JP3911627B2 - EEG signal separation display method and apparatus - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an electroencephalogram signal separation display method and apparatus for separating and outputting each element signal included in an original electroencephalogram signal.
[0002]
[Prior art]
The original signal of the electroencephalogram signal obtained by measuring the brain action potential from the scalp includes a biological signal useful for knowing the brain activity, and can be easily measured. In general, a plurality of signal components based on brain activity are superimposed on the original signal of the electroencephalogram signal, and signal components accompanying artifacts other than brain activity, that is, artifacts are mixed.
[0003]
In a conventional clinical diagnosis using an electroencephalogram signal, an experienced doctor (electroencephalogram interpreter) with many years of experience supervises multiple components based on brain activity and signal components associated with biological activities other than brain activity. The original signal had to be separated and analyzed in the head and diagnosed based on years of experience and skill.
[0004]
Artifacts are bioelectric signals generated by eye movements, heart beats, and the like, and have a large amplitude and a sharp rise on the time axis, so the frequency band is wide and cannot be removed by a frequency filter or the like. For this reason, in recent years, the inventors have developed a method for separating and removing artifacts from the original signal of the electroencephalogram signal, which is called a waveform shape grasping method. The waveform shape grasping method creates a normalized waveform of the artifact waveform, creates an estimated artifact waveform from the normalized waveform and the original signal of the electroencephalogram signal based on the similarity coefficient method or the denormalization method, and This is a method of subtracting and removing the estimated artifact waveform from the original signal (artifact separation method based on eye movement is based on literature: Medical Electronics and Biotechnology. 38-1, 33/41, 2000, based on heartbeat. (Refer to Japanese Patent Application No. 2000-351125 for the method of separating artifacts).
[0005]
However, simply removing artifacts is not sufficient for clinical diagnosis using electroencephalogram signals.
That is, a doctor who is skilled in clinical diagnosis using an electroencephalogram signal makes a diagnosis based on a diagnostic procedure established based on years of experience and skill. The diagnostic procedure is, in other words, the identification of the artifact signal and the electroencephalogram signal, the waveform of the electroencephalogram signal composed of a plurality of specific frequency bands, and the correlation between these electroencephalogram signals, or the electroencephalogram signal and the specific artifact. This is based on the correlation with the signal. The specific frequency band may be a physiologically explainable band, or may be a physiologically unexplained band obtained by skill and experience.
Therefore, in order for an inexperienced doctor to be able to perform a diagnosis equivalent to a doctor with sufficient experience, a method and an apparatus capable of displaying information embodying the diagnostic procedure of a skilled doctor are required.
[0006]
The conventional EEG signal measurement method is to measure and display the original EEG signal signal, which is composed of multiple components based on brain activity and signal components associated with biological activity other than brain activity, or in the original EEG signal signal. Only a signal in a specific frequency band included in the signal is extracted by a frequency filter or the like.
For this reason, only skilled doctors can diagnose from the original signal, inexperienced doctors cannot diagnose, or when only a specific frequency component is extracted, only a part of the diagnostic procedure is displayed and recorded Therefore, there is a problem that even a skilled doctor cannot make a sufficient diagnosis.
In recent years, an independent component analysis method (ICA) is known as a method for measuring an original electroencephalogram signal separately for each signal source element. In this method, the independent component signals that constitute the signal under measurement are identified by mathematical means from the signal under measurement on the time axis, and each element signal is separated and measured and displayed as an uncorrelated signal. is there. However, the component signals separated by this method generally do not match the physiologically meaningful component signals that clinicians need for diagnosis.
[0007]
Thus, the conventional electroencephalogram diagnosis method and apparatus can measure and display only the original signal of an electroencephalogram signal in which a plurality of components based on brain activity and signal components associated with biological activities other than brain activity are superimposed, and a doctor's diagnosis There are issues such as that only a small part of the multiple component signals necessary for the procedure can be measured and displayed, or that it cannot be separated into physiologically meaningful component signals that clinicians need for diagnosis. As a result, there was a problem that an inexperienced doctor could not make a diagnosis equivalent to a doctor who had sufficient experience.
[0008]
[Problems to be solved by the invention]
In view of the above problems, the present invention provides an electroencephalogram diagnosis method and apparatus that can be separated into physiologically meaningful element signals necessary for diagnosis by a clinician and can display these signals in parallel on a time axis. With the goal.
