JP2006110234A - Method for processing brain activity information data, and apparatus thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To propose a method for processing brain activity information data without using the conventional averaging method. <P>SOLUTION: The method for processing the brain activity information data to process brain activity information data which is observed at least one place of a living body includes: a first step for analyzing the brain activity information data by temporal frequency, developing the data in a temporal frequency area, and acquiring a temporal frequency analysis result in each time frequency; a second step for dividing the temporal frequency area where the brain activity information data is developed, into a plurality of small areas having prescribed shapes and sizes; and a third step for obtaining the value of a statistic amount expressing relation between variable numbers on the basis of the temporal frequency analysis result, with respect to the small areas, as regards presence/no presence of electroencephalographic activity generation, estimating the pattern of a electroencephalographic activity related to a problem or a stimulation from the obtained value of the statistic amount, and detecting the electroencephalographic activity. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a brain activity information data processing method and apparatus, and more particularly, to detect brain activity information data without using a conventionally known addition averaging method and classify the data based on a pattern of brain activity. The present invention relates to a brain activity information data processing method and apparatus capable of performing the same.

  In this specification, brain activity information refers to a living body such as an electroencephalogram (electric potential fluctuation obtained by measuring brain activity as electric activity) or a brain magnetic field (magnetic field fluctuation obtained by measuring brain activity as magnetic activity). It means information observed at at least one location.

  Conventionally, it is known that when a stimulus or task is given to a subject, a potential is generated in each part of the brain related to information processing for that stimulus or task. The potentials evoked by the presentation of such stimuli and tasks are called “evoked responses”, “evoked potentials”, or “event-related potentials”, but these evoked responses are generally observed as electroencephalographic activity, It is used for elucidating information processing mechanisms and diagnosing brain functions.

  In general, the electroencephalogram is observed as a comprehensive activity of many nerve cells. In other words, in addition to the evoked response, the background electroencephalogram (the electroencephalogram due to the autonomous brain activity unrelated to the presentation of the stimulus or the task) is also simultaneously observed as the electroencephalogram activity.

By the way, conventionally, since the evoked response is smaller than the background electroencephalogram, it has been considered difficult to detect the evoked response from observation data of one electroencephalogram activity. Conventionally, it has been assumed that the latency and waveform of the evoked response are always constant when a stimulus or task is given under the same conditions.

  Against this background, as a method to extract only the evoked response from the EEG activity, the stimulus and task are presented many times under the same conditions, and the EEG activity obtained in each trial is added based on the stimulus presentation time or the response time of the subject. Then, a method called an averaging method is proposed, in which only the components of the evoked response waveform are obtained by attenuating randomly generated background brain wave components by obtaining the average waveform.

  FIG. 1 shows a conceptual diagram of this averaging method. According to the averaging method, when the stimulus presentation time is set as a reference time, components of the evoked response waveform appear after the reference time (FIG. 1). On the other hand, when the response time of the subject is the reference time, the component of the evoked response waveform appears before the reference time ((B) of FIG. 1).

However, the above-mentioned addition averaging method assumes that it is premised, that is, “when the stimulus or task is given under the same conditions, the latency and waveform of the evoked response are always constant”. There was a problem that the assumption was not realistic.

  In other words, the actual brain has an internal state that changes every moment. Therefore, because attention and concentration on stimuli and tasks cannot be kept constant, and repeated application of the same stimuli and tasks can affect habituation and fatigue, evoked responses even when stimuli and tasks are given under the same conditions The latency and waveform are not always constant.

  For example, FIG. 2 shows the top of the head when the visual stimulus is presented (specifically, when the picture presented to the subject is changed instantaneously as shown in FIG. 3). The electroencephalogram waveform accompanying the electroencephalogram activity for 6 trials measured from (/ Pz position of the 20 method) is shown. In addition, the relationship between the site | part name which shows the position of the electrode in an international 10/20 method, and an electrode symbol is as showing in FIG.

  When FIG. 2 is observed, it can be seen that there are trials in which an α wave (electroencephalogram activity in the vicinity of 10 Hz) is emitted and trials in which it is not emitted 0.25 seconds after the stimulus is presented, even though the same stimulus is given. That is, regarding the α wave, since the same stimulus is repeatedly given as previously assumed, the assumption that it always appears with the same latency and waveform does not hold.

In addition, another problem with the averaging method is that the interrelationship of activities in each part of the brain is not taken into account.

  For example, in FIG. 5, when a visual stimulus is presented (specifically, when the picture presented to the subject is changed instantaneously as shown in FIG. 3), Although the electroencephalogram waveform associated with the electroencephalogram activity measured from the position of Pz in 10/20 method) is shown, as shown in FIG. 5, when an α wave appears at the top of the head 0.25 seconds after the stimulus is presented, it is slightly delayed. The phenomenon that the activity of the θ wave (activity near 5 Hz) appears in the frontal region (see (A) of FIG. 5), and the α wave does not appear in the frontal region unless the α wave appears in the top of the head. There is a phenomenon (see FIG. 5B), but the conventional averaging method does not consider the correlation of the activity between each part of the brain at all. Could be inferior.

The magnetic signal recorded by the brain magnetic field measurement is a measurement of the magnetic signal caused by the potential change generated by the brain, and the electroencephalogram data and the brain magnetic field data basically reflect the same brain activity. The nature of the signal is very similar.

For this reason, there is a problem similar to that in the above-described electroencephalogram processing even in the processing of the brain magnetic field.

Japanese Patent Laid-Open No. 10-80409 “Estimated Waveform Calculation Device” Japanese Unexamined Patent Publication No. 2000-126148 “EEG Data Processing Device and Recording Medium” Nozomi Hoshimiya, Masahiro Ishii, Kei Tsukada, Hideto Ide “Biomedical Engineering”, Morikita Publishing Co., Ltd. 132-139, ISBN 4-627-83210-9

  The present invention has been made in view of the above-described various problems of the prior art, and an object of the present invention is to provide a brain activity information data processing method that does not use the conventional averaging method and its The device is to be provided.

