Electroencephalogram (EEG) analysis device and EEG analysis program, as well as exercise support system and exercise support method.

Abstract

The present invention relates to a brain wave analysis device, a brain wave analysis program, a motion assistance system, and a motion assistance method. A brain wave analysis device (16) comprises a signal acquisition unit (40) that acquires a time series of brain wave signals of a subject (12), and a computation unit (58) that determines an intrinsic frequency correlated to a motion-intention of the subject (12) on the basis of a frequency characteristic related to the time series of the brain wave signals acquired when the subject (12) is at rest.

Description

Technical Field

[0001] The present invention relates to an electroencephalogram analysis device, an electroencephalogram analysis program, a motion support system, and a motion support method.

Background Art

[0002] Conventionally, in various fields including medicine, health, nursing care, and sports, techniques for detecting the movement or state of an analysis subject using various sensors and analyzing the movement state of the analysis subject using the obtained detection data have been known. As an example of this technique, a brain-machine interface that analyzes an electroencephalogram signal to identify a characteristic frequency indicating the movement intention or brain state of an analysis subject and operates a machine according to the signal intensity of the characteristic frequency included in the electroencephalogram signal is known (see Non-Patent Document 1, etc.).

[0003] FIG. 12 is a schematic diagram showing a conventional measurement method regarding the characteristic frequency of an analysis subject. The horizontal axis of the map indicates time (unit: s), and the vertical axis of the map indicates frequency (unit: Hz). Here, "time" means the elapsed time from the resting state to the transition to the movement imagination state, and t = 0 corresponds to the time point of the state transition. Also, the shading of the map indicates the magnitude of the signal intensity corresponding to the combination of time and frequency, and is defined such that the signal intensity increases as the color becomes lighter.

[0004] As understood from this figure, immediately after the transition to the movement imagination state (t > 0), a band-shaped region where the signal intensity is relatively low and extends in the time axis direction occurs on the map. The characteristic frequency corresponding to this band-shaped region is called the individual SMR-ERD frequency (Individual Sensorimotor Rhythm Event-Related Desynchronization Frequency: hereinafter, ISF), and is known to have a high correlation with the movement intention.

Prior Art Documents

Non-Patent Documents

[0005] [Non-Patent Document 1] "Neurophysiological predictor of SMR-based BCI performance", B.Blankertz et al., NeuroImage, Volume 51, p1303-1309, 2010 [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] However, when using the conventional measurement method described above, it is necessary to increase both the number and duration of "trials" taken to transition from a resting state to a motor recall state in order to improve the accuracy of identifying natural frequencies. As a result, there is a problem that not only does the time required for identifying natural frequencies increase, but the mental and physical burden on the person being analyzed also increases.

[0007] Specifically, even assuming a 5-second resting state, a 5-second motor recall state, and 20 trials, the measurement time for the natural frequency would be approximately 3 minutes. In particular, for patients with paralysis, the time required to move the affected area increases, potentially extending the measurement time to, for example, 15 minutes. As this measurement time lengthens, the mental and physical burden on the subject increases.

[0008] The present invention has been made in view of these problems, and its objective is to provide an electroencephalogram (EEG) analysis device and EEG analysis program, as well as a motor support system and motor support method, that can significantly reduce the time required for identification when analyzing the electroencephalogram signals of a subject to be analyzed and identifying natural frequencies that correlate with motor intention or brain state. [Means for solving the problem]

[0009] An electroencephalogram (EEG) analysis device according to a first aspect of the present invention comprises an acquisition unit that acquires a time series of the brainwave signals of a subject to be analyzed, and a calculation unit that determines an intrinsic frequency correlated with the motor intention or brain state of the subject to be analyzed based on the frequency characteristics of the time series of the brainwave signals acquired by the acquisition unit while the subject is at rest.

[0010] An electroencephalogram (EEG) analysis device according to a second aspect of the present invention further comprises a calculation unit that calculates sample values ​​of the peak frequencies in the frequency characteristics, and an estimation unit that uses the population of sample values ​​calculated by the calculation unit to obtain estimated values ​​of the peak frequencies, wherein the calculation unit converts the estimated values ​​obtained by the estimation unit into the natural frequencies according to predetermined conversion rules.

[0011] In a third aspect of the present invention, the electroencephalogram (EEG) analysis device obtains the estimated value by the estimation unit and the calculation unit converts the estimated value into the natural frequency, thereby obtaining the natural frequency without acquiring the electroencephalogram signal of the person being analyzed during the motor intention.

[0012] In the fourth aspect of the present invention, the electroencephalogram analysis device obtains the estimated value by the estimation unit and converts the estimated value into the natural frequency by the calculation unit, thereby obtaining the natural frequency without the subject performing any movement of the paralyzed area.

[0013] In the electroencephalogram (EEG) analysis device according to the fifth aspect of the present invention, the estimation unit uses the population of sample values ​​accumulated by repeatedly performing acquisition by the acquisition unit and calculation by the calculation unit at unit intervals from the start of measurement of the EEG signal to obtain the estimated value based on the successive Bayesian method at each unit interval.

