JP2017209267A - Apparatus, program, and method for diagnosis support of convulsion invagination type acute encephalopathy - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique that supports early diagnosis of the convulsion invagination type acute encephalopathy.SOLUTION: An apparatus (10) for diagnosis support of the convulsion invagination type acute encephalopathy has: a data acquisition unit (11) for acquiring time series data of brain waves measured in a predetermined part on a subject's head in a predetermined time after the convulsion invagination; a calculation unit (12) for calculating a power value of a frequency band previously decided to a predetermined part by performing frequency analysis doing acquired time series data; and a generation unit (13) for generating identification information of the convulsion invagination type acute encephalopathy or the febrile convulsion related to the subject using the calculated power value.SELECTED DRAWING: Figure 2

Description

  The present invention relates to an electroencephalogram analysis technique related to convulsive intussusception type acute encephalopathy.

  Among brain disorders caused by illnesses in children in Japan, the ratio of acute encephalopathy (hereinafter sometimes referred to as AE) is high. AE is an acute encephalopathy with biphasic seizures and late reduced diffusion (hereinafter sometimes referred to as AESD), acute necrotizing encephalopathy (hereinafter sometimes referred to as AE). Yes), HH syndrome (hemiconvulsion-hemiplegia syndrome), Reye syndrome (Reye-like syndrome), encephalitis and encephalopathy with reversible corpus callosal lesion (clinical mild encephalitis / encephalopathy with a reversible splenial lesion; hereinafter referred to as MERS) In some cases), hemorrhagic shock and encephalopathy syndrome (hereinafter sometimes referred to as HSES), and the like. AE is a disease that has a high possibility of remaining sequelae, and in particular, it is known that the residual rate of sequelae of AESD, ANE, and HSES is high (see Non-Patent Document 1).

  On the other hand, there is a technique for analyzing a subject's disease or sleep state by measuring an electroencephalogram. Patent Document 1 below proposes a method for determining whether a subject is a neurodegenerative disease such as Alzheimer's disease by providing specific digital content to the subject and measuring the brain wave of the subject at this time.

JP 2012-157732 A

Hoshino A, et al, “Epidemiology of acute encephalopathy in Japan, with emphasis on the association of viruses and syndrome”, Brain Dev 34: 337-343, 2012

  AESD patients are sent to the hospital in a convulsive state after fever. “Status epilepticus” means a state in which convulsive seizures continue or repeat beyond a certain amount of time (standard is said to be 30 minutes) without recovery of consciousness. On the other hand, among patients who are sent to the hospital in the state of convulsion, there are few residual sequelae, such as febrile seizures (hereinafter sometimes referred to as FS), acute encephalopathy Many patients are caused by other diseases. According to an epidemiological survey in Tottori Prefecture, there are 141 patients mentioned above in a certain 4 years, 12 of them are patients with acute encephalopathy and 70 are patients with febrile seizures Met. Eleven of the 141 people had sequelae and eight of the eleven were patients with acute encephalopathy.

  The present inventors have focused on the importance of determining whether or not the disease of a patient sent to a hospital in the state of convulsion is febrile seizure (FS) at an early stage. Patients who are sent to the hospital due to intussusception generally undergo MRI (Magnetic Resonance Imaging), blood tests, cerebrospinal fluid tests, and the like. However, it is difficult to determine whether or not the cause of the symptom is AESD only from the results of these tests. For example, AESD and FS have the same symptoms in that they become convulsions after fever. At present, doctors have to predict AESD or FS by measuring a patient's brain wave and visually diagnosing the measured brain wave waveform. Patients with AESD then appear abnormal on the MRI image.

  FIG. 1 is a graph showing the electroencephalogram waveforms of a patient with convulsion type acute encephalopathy with encephalopathy remaining as a sequela and a patient with febrile seizure with no sequelae. As shown in FIG. 1, it is difficult for a specialist to discriminate AESD or FS at an early stage in visual diagnosis of an electroencephalogram waveform. Furthermore, according to the visual diagnosis of the early electroencephalogram waveform, there is a possibility that the judgment will be different for each doctor.

  This invention is made | formed in view of such a situation, and provides the technique which assists the early diagnosis of the convulsion status-type acute encephalopathy.

  Each aspect of the present invention employs the following configurations in order to solve the above-described problems.

The first aspect relates to a diagnosis support apparatus for convulsion-type acute encephalopathy. A diagnostic support device for convulsive status acute encephalopathy according to the first aspect is a data acquisition means for acquiring time series data of electroencephalograms measured at a predetermined site on a subject's head within a predetermined time after convulsion And convulsion related to the subject using a frequency analysis of the acquired time-series data to calculate a power value of a predetermined frequency band for the predetermined part, and using the calculated power value Generating means for generating discrimination information for intussusception type acute encephalopathy or febrile seizure.
In the present specification, the specific time “within a predetermined time after the convulsions” is not limited. However, it is desirable that “within a predetermined time after convulsions” is a period (eg, within 72 hours) before the characteristics of convulsive acute encephalopathy appear in other test results such as MRI. During this period, it is difficult to discriminate by other inspections, and the effect of the present invention is increased. Hereinafter, “within a predetermined time of the convulsions type” may be referred to as “disease early stage”.

  The second aspect relates to a diagnostic support program for convulsion type acute encephalopathy executed by a computer. The diagnostic support program for convulsive status acute encephalopathy according to the second aspect is obtained by acquiring time series data of electroencephalograms measured at a predetermined site on the subject's head within a predetermined time after convulsion. By analyzing the frequency of the time-series data, a power value in a predetermined frequency band is calculated for the predetermined site, and using the calculated power value, convulsive acute encephalopathy or heat related to the subject is calculated. The computer is caused to generate seizure discrimination information.

  The third aspect relates to a diagnosis support apparatus for convulsion-type acute encephalopathy. The diagnostic support device for convulsion-type acute encephalopathy according to the first aspect acquires time-series data of electroencephalograms measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion. Data acquisition means; calculation means for calculating a plurality of correlation coefficients of at least one of a time correlation coefficient and a spatial correlation coefficient in the acquired time-series data; and frequency information of the calculated correlation coefficients for the subject's brain wave Distribution acquisition means for acquiring the feature amount of the subject, and generation means for generating discrimination information related to the convulsion type acute encephalopathy of the subject using the acquired feature amount.