  Another object of the present invention is to provide a brain activity information data processing method and apparatus for improving the accuracy of signal detection by using the interrelation between the activities of each part of the brain. .

  Furthermore, the purpose of the present invention is to perform a certain problem such as a recognition problem or discrimination problem such as a presented visual stimulus, auditory stimulus or somatic sensory stimulus, or a psychological problem such as memory or mental arithmetic. It is intended to provide a brain activity information data processing method and apparatus for detecting active brain activity (evoked response) from brain activity information such as an electroencephalogram measured electrically or a magnetic field measured magnetically. .

  Still another object of the present invention is to provide a brain activity information data processing method and apparatus for classifying detected brain activity (evoked response) for each brain activity pattern.

  In order to achieve the above-mentioned object, the present invention is generated in relation to a problem by utilizing the following properties-1 to 4 (see FIG. 5) of brain activity information generated in relation to the problem. In this way, the detected brain activity is detected.

  As described above, since the electroencephalogram and the electroencephalogram have similar properties, in this “Means for Solving the Problems” section, an electroencephalogram is described as an example of brain activity information. To do.

  (1) Property-1: The electroencephalogram signal related to the task has sufficient intensity to be detected from a single trial waveform (waveform for each trial that is not averaged).

  (2) Property-2: Even if the same task is repeatedly given, the activity pattern of the electroencephalogram is not always the same. This is because the electroencephalogram activity pattern is greatly influenced not only by the type of task but also by the internal state of the brain (for example, the degree of attention, familiarity with the task, fatigue, etc.).

  (3) Property-3: Activity of each part of the brain interacts.

  (4) Property-4: The activity of each part of the brain occurs transiently (the generation period is short).

  FIG. 6 is an explanatory diagram showing the nature of the electroencephalogram activity generated in relation to the task. FIG. 6A is an explanatory diagram showing the electroencephalogram activity pattern in the time frequency domain. ) Is an explanatory diagram in the case of including the relationship between the activities of each part of the brain in addition to time and frequency.

  In FIG. 6A, the horizontal axis represents time and the vertical axis represents frequency. Moreover, the figure of (A) of FIG. 6 respond | corresponds when the test subject's reaction time is taken as a reference | standard. If the reference is the stimulus presentation time, the brain wave activity related to the task appears after the reference.

  Referring to FIG. 6A, the presence of a plurality of activity patterns corresponds to property-2. If activity a appears, activity b appears next, and if activity c appears, activity next The appearance of d corresponds to property-3, and the short period during which activity is occurring corresponds to property-4.

  Further, FIG. 6B is an explanatory diagram in the case where the relationship between the activities of each part of the brain is included in addition to time and frequency as described above. That is, in the diagram of FIG. 6B, in addition to the time axis and the frequency axis, another axis is shown, and the brain wave activity pattern is shown in a three-dimensional space. In this case, the added one axis corresponds to the electrode position as shown in FIG. 4 (for example, 0, 1, 2,... In order from the frontal pole portion). Further, in each time frequency region, there is an activity as shown in FIG. 6A, and in addition there is an activity having an interaction between a plurality of electrodes.

Here, the above-mentioned properties -1 to 4 of the electroencephalogram activity were discovered for the first time in the experiments of the present inventors. For property-1, it is easy to determine the presence or absence of an α wave after 0.25 seconds of stimulus presentation in FIG. 2, and for property-2, an α wave appears in 0.25 seconds of stimulus presentation in FIG. In FIG. 5, when an α wave appears at the top of the head, a θ wave appears at the frontal region, and when no α wave appears, no property wave appears. 4 and 4 are experimental grounds that the period of activity in FIGS. 2 and 5 is as short as several hundred milliseconds.

  This invention is made | formed in view of the phenomenon shown by above-mentioned property -1-4 which this inventor discovered.

Next, a method for detecting the electroencephalogram activity as shown in FIG. 6 will be described. Specifically, such detection can be performed by, for example, the following procedures-1 to -3. Note that FIG. 7 will be referred to in the description of the following procedures-1 to 3. This FIG. 7 shows an explanatory diagram of a method for detecting the electroencephalogram activity generated in connection with the problem. (A) is an explanatory diagram in the case of the time frequency domain, and (B) in FIG. 7 is an explanatory diagram in the case of including the relationship of the activity between each part of the brain in addition to the time and the frequency.

  (1) Procedure-1: The electroencephalogram data measured by a known technique is developed on the time frequency domain by continuous wavelet transform, and the amplitude of the signal at each time frequency is obtained. When the electroencephalogram data is recorded from one place on the scalp, the signal is represented on a two-dimensional plane of a time axis and a frequency axis as shown in FIG. On the other hand, when recording is performed from a plurality of locations using a plurality of measurement electrodes, an axis corresponding to the electrode position is added as shown in FIG. 6B, and the electroencephalogram activity is expressed in a three-dimensional space. The numbers 0, 2,... On the axis corresponding to the electrode positions may be arbitrarily assigned to each electrode. In addition, when analyzing using the phase synchronization index representing the relationship of the activity between the electroencephalograms recorded by the two measurement electrodes instead of the amplitude of the signal, on the axis corresponding to the electrode position of FIG. A number is assigned to each set of two measuring electrodes.

  In the above-described procedure-1, continuous wavelet transform is used to develop the measured electroencephalogram data on the time-frequency domain. However, the present invention is not limited to this, and time such as short-section Fourier transform is used. Any frequency analysis technique can be used instead of the continuous wavelet transform.

  In addition, the information obtained as a result of the time-frequency analysis is two pieces of information, that is, the amplitude and the phase of the signal. In the above procedure-1, the amplitude of the signal has been described. Is applicable. Further, since the power of the signal is obtained by squaring the amplitude, the procedure-1 can also be applied when analyzing the power of the signal.