[0014] An electroencephalogram (EEG) analysis device according to a sixth aspect of the present invention further comprises a determination unit that determines whether or not a termination condition is met each time the estimation unit performs an estimation, and the estimation unit terminates the estimation of the peak frequency when the determination unit determines that the termination condition is met.

[0015] In the electroencephalogram (EEG) analysis device according to the seventh aspect of the present invention, the peak frequency is an alpha frequency within the alpha band, the natural frequency is an individual SMR-ERD frequency, and the transformation rule is expressed as an identity function or linear function with the estimated value as an argument.

[0016] In the electroencephalogram (EEG) analysis device according to the eighth aspect of the present invention, the conversion rule is determined according to the person being analyzed.

[0017] In the electroencephalogram (EEG) analysis device according to the ninth aspect of the present invention, the acquisition unit acquires a first time series of the EEG signal measured when the subject is at rest or a second time series of the EEG signal measured when the subject is recalling a movement, and the calculation unit performs a first calculation to determine the natural frequency using only the first time series.

[0018] In the electroencephalogram analysis device according to the tenth aspect of the present invention, the calculation unit switches between performing the first calculation or the second calculation, which determines the natural frequency using both the first time series and the second time series.

[0019] An electroencephalogram (EEG) analysis device according to an eleventh aspect of the present invention further comprises a display unit that presents information to the person being analyzed for requesting a resting state.

[0020] An electroencephalogram (EEG) analysis device according to a twelve aspect of the present invention further comprises a support control unit that controls a motor support device for supporting the movement of the person being analyzed, based on the natural frequencies determined by the calculation unit.

[0021] In the electroencephalogram (EEG) analysis program according to the thirteenth aspect of the present invention, one or more computers are made to perform an acquisition step of acquiring a time series of the brainwave signals of a person to be analyzed, and a calculation step of determining an intrinsic frequency that correlates with the motor intention or brain state of the person to be analyzed, based on the frequency characteristics of the time series of the brainwave signals acquired while the person to be analyzed is at rest.

[0022] A fourteenth aspect of the present invention provides a motor support system comprising: an electroencephalogram (EEG) analysis device as described in the twelfth aspect above; an electroencephalograph that measures the brainwaves of a person being analyzed and supplies the EEG signals obtained to the EEG analysis device; and a motor support device that operates in accordance with the control performed by the EEG analysis device to support the movement of the person being analyzed.

[0023] The method for assisting movement according to the fifteenth aspect of the present invention is a method using a system including an electroencephalograph that measures the electroencephalogram of a subject to output an electroencephalogram signal, an electroencephalogram analysis device that analyzes the electroencephalogram signal supplied from the electroencephalograph, and a movement assistance device that assists the movement of the subject by operating according to the control performed by the electroencephalogram analysis device, including: an acquisition step in which the electroencephalogram analysis device measures the resting electroencephalogram of the subject using the electroencephalograph and acquires the time series of the electroencephalogram signal; a calculation step in which the electroencephalogram analysis device obtains a natural frequency correlated with the movement intention or brain state of the subject based on the frequency characteristics regarding the time series of the acquired electroencephalogram signal; a calibration step in which calibration is performed to set the obtained natural frequency as a calibration parameter of the movement assistance device; and an assistance step in which the movement of the subject is assisted by controlling the operation of the movement assistance device in which the calibration has been performed.

Advantages of the Invention

[0024] According to the present invention, when analyzing the electroencephalogram signal of a subject to identify the natural frequency correlated with the movement intention, the time required for the identification can be significantly reduced.

Brief Description of the Drawings

[0025] <H000095>It is an overall configuration diagram of a BMI system incorporating an electroencephalogram analysis device according to an embodiment of the present invention. [Figure 2] It is a functional block diagram of the processor and memory shown in FIG. 1. [Figure 3] It is a flowchart regarding a method for assisting movement using the BMI system shown in FIG. 1. <0H000101>It is a flowchart regarding the analysis operation by the electroencephalogram analysis device of FIG. 1. [Figure 5] It is a diagram showing the time change of the electroencephalogram signal at rest. [Figure 6] It is a diagram showing a method for calculating the IAF sample value. [Figure 7] It is a diagram showing an example of the convergence of the IAF estimated value by the sequential Bayesian method. [Figure 8] This diagram schematically illustrates the time-saving effect of truncation processing during convergence. [Figure 9] This diagram shows the relationship between the IAF and the ISF. [Figure 10] This figure shows the probability density distribution of the deviation between IAF and ISF. [Figure 11] This is a schematic diagram illustrating the effects of the electroencephalogram (EEG) analysis method in this embodiment. [Figure 12] This is a schematic diagram illustrating a conventional measurement method for the natural frequencies of the subject being analyzed. [Modes for carrying out the invention]

[0026] Embodiments of the present invention will be described below with reference to the attached drawings. To facilitate understanding of the description, the same reference numerals are used for identical components and steps in each drawing whenever possible, and redundant explanations are omitted.