  The fourth aspect relates to a diagnostic support program for convulsive status acute encephalopathy executed by a computer. The diagnostic support program for convulsion-type acute encephalopathy according to the second aspect acquires time-series data of electroencephalograms measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion. , Calculating a plurality of correlation coefficients of at least one of the time correlation coefficient and the spatial correlation coefficient in the acquired time series data, acquiring the frequency information of the calculated correlation coefficient as a feature quantity of the subject's brain wave, Using the acquired feature amount, the computer is caused to generate discrimination information related to the convulsive status encephalopathy of the subject.

  In addition, another aspect of the present invention is a diagnosis support method for convulsive acute encephalopathy executed by the diagnosis support apparatus according to the first or third aspect. In addition, a computer-readable recording medium that records the diagnosis support program according to the second or fourth aspect may be used. This recording medium includes a non-transitory tangible medium.

  According to each of the above aspects, early diagnosis of convulsion-type acute encephalopathy can be supported.

It is a graph which respectively shows the electroencephalogram waveform of the patient of the convulsion type | formula acute encephalopathy (AESD) in which the cerebral disorder remained as a sequela, and the patient of the febrile seizure (FS) in which the sequelae did not remain. It is a figure which shows notionally the apparatus structural example of the diagnosis assistance apparatus in 1st embodiment. It is a figure which shows notionally the process structural example of the diagnosis assistance apparatus in 1st embodiment. It is a flowchart which shows the operation example of the diagnosis assistance apparatus in 1st embodiment. It is a figure which shows notionally the electrode arrangement | positioning and the electroencephalogram measurement method in an Example. It is a figure which shows the power spectrum of the convulsion status-type acute encephalopathy (AESD) and febrile convulsions (FS). It is a table | surface which shows the statistics of the power value (spectral intensity | strength) obtained from the electroencephalogram of a convulsion intussusception type acute encephalopathy (AESD) patient group, a febrile convulsion (FS) patient group, and a healthy subject group. It is a box-and-whisker chart showing statistics of power values (spectral intensity) obtained from electroencephalograms of a convulsive intussusception type acute encephalopathy (AESD) patient group, a febrile seizure (FS) patient group, and a healthy subject group. It is a figure which shows the statistics of the power value (spectral intensity | strength) obtained from the electroencephalogram of a convulsion intussusception type acute encephalopathy (AESD) patient group, a febrile convulsion (FS) patient group, and a healthy subject group. It is a box-and-whisker diagram showing statistics of predetermined calculation results of power values (spectral intensities) obtained from electroencephalograms of patients with convulsive intussusception type acute encephalopathy (AESD), febrile convulsions (FS) patients and healthy subjects. It is a graph showing the relationship between the electroencephalogram measurement time of each patient of an AESD patient (7 persons) and FS patient (15 persons), and the power value of the frequency band in the electroencephalogram. It is a graph which shows the time change of the power value of alpha wave in each patient of AESD and FS. It is a figure which shows notionally the time correlation and spatial correlation in 2nd embodiment. It is a figure which shows notionally the process structural example of the diagnosis assistance apparatus in 2nd embodiment. It is a figure which shows the example of the received extraction time point and object electroencephalogram data. It is a figure which shows the example of the pair of the single time interval in a different time and a different measurement part. It is a figure which shows the example of the generation frequency for every time correlation coefficient acquired regarding one measurement site | part. It is a figure which shows the creation example of the frequency information (histogram) of a time correlation coefficient. It is a figure which shows the example which produces | generates the frequency information of a correlation coefficient combining the frequency information of a time correlation coefficient, and the frequency information of a spatial correlation coefficient. It is a flowchart which shows the whole operation | movement of the diagnosis assistance apparatus in 2nd embodiment. It is a flowchart which shows the detailed operation | movement of (S54) and (S55) shown by FIG. It is a flowchart which shows schematic operation | movement of the diagnosis assistance apparatus in the modification of 2nd embodiment. It is a figure which shows the histogram of a time correlation coefficient. It is a figure which shows the histogram of a spatial correlation coefficient. It is a figure which shows the histogram which integrated the histogram of the time correlation coefficient, and the histogram of the spatial correlation coefficient.

  Embodiments of the present invention will be described below. In addition, each embodiment given below is an illustration, respectively, and this invention is not limited to the structure of each following embodiment.

[First embodiment]
Hereinafter, the diagnosis support apparatus in the first embodiment will be described. The diagnosis support apparatus according to the first embodiment generates discrimination information regarding convulsive status encephalopathy (AESD) or febrile seizure (FS).

〔Device configuration〕
FIG. 2 is a diagram conceptually illustrating a device configuration example of the diagnosis support device 10 in the first embodiment. The diagnosis support apparatus 10 is a so-called computer and has a hardware element group as shown in FIG. The diagnosis support apparatus 10 may be a general-purpose computer such as a PC (Personal Computer), or may be a dedicated computer such as an electroencephalograph or a dedicated device capable of communicating with the electroencephalograph. As shown in FIG. 2, the diagnosis support apparatus 10 includes a central processing unit (CPU) 1, a memory 2, an input / output interface (I / F) unit 3, a communication unit 4, and the like. The CPU 1 is connected to other hardware elements through a communication line such as a bus. The hardware element group illustrated in FIG. 2 can also be collectively referred to as an information processing circuit.

The CPU 1 may be a general CPU, or instead of or in addition to at least one of an application specific integrated circuit (ASIC), a DSP (Digital Signal Processor), a GPU (Graphics Processing Unit), and the like. May be included.
The memory 2 is a RAM (Random Access Memory), a ROM (Read Only Memory), or an auxiliary storage device (such as a hard disk).

  The input / output I / F unit 3 can be connected to a user interface device such as a display device (not shown) or an input device (not shown). The display device is a device that displays a screen corresponding to the drawing data processed by the CPU 1, such as an LCD (Liquid Crystal Display) or a CRT (Cathode Ray Tube) display. The input device is a device that receives an input of a user operation such as a keyboard and a mouse. The display device and the input device may be integrated and realized as a touch panel.

  The communication unit 4 performs wireless communication or wired communication, and communicates with other computers and devices such as an electroencephalograph. The communication unit 4 may include a USB (Universal Serial Bus) unit. The communication unit 4 can also exchange data with a portable recording medium such as a USB memory, a CD (Compact Disk), a DVD (Digital Versatile Disc), or a Blu-ray disc. The communication method supported by the communication unit 4 is not limited.

