JP2842409B2 - Evoked waveform calculator - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal analyzing and processing apparatus, and more particularly to a method for inputting biological information data such as brain waves, brain magnetic fields, myoelectric potentials and the like, and a trigger signal, which are observed at least at one place on the surface of a living body. In addition, the present invention relates to an evoked waveform calculation device for obtaining an evoked waveform or an event-related waveform for a trigger signal.

[0002]

2. Description of the Related Art Generally, a signal emitted from a living body is very weak and is easily affected by electric noise or magnetic noise. For example, when trying to observe the activity in the brain as an electroencephalogram, the signal for a specific stimulus or response is very weak and it is very difficult to extract it.

Conventionally, a method called an "averaging method" has been used in order to emphasize a biological signal for a specific stimulus or response. That is, a certain external stimulus is repeated a plurality of times, and time-series data of a biological signal obtained for each stimulus is collected. These are aligned at the time of stimulation and added, and the result is divided by the number of times to obtain an averaged time series data.

[0004] Further, when a signal corresponding to the response of the subject is used instead of the external stimulus to obtain time-series data of a biological signal corresponding to the signal, the calculation can be performed in a similar manner. For example, assuming that the time at which a certain external stimulus is given or the time at which the subject's reaction occurs is 0, and the data at the time t observed at the i-th time is d _i (t), the data is obtained for n stimuli. The averaging data a (t) of the time-series data of the biological information is represented by the following equation (1).

[0005]

(Equation 1)

In the time-series data obtained by averaging, the amplitude of a component corresponding to a certain external stimulus or the response of the subject increases n times by adding n times, but the non-corresponding component increases only n ^1/2 times. Therefore, by increasing n, the S / N ratio (signal-to-noise ratio) of the target component increases.

The time series data obtained in this way is
The analysis was performed by referring to these as "evoked waveforms" or "event-related waveforms" and as biological signals for specific stimuli and responses.

FIG. 12 is a block diagram showing the configuration of the above-described conventional induced waveform calculating apparatus. Referring to FIG. 12, in the conventional method, a data storage unit 2 that receives biological information data 11 and a trigger signal and stores data for a predetermined period from a trigger, and an addition and average that averages data 21 from data storage unit 2 And an averaging unit 3.

[0009]

In the above-mentioned conventional method, when the number of measurements n is a relatively small value, the S / N ratio does not increase so much, and it is difficult to extract the induced waveform. There is a problem that.

Conventionally, the assumed noise model is static, but in reality, the parameters representing the noise model are dynamically changing. For this reason, when even one input data having a very large noise is to be added, it is difficult to obtain a good induced waveform even if the number of additions is relatively large.

Accordingly, the present invention has been made in view of the above circumstances, and an object of the present invention is to solve the above-described problem and to determine whether or not an observed biological signal is data suitable for averaging. An object of the present invention is to provide an evoked waveform calculation apparatus that outputs a good evoked waveform by calculating input-only evoked waveforms using only matched data by removing input data including noise and the like.

[0012]

In order to achieve the above object, the present invention provides a biological information data such as an electroencephalogram, a brain magnetic field, and a myoelectric potential measured at at least one place on the surface of a living body, a trigger signal, In the evoked waveform calculation device for inputting the evoked waveform and the event-related waveform for the trigger signal,
The biological information data and the trigger signal are input, and the biological information data of a predetermined time width is accumulated from a point in time according to the trigger signal, and the data regarding the averaging is determined according to the determination information regarding the data selection. Accumulation means for outputting, data for the averaging, and averaging means for calculating an average value for each time and for each measurement position,
Input data analysis means for inputting the biological information data and calculating a feature amount of the input data ;
Effectiveness meter that inputs the averaging result and calculates its effectiveness
Calculating means, and the feature amount of the input data and / or the addition
Data selection means for determining from the effectiveness of the average result whether the input biometric information data is data suitable for addition,
An evoked waveform calculation device characterized by comprising:

[0013]