[0027] [Configuration of BMI System 10] <Overall Structure> Figure 1 is an overall configuration diagram of a brain-machine interface system (hereinafter referred to as the BMI system 10) incorporating an electroencephalogram (EEG) analysis device 16 in one embodiment of the present invention. The BMI system 10 is configured to analyze the brain waves emitted by the person being analyzed 12 and to support the movements of the person being analyzed 12 based on the analysis results. Specifically, the BMI system 10 includes an electroencephalograph 14, an EEG analysis device 16, and a movement support device 18.

[0028] The electroencephalograph 14 is, for example, a headset configured to measure brain waves emitted from the head 12h of the subject 12. The electroencephalograph 14 outputs the detected electrical signals via electrodes (not shown) to the electroencephalograph 16.

[0029] The electroencephalogram (EEG) analysis device 16 is a computer configured to analyze the motor intentions and brain states such as fatigue and cognition of the subject 12 based on the EEG signals measured by the electroencephalograph 14. Specifically, this EEG analysis device 16 includes an operation unit 22, a display unit 23, a sensor controller 24, a processor 26, and a memory 28.

[0030] The control unit 22 is configured to allow various operations to be performed by users, including the person being analyzed 12 and medical professionals. The control unit 22 may include, for example, input devices such as operation buttons and a microphone, or output devices such as a display panel and a speaker.

[0031] The presentation unit 23 is an output device that, in response to a command from the processor 26, presents information (hereinafter also referred to as "request information") to the person being analyzed 12 requesting a resting state. The presentation unit 23 consists of, for example, a display panel, a lamp, a speaker, etc. Examples of ways in which the request information is presented include written or voice guidance, lighting up a lamp, and outputting various types of sounds. The entity presenting the request information is not limited to the presentation unit 23 of the electroencephalogram (EEG) analyzer 16, but may be a person other than the person being analyzed 12 (for example, the operator of the EEG analyzer 16).

[0032] The sensor controller 24 is a control circuit that performs various controls on the electroencephalograph 14. The sensor controller 24 can perform various signal processing, including sampling processing including sensor synchronization, low-pass filtering, and A / D conversion. As a result, the sensor controller 24 acquires electrical signals (i.e., electroencephalogram signals) representing the brainwaves of the person being analyzed 12 at predetermined sampling intervals and supplies these electroencephalogram signals to the processor 26. Specifically, the sampling interval can take any value in the range of tens to hundreds of milliseconds.

[0033] The processor 26 comprehensively controls each component of the electroencephalogram (EEG) analysis device 16. The processor 26 may be a general-purpose processor including a CPU (Central Processing Unit) or an MPU (Micro-Processing Unit), or it may be a dedicated processor including an FPGA (Field Programmable Gate Array) or a GPU (Graphics Processing Unit).

[0034] Memory 28 is a non-transient storage medium that includes ROM (Read Only Memory) and RAM (Random Access Memory), and stores the programs and data necessary for the processor 26 to control each component.

[0035] The exercise support device 18 is a device for supporting or assisting the movement of a target body part (in the example shown in this figure, the arm 12a) by the person being analyzed 12. Examples of target body parts include various body parts that perform extension / flexion movements, such as the hand, foot, fingers, knee, and elbow, in addition to the arm 12a. The exercise support device 18 may be a "wearable robot" that assists the extension / flexion movement of a body part by the person being analyzed 12 through the driving of an actuator, or it may be an "illusion-inducing device" that assists the extension / flexion movement of a body part by the patient by providing illusionary stimuli through vision or touch.

[0036] <Function Block> Figure 2 is a functional block diagram of the processor 26 and memory 28 shown in Figure 1. The processor 26 reads and executes the electroencephalogram (EEG) analysis program from the memory 28, thereby functioning as a signal acquisition unit 40 (corresponding to the "acquisition unit"), a frequency identification unit 42, and a support control unit 44.

[0037] The signal acquisition unit 40 acquires a time series of electroencephalogram (EEG) signals from the subject 12 via the sensor controller 24 (Figure 1). This allows for the sequential acquisition of EEG signals within a unit period. The unit period may be equal in length to the sampling interval, or it may be an integer multiple of the sampling interval. The EEG signal time series includes signals measured when the subject 12 is at rest (hereinafter also referred to as the "first time series") and signals measured when the subject 12 is recalling motor movements (hereinafter also referred to as the "second time series").

[0038] The frequency identification unit 42 identifies natural frequencies correlated with the motor intentions of the subject 12 by analyzing the electroencephalogram (EEG) signals acquired by the signal acquisition unit 40. Examples of natural frequencies include the individual ERD frequency (Event-related desynchronization frequency) and the individual SMR-ERD frequency (i.e., ISF).

[0039] Here, "SMR-ERD frequency" refers to the frequency within the 8-13 Hz alpha band of scalp electroencephalography (EEG), measured near the motor cortex, where event-related desynchronization (ERD), a motor-related response, is strong. It is known that SMR-ERD frequencies vary from person to person, fluctuating within the 8-13 Hz range. Therefore, to clearly indicate that it is a frequency unique to each individual, it is sometimes called the individual SMR-ERD frequency (i.e., ISF).

[0040] Specifically, this frequency identification unit 42 comprises a preprocessing unit 50, a calculation unit 52, an estimation unit 54, a determination unit 56, and an arithmetic unit 58.