  (2) Procedure-2: As shown in FIGS. 7A and 7B, the time-frequency region is divided into a plurality of small regions having a predetermined shape and size.

  As the shape of the small region, for example, an arbitrary shape such as a rectangular shape such as a rectangle or a square, another polygonal shape such as a triangle or a pentagon, or a shape partitioned by a curve can be appropriately selected. . In the example shown in FIG. 7, each small region is formed in a rectangular shape having the same shape in consideration of simplification of the subsequent processing.

  The size of the small region may be set to an arbitrary size as long as it is smaller than the size of the electroencephalogram activity on the time frequency region.

  In addition, since the magnitude of the electroencephalogram activity varies depending on the subject and the subject to be presented to the subject, in actual analysis, actually observe the waveform of the electroencephalogram on the time frequency domain as shown in FIG. It is preferable to determine the size of.

  (3) Procedure-3: As for the presence / absence of EEG activity occurrence, a mutual information amount, which is a statistic representing the relationship between variables, is obtained for each small area created in Procedure-2, and the pattern of EEG activity is estimated. To detect EEG activity. Specifically, for example, the estimation of the pattern of the electroencephalogram activity related to the task or the stimulus from the obtained statistic value is performed as follows. First, the pseudo presentation time (or response time) is set to be the same as the number of stimuli (or subject responses) that were actually presented, regardless of the presentation time of the task or the response time of the subject. Similarly, the mutual information amount is obtained as a reference. When this operation is performed a plurality of times (about 200 to 2000 times), it is possible to estimate the probability distribution of the values taken by the mutual information when the electroencephalogram activity is not related to the stimulus or the task. The distribution at this pseudo presentation time (or reaction time) is compared with the mutual information value when the stimulus is actually presented, and the value is statistically significant (for example, determined with a risk factor of 0.5%). )), It is considered that the brain wave activity related to the position of the two areas for which the mutual information amount is obtained is generated by the stimulation or the presentation of the task.

  In the above procedure-3, the mutual information is used as a statistic used in the analysis. However, the present invention is not limited to this, and if it represents the relationship between variables, it is replaced with the mutual information. A correlation coefficient or a ratio between occurrence probability and conditional probability may be used. Note that the mutual information amount is obtained by taking the logarithm of the ratio between the occurrence probability and the conditional probability.

As described above, in Procedures 1 to 3, the time-frequency region is divided into a plurality of small regions having a predetermined shape (for example, a rectangle) and a size, and the electroencephalogram activity between the plurality of small regions. For the relationship between occurrence and non-occurrence, the occurrence probability p (y) is compared with the conditional probability p (y | x), or a statistic indicating the relationship between variables such as mutual information l (x, y) is used. For example, the brain wave activity is detected by estimating the pattern of the brain wave activity (see FIG. 7). FIG. 7A shows a case where brain wave activity is detected by dividing into small areas on a two-dimensional plane in the time frequency domain, and FIG. 7B shows each part of the brain in addition to time and frequency. It shows a case where the EEG activity is detected by dividing it into small areas on a three-dimensional space including the relationship of the activities between them. When taking into account the relationship of activity between each part of the brain as shown in FIG. 7B, the statistic is obtained across a plurality of electrodes.

That is, the invention according to claim 1 of the present invention is a brain activity information data processing method for processing brain activity information data observed in at least one place of a living body, wherein a plurality of time frequency regions related to brain activity information data are set. Divide into multiple small areas with a predetermined shape and size, and based on the relationship between the presence or absence of brain activity information data between the multiple small areas, estimate the pattern of brain wave activity related to tasks and stimuli In this way, EEG activity is detected.

  According to a second aspect of the present invention, there is provided a brain activity information data processing method for processing brain activity information data observed in at least one place of a living body, and performing time-frequency analysis on the brain activity information data. A first step of expanding the frequency domain to obtain a time frequency analysis result at each time frequency, and the time frequency domain in which the brain activity information data is expanded into a plurality of small areas having a predetermined shape and size. Regarding the second step of dividing and the presence / absence of electroencephalogram activity occurrence, a value of a statistic representing a relationship between variables is obtained between the plurality of small regions based on the time frequency analysis result, and the obtained statistic And a third step of detecting the electroencephalogram activity by estimating the electroencephalogram activity pattern related to the task or the stimulus from the value of.

  The invention according to claim 3 of the present invention is the invention according to claim 2 of the present invention, wherein the statistic representing the relationship between variables in the third step is a mutual information amount, a correlation. It is the ratio of the number or occurrence probability to the conditional probability.

  According to a fourth aspect of the present invention, in the invention according to any one of the first, second, and third aspects of the present invention, the plurality of small regions are rectangles having the same size. It is designed to have a shape.

  According to a fifth aspect of the present invention, in the brain activity information data processing apparatus for processing the brain activity information data observed at at least one place of the living body, the brain activity information data is subjected to time-frequency analysis and time is analyzed. Means for expanding the frequency domain and obtaining a time frequency analysis result at each time frequency; and means for dividing the time frequency domain in which the brain activity information data is expanded into a plurality of small areas having a predetermined shape and size. In addition, regarding the presence or absence of the occurrence of electroencephalogram activity, a statistic value representing the relationship between the variables is obtained between the plurality of small regions based on the time frequency analysis result, and a problem or stimulus is determined from the obtained statistic value. And a means for detecting the electroencephalogram activity by estimating the pattern of the electroencephalogram activity related to.