Embodiments of the present invention will be described below. In a preferred embodiment of the present invention, the biological information data (11 in FIG. 1) and the trigger signal (FIG.
12) is input, biometric information data of a predetermined time width is accumulated from the time corresponding to the trigger signal, and the determination information (41 in FIG. 1) regarding data selection from the data selection means (4 in FIG. 1). ), Data storage means (2 in FIG. 1) for outputting data relating to the averaging (21 in FIG. 1).
And data (21 in FIG. 1) relating to the averaging from the data storage means (2 in FIG. 1), and the averaging means (3 in FIG. 1) for calculating the average value for each time and each measurement position.
And input data analysis means (5 in FIG. 1) for inputting biometric information data and calculating the characteristic amount of the input data, and data for inputting the characteristic amount of the input data and determining whether the data is suitable for addition Selecting means (4 in FIG. 1).

According to a second preferred embodiment of the present invention, the biological information data and the trigger signal are inputted, and the biological information data of a predetermined time width is accumulated from a time point corresponding to the trigger signal. The data storage means (2 in FIG. 2) for outputting data on the averaging in accordance with the determination information on the data selection from the data selection means (4 in FIG. 2), and the data on the averaging are input and And an addition average (3 in FIG. 2) for calculating an average value, an effectiveness calculation means (6 in FIG. 2) for inputting the result of the addition and calculating the effectiveness thereof for each measurement position, and Data selecting means (4 in FIG. 2) for inputting the validity and determining whether the biological information data is data suitable for addition.

In the embodiment of the present invention,
As in the above embodiment, input data analysis means for calculating the characteristic amount of the input biological information data (see 5 in FIG. 1)
And a validity calculating means (refer to 6 in FIG. 2) for calculating the validity of the averaging result, and in the third embodiment, the input data analyzing means It has both of the effectiveness calculation means (see FIG. 3), and based on the data output from these means,
Data selection means (4 in FIG. 3) for determining whether the input biometric information data is data suitable for addition is provided.

The principle of the evoked waveform calculator according to the embodiment of the present invention will be described.

In the first and third embodiments of the present invention, the time series of the input biological information data is analyzed to determine whether the data is suitable for addition. Analysis methods include a method based on frequency analysis, a method based on wavelet analysis, a method based on template matching, and a method based on covariance calculation.

Further, in the second and third embodiments of the present invention, the time series of the input biometric information data is temporarily added to the averaged data and averaged, and the obtained averaged result is analyzed. Then, it is determined whether the input data is data suitable for addition. Analysis methods include a method based on frequency analysis, a method based on wavelet analysis, a method based on template matching, and a method based on covariance calculation.

Therefore, it is possible to select only the input waveform having less noise, and it is possible to calculate a good induced waveform. In addition, since data including large noise can be excluded from addition targets, a good induced waveform can be calculated from a small number of input data.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS In order to explain the above-described embodiment of the present invention in more detail, an embodiment of the present invention will be described below with reference to the drawings.

[0021]

Embodiment 1 FIG. 1 shows a first embodiment of an evoked waveform calculating apparatus according to the present invention.
FIG. 3 is a block diagram showing a configuration of the example. In the embodiment described below, an example will be described in which an external stimulus such as light or sound is given to a subject, brain wave data from the subject is input, and an induced brain wave is output. Also,
In the following description, the biological information data 11 is time-series data of brain waves observed from at least one electrode placed on the subject's head. However, the biological information data 11
In addition to using brain wave data, brain magnetic field data, myoelectric potential data, and the like can be used.

As the external stimulus, for example, a flash or a beep sound can be used. The trigger signal 12 is set to have a relation with an external stimulus, for example, as shown in FIG. As the trigger signal 12, in addition to using a signal related to an external stimulus, it is also possible to use a value corresponding to a subject's spontaneous reaction due to blinking or a switch held by the subject.

Referring to FIG. 1, in the present embodiment, data storage unit 2 receives biological information data 11 and trigger signal 12 and, for example, determines a predetermined value from the time when trigger signal 11 rises from 0V to 5V. The time series of the biological information data 11 in the time section that has been set is stored. Further, it is also possible to accumulate the time series of the biological information data from a point in time before the trigger signal 11 rises to 5V or a point in time after the trigger signal 11 is shifted.