[0041] The preprocessing unit 50 performs preprocessing necessary for calculating the Individual Alpha Frequency (IAF) on the time series of electroencephalogram signals acquired by the signal acquisition unit 40. This preprocessing includes, for example, [1] "filtering" including a moving average, [2] "frequency conversion" including FFT (Fast Fourier Transform), and [3] "trend removal" to remove 1 / f noise from the frequency characteristics (or power spectrum).

[0042] Here, "alpha frequency" refers to the frequency that shows a peak in signal intensity within the 8-13 Hz alpha band of scalp electroencephalography (EEG), which reflects the collective activity of brain nerve cells. It is known that alpha frequencies vary from person to person and fluctuate within the 8-13 Hz range. Therefore, to clearly indicate that it is a frequency unique to each individual, it is sometimes called the individual alpha frequency (i.e., IAF).

[0043] The calculation unit 52 calculates the IAF of the subject 12 from the frequency characteristics of the electroencephalogram signal obtained by the preprocessing unit 50, thereby determining sample values ​​for each time series (hereinafter also referred to as "IAF sample values"). Specifically, the calculation unit 52 detects the maximum peak within a specific band (in this case, the alpha band) in the frequency characteristics and calculates the frequency corresponding to that maximum peak as the IAF sample value. In addition to the alpha band (8-13 Hz), at least one of the delta band (1-3 Hz), theta band (3-7 Hz), beta band (14-30 Hz), and gamma band (30 Hz or higher) can be selected as the specific band depending on the subject of analysis.

[0044] The estimation unit 54 uses the population of IAF sample values ​​calculated by the calculation unit 52 to estimate the IAF of the subjects 12, thereby obtaining estimated values ​​for each population (hereinafter also referred to as "IAF estimates"). Various statistical methods, including Bayesian methods and sequential Bayesian methods (or Kalman filters), can be used as estimation methods. For example, the estimation unit 54 may use the population accumulated from the start of electroencephalogram (EEG) signal measurement to obtain estimates based on the sequential Bayesian method for each unit period.

[0045] The determination unit 56 determines whether the IAF estimate satisfies the termination condition each time the estimation unit 54 performs an estimation, and if the termination condition is met, it instructs the estimation unit 54 to terminate the estimation process. Examples of termination conditions include: [Condition 1] the IAF estimate has converged (for example, the amount of change has fallen below a threshold), [Condition 2] the number of estimations performed by the estimation unit 54 has exceeded a threshold, [Condition 3] time has elapsed since the start of measurement of electroencephalogram signals, and [Condition 4] a combination of the above conditions 1 to 3.

[0046] When the determination unit 56 determines that the termination condition is met, the calculation unit 58 converts the most recently obtained IAF estimate into an eigenfrequency (in this case, ISF) correlated with the motion intention of the subject being analyzed 12, according to a predetermined conversion rule, thereby obtaining a converted value (hereinafter, ISF converted value) for each analysis operation. This conversion rule may be a rule common to all subjects being analyzed 12, or it may be a rule different depending on the subject being analyzed 12. Furthermore, if the conversion rule is a function that takes IAF as an argument, the shape of the function may be [1] a linear function such as an identity function or a linear function, or [2] a nonlinear function such as a polynomial function with a power of 2 or greater or an exponential function.

[0047] The calculation unit 58 performs calculation processing to determine the ISF using only the first time series described above (hereinafter referred to as the "first calculation"), but it may also perform calculation to determine the ISF using both the first and second time series (hereinafter referred to as the "second calculation"). In this case, the calculation unit 58 may switch between the first and second calculations as needed. The two types of calculations may be switched manually, for example, via input operations from the operation unit 22 (Figure 1), or they may be switched automatically based on the analysis results of the electroencephalogram signals acquired by the signal acquisition unit 40.

[0048] The support control unit 44 performs control (hereinafter also referred to as "support control") of the motor support device 18 based on the ISF conversion value obtained by the calculation unit 58. The support control unit 44 is composed of a setting unit 60 that sets an ISF setting value suitable for the person being analyzed 12, and a determination unit 62 that determines the control amount of the motor support device 18 using the frequency characteristics of the electroencephalogram signal and the ISF setting value set in the setting unit 60.

[0049] Meanwhile, the memory 28 stores the data subject information 70, frequency information 72, and conversion information 74, each associated with the data subject.

[0050] The subject information 70 includes various information relating to the subject 12, such as the subject 12's identification information / personal information, the subject 12's diagnosis results / recovery status, and the type and usage history of the exercise support device 18. The frequency information 72 includes various information relating to the peak frequency or natural frequency, such as the IAF sample value, IAF estimate, and ISF transformed value. The transformation information 74 includes various information that allows for the identification of the transformation rule, such as the type of function shape, coefficients, order, and LUT (lookup table).

[0051] [BMI System 10 Operation] The BMI system 10 in this embodiment is configured as described above. Next, the operation of the BMI system 10, and more specifically the motor support operation by the electroencephalogram (EEG) analysis device 16, will be explained with reference to the flowcharts in Figures 3 and 4 and Figures 5 to 11.