  However, the hardware configuration of the diagnosis support apparatus 10 is not limited to the example shown in FIG. The diagnosis support apparatus 10 may include other hardware elements not shown. For example, when the diagnosis support apparatus 10 is realized as an electroencephalograph, the diagnosis support apparatus 10 may include an electrode or the like for measuring an electroencephalogram. The number of hardware elements is not limited to the example in FIG. For example, the diagnosis support apparatus 10 may have a plurality of CPUs 1.

  A diagnosis support program 7 is stored in the memory 2 of the diagnosis support apparatus 10. The diagnostic support program 7 may be stored in advance in the ROM (memory 2), or from a portable recording medium such as a CD (Compact Disc) or a memory card, or another computer on the network via the communication unit 4. May be installed and stored in the memory 2. The diagnosis support program 7 may be sold in a state of being stored in a recording medium that is readable by a computer, or may be sold in a state of being stored in a server device on the Internet.

[Processing configuration]
FIG. 3 is a diagram conceptually illustrating a processing configuration example of the diagnosis support apparatus 10 in the first embodiment. The diagnosis support apparatus 10 has software elements as shown in FIG. Specifically, the diagnosis support apparatus 10 includes a data acquisition unit 11, a calculation unit 12, a generation unit 13, and the like. Each of these software elements is realized, for example, by executing a diagnosis support program 7 stored in the memory 2 by the CPU 1.

  The data acquisition unit 11 acquires time-series data of electroencephalograms measured at a predetermined site on the subject's head within a predetermined time after convulsion (initial stage of disease). The meanings of “convulsions” and “within a predetermined time after convulsions (early disease)” are as described above.

  It is desirable that the “predetermined part on the head” conforms to the electrode arrangement defined by the International 10-20 Act. However, as long as the position on the head can be standardized in all subjects, the method for standardizing the part is not limited. For example, when the measurement is performed by the monopolar derivation method, the predetermined part is a certain position on the head. Further, when the measurement is performed by the bipolar derivation method, the predetermined part is between two positions where the potential difference on the head is measured.

  In the case where a plurality of candidate parts on the head for measuring the electroencephalogram are provided, the data acquisition unit 11 may acquire the identification number (part ID) of the part where the electroencephalogram is measured, together with the time-series data of the electroencephalogram. . Further, when the diagnosis support apparatus 10 is an electroencephalograph, the data acquisition unit 11 identifies which part of the data is the data by specifying the electrode that is the transmission source of the potential difference signal among the plurality of electrodes. be able to. The candidate site is, for example, between F3 and C3, between F4 and C4, between C3 and P3, and between C4 and P4 as defined by the International 10-20 Law. Specific examples of the predetermined part for measuring the electroencephalogram will be described in the section of the examples.

  The “time series data of the electroencephalogram” is data indicating a temporal change of the potential difference, and is discrete data including a plurality of pairs of the potential difference and the measurement time.

  The data acquisition unit 11 can acquire the time series data of the electroencephalogram measured by the electroencephalograph from the electroencephalograph or another computer via the communication unit 4. Moreover, the data acquisition part 11 may acquire the data from the portable recording medium on which the time series data of the electroencephalogram were recorded. When the diagnosis support apparatus 10 is an electroencephalograph, the data acquisition unit 11 acquires time-series data of electroencephalograms by sampling the potential difference signal obtained from the electrodes at a set sampling frequency.

  The calculation unit 12 performs frequency analysis on the time series data acquired by the data acquisition unit 11 to calculate a power value in a predetermined frequency band for the predetermined part. When the part where the electroencephalogram is measured is fixed, the calculation unit 12 may hold in advance information on a frequency band that is a target of power value calculation. When there are a plurality of candidate regions for which an electroencephalogram is measured, the calculation unit 12 holds a correspondence relationship between the region and the frequency band in advance, and is acquired by the data acquisition unit 11 from this correspondence relationship. The frequency band corresponding to the region ID may be specified.

  The calculation unit 12 uses, for example, Fourier transform (Fast Fourier Transform (FFT), Discrete Fourier Transform (DFT), etc.) as frequency analysis. However, if a power value in a specific frequency band can be obtained, other methods such as wavelet transform may be used for frequency analysis. The calculation unit 12 calculates a power value in a predetermined frequency band for the above-described predetermined part where the electroencephalogram is measured, from the power value (spectral intensity) obtained with a predetermined frequency resolution. For example, the calculation unit 12 calculates an average power value in a frequency band corresponding to a predetermined part. The frequency band predetermined for the predetermined part is set to, for example, an α wave band (8 Hz to 12 Hz) or a γ wave band (20 Hz to 40 Hz) as exemplified in the section of the embodiment. In addition, the frequency band may be set to a predetermined bandwidth (such as a combined bandwidth of a β wave and a γ wave) in a fast wave component (a component faster than a β wave (13 Hz or more)).

  The generation unit 13 uses the power value calculated by the calculation unit 12 to generate discrimination information regarding convulsive status encephalopathy (AESD) or febrile seizure (FS) related to the subject. The discrimination information is information generated using the power value calculated by the calculation unit 12, and if the doctor who has seen the information can discriminate and predict whether the subject is AESD or FS, Specific contents are not limited. For example, the generation unit 13 generates discrimination information including the calculated power value, the statistical value of the power value of the AESD patient, and the statistical value of the power value of the FS patient. The statistical value is, for example, any one of an average power value, a combination of an average power value and a variance value, and a combination of an average power value and a standard deviation. Each average power value of the AESD patient and the FS patient can be calculated by collecting power values obtained from the brain waves at the initial stage of each patient's disease on a performance basis and averaging them for each frequency band. A doctor who sees the discrimination information can make a discrimination prediction by determining which average power value the calculated power value is closer to.

  Further, the generation unit 13 may use the calculated power value as the discrimination information as it is. In this case, the statistical values of the power values of the AESD patient and the FS patient may be provided separately so as to be referred to the doctor by printed matter or display. The generation unit 13 can automatically determine which average power value the calculated power value is close to, and can generate determination information indicating the result as a character string or a numerical value. In this case, the generation unit 13 can generate discrimination information including a character string such as “subject has a high probability of convulsive acute encephalopathy” or “subject has a high possibility of febrile seizure”. .