The data storage section 2 outputs data 21 relating to the averaging to be averaged from the stored biological information data in response to the data selection signal 41. As the data storage unit 2, for example, a magnetic disk device, a semiconductor memory device, or the like can be used.

The averaging unit 3 inputs data 21 relating to averaging and outputs averaging waveform data 31. If the data 21 regarding the averaging is a time series of n pieces of biological information data to be added, the averaging waveform data 31 is calculated for each arrangement location. For example, FIG.
When the addition target data measured in response to the trigger signal shown in FIG. 3 is input as the data 21 relating to the averaging, an example of the averaging waveform data 31 is as shown in FIG. 6, for example.

[0026] The measured value at time t of the i-th addition target data measured at the position j v _i and ^j (t), arithmetic mean waveform data avg ^j position j (t), the following equation (2)
Given by

[0027]

(Equation 2)

Alternatively, it is also possible to perform the averaging by making the averaging unit 3 a mechanism for holding the previously calculated averaging data and the number of additions k up to the previous time. In this case, one piece of biological information data to be added may be used as the data 21 regarding the averaging.

Assuming that the previously calculated averaged data at time t at position j is p ^j (t) and one data to be added is s ^j (t), the averaged waveform data avg ^j ( t) is given by the following equation (3).

[0030]

(Equation 3)

As the averaging unit 3, for example, a personal computer or a workstation can be used.

The data selection unit 4 receives the analysis information 51 on the input data and outputs a result of determination as to whether the input data is data suitable for addition. For example, the feature amount of the input data exceeds a predetermined threshold, or
If the value is less than the threshold value, it is determined that the input data is not to be subjected to averaging. As the data selection unit 4, for example, a personal computer or a workstation can be used.

The input data analysis unit 5 stores the biological information data 1
1 to perform analysis, calculate the feature amount of the input data, and output analysis information 51 on the input data. As the input data analysis unit 5, for example, a personal computer, a workstation, or the like can be used.

Next, referring to the flowchart of FIG.
The basic operation of the present embodiment will be described.

First, the data storage unit 2 monitors the trigger signal 12 (step 101), and when the trigger signal is input, inputs and stores the biological information data (step 10).
2).

Next, the input data analysis unit 5 calculates the feature amount of the input data, and outputs analysis information 51 on the input data to the data selection unit 4 (step 103).

The data selection section 4 determines whether or not the stored biological information data is suitable for addition based on the analysis information 51 on the data (step 10).
4). If it is determined that the data is suitable for addition, the data storage unit 2 registers the newly stored data as an addition target (step 105). When the experiment is completed (step 106), the averaging unit 3 receives the data 21 about the averaging output from the data storage unit 2, calculates and outputs an averaging waveform using only data suitable for addition (step 107). ).

As shown in the flowchart of FIG. 8, every time the biometric information data determined to be suitable for addition is registered (step 105 in FIG. 8), an average waveform is calculated and output (step 105 in FIG. 8). 8 may be performed.

Next, an example of an input data analysis method and an averaging object determination method will be described. The following methods can be used alone or in combination.

(1) Method using frequency analysis : This method is a method in which frequency analysis is performed on input data using Fourier transform, and the obtained frequency spectrum is used as a feature amount of the input data. The Fourier transform itself is a known technique, and is described in, for example, Reference 1 (E. Oran Brigham, The Fast
Fourier Transform, Prentice-Hall, Inc., 1974).

As a method of judging the object of averaging, for example, 20
For example, if the frequency component of more than Hz exceeds a certain threshold, the frequency component is not added.

(2) Method using wavelet analysis :
In this method, input data is filtered using a wavelet transform, and the obtained parameters and filtering results are used as feature amounts. The wavelet transform itself is a known technique.
arles K. Chui, “An Introduction to Wavelets”, Aca
demic Press, 1992), but here I will explain the method a little in detail.

Assuming that the value of the waveform measured at time t is represented by χ (t), the expansion coefficient α (a, b) of the wavelet transform for the waveform χ (t) is given by the following equation (4). Can be

[0044]

(Equation 4)

Here, ψ _{a, b} (t) is given by the following equation (4-a), and ψ (t) is a basis function of expansion called a basic wavelet.