[0052] <Explanation of exercise support methods> Figure 3 is a flowchart of the exercise support method using the BMI system 10 shown in Figure 1. In step SP100, the "attachment" process is performed in which the electroencephalograph 14 is attached to the head of the person being analyzed 12. In step SP102, the "start" process is performed in which monitoring of the brain waves emitted by the person being analyzed 12 is started. In step SP104, the "confirmation" process is performed to check whether the person being analyzed 12 is in a resting state. If the person being analyzed 12 is not in a resting state (step SP104: NO), the process remains at step SP104 until the person being analyzed 12 is in a resting state. On the other hand, if the person being analyzed 12 is in a resting state (step SP104: YES), the process proceeds to the next step SP106.

[0053] In step SP106, a "calibration" process is performed to calibrate the motor support device 18. Specifically, the electroencephalogram (EEG) analyzer 16 analyzes the EEG signals sequentially supplied from the electroencephalograph 14 from step SP102 onward to determine the natural frequency (i.e., ISF) of the person being analyzed 12, and sets the value of this ISF as the calibration parameter for the motor support device 18.

[0054] In step SP108, a "support" process is performed to provide neurorehabilitation to the subject 12. Specifically, the electroencephalogram (EEG) analyzer 16 controls the operation of the motor support device 18, which was calibrated in step SP106. As a result, the motor support of the subject 12 is provided through the operation of the motor support device 18.

[0055] <Analysis operation by electroencephalogram (EEG) analyzer 16> Figure 4 is a flowchart of the analysis operation performed by the electroencephalogram (EEG) analysis device 16 shown in Figure 1. For example, prior to rehabilitation of the arm 12a of the subject 12, the EEG analysis device 16 executes this flowchart to identify the subject 12's natural frequency (i.e., ISF). This flowchart is not limited to being performed only once at the start of rehabilitation, but may be executed one or more times during the rehabilitation as needed. Furthermore, this operation can be directly confirmed by analyzing the EEG analysis program, or indirectly confirmed by comparing the ideal signal waveform simulated and input to the electroencephalograph 14 with the output result from the EEG analysis device 16.

[0056] In step SP10 of Figure 4, the signal acquisition unit 40 acquires the time series of electroencephalogram signals within a unit period via the electroencephalograph 14 and the sensor controller 24.

[0057] Figure 5 shows the time evolution of electroencephalogram (EEG) signals during rest. The horizontal axis of the graph represents time (in seconds), and the vertical axis represents electroencephalogram (in mV). As can be seen from this figure, EEG signals have a complex waveform that fluctuates finely above and below a baseline value.

[0058] In step SP12 of Figure 4, the frequency identification unit 42 (more specifically, the preprocessing unit 50) performs preprocessing on the time series of electroencephalogram signals acquired in step SP10. This allows the frequency characteristics of the electroencephalogram signals to be obtained.

[0059] In step SP14, the frequency identification unit 42 (more specifically, the calculation unit 52) ​​calculates IAF sample values ​​from the frequency characteristics obtained by the preprocessing in step SP12.

[0060] Figure 6 shows the method for calculating the IAF sample value. The horizontal axis of the graph represents frequency (unit: Hz), and the vertical axis represents signal strength (unit: dimensionless). As can be seen from this figure, this frequency response has a maximum peak around 11 Hz. In other words, the IAF sample value is calculated as 11 Hz.

[0061] In step SP16 of Figure 4, the processor 26 checks whether the timing for estimating the IAF (hereinafter referred to as the "estimation timing") has arrived. If the estimation timing has not yet arrived (step SP16: NO), the processor 26 returns to step SP10 and sequentially repeats the execution of steps SP10 to SP16 until the timing arrives. On the other hand, if the estimation timing has arrived (step SP16: YES), the processor 26 proceeds to the next step SP18.

[0062] In step SP18, the frequency identification unit 42 (more specifically, the estimation unit 54) estimates the IAF by applying the successive Bayesian method to the population accumulated through the successive calculations in step SP14. For example, according to the Bayesian method, the posterior probability p(IAF=A|Data) is calculated using the prior probability p(IAF=A) according to the following equation (1). Note that in the case of the successive Bayesian method, the previous posterior probability is considered as the prior probability for the current posterior probability, and the current posterior probability is calculated accordingly.

[0063]

number

[0064] In step SP20, the frequency identification unit 42 (more specifically, the determination unit 56) determines whether the IAF estimate obtained in step SP18 satisfies the termination condition. If the termination condition is not met (step SP20: NO), the processor 26 returns to step SP10 and sequentially repeats the execution of steps SP10 to SP20 until the termination condition is met.

[0065] Figure 7 shows an example of the convergence of IAF estimates using the successive Bayesian method. The horizontal axis of the graph represents time (in s), and the vertical axis represents the estimation error (in Hz). Here, "time" corresponds to the elapsed time from the start of the estimation using the successive Bayesian method (t=0). "Estimation error" corresponds to the value obtained by subtracting the actual value from the estimated IAF (i.e., the deviation).