[Operation example / Information processing flow]
FIG. 4 is a flowchart showing an operation example of the diagnosis support apparatus 10 in the first embodiment. The diagnosis support apparatus 10 according to the first embodiment executes the information processing flow illustrated in FIG. 4 by executing the diagnosis support program 7 by the CPU 1 as described above. In other words, the CPU 1 loads and executes the diagnosis support program 7 from the memory 2 to realize an information processing flow as shown in FIG. 4 in cooperation with other hardware elements.

  The CPU 1 (diagnosis support apparatus 10) acquires time-series data of electroencephalograms measured at a predetermined site on the subject's head within a predetermined time after convulsion (initial stage of disease) (S41). When there are a plurality of candidate regions for measuring the electroencephalogram, the CPU 1 may acquire the region ID that is the identification number of the region where the electroencephalogram is measured, together with the time-series data of the electroencephalogram. The specific content of (S41) is the same as the processing content of the data acquisition unit 11. For example, the CPU 1 can acquire time-series data of an electroencephalogram measured by an electroencephalograph from the electroencephalograph, another computer, or a portable recording medium via the communication unit 4. When the diagnosis support apparatus 10 is an electroencephalograph, the CPU 1 acquires time series data of electroencephalograms by processing a potential difference signal obtained from the electrodes.

  The CPU 1 performs frequency analysis on the time-series data acquired in (S41), thereby calculating a power value in a predetermined frequency band for a predetermined part where the electroencephalogram is measured (S42). The specific content of (S42) is the same as the processing content of the calculation part 12. As described above, when there are a plurality of candidate regions for measuring brain waves, the CPU 1 specifies the frequency band corresponding to the region ID acquired in (S41), and calculates the power value of the frequency band. Also good.

  The CPU 1 uses the power value calculated in (S42) to generate discrimination information regarding convulsive status encephalopathy (AESD) or febrile seizure (FS) related to the subject (S43). The contents of the discrimination information to be generated are as described above. That is, the specific content of (S43) is the same as the processing content of the generation unit 13.

  The CPU 1 may display the generated discrimination information on a display device (not shown) connected to the input / output I / F unit 3 or print it on a printer (not shown) via the communication unit 4. You can also. Further, the CPU 1 can store the discrimination information in a file and output the file to another computer or a portable recording medium via the communication unit 4. The method of presenting the generated discrimination information to the doctor is not limited.

  As the first embodiment, it is also possible to realize an AESD diagnosis support method. The AESD diagnosis support method in the first embodiment is executed by at least one computer or CPU such as the above-described diagnosis support apparatus 10 or CPU 1. For example, when the CPU 1 executes the diagnosis support program 7, the AESD diagnosis support method including the information processing flow shown in FIG. 4 can be realized.

[Operation and effect of the first embodiment]
As described above, in the first embodiment, the time series data of the brain wave of the subject in the early stage of the disease is frequency-analyzed, and the power value in the predetermined frequency band is calculated for the predetermined part where the brain wave is measured. . Then, using the power value, discrimination information regarding AESD or FS regarding the subject is generated.

The present inventors have focused on the importance of identifying early whether the disease of a patient sent to a hospital in the state of convulsion is febrile convulsions (FS) or convulsive contusion type acute encephalopathy (AESD). did. It is difficult to determine whether the cause of convulsions is AESD or FS only from the results of tests such as MRI, blood tests, and cerebrospinal fluid tests. AESD and FS are convulsive after fever. The symptoms are the same. At present, doctors have to predict AESD or FS by measuring a patient's brain wave and visually diagnosing the measured brain wave waveform.
Then, as described in detail in the Examples section, the inventors of the present invention measured the power values of a specific frequency band between the AESD patient and the FS patient in an electroencephalogram measured at a specific site on the head in the early stage of the disease. It was demonstrated that there was a significant difference between them. Thereby, the present inventors use the power value in the frequency band determined in advance for the predetermined part where the electroencephalogram is measured as in the first embodiment, so that the subject is AESD or FS. It was newly found that it is possible to predict and discriminate.

  As described above, according to the first embodiment, by using the power value calculated by frequency analysis, it is possible to generate discrimination information regarding AESD or FS regarding a subject in the early stage of the disease. This discrimination information is information that enables quantitative discrimination because it is generated using objective numerical values. Thereby, according to the first embodiment, it is possible to perform quantitative discrimination prediction of AESD or FS, which is difficult in other examinations or visual diagnosis of an electroencephalogram, that is, the early stage of the disease. Diagnosis can be supported.

[First modification]
In the first embodiment described above, the discrimination information may be generated using power values in a plurality of frequency bands. That is, a plurality of frequency bands may be determined in advance for a specific part where an electroencephalogram is measured. Such a form will be described as a first modification, focusing on the content different from the first embodiment described above.

  In the first modification, the calculation unit 12 calculates a power value for each of a plurality of frequency bands determined in advance with respect to a predetermined part where an electroencephalogram is measured. When the part where the electroencephalogram is measured is fixed, the calculation unit 12 may hold in advance information on a plurality of frequency bands that are power value calculation targets. In addition, when there are a plurality of candidate regions for which an electroencephalogram is measured, the calculation unit 12 holds a correspondence relationship between the region and a plurality of frequency bands in advance, and the data acquisition unit 11 determines from the correspondence relationship. What is necessary is just to identify the several frequency band corresponding to acquired site | part ID.

  The plurality of frequency bands may be obtained by dividing a certain continuous frequency band, or may be bands separated from each other. For example, the plurality of frequency bands are set to an α wave band (8 Hz to 12 Hz) and a γ wave band (20 Hz to 40 Hz).

  The generation unit 13 generates the determination information using the plurality of power values calculated for the plurality of frequency bands by the calculation unit 12. Also in the first modification, the specific content of the discrimination information is not limited. For example, the generation unit 13 generates discrimination information including the calculated power values, and statistical values of the power values of the AESD patients and the statistical values of the power values of the FS patients for each of the frequency bands. To do. Further, the generation unit 13 may use the calculated plurality of power values as discrimination information as they are. In this case, the statistical value of the power value for each of the plurality of frequency bands of the AESD patient and the FS patient may be provided so as to be referred to a doctor separately by printed matter or display. Further, the generation unit 13 can automatically determine which of the calculated power values is close to which average power value, and can generate determination information indicating the result as a character string or a numerical value. .