[0046]

(Equation 5)

This basic wavelet is localized in a region centered on the origin, and its Fourier component is also localized with a width around one point in the frequency space. The parameter a specifies the expansion and contraction of the basic wavelet function in the time direction, and corresponds to the frequency of the Fourier transform. The parameter b does not correspond to the Fourier transform, but specifies parallel movement in the time direction.

For example, a signal of a high frequency component is important in a waveform near a certain time t ₁ , but a low frequency component is important at another time t ₂ , and a high frequency component may be regarded as noise. If this data is filtered using Fourier transform to remove high frequency component noise, information near time t ₁ will be lost.

On the other hand, in the filtering using the wavelet transform, filtering can be performed for different frequency bands at each time, so that noise can be effectively removed without losing important signals. Becomes Regarding the filtering using the wavelet transform, for example, reference 3 (Oliver Bertrand et al.,
“Time-Frequency Digital Filtering Based on an Inv
ertible Wavelet Transform: An Application to Evoke
d Potentials ”, IEEE TRANSACTIONS ON BIOMEDICAL EN
GINEERING, VOL. 41, NO. 1, JANUARY 1994).

As a method of judging the object of addition, for example, there is a method of making the object to be added when the waveform amplitude of the filtering result exceeds a certain threshold value, or when the parameter is within a predetermined range. .

(3) Method Using Template Collation : This method is a method in which input data is collated with a template prepared in advance, and the degree of collation is used as a feature amount. Let the time series of the template be c (t) and the input data be v
Assuming that (t), the collation degree e can be calculated as the sum of the square errors between the time series v _i (t) and c _i (t), for example, as in the following equation (5).

[0052]

(Equation 6)

Here, L represents the length of the template.

As a method of calculating the degree of collation, a neural network can be used. In template matching using a neural network, for example, a hierarchical network (input layer, intermediate layer 1,
2 and an output layer) or a recurrent type network as shown in FIG. First, prepare one or more time series representing templates,
Learning is performed by correcting the coupling coefficient in the network.
For example, first, an index τ representing goodness as a template is assigned to each template. As an assignment method, for example, there is a subjective method. Each template and τ
Is used as a teacher signal, the value of each template is given to the input layer, and the corresponding value of τ is given to the output layer to correct the coupling coefficient in the neural network. As a method of correcting the coupling coefficient, for example, an error back propagation learning method or the like is used. A specific correction method using the backpropagation learning method is described in, for example, Reference 4 (DERumelhart et al., “Parallel
Distributed Processing, vol. 1 ”, MIT press, pp.
318-362, 1986).

Next, when input data is given to the input layer of the learned neural network, the collation degree of the input data is output from the output layer.

As a method of judging the averaging target, for example, there is a method of excluding the averaging target from the addition target when the collation degree exceeds a certain threshold.

(4) Method using covariance : This method is a method of calculating a variance-covariance matrix with respect to an arrangement point of input data and using the value as a feature value. For example, if the input data at the time t measured at the arrangement point k is v ^k (t), the element s _ij of the variance-covariance matrix is represented by the following equation (6).

[0058]

(Equation 7)

Here, L represents the length of the input data, and v ^k} represents the average value of the input data at the arrangement point k.

As a method of judging the averaging target, for example, a correlation coefficient r _ij between the arrangement points i and j is calculated from a variance-covariance matrix, and when the value exceeds a certain threshold value, the addition target is calculated. There is a method of excluding from. The correlation coefficient r _ij is
It is given by the following equation (7).

[0061]

(Equation 8)

By using the above-described apparatus, even if there is input data containing noise, it can be efficiently excluded, and a good induced waveform can be calculated.

[0063]

Embodiment 2 FIG. 2 is a block diagram showing a configuration of a second embodiment of the present invention. 2, the same elements as those in FIG. 1 are denoted by the same reference numerals. In the following, description of portions common to the present embodiment and the first embodiment will be omitted, and differences will be described.