[0066] The solid line graph shows the average value of data from 180 healthy adults. The lower dashed line graph shows the boundary line of "mean - 2σ" (σ: standard deviation), and the upper graph shows the boundary line of "mean + 2σ". As can be seen from this figure, although the estimation error is large at the start, it decreases exponentially over time, and eventually converges to zero or a value close to it.

[0067] For example, assuming that the behavior of the IAF estimate follows a Gaussian distribution, setting the estimation time to 26 seconds allows the estimation error for 95.7% of healthy adults (i.e., within the 4σ range) to be kept within 1 Hz. This significantly reduces the time required compared to conventional measurement methods (several minutes to over ten minutes) as illustrated in Figure 12.

[0068] Figure 8 schematically illustrates the time-saving effect of truncation during convergence. The axis of the histogram represents the estimated time required for an individual's estimation error (see Figure 6) to fall within 1 Hz, i.e., the convergence time (in seconds). According to this histogram, the frequency is highest when the convergence time is within 4 seconds, and gradually decreases as the convergence time increases.

[0069] As can be seen from this figure, the median convergence time corresponds to "11s," and the 2σ boundary of the convergence time corresponds to "30s." In other words, if the estimation error for 95.7% of healthy adults is to be kept within 1Hz, the estimation time can be further reduced by approximately 2 / 3 (30-11=19s) by introducing the truncation process at convergence (Condition 1 described above).

[0070] On the other hand, if the termination condition is met by returning to step SP20 in Figure 4 (step SP22: NO), the processor 26 proceeds to the next step SP22.

[0071] In step SP22 of Figure 4, the frequency identification unit 42 (more specifically, the calculation unit 58) converts the IAF estimate obtained most recently in step SP18 into an ISF according to the conversion rule specified by the conversion information 74. Specifically, the ISF is converted according to the following equation (2), where g(·) is an arbitrary function.

[0072]

number

[0073] Figure 9 shows the relationship between ISF and IAF. This figure is based on the number of IAF and ISF data pairs counted for 187 healthy adults, and the lighter the color (closer to white), the higher the count. As can be seen from this figure, ISF has a certain positive correlation (r=0.62) with IAF.

[0074] Figure 10 shows the probability density distribution for the deviation between IAF and ISF. Here, "deviation" is defined as Δ = IAF - ISF (unit: Hz). As can be seen from this figure, this distribution has a maximum probability at a deviation of Δ = 0 Hz.

[0075] In other words, the ISF may be calculated according to equation (3) below, focusing on the relationship shown in Figures 9 and 10. In this case, g(x) = x in equation (2), which represents the so-called identity transformation.

[0076]

number

[0077] In step SP22 of Figure 4, the support control unit 44 (more specifically, the setting unit 60) sets the ISF conversion value obtained in step SP20. This allows the support control unit 44 to perform control on the motor support device 18 that is appropriate for the person being analyzed 12.

[0078] <Effects of the above method> Figure 11 is a schematic diagram illustrating the effects of the electroencephalogram (EEG) analysis method in this embodiment. The axis of the graph represents the time (in seconds) from the start of EEG measurement. The upper bar graph shows the breakdown of ISF identification time in the "Comparative Example" (see Figure 12). The lower bar graph shows the breakdown of ISF identification time in the "Example" (see Figures 3 and 4).

[0079] In the "conventional example," if T1 is the time to reach a resting state, T2 is the time to maintain a motor recall state, and 20 measurement trials are performed, the measurement time for the natural frequency is 20Tc (where Tc = T1 + T2). In particular, in the case of subject 12 with paralysis, the time required to move the affected area becomes longer, and the measurement time can be, for example, 200 to 1000 seconds. As this measurement time increases, the mental and physical burden on subject 12 increases.

[0080] On the other hand, in the "Example," if we define T3 as the total time required from reaching a resting state, starting ISF estimation, and completing the estimation, then, as already explained in Figures 7 and 8, T3 is approximately 30 seconds. In other words, ISF can be identified in a short time without acquiring electroencephalogram signals during the subject 12's motor attempt, or without the subject 12 performing any movement of the paralyzed area. This significantly reduces the mental and physical burden on the subject 12.

[0081] [Summary of Embodiments] As described above, the exercise support system in this embodiment (here, the BMI system 10) comprises an electroencephalogram (EEG) analyzer 16, an electroencephalograph 14 that measures the brainwaves of a subject 12 and supplies the resulting EEG signals to the EEG analyzer 16, and an exercise support device 18 that operates according to the control performed by the EEG analyzer 16 to support the movements of the subject 12.

[0082] This electroencephalogram (EEG) analysis device 16 includes an acquisition unit (here, a signal acquisition unit 40) that acquires a time series of the brainwave signals of the person being analyzed 12, and a calculation unit 58 that determines natural frequencies correlated with the motor intention or brain state of the person being analyzed 12 based on the frequency characteristics of the time series of acquired brainwave signals.

[0083] Furthermore, according to the electroencephalogram (EEG) analysis method and EEG analysis program in this embodiment, one or more computers (or processor 26) perform an acquisition step (SP10) to acquire a time series of the brainwave signals of the subject 12 being analyzed, and a calculation step (SP22) to determine the natural frequencies correlated with the motor intentions of the subject 12 based on the frequency characteristics of the time series of brainwave signals acquired while the subject 12 was at rest.