  The usage of the calculated plurality of power values is not limited to the above example. The production | generation part 13 can also perform the further calculation with respect to the calculated several power value, and can also generate | occur | produce the discrimination information based on the value obtained as a result of the calculation. For example, a value obtained by dividing the power value of a certain frequency band by the power value of another frequency band may be directly used, or the total value of the power values of two or more frequency bands may be at least one. A value obtained by dividing the sum of the power values of two or more frequency bands in which two frequency bands are different may be directly used. As will be described later in the embodiment, for example, at least one of the ratio of the power value of the δ wave to the α wave, the ratio of the total power value of the δ wave and the θ wave to the total power value of the α wave and the β wave, etc. Can be used to generate discrimination information.

In the operation example or the information processing flow of the diagnosis support apparatus 10 in the first modification, the following modifications are made from FIG.
In (S <b> 42), the CPU 1 calculates a power value for each of a plurality of frequency bands determined in advance for a predetermined part where the electroencephalogram is measured.
In (S43), CPU1 produces | generates the discrimination | determination information whether it is FS or AESD regarding a test subject using the some power value calculated by (S42).

  As described above, in the first modification, discrimination information is generated using a plurality of power values related to a plurality of frequency bands. That is, in the first modification, the power value of the subject and the statistical values of the power values of the AESD patient and the FS patient are compared for each of a plurality of frequency bands to determine whether AESD or FS. Become. Therefore, according to the first modification, it is possible to improve the discrimination accuracy between AESD and FS.

[Second modification]
In the first embodiment described above, the discrimination information may be generated using power values in one or more frequency bands in each brain wave at a plurality of sites. That is, the electroencephalogram to be subjected to frequency analysis may be measured from a plurality of parts on the head, and one or more frequency bands may be determined in advance for each part. Such a form will be described as a second modification, focusing on the content different from the first embodiment described above.

  The data acquisition unit 11 acquires time series data of brain waves at a plurality of predetermined sites on the subject's head. The data acquisition unit 11 may acquire a part ID for each time-series data of the electroencephalogram. Further, when the diagnosis support apparatus 10 is an electroencephalograph, the data acquisition unit 11 identifies which part of the data is the data by specifying the electrode that is the transmission source of the potential difference signal among the plurality of electrodes. be able to. It is desirable to measure the electroencephalograms at a plurality of predetermined sites at the same timing. This is because the measurement conditions match among a plurality of parts, so that improvement in discrimination accuracy can be expected. However, the electroencephalogram may be measured in order for each predetermined part. Specific examples of the plurality of predetermined portions will be described in the example section.

  The calculation unit 12 calculates the power values of one or a plurality of frequency bands determined in advance for each predetermined part in the electroencephalograms of the plurality of predetermined parts. One or a plurality of the same frequency bands may be determined in advance for each predetermined part. Further, a different number of frequency bands may be determined in advance for each predetermined part. For example, an α wave band and a γ wave band are determined in advance for a portion between F4 and C4, and only a γ wave band is determined in advance for a portion between C4 and P4. Furthermore, different frequency bands may be determined in advance for the first predetermined portion and the second predetermined portion included in the plurality of predetermined portions. For example, an α wave band is determined in advance for a portion between F4 and C4, and a γ wave band is determined in advance for a portion between C4 and P4.

  The production | generation part 13 produces | generates the above-mentioned discrimination | determination information using the power value calculated about the said several predetermined site | part and the said 1 or several frequency band. Also in the second modification, the specific content of the discrimination information is not limited. For example, the generation unit 13 generates discrimination information including a plurality of calculated power values, a statistical value of the power value of the AESD patient and a statistical value of the power value of the FS patient for each predetermined region and each frequency band. Further, the generation unit 13 may use the calculated plurality of power values as discrimination information as they are. Further, the generation unit 13 automatically determines which average power value the calculated plurality of power values are close to for each predetermined part and each frequency band, and determination information indicating the result as a character string or a numerical value Can also be generated.

  Furthermore, as illustrated in the first modification, the generation unit 13 can perform a predetermined calculation on the plurality of calculated power values and generate discrimination information based on the value obtained by the calculation. As the calculation, the ratio of power values in different frequency bands in the same part may be calculated, or the ratio of power values in the same frequency band in different parts may be calculated.

In the operation example or information processing flow of the diagnosis support apparatus 10 in the second modification, the following modifications are made from FIG.
In (S41), the CPU 1 obtains time series data of electroencephalograms at a plurality of predetermined sites on the subject's head.
In (S42), the CPU 1 calculates the power values of one or a plurality of frequency bands determined in advance for each predetermined region in the electroencephalogram of each of the plurality of predetermined regions.
In (S43), CPU1 produces | generates the discrimination | determination information regarding AESD or FS regarding a test subject using the several power value calculated about the said several predetermined site | part and the said one or several frequency band.

  As described above, in the second modification, for each electroencephalogram measured at each of a plurality of predetermined sites on the head, power values of one or a plurality of frequency bands determined in advance for each predetermined site are calculated, Discrimination information is generated using such a plurality of power values. That is, in the second modified example, the power value of the subject and the statistical values of the power values of the AESD patient and the FS patient are compared for each of a plurality of parts on the head and a plurality of frequency bands, It is determined whether it is FS. Therefore, according to the second modification, it is possible to improve the discrimination accuracy between AESD and FS.

[Other examples of risk assessment information]
In the first embodiment described above, a known statistical classification method may be used to automatically determine whether the subject is FS or AESD.

  Specifically, the generation unit 13 statistically classifies the calculated power value into AESD or FS by comparing the AESD learning data and the FS learning data with the power value calculated by the calculation unit 12. Various statistical classification methods exist, and they are used in various technical fields such as machine learning, pattern matching, and data clustering. Each learning data of AESD and FS is, for example, a set of power values obtained by the same method (predetermined site and disease initial stage) as that of the subject for each patient diagnosed with AESD or FS. As the statistical classification method, for example, SVM (support vector machine), linear discriminant analysis, k-nearest neighbor method, and the like can be used.

  In the comparison between the learning data and the calculated power value, for example, the generation unit 13 calculates the similarity to the AESD learning data and the similarity to the FS learning data, respectively. The Euclidean distance may be used for this similarity. The production | generation part 13 classify | categorizes a test subject's power value in the direction which shows a similarity larger than a predetermined threshold value among AESD and FS.