[0009]
[Means for Solving the Problems]
In order to achieve the above object, the method for separating and displaying an electroencephalogram signal according to the present invention converts the original signal of the electroencephalogram signal into a plurality of element signals resulting from brain activity and biological activities other than brain activity that can physiologically identify the cause of occurrence. The separated element signals are separated, and these separated element signals are simultaneously displayed and recorded.
According to this method, since all the electroencephalogram element signals necessary for physiologically meaningful electroencephalogram diagnosis are separated and displayed at the same time, the waveform of the electroencephalogram element signal and the correlation between these electroencephalogram element signals can be determined. It is possible to know and avoid mistakenly diagnosing elemental signals caused by biological activities other than brain activity as brain waves, and inexperienced doctors can make a diagnosis equivalent to a doctor with sufficient experience .
[0010]
In addition, the method for separating and displaying an electroencephalogram signal according to the present invention includes a plurality of element signals resulting from brain activity, and an element signal resulting from biological activity other than brain activity that can physiologically identify the cause of generation. A plurality of element signals caused by brain activity and element signals necessary for diagnosis among element signals caused by biological activities other than brain activity are simultaneously displayed and recorded.
According to this method, all the electroencephalogram element signals necessary for physiologically meaningful electroencephalogram diagnosis are separated and displayed, and the element signals resulting from biological activities other than brain activity that can identify the cause of physiological occurrence, that is, Since the specific artifact is also separated and displayed at the same time, the correlation between the electroencephalogram signal necessary for the diagnosis and the specific artifact signal can be known.
[0011]
Further, the electroencephalogram signal separation display device of the present invention is caused by brain activity from an electroencephalogram signal from which an artifact separation unit that separates an element signal caused by biological activity other than brain activity from an original signal of the electroencephalogram signal, and an electroencephalogram signal from which the artifact is separated. And a plurality of electroencephalogram separation units for separating the plurality of element signals, and the separated element signals are displayed simultaneously.
The artifact separation unit includes a filter that performs signal processing on the original EEG signal input to the separation unit based on a waveform shape grasping method, and an adder that performs signal processing by subtracting the output from the input of the filter. It is software or hardware.
In addition, the waveform shape grasping method creates a normalized waveform of the artifact waveform, creates an estimated artifact waveform from this normalized waveform and the original signal of the electroencephalogram signal based on the similarity coefficient method or the denormalization method, It is characterized by subtracting the estimated artifact waveform from the original EEG signal.
Preferably, the electroencephalogram separation unit is software or hardware including a filter that performs signal processing for frequency band extraction and an adder that performs signal processing for subtracting the output from the input of the filter.
The electroencephalogram separation unit may be software or hardware that performs signal processing of Fourier transform / inverse Fourier transform.
According to this apparatus, the artifact separation unit operates based on a waveform shape grasping method (reference: medical electronics and biotechnology. See 38-1, 33/41, 2000, Japanese Patent Application No. 2000-351125). And an artifact having a sharp rise on the time axis can be displayed separately for each source.
Further, since the operation of these separation units can be performed by an algorithm configured by computer software, it is only necessary to install this software in an existing digital electroencephalograph. In the case of an existing analog electroencephalograph, a computer equipped with an A / D and D / A conversion device may be added.
Thus, according to the apparatus of the present invention, the electroencephalogram signal separation and display method of the present invention can be realized at low cost.
[0012]
According to the method and apparatus for separating and displaying an electroencephalogram signal of the present invention having the above-described configuration, it is possible to visually grasp an operation equivalent to a doctor who has sufficient experience. Therefore, it is possible to prevent a diagnosis error of an inexperienced doctor or clinical laboratory technician, and to reduce the burden on the doctor when diagnosing an electroencephalogram.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. The same numbers in the drawings indicate the same or equivalent.
FIG. 1 is a diagram showing the configuration of an electroencephalogram signal separation display method and apparatus according to the present invention.
The brain wave signal separation display method and apparatus of FIG. 1 inputs an original signal 1 of an electroencephalogram signal recorded from above the scalp to a filter group 2, and an element signal included in the original signal of the electroencephalogram signal by a filter in the filter group 2. 31, 32, 3N, and residual signal 3E are separated and output in parallel.