  According to a sixth aspect of the present invention, in the brain activity information data processing apparatus for processing brain activity information data observed in at least one place of a living body, brain activity information data for inputting brain activity information data is provided. Input means, task performance time data input means for inputting time data relating to task performance, area division parameter input means for inputting parameters for dividing the time-frequency domain into small areas of a predetermined shape and size, and electroencephalogram An electroencephalogram activity occurrence determination threshold value input means for inputting a threshold value for determining whether or not an activity has occurred, and the brain activity information data input from the brain activity information data input means are expanded on a time-frequency region, Time frequency analysis means for obtaining a time frequency analysis result at the time frequency, parameters input from the region division parameter input means, and Based on the threshold value input from the electroencephalogram activity occurrence determination threshold value input means, using the time frequency analysis result on the time frequency domain developed by the time frequency analysis means, the parameter input from the region division parameter input means A contingency table creating means for creating a contingency table for the generation of electroencephalogram activity in each small region indicated by A statistic calculating means for calculating the value of the statistic, a probability distribution creating means for creating a probability distribution of values taken by the statistic based on the pseudo data, a probability distribution created by the probability distribution creating means, and the statistic computing means The calculated statistic value is compared, and the comparison result indicates that the statistic value calculated by the statistic calculating unit is the probability distribution generated by the probability distribution generating unit. If it is statistically significant, it is considered that the brain wave activities related to the positions of the plurality of small areas for which the above-mentioned statistic values are obtained are generated by the presentation of the task or the stimulus, An activity pattern estimation means for estimating that the positional relationship indicates an activity pattern of the electroencephalogram activity; an activity pattern of the electroencephalogram activity estimated by the activity pattern estimation means; and a time frequency analysis result obtained by the time frequency analysis means. On the basis of this, a detection classification means for detecting and classifying the activity pattern of the electroencephalogram activity is provided.

  Further, the invention according to claim 7 of the present invention is the invention according to any one of claims 5 or 6 of the present invention, wherein the statistic is a mutual information amount, a correlation coefficient or an occurrence probability. And the conditional probability.

  The invention according to claim 8 of the present invention is the invention according to any one of claims 5, 6 or 7 of the present invention, wherein the plurality of small regions are rectangles having the same size. It is designed to have a shape.

  Since the present invention is configured as described above, the following excellent effects can be obtained.

  That is, the present invention detects EEG activity related to a stimulus or a task without using the averaging method, so that when the stimulus or the task is given under the same conditions, the latency and waveform of the evoked response are always constant. The assumption that there is no need. Therefore, even if multiple types of electroencephalographic activity are related to the performance of the same task, and they occur according to the internal state such as the subject's attention and fatigue, it is possible to detect the activity of each pattern individually. There is an excellent effect.

  Hereinafter, an example of an embodiment of a brain activity information data processing method and apparatus according to the present invention will be described in detail with reference to the accompanying drawings.

First, FIG. 8 is a block diagram showing the overall system configuration of a brain activity information data processing apparatus according to the present invention that detects and classifies brain wave activities related to a task.

  The brain activity information data processing apparatus shown in FIG. 8 is constructed by, for example, a computer system, and is constituted by each processing means indicated by reference numerals 1-13. The processing means indicated by reference numerals 1 to 4 constitute input means, the processing means indicated by reference numerals 5 to 12 constitute analysis means, and the processing means indicated by reference numeral 13 constitutes output means. Details of the functions and roles of the processing units indicated by reference numerals 1 to 13 will be described below.

  The processing means indicated by reference numeral 1 is an electroencephalogram data input unit, and performs a process of inputting electroencephalogram data. The electroencephalogram data input unit 1 includes, for example, a recording electrode, a differential amplifier, a band-pass filter, an AD converter, and the like. A process of inputting the statistical value distribution map creation unit 10 based on data (hereinafter simply referred to as “creation unit 10” as appropriate) is performed.

  Here, when the electroencephalogram data is recorded from a plurality of locations instead of one location on the scalp, the electroencephalogram data for the measurement location used for analysis is input.

  The processing means indicated by reference numeral 2 is a stimulus presentation time / subject response time data input unit, and time data relating to task performance such as stimulus and task presentation time, subject response time, etc. are time frequency analysis unit 5 and creation unit 10. Process to input to.

  The processing means indicated by reference numeral 3 is an area division parameter input unit, and divides the time frequency area into small areas (for example, rectangular small areas) having a predetermined shape and size, and brain activity information data A user of the processing apparatus sets parameters for the size and range of each small region as follows, which are referred to as a “partition table creation unit 7” (hereinafter simply referred to as “partition table creation unit 7”) and a creation unit 10 for generation of electroencephalogram activity. It is a device for inputting to. At the time of this input, the size of the small region may be set to an arbitrary size as long as it is smaller than the size of the electroencephalogram activity on the time frequency region. Since the magnitude of the electroencephalogram activity varies depending on the subject and the subject presented to the subject, in actual analysis, actually observe the waveform of the electroencephalogram on the time frequency domain as shown in FIG. 2 to determine the size of the small area. It is preferable to determine.

  The processing means indicated by reference numeral 4 is an electroencephalogram activity occurrence determination threshold value input unit, and a threshold value for the user of the brain activity information data processing device to determine whether or not an electroencephalogram activity has occurred is generated by the split table creation unit 7 and the creation This is an apparatus for inputting to the unit 10. Here, the threshold value input by the brain wave activity occurrence determination threshold value input unit 4 is, for example, when the time frequency analysis unit 5 analyzes the signal power and amplitude by focusing on the average power and average amplitude values of all data. input. When analysis is performed using a phase synchronization index that represents the relationship between the activities of electroencephalograms recorded by two measurement electrodes, the phase synchronization index takes a value from 0 to 1, and when the value is close to 0, two The electroencephalogram activity represents a random activity, and when the value is close to 1, it represents that the two electroencephalogram activities are operating in synchronization, so that a threshold for determining whether or not the synchronous activity has occurred Is set to an intermediate value of 0.6.

  In this embodiment, an intermediate value of 0.6 is set as a threshold value for determining the presence or absence of the occurrence of synchronous activity, but the present invention is not limited to this. Any intermediate value between about 4 and 0.6 may be set.