The data selection section 4 receives the analysis information 61 on the averaged waveform data and outputs a result of determination as to whether or not the input data is data suitable for addition. For example, when the validity of the averaging waveform exceeds a predetermined threshold or is less than the threshold, it is determined that input data added to the averaging is excluded from the averaging target. As the data selection unit 4, for example, a personal computer or a workstation can be used.

The effectiveness calculating section 6 calculates the averaged waveform data 3
1 to perform analysis, calculate the validity of the averaged waveform data, and obtain analysis information 6 on the averaged waveform data.
Outputs 1. For example, a personal computer or a workstation can be used as the effectiveness calculating unit 5.

Next, the basic operation of this embodiment will be described with reference to the flowchart of FIG.

First, the data storage unit 2 monitors the trigger signal 12 (step 201), and when the trigger signal is input, inputs and stores the biological information data (step 20).
2). The averaging unit 3 receives the data 21 about the averaging output from the data storage unit 2, temporarily adds the newly input data to the addition target, calculates and outputs an averaging waveform (step 203).

Next, the validity calculating section 6 calculates the validity of the averaged waveform, and outputs the analysis information 61 relating to the averaged waveform data to the data selecting section 4 (step 204).

The data selection section 4 determines whether the newly stored biological information data is suitable for addition based on the analysis information 61 on the averaged waveform data (step 205). If it is determined that the data is suitable for addition, the data storage unit 2 registers the newly stored data as an addition target (step 206).

When the experiment is completed (step 207), an added waveform is calculated from the input data to be added (step 208), and the output is completed.

FIG. 11 shows a processing method for determining the effectiveness of averaging while conducting an experiment. As a process 1, input data is accumulated during the experiment, and after the experiment is completed. As a process 2, an averaging is performed from the individual accumulated data, and the effectiveness of the averaging waveform is calculated.
Needless to say, the processing may be performed by a step process) method.

Next, an example of a method of calculating the effectiveness of the averaged waveform and a method of judging the average of the average will be described. The following methods can be used alone or in combination. Although these methods have the same basic principle as the input data analysis method described in the first embodiment, a detailed description thereof will be omitted to avoid duplication.

(1) Method Using Frequency Analysis : This method has the same operation principle as that of the input data analysis method in the first embodiment, except that the input data is replaced with the averaging waveform data.

(2) Method using wavelet analysis :
This method has the same operating principle as that of the input data analysis method in the first embodiment, in which the input data is replaced with the averaged waveform data.

(3) Method using template collation : This method has the same operation principle as that of the input data analysis method in the first embodiment, except that the input data is replaced with the average waveform data.

(4) Method using covariance :
In the method of analyzing input data in the first embodiment, the operation principle is the same as that in which the input data is read as averaging waveform data.

(5) Method of comparing the previous tendency : The validity obtained by the above method is stored, and the validity calculated for new input data is, for example, the stored validity. This is a method in which input data having a degree higher than the degree is added.

By using the above-described apparatus, even if there is input data containing noise, it can be efficiently excluded, and a good induced waveform can be calculated.

[0079]

Embodiment 3 FIG. 3 is a block diagram showing the configuration of a third embodiment of the present invention. 3, the same elements as those in FIGS. 1 and 2 are denoted by the same reference numerals. This embodiment includes both the input data analysis unit 5 in the first embodiment and the effectiveness calculation unit 6 in the second embodiment. The synergistic operation and effect of the operation and effect of the example are achieved. Note that the description of each configuration is the same as that described in each of the above embodiments, and thus description thereof will be omitted.

[0080]

As described above, according to the present invention,
Excellent by judging whether the observed biological signal is data suitable for averaging, excluding input data including noise, etc., and calculating the evoked waveform only with the appropriate data. Thus, it is possible to output an evoked waveform.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a second exemplary embodiment of the present invention.

FIG. 3 is a block diagram illustrating a configuration of a third exemplary embodiment of the present invention.

FIG. 4 is a diagram for explaining an embodiment of the present invention, and is a diagram illustrating an example showing a relationship between an external stimulus and a trigger signal.