  The generation unit 13 generates discrimination information including the above-described statistical classification result or statistical classification basis data. The result of the statistical classification indicates whether the calculated power value of the subject belongs to AESD or FS. The statistical classification ground data is data used as a ground for determining the affiliation of the power value of the subject. For example, the generation unit 13 uses the similarity between the learning data and the calculated power value as the basis data.

  Examples will be given below to explain the above-mentioned contents in more detail. The present invention is not limited by the following examples.

  The present inventors have analyzed the past electroencephalogram data measured from a pediatric patient sent to the hospital by convulsion, and obtained the following results. Specifically, from December 2001 to January 2014, pediatric patients (68 patients) sent to multiple target hospitals, including Tottori University Hospital, within 120 hours after convulsion EEG data (72 pieces) measured for 20 to 30 minutes were used.

  FIG. 5 is a diagram conceptually illustrating an electrode arrangement and an electroencephalogram measurement method in the embodiment. In FIG. 5, the part surrounded by a circle shows the electrode arrangement according to the International 10-20 method. In this example, the electroencephalogram data was measured by the bipolar derivation method with the electrode arrangement according to the International 10-20 method shown in FIG. That is, between Fp1 and F3, between Fp2 and F4, between F3 and C3, between F4 and C4, between C3 and P3, between C4 and P4, between P3 and O1. And electroencephalogram data indicating the time displacement of each potential difference between P4 and O2 was measured. As measurement parameters, a time constant of 0.3 seconds, a high frequency filter (60 Hz), a sensitivity of 10 μV / mm, and a sampling frequency of 200 Hz or 500 Hz were used.

FIG. 6 is a diagram showing power spectra of convulsive status encephalopathy (AESD) and febrile seizure (FS).
In this example, fast Fourier transform was applied as frequency analysis to each of the above-described brain wave data, and a power spectrum as shown in FIG. 6 was calculated. Specifically, the spectral intensity was calculated for a frequency band from 1 Hz to 60 Hz with a step width of 0.4 Hz at a sampling frequency of 200 Hz and a step width of 0.2 Hz at a sampling frequency of 500 Hz.
FIG. 6 shows an example of spectral intensities obtained from the brain waves of 18 AESD patients among the above 68 pediatric patients and an example of spectral intensities obtained from the brain waves of 20 FS patients. Yes.

  In the present embodiment, the calculated spectrum intensity is δ wave (1 Hz to 3 Hz), θ wave (4 Hz to 7 Hz), α wave (8 Hz to 12 Hz), β wave (13 Hz to 19 Hz), γ wave. Each frequency band (20 Hz to 40 Hz) was averaged. In FIG. 6, each frequency band region is divided and represented.

FIG. 7 is a table showing statistics of power values (spectral intensity) obtained from the electroencephalograms of the AESD patient group, the FS patient group, and the healthy subject group. The table in FIG. 7 shows the statistical values of the power values for each part where the electroencephalogram was measured (electrode arrangement according to the International 10-20 method) and frequency bands (δ wave, θ wave, α wave, β wave, γ wave). Shows the relationship. In the notation “A ± B” in FIG. 7, “A” indicates the average power value of the corresponding population, and “B” indicates the standard deviation of the corresponding population. The subscript “a” of “A ± B” indicates a significance level with a p value of less than 0.01, and the suffix “b” indicates a significance level with a p value of less than 0.05.
The AESD patient group was formed by 6 of the 18 AESD patient groups described above, and the FS patient group was formed by the 20 persons described above. A group of healthy subjects was formed by 13 healthy children, measured at Tottori University Hospital, apart from the above-mentioned pediatric patients.

FIG. 8 is a box-and-whisker diagram showing statistics of power values (spectral intensity) obtained from the electroencephalograms of the AESD patient group, the FS patient group, and the healthy subject group. The abscissa shows four sites (between Fp2 and F4, between F4 and C4, between C4 and P4, and between P4 and O2), and on the ordinate, the power value (spectral intensity). )It is shown.
FIG. 8A shows the statistical value of the power value of the α wave at each of the four parts described above, and FIG. 8B shows the statistical value of the power value of the β wave at the same part of FIG. C) shows the statistical value of the power value of the γ wave in each same part. In each figure, three box whiskers (AESD patient group, FS patient group, and healthy person group) are shown for each part. The upper and lower ends of the whiskers indicate the maximum and minimum power values, and the lower, middle, and upper ends of the box indicate the first quartile, median, and third quartile. In addition, among the connecting lines connecting the box whiskers, the connecting line with one asterisk (*) indicates a significance level with a p value of less than 0.05, and two asterisks (**) are attached. The connecting line shows a significance level with a p-value less than 0.01.

FIG. 9 is a diagram showing statistics of power values (spectral intensity) obtained from the electroencephalograms of the AESD patient group, the FS patient group, and the healthy subject group. The abscissa shows four sites (between Fp2 and F4, between F4 and C4, between C4 and P4, and between P4 and O2), and on the ordinate, the power value (spectral intensity). )It is shown.
FIG. 9A shows the statistical value of the power value of the α wave at each of the four parts described above, and FIG. 9B shows the statistical value of the power value of the β wave at the same part of FIG. C) shows the statistical value of the power value of the γ wave in each same part. In each figure, three marked vertical lines are shown for each part. The upper and lower lines of the upper and lower lines indicate ± standard deviation, and the mark indicates the average value. Also, of the connecting lines connecting the marked upper and lower lines, the connecting line marked with an asterisk (*) indicates a significance level with a p-value less than 0.05, and two asterisks (**) are attached. The connected line shows a significance level with a p-value less than 0.01.