  The processing means indicated by reference numeral 5 is a time-frequency analysis unit, which develops the electroencephalogram data input from the electroencephalogram data input unit 1 on the time-frequency domain by continuous wavelet transform, and uses the amplitude of the signal at each time frequency as waveform data. It is calculated and transferred to a waveform data storage unit 6 (hereinafter simply referred to as “waveform data storage unit 6” as appropriate) and a creation unit 10 in the time-frequency domain described later. In addition, when analyzing the activity relationship between the electroencephalograms recorded by the two measurement electrodes, the phase synchronization index in the time frequency domain is calculated as waveform data.

  The processing means indicated by reference numeral 6 is a waveform data storage unit on the time frequency domain, and on the time frequency domain transferred from the time frequency analysis unit 5 until the calculation in the activity pattern estimation unit 11 described later is completed. Store the waveform data in the storage area. When the calculation in the activity pattern estimation unit 11 is completed, the waveform data storage unit 6 converts the stored waveform data into an EEG activity detection / classification unit 12 (hereinafter simply referred to as “electroencephalogram activity”) related to a problem to be described later. It is appropriately referred to as a detection / classification unit 12 ”.

  The processing means indicated by reference numeral 7 is a contingency table creation unit for the generation of electroencephalogram activity, and based on the parameters and threshold values input from the region segmentation parameter input unit 3 and the electroencephalogram activity generation determination threshold value input unit 4, a time frequency analysis unit Using the waveform data in the time-frequency region transferred from 5, a contingency table for generating the electroencephalogram activity in each set region is created. The contingency table here means a table (see FIG. 9) that summarizes the frequencies of combinations of categories of a plurality of qualitative variables generally used in multivariate analysis and the like.

  Note that the contingency table shown in FIG. 9 is used when analyzing two regions, region-x and region-y. When analyzing three or more regions simultaneously, a contingency table corresponding to three or more variables is used. Note that n in the contingency table shown in FIG. 9 represents the number of applied stimuli or the number of times the subject has reacted.

  Here, whether or not the electroencephalogram activity is generated in each small area is determined by determining the average value of the waveform data (signal amplitude, synchronization index, etc.) obtained by the time-frequency analysis unit 5 in the small area or a representative value similar thereto. And the magnitude relationship between the threshold value input from the brain wave activity occurrence determination threshold value input unit 4 and the comparator in the contingency table creation unit 7 are compared. The created contingency table is stored in a storage device in the contingency table creation unit 7.

  The processing means indicated by reference numeral 8 is a statistic calculation unit, which takes out a necessary partition table from the storage device in the partition table creation unit 7 and is mutual information that is a statistic representing a relationship between each small region regarding the occurrence of brain wave activity. Calculate quantity. The calculation result is transferred to a statistical value storage unit 9 (hereinafter simply referred to as “statistical value storage unit 9” as appropriate).

  The processing means indicated by reference numeral 9 is a statistic value storage unit, and stores the statistic value transferred from the statistic calculation unit 8 in the storage area until the calculation in the creation unit 10 is completed. When the calculation in the creation unit 10 is completed, the statistic value storage unit 9 transfers the stored statistic value to the activity pattern estimation unit 11 described later.

  The processing means indicated by reference numeral 10 is a section for creating a distribution diagram of statistical values based on pseudo data, and is a pseudo presentation that is completely independent of the presentation time of the task and the response time of the subject using a random number generator. It is composed of a device for creating time (or reaction time), a device having the same functions as the time frequency analysis unit 5, the contingency table creation unit 7 and the statistic calculation unit 8. Here, the pseudo presentation time (or reaction time) is set at random as many as the number of stimuli (or subject responses) actually presented input from the stimulus presentation time / subject response time data input unit 2. Using the pseudo time as a reference, the value of the statistic is obtained in the same manner as the actual data using a device having the same functions as the time frequency analysis unit 5, the contingency table creation unit 7 and the statistic calculation unit 8. When this operation is performed a plurality of times (about 200 to 2000 times), it is possible to create a probability distribution of values taken by the statistic when the electroencephalogram activity is not related to the stimulus or the task. Then, the creation unit 10 transfers the probability distribution thus created to the activity pattern estimation unit 11.

  The processing means indicated by reference numeral 11 is an activity pattern estimation unit, and compares the probability distribution transferred from the creation unit 10 with the value of the statistic about the actual data transferred from the statistic value storage unit 9. Then, the activity pattern estimation unit 11 indicates that the value from the statistic value storage unit 9 is statistically significant in the probability distribution (for example, determination is made with a risk factor of 0.5%) based on the comparison result. If there is, it is considered that the electroencephalogram activities related to the positions of the plurality of areas for which the statistics are obtained are generated by the presentation of the stimulus or the task. The positional relationship between the plurality of regions indicates the activity pattern of the electroencephalogram activity. The activity pattern of the electroencephalogram activity detected here is stored in the storage device of the activity pattern estimation unit 11.

  The processing means indicated by reference numeral 12 is a detection / classification unit for brain wave activity related to the task, and the activity pattern of the brain wave activity stored in the storage device in the activity pattern estimation unit 11 and the storage device in the waveform data storage unit 6. Is used to detect and classify which activity pattern of electroencephalogram activity appears in which trial using the waveform data in the time-frequency domain stored in. This process is performed using, for example, a template matching method which is one of pattern classification methods. Then, the result of the detection and classification of the electroencephalogram activity as the processing result of the detection / classification unit 12 is transferred to the result output unit 13.

  The processing means indicated by reference numeral 13 is a result output unit, and is constituted by, for example, a printer device or a display device. As a result, the result of the detection and classification of the electroencephalogram activity, which is the processing result of the detection / classification unit 12, is output from the output unit 13 to a form or a screen.

Next, the operation of the brain activity information data processing apparatus according to the present invention will be described with reference to the flowchart shown in FIG.