[0084] Furthermore, according to the exercise support method and exercise support program in this embodiment, one or more computers (or processor 26) further perform, in addition to the acquisition step (SP10) and calculation step (SP22) described above, a calibration step (SP24) which performs calibration to set the obtained natural frequency as a calibration parameter for the exercise support device 18, and an assistance step (SP108) which supports the movement of the person being analyzed 12 by controlling the operation of the exercise support device 18 after calibration has been performed.

[0085] In this way, by determining the natural frequencies correlated with the motor intention or brain state of the subject 12 based on the frequency characteristics of the time series of the electroencephalogram (EEG) signal acquired while the subject 12 is at rest, it becomes possible to identify the natural frequencies using the acquired EEG signal without requiring the subject 12 to maintain a resting state, i.e., without requesting any additional movement. This significantly reduces the time required to identify natural frequencies correlated with motor intention or brain state when analyzing the EEG signal of the subject 12.

[0086] Furthermore, the electroencephalogram analysis device 16 further includes a calculation unit 52 that calculates sample values ​​of peak frequencies in the frequency characteristics, and an estimation unit 54 that obtains estimated values ​​of peak frequencies using the population of calculated sample values. The calculation unit 58 may convert the estimated values ​​obtained by the estimation unit 54 into natural frequencies according to predetermined conversion rules.

[0087] In other words, the electroencephalogram (EEG) analyzer 16 may determine the natural frequency without acquiring EEG signals from the subject 12 during motor attempt, or without the subject 12 performing any movement of the paralyzed area, by having the estimation unit 54 obtain an estimated peak frequency and the calculation unit 58 convert the estimated value into a natural frequency. This significantly reduces the mental and physical burden on the subject 12.

[0088] Alternatively, the estimation unit 54 may use a population of sample values ​​accumulated by repeatedly performing acquisition by the signal acquisition unit 40 and calculation by the calculation unit 52 at unit intervals from the start of measurement of electroencephalogram signals to obtain estimated values ​​based on the successive Bayesian method at unit intervals.

[0089] Furthermore, if the electroencephalogram analysis device 16 further includes a determination unit 56 that determines whether or not the termination condition is met each time the estimation unit 54 performs an estimation, the estimation unit 54 may terminate the peak frequency estimation when the determination unit 56 determines that the termination condition is met. By terminating the estimation process when the termination condition is met, the time required to identify the natural frequency can be further reduced.

[0090] Furthermore, if the peak frequency is the alpha frequency within the alpha band (i.e., IAF) and the natural frequency is the individual SMR-ERD frequency (i.e., ISF), the transformation rule may be expressed as an identity function or linear function with the IAF estimate as an argument. Using a simpler transformation rule can speed up the calculation.

[0091] Furthermore, the conversion rules may be determined according to the subject being analyzed 12. This allows for a conversion that better matches the characteristics of the subject being analyzed 12.

[0092] Furthermore, when the signal acquisition unit 40 acquires a first time series of electroencephalogram signals measured when the subject 12 is at rest, or a second time series of electroencephalogram signals measured when the subject 12 is recalling a motor movement, the calculation unit 58 may perform a first calculation to determine the natural frequency using only the first time series. In addition, the calculation unit 58 may switch between performing the first calculation described above, or a second calculation to determine the natural frequency using both the first and second time series.

[0093] Furthermore, the electroencephalogram (EEG) analysis device 16 may further include a presentation unit that presents information to the person being analyzed 12 requesting a resting state. The EEG analysis device 16 may also further include a support control unit 44 that controls a motor support device 18 to assist the movements of the person being analyzed 12 based on the natural frequencies determined by the calculation unit 58. This enables analysis or motor support for the person being analyzed 12.

[0094] [Differentiation] It should be noted that the present invention is not limited to the embodiments described above, and can be freely modified without departing from the spirit of the invention. Alternatively, the various components may be combined in any way that does not create a technical inconsistency. Alternatively, the execution order of each step constituting the flowchart may be changed as long as it does not create a technical inconsistency.

[0095] In the embodiment described above, the case in which the electroencephalogram (EEG) analysis device 16 analyzes EEG signals and controls the motor support device 18 was used as an example, but the system configuration is not limited to this. For example, the analysis unit and the control unit may be provided separately and configured to exchange necessary data with each other via wired or wireless communication. In this case, a cloud-based or on-premise server device may perform the analysis processing. In particular, if the server device is cloud-based, the server device may be a group of computers that make up a distributed system.