As shown in FIGS. 7, 8, and 9, a significant difference between the AESD patient group and the FS patient group was recognized in the power value relating to the combination of the region and the frequency band shown below.
Γ wave at the site between FP2 and F4 γ wave at the site between F3 and C3 α wave and γ wave at the site between F4 and C4 γ at the site between C4 and P4 In addition, the significant difference between the FS patient group and the healthy subject group was recognized by the power value which concerns on the combination of a site | part and a frequency band shown next.
Gamma wave in the region between F3 and C3 Gamma wave in the region between F4 and C4 Gamma wave in the region between C4 and P4 β wave and γ in the region between P4 and O2 wave

Thus, this example demonstrates that there is a significant difference between the AESD patient and the FS patient in the power value of the specific frequency band in the electroencephalogram measured at the specific site on the head in the early stage of the disease. It was. As a result, the usefulness of the discrimination information for AESD or FS generated using a power value in a predetermined frequency band with respect to the predetermined part where the electroencephalogram is measured is also confirmed.
In the above-described first embodiment, for example, any one of the above-described combinations of a predetermined portion and a frequency band in which a significant difference is recognized can be used.
In the first modification, for example, a combination of a portion between F4 and C4 and frequency bands of α waves and γ waves can be used.
In the second modification, any one of a part between FP2 and F4, a part between F3 and C3, a part between F4 and C4, and a part between C4 and P4, Combinations with frequency bands can be used, and all combinations with significant differences can be used.

  Furthermore, the present inventors predict whether a subject is AESD or FS by using a power value ratio between a plurality of predetermined frequency bands with respect to a predetermined part where an electroencephalogram is measured. I also found a new possibility.

FIG. 10 is a box-and-whisker diagram showing statistics of predetermined calculation results of power values (spectral intensity) obtained from the electroencephalograms of the AESD patient group, the FS patient group, and the healthy subject group. The horizontal axis shows each of the four parts (between Fp2 and F4, between F4 and C4, between C4 and P4, and between P4 and O2), and the vertical axis shows the power value ratio. Has been.
FIG. 10A shows the statistical value of the power value ratio of the δ wave to the α wave in each of the above four parts, and FIG. 10B shows the total power value of the δ wave and the θ wave in the same part. The statistical value of the ratio to the total power value of α wave and β wave is shown. In each figure, three box whiskers (AESD patient group, FS patient group, and healthy person group) are shown for each part. The upper and lower ends of the whiskers indicate the maximum and minimum values, and the lower, middle and upper ends of the box indicate the first quartile, median and third quartile. In addition, among the connection lines connecting the box whiskers, a connection line marked with an asterisk (*) indicates a significance level with a p value of less than 0.05.

  In the example of FIG. 10, the ratio of the power value of the δ wave to the α wave in the region between F4 and C4, and the δ wave and θ in each region between F4 and C4 and between C4 and P4. A significant difference between the AESD patient group and the FS patient group is recognized in the ratio of the total wave power value to the total power value of the α wave and the β wave.

  Thus, according to the present example, it was proved that there is a significant difference between the AESD patient and the FS patient in the power value ratios of a plurality of specific frequency bands. As a result, the usefulness of the discrimination information whether AESD or FS is generated using the power value ratios of a plurality of frequency bands determined in advance for a predetermined part where the electroencephalogram is measured is also confirmed.

  In addition, the present inventors also analyzed the electroencephalogram data measured within 24 hours after the convulsion, and in the AESD patient group, between F3 and C3, between F4 and C4, between P3 and O1. It has been confirmed that the power value of each frequency band of the θ wave, α wave, β wave, and γ wave in the region between them is lower than that of the FS patient group. Furthermore, the power value ratio of the AESD patient group to the α wave of δ wave, and the ratio of the total power value of the δ wave and θ wave to the total power value of the α wave and β wave are the FS patient group at each site. It has been confirmed that it is higher than

Furthermore, it was demonstrated that it is possible to discriminate between AESD patients and FS patients by changing the power value of each frequency band within 24 hours after the convulsion.
In Example 2, EEG data of 7 AESD patients and 15 FS patients measured within 24 hours after convulsion were used.
FIG. 11A is a graph showing the relationship between the electroencephalogram measurement time of each of AESD patients (7 patients) and FS patients (15 patients) and the power value of the α wave frequency band in the electroencephalogram. FIG. 11B is a graph showing the relationship between the electroencephalogram measurement time of each of AESD patients (7 patients) and FS patients (15 patients) and the power value ratio of the δ wave to the α wave in the electroencephalogram. . FIG. 11C shows the EEG measurement time of each of AESD patients (7 patients) and FS patients (15 patients), and the total power value of δ wave and θ wave, α wave and β wave of the brain wave. It is a graph showing the relationship with the ratio with respect to a total power value.
In each figure, black circles show data for AESD patients, and white circles show data for FS patients.

According to FIG. 11A, in the electroencephalogram of an AESD patient, the power value of the α wave gradually decreases within 24 hours after the convulsion. In other words, it is expressed that the power value of the α wave has a negative correlation with the elapsed time after the convulsion. On the contrary, in the electroencephalogram of the FS patient, the power value of the α wave gradually increases within 24 hours after the convulsion. In other words, it is expressed that the power value of the α wave has a positive correlation with the elapsed time after the convulsion.
According to FIG. 11B, in the electroencephalogram of the AESD patient, the power value ratio of the δ wave to the α wave increases within 24 hours after the convulsion. In other words, the ratio of the power value of the δ wave to the α wave has a positive correlation with the elapsed time after convulsion. On the other hand, in the electroencephalogram of the FS patient, the ratio of the power value of the δ wave to the α wave decreases within 24 hours after the convulsion. In other words, the power value ratio of the δ wave to the α wave has a negative correlation with the elapsed time after convulsion.
According to FIG. 11C, in the electroencephalogram of the AESD patient, the ratio of the total power value of the wave and the θ wave to the total power value of the α wave and the β wave increases within 24 hours after the convulsion. In other words, the ratio of the total power value of the δ wave and the θ wave to the total power value of the α wave and the β wave has a positive correlation with the elapsed time after the convulsions. On the other hand, in the electroencephalogram of FS patients, the ratio decreases within 24 hours after convulsion. In other words, the ratio of the total power value of the δ wave and the θ wave to the total power value of the α wave and the β wave has a negative correlation with the elapsed time after the convulsions.
Further, according to FIG. 11, it can be seen that in the electroencephalogram of an AESD patient, the α wave component and the β wave component continue to decrease over time, and the slow wave component increases with time.

In FIG. 11, the time distribution of the electroencephalogram power values in the AESD patient group and the FS patient group was confirmed, but the temporal change in the electroencephalogram power values in each of the AESD and FS patients was also confirmed.
FIG. 12 is a graph showing the time change of the power value in the α wave frequency band in each patient of AESD and FS. In FIG. 12, electroencephalograms measured twice at time intervals within 24 hours after convulsion were analyzed in each of two AESD patients and two FS patients. The solid line shows the data change of the AESD patient, and the broken line shows the data change of the FS patient.