  That is, this flowchart shows a case where the present invention is applied to actual electroencephalogram data. In an electroencephalogram activity measurement experiment, as shown in FIG. Present the time and instruct the subject to press the button at hand whenever the apparent orientation of the cube changes. The time when the subject presses the button is regarded as the subject's reaction time.

  In the following description, the relationship of the brain wave synchronization activity between the frontal region and the parietal region related to the change in the apparent orientation of the cube will be described as an example.

When the operation of the brain activity information data processing apparatus is started, first, the process of step S1002 is performed. In the process of step S1002, two electroencephalogram data measured for 12 minutes from the frontal region and the top of the head and the subject's reaction time data (the number of responses is 95 times in 12 minutes) and the electroencephalogram data input unit 1 and the stimulus presentation, respectively Input to the time / subject response time data input unit 2. The data input to the electroencephalogram data input unit 1 and the stimulus presentation time / subject response time data input unit 2 are transmitted to the time frequency analysis unit 5 and the electroencephalogram data input unit 1 and the stimulus presentation time / subject response time data input unit 2, respectively. It is transferred to the creation unit 10.

  When the process of step S1002 described above is completed, the process proceeds to step S1004, and the time-frequency analysis unit 5 obtains the front-head-top phase synchronization index on the time-frequency domain. FIG. 12 shows an example of a waveform of the phase synchronization index on the time frequency domain, and in FIG. 12, time 0.0 corresponds to the response time of the subject. Further, the synchronization index takes a value between 0 and 1, and the index takes a value close to 1 or 1 when the frontal-parietal brain wave activity is synchronized, and close to 0 or 0 otherwise. Takes a value. The waveform data in the time frequency domain thus determined by the time frequency analysis unit 5 is transferred to the waveform data storage unit 6 and the contingency table creation unit 7.

  When the process of step S1004 is completed, the process proceeds to step S1006, and the time-frequency region is divided into small regions having a predetermined shape and size (in this embodiment, the shape is a rectangular shape). The parameters for the size and range of each small region are input to the region segmentation parameter input unit 3 and a threshold value for determining whether or not an electroencephalogram activity has occurred is input to the electroencephalogram activity occurrence determination threshold input unit 4. Here, the size of the small region may be set to an arbitrary size as long as it is smaller than the size of the electroencephalogram activity on the time frequency region. For example, the waveform data shown in FIG. 12 is extracted from the storage device of the waveform data storage unit 6 to examine the magnitude of the electroencephalogram activity. In this embodiment, the small area is a rectangle of 0.025 seconds × 1.25 Hz.

  In addition, it is expected that EEG activity related to the task appears before the subject's reaction time, and it is known that EEG activity in the 30 Hz to 40 Hz band is related to object recognition. In the form, the range shown in FIG. 12 is divided into small areas. The number of small regions is 720, and the threshold value for determining whether or not the electroencephalogram activity occurs is set to 0.6.

  When the process of step S1006 described above is completed, the process proceeds to step S1008, and the contingency table creation unit 7 checks the relationship between each region regarding the occurrence of brain wave activity. In this embodiment, since the relationship between the two areas has been examined, the contingency table shown in FIG. 9 is created for all combinations of two small area groups selected from a total of 720.

  When the process in step S1008 is completed, the process proceeds to step S1010. In order to quantitatively investigate the relationship between the two areas, the contingency table stored in the storage device of the contingency table creating unit 7 is used. Mutual information, which is a statistic representing the relationship between variables, was obtained.

  FIG. 13 shows an example in which mutual information between two small areas is determined for the generation of electroencephalogram activity, which is the mutual information between the white small area pointed to by the triangle and the other areas. The amount is shown. The darker the color, the stronger the association between the subregion and the white subregion pointed to by the triangle with respect to the occurrence of electroencephalogram activity.

  The statistics obtained by the contingency table creation unit 7 are transferred to and stored in the storage area of the statistics calculation unit 8.

  When the process in step S1010 is completed, the process proceeds to step S1012. In the creation unit 10, data about a pseudo reaction time that is completely unrelated to the reaction time of the subject is first created using a random number generator. To do. Here, the number of reaction times in the pseudo data is matched with the number of actual reaction times input from the stimulus presentation time / subject response time data input unit 2. Next, based on this pseudo reaction time data, the time frequency analysis unit 5, the contingency table creation unit 7 and the statistic calculation unit 8 were used in the creation unit 10 having the same functions, and were obtained in step S1010. Find the same statistic value as. This operation is repeated 1024 times to create a probability distribution of values taken by the statistics when the electroencephalogram activity is not related to the task. The probability distribution thus created is transferred to the activity pattern estimation unit 11.

  When the process of step S1012 described above is completed, the process proceeds to the process of step S1014. In the activity pattern estimation unit 11, the value of the statistic and the electroencephalogram activity respectively transferred from the statistic storage unit 9 and the creation unit 10 are related to the task. The brain wave activity pattern is estimated using the probability distribution of the values taken by the statistics when not. For example, the mutual information value of the black small area near (−0.4 seconds, 25 Hz) in FIG. 13B near (−0.2 seconds, 40 Hz) in FIG. Significance can be detected at a risk factor of 0.5% (values are 0.1% and 0.2%, respectively). Therefore, when an electroencephalogram activity is generated in association with the task in the small white area pointed to by the triangle in FIG. 13, in the case of FIG. 13A, in the vicinity of (−0.2 seconds, 40 Hz), FIG. In the case of, it can be seen that electroencephalogram activity continues to occur in the vicinity of (−0.4 seconds, 25 Hz). Therefore, it is presumed that there are two types of electroencephalogram activity patterns shown in FIG. The estimated activity pattern is transferred to the detection / classification unit 12.