[0096] [Explanation of symbols] 10...BMI system (exercise support system), 12...Subject being analyzed, 14...Electroencephalograph, 16...Electroencephalogram analysis device (computer), 18...Exercise support device, 26...Processor, 28...Memory, 40...Signal acquisition unit (acquisition unit), 42...Frequency identification unit, 44...Exercise support unit, 50...Preprocessing unit, 52...Calculation unit, 54...Estimation unit, 56...Determination unit, 58...Calculation unit

Claims

1. An acquisition unit that acquires the time series of the brainwave signals of the subject being analyzed, A calculation unit calculates a sample value of the peak frequency in the frequency characteristics of the time series of the electroencephalogram signal acquired by the acquisition unit while the subject of analysis is at rest, An estimation unit that uses the population of sample values ​​calculated by the calculation unit to obtain an estimated value of the peak frequency, A calculation unit converts the estimated value obtained by the estimation unit into the natural frequency of the person being analyzed, according to a conversion rule that shows the relationship between the peak frequency and the natural frequency correlated with the motor intention or brain state. An electroencephalogram (EEG) analysis device equipped with the following features.

2. The conversion rule is a rule based on the relationship in healthy adults, The electroencephalogram (EEG) analysis device according to claim 1.

3. The electroencephalogram (EEG) analysis device according to claim 1, wherein the estimation unit obtains the estimated value and the calculation unit converts the estimated value into the natural frequency, thereby determining the natural frequency without acquiring the electroencephalogram signal of the person being analyzed during the motor intention.

4. The electroencephalogram (EEG) analysis apparatus according to claim 1, wherein the estimation unit obtains the estimated value and the calculation unit converts the estimated value into the natural frequency, thereby determining the natural frequency without the subject performing any movement of the paralyzed area.

5. The estimation unit uses the population of sample values ​​accumulated by repeatedly performing acquisition by the acquisition unit and calculation by the calculation unit at each unit period from the start of measurement of the electroencephalogram signal to obtain the estimated value based on the sequential Bayes method at each unit period. The electroencephalogram (EEG) analysis device according to claim 1.

6. The system further includes a determination unit that determines whether or not the termination condition is met each time the estimation unit performs an estimation, The estimation unit terminates the estimation of the peak frequency if the determination unit determines that the termination condition is met. The electroencephalogram (EEG) analysis device according to claim 5.

7. The aforementioned peak frequency is the alpha frequency within the alpha band. The aforementioned natural frequency is the individual SMR-ERD frequency. The transformation rule is expressed as an identity function or linear function that takes the estimated value as an argument. The electroencephalogram (EEG) analysis device according to claim 1.

8. The aforementioned conversion rules are determined according to the subject being analyzed. The electroencephalogram (EEG) analysis device according to claim 1.

9. The acquisition unit acquires a first time series of the electroencephalogram (EEG) signal measured when the subject is at rest, or a second time series of the EEG signal measured when the subject is recalling a movement. The electroencephalogram (EEG) analysis device according to claim 1, wherein the calculation unit performs a first calculation to determine the natural frequency using only the first time series.

10. The calculation unit performs the first calculation, or the second calculation which determines the natural frequency using both the first time series and the second time series, by switching between the two. The electroencephalogram (EEG) analysis device according to claim 9.

11. The system further includes a display unit that presents information to the subject being analyzed, requesting that they remain at rest. The electroencephalogram (EEG) analysis device according to claim 1.

12. The system further includes a support control unit that performs control over a motion support device to assist the motion of the subject based on the natural frequency converted by the calculation unit, which is correlated with the motion intention of the subject being analyzed. The electroencephalogram (EEG) analysis device according to claim 1.

13. The acquisition step involves obtaining a time series of the subject's electroencephalogram (EEG) signals, A calculation step of calculating a sample value of the peak frequency in the frequency characteristics of the time series of the electroencephalogram signal acquired while the subject of analysis was at rest, An estimation step to obtain an estimated value of the peak frequency using the population of the calculated sample values, A calculation step of converting the obtained estimate into the natural frequency of the person being analyzed, according to a conversion rule that shows the relationship between the peak frequency and the natural frequency that correlates with the motor intention or brain state, An electroencephalogram (EEG) analysis program that runs on one or more computers.

14. The electroencephalogram analysis device according to claim 12, An electroencephalograph that measures the brainwaves of a subject and supplies the resulting brainwave signals to the electroencephalograph, A motor support device that operates in accordance with the control performed by the electroencephalogram analysis device, thereby supporting the movement of the person being analyzed. An exercise support system equipped with the following features.

15. A method for providing motor support using a system comprising: an electroencephalograph that measures the brain waves of a subject and outputs an electroencephalogram signal; an electroencephalogram analysis device that analyzes the electroencephalogram signal supplied from the electroencephalograph; and a motor support device that operates according to the control performed by the electroencephalogram analysis device to support the movement of the subject, The aforementioned electroencephalogram analysis device, The acquisition step involves using the electroencephalograph to measure the resting electroencephalogram of the subject to be analyzed and obtaining a time series of the electroencephalogram signals. A calculation step of calculating a sample value of the peak frequency in the frequency characteristics of the time series of the acquired electroencephalogram signal, An estimation step to obtain an estimated value of the peak frequency using the population of the calculated sample values, A calculation step of converting the obtained estimated value to the natural frequency of the person being analyzed, according to a conversion rule that shows the relationship between the peak frequency and the natural frequency correlated with the motion intention, A calibration step in which the converted natural frequency is set as a calibration parameter of the motion support device, A support step that supports the movement of the person being analyzed by controlling the operation of the movement support device on which the calibration has been performed, A method of providing exercise support.

Images