[0014]
The original signal 1 of the electroencephalogram signal is generally composed of a plurality of element wave components. If the original signal is U (t) and each element wave signal is Si (t), U (t) = S ₁ (t) + S ₂ ( t) +... + S _N (t) + S _E (t).
Here, the element wave signals S ₁ (t), S ₂ (t), S ₃ (t),..., S _N (t) are electric signals (artifacts) and generations associated with biological movements other than brain waves, respectively. The origins correspond to different electroencephalogram signals, each of which is a signal unit separable from a physiological standpoint, and S _E (t) is a residual signal of the other components. A filter considering the characteristics of each element wave signal S _i (t) included in the original signal is configured, and signal separation is realized by extracting the element wave signal S _i (t) from the original signal.
[0015]
FIG. 2 is a diagram showing a configuration example of the filter group of the present invention.
Here the filter 41, 42, 43 ,,,, 4N, each element wave signal _{S 1 (t), S 2} (t), S 3 (t) ,,,, S N for extracting a (t) It is a filter. The filter 41 is a filter for extracting an artifact described later, and the element wave signal S ₁ (t) is a signal composed of an artifact component. The original signal 1 (U (t)) is input to the filter 41 and the adder 51, the output signal S ₁ (t) of the filter 41 as the element signal 31 is input to the adder 51, the difference is calculated, and the signal U ₁ (t) is obtained.
[0016]
Expressed as a general form, the signal U _i-1 (t) is input to the filter 4i and the adder 5i, the output signal S _i (t) of the filter 4i as the element signal 3i is input to the adder 5i, and the difference (U _i-1 (t) -S _i (t)) is calculated to obtain a signal U _i (t). However, i is 1 to N, U ₀ (t) = U (t), and U _N (t) = S _E (t). Each element wave signal S ₁ (t), S ₂ (t), S ₃ (t),,, S _N (t) and S _E (t) separated from the original signal is an element signal that is a time-series signal. 31, 32, 33,..., 3N, and output as a residual signal 3E.
[0017]
Each of the filters 41, 42, 43,..., 4N is not limited in configuration, and may be a nonlinear filter. Moreover, although each filter 42, 43, ..., 4N in FIG. 2 is set as the structure connected in series, it is also possible to arrange | position in parallel.
[0018]
Next, an artifact separation filter will be described.
In general, artifacts included in a measured original electroencephalogram signal interfere with diagnosis in clinical diagnosis using an electroencephalogram, and thus must be separated from the original signal. There are various artifacts included in the original electroencephalogram signal, such as those caused by eye movement, those caused by electrocardiogram, those caused by body movement, and those caused by muscle activity. By constructing a filter for separating each of these various artifacts, it is possible to accurately extract the electroencephalogram component itself. Therefore, in the filter 41 in FIG. 2, artifacts are first separated from the electroencephalogram signal, but by connecting filters for separating the respective artifacts in series as the filter 41, a plurality of artifacts can be simultaneously obtained. A separating filter can be constructed.
[0019]
The artifact is a waveform having a large amplitude and a sharp rising on the time axis, so that the frequency band is wide, and if it is removed by a frequency filter or the like, element signals useful for diagnosis are separated and cannot be used.
In such a case, if the waveform shape grasping method developed by the present inventors is used, it can be separated for each cause of occurrence of the artifact.
The filter for artifact separation according to the present invention uses a waveform shape grasping method. For example, the configuration of the filter 41 that extracts the artifact caused by eye movement, that is, the blink artifact component S ₁ (t), is described in the literature [Tsuyoshi Sugi, Masatoshi Nakamura, Akio Ikeda, Takashi Nagahama, Hiroshi Shibasaki: Normalized addition Real-time processing of blink artifact separation mixed in an electroencephalogram recording by averaging, medical electronics and biotechnology, 38-1, 33/41, 2000].
[0020]
Filters 42, 43,..., 4N for extracting the electroencephalogram element wave signals S ₂ (t), S ₃ (t),..., S _N (t) are filters for extracting a predetermined frequency band. For example, a primary or secondary low-pass filter, a band-pass filter, a high-pass filter, or the like is used. A typical example is a Butterworth filter. Furthermore, it is also possible to comprise an FIR filter (Finite Impulse Response Filter) having a linear phase characteristic.
It is also possible to extract a more accurate frequency band by using Fourier transform / inverse Fourier transform instead of each filter.