  When the process of step S1014 is completed, the process proceeds to step S1016, and the detection / classification unit 12 uses data on the electroencephalogram activity pattern transferred from the activity pattern estimation unit 11 as one of pattern classification methods. A certain template matching method is used to detect the electroencephalogram activity generated in association with the task and classify the pattern from the waveform data in the time frequency domain transferred from the waveform data storage unit 6. FIG. 15 shows an example of template matching. This is because one of the waveform data transferred from the waveform data storage unit 6 (the waveform data shown in FIG. 12) is the activity pattern of the electroencephalogram activity transferred from the activity pattern estimation unit 11 (the activity pattern shown in FIG. 14). ). In FIG. 15, since the value of the synchronization index at the position corresponding to pattern-a is large, it can be seen that in this trial, the electroencephalogram activity characterized by pattern-a occurred in association with the task. The detection / classification result of the electroencephalogram activity in the detection / classification unit 12 is transferred to the result output unit 13.

  When the process of step S1016 is completed, the process proceeds to step S1018, and the result output unit 13 outputs the result transferred from the detection / classification unit 12 to a form or a screen.

Here, in order to show the superiority of the present invention, an application example of the conventional method will be described with reference to FIG.

  That is, FIG. 16 shows the result of applying the conventional method (addition averaging method) assuming that the latency and waveform of the evoked response are always constant when a stimulus or task is given under the same conditions. In addition, this FIG. 16 is the addition average waveform calculated | required on the basis of the test subject's reaction time.

  If the above conventional assumptions are correct, the synchronization index should show a value close to 1 at the position where the brain wave activity related to the task exists, but the values of the synchronization index are all smaller than 1. EEG activity cannot be detected.

  As described in the above-mentioned “Background Art” section and the above-described embodiment of the present invention, even if the same problem is given, the same EEG activity does not always appear. It is considered that the assumption is not valid, and therefore the electroencephalogram activity cannot be detected by the conventional method (additive averaging method).

As described above, unlike the prior art, the present invention can detect the electroencephalogram activity that could not be detected so far by taking into account the characteristics of the electroencephalogram at the time of signal detection.

  Since the present invention has an excellent ability to detect an electroencephalogram signal as compared with a method using a conventional method (addition averaging method), an examination using an electroencephalogram measurement that can be easily used in a medical field by using the present invention. It becomes possible to construct a system and a diagnostic system. For example, if an inspection / diagnosis system built using the present invention is incorporated into a small electroencephalograph and distributed to the patient, the analysis result of the electroencephalogram data measured by the patient at home can be sent to the hospital using the Internet. Advanced home medical care and remote medical care will be realized.

  In addition, the fMRI and CT scans that are currently commonly used in hospital examinations and diagnosis are for examining brain active sites and tumor sites. The examination / diagnosis system constructed using the present invention If it becomes possible to diagnose from the activity of the electroencephalogram instead of the part by using this, it will be possible to carry out accurate examination / diagnosis with a small and small examination / diagnosis system compared with the current system using fMRI or CT scan. become.

The embodiment described above can be modified as shown in the following (1) to (4).

  (1) In the above-described embodiment, the case where brain waves are processed as brain activity information using the brain activity information data processing apparatus according to the present invention has been described. However, the present invention is not limited to this. . That is, as described above, the magnetic signal recorded by the magnetoencephalogram measurement is a measurement of the magnetic signal caused by the potential change generated by the brain, and the electroencephalogram data and the magnetoencephalogram data are basically the same brain activity. Therefore, the present invention can also be applied to the case where a brain magnetic field is processed as brain activity information.

  (2) The shape and size (dimensions) described in the above-described embodiment, for example, the shape and size of a small region are not limited to the description in the above-described embodiment, and may be changed as appropriate. Of course it is good.

  (3) In the above-described embodiment, the time-frequency analysis unit 5 uses continuous wavelet transform to develop the electroencephalogram data on the time-frequency domain. However, the present invention is not limited to this. Yes, any time-frequency analysis technique such as short interval Fourier transform can be used instead of continuous wavelet transform. The time frequency analysis results are two pieces of information, ie, the amplitude and phase of the signal. The time frequency analysis unit 5 has described the amplitude of the signal. Can be applied similarly. Further, since the power of the signal is obtained by squaring the amplitude, the time frequency analysis unit 5 can also be applied when analyzing the power of the signal.

  (4) You may make it combine suitably the embodiment shown above and the modification shown in said (1) thru | or (3).

  The present invention is expected to be applied to clinical research in the field of medical treatment and welfare or basic research using brain waves such as brain science and biotechnology. Specifically, the present invention is used as a diagnostic device using brain wave measurement. be able to.

FIG. 1 shows a conceptual diagram of the addition averaging method, (A) shows a case where addition averaging is performed based on the stimulus presentation time, and (B) shows a case where addition averaging is performed based on the reaction time of the subject. FIG. 2 is an electroencephalogram waveform diagram associated with electroencephalogram activity for six trials measured from the top of the brain activity generated by the presentation of the visual stimulus (Pz position in the international 10/20 method). FIG. 3 is an explanatory diagram showing the stimulus given to the subject. FIG. 4 is an explanatory diagram showing the relationship between the part name indicating the position of the electrode and the electrode symbol in the international 10/20 method. FIG. 5 shows an electroencephalogram waveform associated with electroencephalogram activity measured from the top of the brain activity (position of Pz in the international 10/20 method) generated by the presentation of a visual stimulus, and (A) shows an α wave at the top of the head. (B) shows the average waveform of trials in which no alpha wave appears at the top of the head. 6A and 6B are explanatory diagrams showing the nature of the electroencephalogram activity generated in relation to the task. FIG. 6A is an explanatory diagram showing an electroencephalogram activity pattern in the time-frequency domain, and FIG. It is explanatory drawing at the time of including the relationship of the activity between each site | part of a brain in addition to a frequency. FIG. 7 is an explanatory diagram for a method of detecting an electroencephalogram activity generated in association with a task, (A) is an explanatory diagram in the case of a time frequency domain, and (B) is in addition to time and frequency. It is explanatory drawing at the time of including the relationship of the activity between each site | part of a brain. FIG. 8 is a block diagram showing the overall system configuration of a brain activity information data processing apparatus according to the present invention that detects and classifies brain wave activities related to a task. FIG. 9 is an explanatory diagram of a contingency table created by the contingency table creation unit for the generation of electroencephalogram activity. FIG. 10 is a flowchart showing the operation of the brain activity information data processing apparatus according to the present invention. FIG. 11 is an explanatory diagram showing an experiment subject in an electroencephalogram activity measurement experiment. FIG. 12 is a waveform diagram showing an example of a waveform in the time-frequency domain of the phase synchronization index between the frontal region and the parietal region. FIG. 13 is an explanatory diagram showing an example in which the mutual information amount between two small regions is obtained with respect to the generation of electroencephalogram activity. FIG. 14 is an explanatory diagram showing an electroencephalogram activity pattern estimated in step S1014. FIG. 15 is an explanatory diagram illustrating an example of detecting an electroencephalogram activity pattern related to a task. FIG. 16 is an explanatory diagram showing a result of applying the conventional method (addition averaging method).