FIG. 5 is a diagram for explaining the embodiment of the present invention, and is a diagram schematically illustrating an example of a signal related to averaging.

FIG. 6 is a diagram for explaining an embodiment of the present invention, and is a diagram schematically illustrating an example of averaged waveform data.

FIG. 7 is a flowchart illustrating an example of an operation according to the first exemplary embodiment of the present invention.

FIG. 8 is a flowchart illustrating an example of an operation according to the first exemplary embodiment of the present invention.

FIG. 9 is a diagram for explaining an embodiment of the present invention, and is a diagram schematically illustrating an example of a configuration of a neural network.

FIG. 10 is a diagram for explaining an embodiment of the present invention;
It is a figure which shows an example of a structure of a recurrent type network typically.

FIG. 11 is a flowchart illustrating an example of an operation according to the second exemplary embodiment of the present invention.

FIG. 12 is a block diagram showing a configuration of a related art induced waveform calculation device.

[Explanation of symbols]

 Reference Signs List 2 data storage unit 3 averaging unit 4 data selection unit 5 input data analysis unit 6 validity calculation unit 11 biological information data 12 trigger signal 21 data on addition averaging 31 averaged waveform data 41 judgment information on data selection 51 analysis on input data Information 61 Analysis information on averaging waveform data

Claims (11)

(57) [Claims] 1. A trigger signal and biological information data such as an electroencephalogram, a brain magnetic field, and a myoelectric potential measured at at least one location on the surface of a living body are inputted, and an evoked waveform or an event-related waveform corresponding to the trigger signal is input. In the evoked waveform calculating apparatus for obtaining the biological information data and the trigger signal as inputs, accumulate the biological information data of a predetermined time width from the time according to the trigger signal, to the determination information regarding data selection In response, accumulation means for outputting data about the averaging, data for the averaging, input, and averaging means for calculating an average value for each time and for each measurement position, and inputting the biometric information data. Input data analysis means for calculating the characteristic amount of the data, and the averaging result from the averaging means,
An effectiveness calculating means for calculating the effectiveness, and a feature amount of the input data and / or a result of the averaging result.
An evoked waveform calculation device, comprising: data selection means for determining whether or not the input biological information data is data suitable for addition from the validity . 2. Inputting biological information data and a trigger signal,
Accumulating means for accumulating biological information data of a predetermined time width from a time point corresponding to the trigger signal and outputting data relating to averaging in accordance with determination information relating to data selection; inputting data relating to averaging, For each time and each measurement position, an averaging means for calculating an average value, an averaging result is input, and an effectiveness calculating means for calculating the effectiveness thereof. And a data selecting means for determining whether or not the data is suitable for addition. 3. The evoked waveform calculation device according to claim 1, wherein a frequency analysis device is used as said input data analysis means. 4. The evoked waveform calculation device according to claim 1, wherein a wavelet analysis device is used as said input data analysis means. 5. The evoked waveform calculation apparatus according to claim 1, wherein a template matching device is used as said input data analysis means. 6. The evoked waveform calculator according to claim 1, wherein a covariance calculator is used as said input data analysis means. Wherein said as an effective measure calculation means, induced waveform computing apparatus according to claim 1 or 2, wherein the using frequency analyzer. Wherein said as an effective measure calculation means, induced waveform computing apparatus according to claim 1, wherein the use of wavelet analysis device. Wherein said as an effective measure calculation means, induced waveform computing apparatus according to claim 1, wherein the use of a template matching apparatus. Wherein said as an effective measure calculation means, induced waveform computing apparatus according to claim 1, wherein the use of covariance calculation unit. 11. An accumulating means for accumulating input data for a predetermined period and outputting data relating to averaging, said accumulating means.
That outputs the result of averaging from the data output from
A signal analyzing processing system comprising an average means, a predetermined feature amount for said input data, and / or, prior to
Means for calculating the effectiveness of the serial averaging result, it is determined whether or not data to which the input data is suitable for addition on the basis of the calculated value, the the data suitable for addition
Signal analyzing processing system characterized by comprising a means for selecting control such averaging process at the output to the averaging means from the storing means is made.