. It is a figure explaining the difference of the equivalent electric current dipole position in the case of the addition average frequency | count of 20 times in a primary visual cortex. It is a figure explaining the difference of the equivalent electric current dipole position in the case of 50 addition average frequency | counts in a primary visual cortex. It is a figure explaining the difference of the equivalent electric current dipole position in the case of the addition average frequency | count of 100 times in a primary visual cortex. It is a figure explaining the difference of the equivalent electric current dipole position using the data from which the magnetocardiogram component in the primary visual cortex was removed. It is a figure which shows the standard deviation of an addition average frequency | count and an MRI coordinate system XYZ axis. It is a figure explaining the influence which the isomagnetic field diagram of a magnetocardiogram component has on equivalent current dipole estimation. It is screen explanatory drawing of a magnetocardiogram component removal program. It is a figure explaining the binary file structure which a program can read. It is a figure explaining the waveform distortion by phase distortion. It is a figure which shows a typical magnetocardiogram waveform. It is a figure explaining the result of having tried extraction of a trigger using correlation. It is a figure which shows the frequency distribution of the magnetic field data which has not applied the 60Hz band elimination filter at the time of measurement. It is a figure explaining the result of the hum noise removal using a band elimination filter. It is a figure explaining the result of hum noise removal using this method.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 biological signal measurement part, 20 noise signal extraction part, 30 noise generation time acquisition part, 40 noise removal process part, 50 biological measurement signal storage part, 60 noise signal storage part, 100 brain magnetic field measuring apparatus.

Claims (5)

Recording a biological measurement signal measured from a living body in a memory together with information on the generation time of a noise signal having a time correlation;
The time correlation is obtained by taking an addition average in synchronization with the generation time of the noise signal with reference to information on the generation time of the noise signal having the time correlation recorded in the memory with respect to the biological measurement signal. Extracting a characteristic noise signal;
By subtracting the extracted noise signal from the biological measurement signal according to the generation time of the noise signal with reference to the information on the generation time of the noise signal having a time correlation recorded in the memory, Obtaining a biological measurement signal from which the noise signal has been removed and recording it in a memory.   When the averaging is performed in synchronization with the generation time of the time-correlated noise signal with respect to the biological measurement signal, a part of the time range in which the averaging is taken according to the fluctuation in the time direction of the noise signal The noise removal method according to claim 1, wherein the averaging is performed by overlapping. A measurement unit for measuring a signal generated from a living body;
An acquisition unit that acquires information regarding the generation time of a noise signal having a time correlation that occurs when measuring a biological measurement signal by the measurement unit;
A storage unit that stores the biological measurement signal measured by the measurement unit together with information related to the generation time of the time-correlated noise signal;
A noise signal extraction unit that extracts the time-correlated noise signal by taking an addition average in synchronization with the generation time of the time-correlated noise signal with respect to the biological measurement signal;
A noise removal processing unit that obtains a biological measurement signal from which the noise signal has been removed by subtracting the extracted noise signal from the biological measurement signal in accordance with the occurrence time of the time-correlated noise signal. A biological information measuring device comprising: A measurement unit for measuring a brain magnetic field signal generated from a living body,
An acquisition unit for acquiring information on the occurrence time of a heartbeat when measuring the brain magnetic field signal;
A storage unit for storing the cerebral magnetic field signal measured by the measurement unit together with information on the occurrence time of the heartbeat;
With respect to the brain magnetic field signal, a noise signal extraction unit that extracts a noise signal due to the magnetocardiogram by taking an addition average in synchronization with the time of occurrence of the heartbeat;
A noise removal processing unit that obtains a brain magnetic field signal from which the noise signal has been removed by subtracting the extracted noise signal from the brain magnetic field signal in accordance with the time of occurrence of a heartbeat. Brain magnetic field measuring device. Recording a biological measurement signal measured from a living body in a memory together with information on the generation time of a noise signal having a time correlation;
The time correlation is obtained by taking an addition average in synchronization with the generation time of the noise signal with reference to information on the generation time of the noise signal having the time correlation recorded in the memory with respect to the biological measurement signal. Extracting a characteristic noise signal;
By subtracting the extracted noise signal from the biological measurement signal according to the generation time of the noise signal with reference to the information on the generation time of the noise signal having a time correlation recorded in the memory, A program for causing a computer to execute a step of acquiring a biological measurement signal from which a noise signal has been removed and recording it in a memory.