Explanation of symbols

1 EEG data input unit 2 Stimulus presentation time / subject response time data input unit 3 Domain division parameter input unit 4 EEG activity generation determination threshold value input unit 5 Time frequency analysis unit 6 Waveform data storage unit in time frequency domain 7 EEG activity generation Contingency table creation unit 8 Statistical calculation unit 9 Statistical value storage unit 10 Statistical value distribution map creation unit based on pseudo data 11 Activity pattern estimation unit 12 Detection of brain wave activity related to task Classification part 13 Result output part

Claims (8)

In a brain activity information data processing method for processing brain activity information data observed in at least one part of a living body,
Dividing the time-frequency region related to the brain activity information data into a plurality of small regions having a plurality of predetermined shapes and sizes;
Brain activity information data processing characterized in that brain wave activity is detected by estimating a pattern of brain wave activity related to a task or a stimulus based on the relationship between presence or absence of brain activity information data among the plurality of small regions Method. In a brain activity information data processing method for processing brain activity information data observed in at least one part of a living body,
A first step of analyzing the brain activity information data by time-frequency analysis and expanding it on a time-frequency region, and obtaining a time-frequency analysis result at each time frequency;
A second step of dividing the time-frequency region in which the brain activity information data is developed into a plurality of small regions having a predetermined shape and size;
Regarding the presence or absence of electroencephalogram activity, a value of a statistic representing a relationship between variables is obtained between the plurality of small regions based on the time-frequency analysis result, and related to a task or a stimulus from the obtained statistic value And a third step of detecting an electroencephalogram activity by estimating a pattern of the electroencephalogram activity performed. The brain activity information data processing method according to claim 2,
The brain activity information data processing method, wherein the statistic representing the relationship between variables in the third step is a mutual information amount, a correlation coefficient, or a ratio between an occurrence probability and a conditional probability. In the brain activity information data processing method according to any one of claims 1, 2, or 3,
The plurality of small regions have a rectangular shape having the same size. The brain activity information data processing method, wherein: In a brain activity information data processing device for processing brain activity information data observed at at least one location of a living body,
A means of analyzing the brain activity information data by time-frequency analysis and expanding it on the time-frequency domain, and obtaining a time-frequency analysis result at each time frequency;
Means for dividing the time-frequency region in which the brain activity information data is developed into a plurality of small regions having a predetermined shape and size;
Regarding the presence or absence of electroencephalogram activity, a value of a statistic representing a relationship between variables is obtained between the plurality of small regions based on the time-frequency analysis result, and related to a task or a stimulus from the obtained statistic value A brain activity information data processing apparatus comprising: means for estimating a brain wave activity pattern and detecting brain wave activity. In a brain activity information data processing device for processing brain activity information data observed at at least one location of a living body,
Brain activity information data input means for inputting brain activity information data;
Task performance time data input means for inputting time data related to task performance;
Area division parameter input means for inputting parameters for dividing the time-frequency area into small areas of a predetermined shape and size;
EEG activity occurrence determination threshold value input means for inputting a threshold value for determining whether or not EEG activity has occurred;
Time frequency analysis means for expanding the brain activity information data input from the brain activity information data input means on a time frequency region and obtaining a time frequency analysis result at each time frequency;
Based on the parameter input from the region division parameter input means and the threshold value input from the electroencephalogram activity occurrence determination threshold value input means, using the time frequency analysis result on the time frequency domain developed by the time frequency analysis means. , A contingency table creating means for creating a contingency table for generating an electroencephalogram activity in each small region indicated by the parameter input from the region segmentation parameter input means;
A statistic calculator that calculates a value of a statistic indicating a relationship between each small region for the electroencephalogram activity generation indicated by the contingency table created by the contingency table creating means;
A probability distribution creating means for creating a probability distribution of values taken by a statistic based on pseudo data;
The probability distribution created by the probability distribution creating means is compared with the statistic value computed by the statistic computing means, and the comparison result is the statistical value computed by the statistic computing means. When the statistical distribution is statistically significant in the probability distribution created by the means, it is considered that the brain wave activities related to each other have occurred at the positions of the plurality of small areas for which the value of the statistic is obtained, Activity pattern estimation means for estimating that the positional relationship of the regions indicates an activity pattern of electroencephalogram activity;
Detecting and classifying means for detecting and classifying the activity pattern of the electroencephalogram activity based on the activity pattern of the electroencephalogram activity estimated by the activity pattern estimation means and the time frequency analysis result obtained by the time frequency analysis means. Characteristic brain activity information data processing device. In the brain activity information data processing device according to any one of claims 5 and 6,
The statistic is a mutual information amount, a correlation coefficient, or a ratio between an occurrence probability and a conditional probability. In the brain activity information data processing device according to any one of claims 5, 6 and 7,
The plurality of small regions have a rectangular shape having the same size. The brain activity information data processing device.