The band center wavelength and bandwidth of the filter can be arbitrarily set so that element signal waves that are sufficiently meaningful in clinical diagnosis can be extracted.
[0021]
FIG. 3 is a diagram showing a parallel configuration example of the filter of the present invention.
Here, the filter 41 is the same as the filter described in FIG. 2 and is a filter for extracting artifacts. The filters 42, 43,..., 4N are filters for extracting the element signals 32, 33, 3N from the signal from which artifacts have been removed. The original signal 1 (U (t)) is input to the filter 41 and the adder 51, the output signal S ₁ (t) of the filter 41 as the element signal 31 is input to the adder 51, the difference is calculated, and the artifact Separate to obtain the signal U ₁ (t). The signal U ₁ (t) is input to the filters 42, 43,..., 4N, and element signals 32, 33,.
[0022]
FIG. 4 is a diagram showing a configuration example of the Fourier transform / inverse Fourier transform apparatus of the present invention.
Here, the filter 41 is the same as the filter described in FIG. 2, and the Fourier transform / inverse Fourier transform device 7 serves as a filter for extracting the element signals 32, 33, 3N. The original signal 1 (U (t)) is input to the filter 41 and the adder 51, the output signal S ₁ (t) of the filter 41 as the element signal 31 is input to the adder 51, the difference is calculated, and the artifact Separate to obtain the signal U ₁ (t). The signal U ₁ (t) is stored in the memory 6, and the parallelized data 8 is input to the Fourier transform / inverse Fourier transform device 7. The Fourier transform / inverse Fourier transform device 7 performs Fourier transform on the parallelized data 8 and performs inverse Fourier transform on a predetermined band corresponding to each of the element signals 32, 33, 3N, and the residual signal 3E, respectively. Element signals 32, 33,..., 3N and a residual signal 3E are output.
[0023]
Next, an embodiment in which among blinking artifacts having the highest mixing frequency and large amplitude among the artifacts mixed in the original electroencephalogram signal, the brain wave component is separated into four element signals will be described.
FIG. 5 shows an implementation in which the original signal U (t) is decomposed into five element wave signals S ₁ (t), S ₂ (t), S ₃ (t), S ₄ (t), and S _E (t). It is a figure which shows an example.
S ₁ (t) is a blink artifact component caused by eye movement, S ₂ (t) is an α wave component (8 to 13 Hz), S ₃ (t) is a δ wave component (0.5 to ₄ Hz), S ₄ ( t) is a θ wave component (4 to 8 Hz), and S _E (t) is a high frequency component (13 Hz or more) of β wave or more.
[0024]
In clinical diagnosis, a doctor separates and captures an artifact in the original EEG signal and the EEG signal, and then interprets the EEG signal. The method of reading the electroencephalogram signal differs slightly depending on the doctor, but it is roughly divided into three bands, slow wave (δ wave, θ wave) band, α wave band, and fast wave (β wave) band. Is being read. According to the present invention, it is possible to take a filter configuration corresponding to the reading procedure in this clinical diagnosis of a doctor, and it is possible to separate and capture an electroencephalogram signal for each element. In FIG. 5, it can be seen that each element included in the original electroencephalogram signal is appropriately separated.
[0025]
In general, an electroencephalogram records a plurality of channels simultaneously in order to see the characteristics of the entire head. It is easy to apply the present invention to multi-channel processing, and it is possible to obtain the distribution on the scalp for each separated element, which is the same as what a doctor envisions in the head during inspection interpretation, Application to advanced clinical diagnosis becomes possible.
[0026]
Next, the influence of the type of filter that extracts the electroencephalogram element signal will be described.
FIG. 6 is a diagram showing a time series signal waveform and a periodogram of element signals when a first-order Butterworth bandpass filter is used.
FIG. 7 is a diagram showing time series signal waveforms and periodograms of element signals when a second-order Butterworth bandpass filter is used.
FIG. 8 is a diagram showing time series signal waveforms and periodograms of element signals when a Fourier / Fourier inverse transform device is used.
6-8, the left figure shows a time-series signal waveform, and is an original signal 1 of an electroencephalogram signal, element signals 32, 33, 34, 35, 36, 37, and a residual signal 3E in order from the top. Further, the right figure is a periodogram (frequency spectrum) corresponding to the time-series signal waveform shown in the left figure. The band configuration of the filter group 2 includes a filter 41 for removing blink artifacts, a filter 42 for extracting a low frequency component (˜0.5 Hz), and a δ wave component (0.5 To 43 Hz, a filter 44 to extract the θ wave component (4 to 8 Hz), a filter 45 to extract the α wave component (8 to 13 Hz), and a filter 46 to extract the β wave component (13 to 30 Hz). This is a filter 47 for extracting high-frequency components (30 Hz to).
[0027]
As can be seen from FIGS. 6 to 8, the waveform of the residual signal 3E changes depending on the type of filter to be processed, but the other element signals hardly change and can be used as element signals necessary for a doctor's clinical diagnosis.
[0028]
【The invention's effect】
As can be understood from the above description, according to the method and apparatus for separating and displaying an electroencephalogram signal of the present invention, it is possible to visually grasp a work equivalent to a doctor who has obtained sufficient experience, and therefore, inexperienced. A doctor can make a diagnosis equivalent to a doctor who has sufficient experience, can prevent a diagnostic error of a clinical laboratory technician, and can reduce the burden on the doctor at the time of diagnosis of an electroencephalogram.
[Brief description of the drawings]
FIG. 1 is a conceptual diagram showing an electroencephalogram signal separation table display method and apparatus according to the present invention.
FIG. 2 is a diagram illustrating a configuration example of a filter group of the present invention.
FIG. 3 is a diagram illustrating a parallel configuration example of a filter according to the present invention.
FIG. 4 is a diagram illustrating a configuration example of a Fourier transform / inverse Fourier transform apparatus according to the present invention.
FIG. 5 is a signal waveform diagram in which the electroencephalogram signal mixed with the blink artifact of the present invention is separated as N = 4 and output in parallel.
FIG. 6 is a diagram showing a time-series signal waveform and a periodogram using a first-order Butterworth filter according to an embodiment of the present invention.
FIG. 7 is a diagram showing a time-series signal waveform and a periodogram using a second-order Butterworth filter according to an embodiment of the present invention.
FIG. 8 is a diagram displaying a time-series signal waveform and a periodogram using the Fourier transform / inverse Fourier transform apparatus according to the embodiment of the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Original signal 2 Filter group 2a Artifact separation part 2b EEG separation part 31, 32, 3N, 3E Element signal 41, 42, 43, ... 4N filter 51, 52, 53, ..., 5N Adder 6 Memory 7 Fourier transform / inverse Fourier transform device 8 Parallelized data

Claims (7)

The original EEG signal is separated into multiple element signals resulting from brain activity and element signals originating from biological activities other than brain activity that can be identified physiologically, and these separated element signals are simultaneously A method of separating and displaying an electroencephalogram signal, comprising displaying and recording. The original signal of the electroencephalogram signal is separated into a plurality of element signals resulting from brain activity and an element signal resulting from biological activity other than brain activity that can physiologically identify the cause of occurrence, An electroencephalogram signal separation and display method characterized by simultaneously displaying and recording an element signal and an element signal necessary for diagnosis among element signals resulting from biological activities other than the brain activity. Artifact separation unit that separates element signals resulting from biological activities other than brain activity from the original EEG signal signal, and multiple brain wave separations that separate multiple element signals caused by brain activity from the above-mentioned brain wave signals from which artifacts have been separated An electroencephalogram signal separation display device, wherein the separated element signals are simultaneously displayed. The artifact separation unit includes a filter that performs signal processing on the original EEG signal input to the separation unit based on a waveform shape grasping method, and an adder that performs signal processing by subtracting the output from the input of the filter. The electroencephalogram signal separation display device according to claim 3, which is software or hardware. The waveform shape grasping method creates a normalized waveform of the artifact waveform, creates an estimated artifact waveform from the normalized waveform and the original signal of the electroencephalogram signal based on the similarity coefficient method or the denormalization method, 5. The electroencephalogram signal separation display device according to claim 4, wherein the estimated artifact waveform is subtracted from the original signal of the electroencephalogram signal for separation. The electroencephalogram separation unit is software or hardware including a filter that performs signal processing for frequency band extraction and an adder that performs signal processing for subtracting the output from the input of the filter. 4. The electroencephalogram signal separation display device according to 3. 4. The electroencephalogram signal separation display device according to claim 3, wherein the electroencephalogram separation unit is software or hardware that performs signal processing of Fourier transform / inverse Fourier transform.

Images