  According to FIG. 12, in both of the AESD patients, the power value of the α wave decreases within 24 hours after the convulsion. On the other hand, in both of the FS patients, the power value of the α wave increased significantly within 24 hours after the convulsion. This result is the same as the result of FIG. 11A, and it is confirmed that the correlation can be said in FIG. 11 for each patient of AESD and FS.

  Thus, according to the present example, it was demonstrated that the temporal change in the power value of the electroencephalogram in the early stage of the disease (for example, within 24 hours) can be an index for discriminating between AESD and FS. Therefore, in addition to or in addition to the first embodiment and the modification described above, the determination information may be generated using a time change of a power value in a predetermined frequency band for a predetermined part. Specifically, the power value for each time may be included in the determination information, information indicating an increasing tendency or a decreasing tendency may be included in the determination information, and the patient showing the increasing tendency is FS and decreases. A patient who shows a trend may generate discrimination information that predicts that the patient has AESD. The accuracy of the generated discrimination information may be determined depending on whether the results obtained in the first embodiment and the modification described above are the same as the results obtained in this example.

[Second Embodiment]
Hereinafter, the diagnosis support apparatus according to the second embodiment will be described.
First, an outline of diagnosis support performed by the diagnosis support apparatus according to the second embodiment will be described.
The diagnosis support apparatus according to the second embodiment acquires time series data of brain waves measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion, and in the acquired time series data A plurality of correlation coefficients of at least one of a temporal correlation coefficient and a spatial correlation coefficient are calculated, frequency information of the calculated correlation coefficient is acquired as a feature quantity of the subject's brain wave, and the acquired feature quantity is used. The discriminating information regarding the convulsive status encephalopathy (AESD) of the subject is generated.

The meanings of “convulsion status”, “within a predetermined time after convulsion status (disease early stage)”, “predetermined region on the head”, and “time series data of electroencephalogram” are as described above.
In the second embodiment, time series data of brain waves measured at one predetermined site may be used, or time series data of brain waves measured at a plurality of predetermined sites may be used. However, for highly accurate discrimination, it is preferable to use data measured at a plurality of predetermined sites. In the following description, the predetermined part where the electroencephalogram is measured may be referred to as a measurement part.

FIG. 13 is a diagram conceptually showing temporal correlation and spatial correlation in the second embodiment. FIG. 13 shows an example of time-series data of electroencephalograms measured at N predetermined sites on the subject's head.
Here, the “time correlation coefficient” is the strength of the relationship (similarity) of waveforms between different time points (time) on the time axis in the electroencephalogram time-series data measured at one predetermined site. It is an index value indicating In FIG. 13, since the time axis is taken from the left to the right of the page, the time correlation coefficient indicates the strength of the relationship between the waveforms in the horizontal direction.
The “spatial correlation coefficient” is an index value indicating the strength of waveform relationship (similarity) between time-series data of electroencephalograms measured at two different predetermined sites. In FIG. 13, since time series data of the electroencephalogram for each measurement site is shown in the vertical direction on the paper surface, the spatial correlation coefficient indicates the strength of the waveform relationship between the upper and lower sides.
In the “time correlation coefficient” and the “spatial correlation coefficient”, the time length of the waveform compared between different time points or between different measurement sites is arbitrary. The amplitude values of electroencephalograms measured at a certain time point may be compared, waveform data having the same time length may be compared, or waveform data having different time lengths may be compared. In this embodiment, as will be described later, an example in which waveform data having the same time length (single time interval) is compared to calculate a time correlation coefficient or a spatial correlation coefficient is shown.

  The “correlation coefficient frequency information” is information indicating the occurrence frequency of the time correlation coefficient or the spatial correlation coefficient. Examples of the occurrence frequency include the number of occurrences or occurrence rate for each coefficient value, the number of occurrences or occurrence rate for each predetermined value range of the coefficient, and the like. For example, a plurality of calculated time correlation coefficients or spatial correlation coefficients are in a plurality of predetermined ranges (−1.0 or more and less than −0.25, −0.25 or more and less than −0.5, etc.). The number of coefficients classified and belonging to each value range is counted as “frequency information of correlation coefficients”. Further, the calculated plurality of temporal correlation coefficients or spatial correlation coefficients are each rounded, and the number of each coefficient value after the processing is counted as “frequency information of correlation coefficients”.

  The “frequency information of the correlation coefficient” is used as a feature amount (information indicating the feature) of the subject's brain wave, and discrimination information regarding the subject's AESD is generated. The discriminating information is information generated using the characteristic amount (correlation coefficient frequency information) of the subject's brain wave, and can be any information that allows the doctor who has seen the information to discriminate and predict something about AESD. The specific contents are not limited. The discriminating information may be information that helps the doctor to determine whether or not the subject has AESD, and the doctor discriminates whether or not the subject has AESD that is likely to have sequelae. It may be information that helps you. Specific contents of this discrimination information will be described later.

  As shown in this summary, the inventors have found that either or both of the time correlation coefficient and the spatial correlation coefficient of the subject's electroencephalogram after convulsion are characteristic of the electroencephalogram related to AESD. By using the feature amount, a new idea that doctors can assist diagnosis of AESD of subjects is obtained. Such an idea is embodied in the second embodiment, and by using either one or both of the time correlation coefficient and the spatial correlation coefficient of the subject's brain wave as a feature amount, some discrimination regarding the subject's AESD is performed. The discrimination information that can be made can be generated. Since the feature amount is an objective numerical value, according to the second embodiment, it is possible to quantitatively determine AESD.

  Hereinafter, the diagnosis support apparatus in the second embodiment will be described in more detail. In the following description, for convenience of explanation, a case where both the time correlation coefficient and the spatial correlation coefficient are used as feature amounts is exemplified. However, in the following description, even if any one of the time correlation coefficient and the spatial correlation coefficient is used as the feature amount, the above-described effects can be obtained.

〔Device configuration〕
The diagnosis support apparatus in the second embodiment is a so-called computer, and has a hardware element group as shown in FIG. Each hardware element is as described in the first embodiment. The diagnosis support apparatus in the second embodiment may be a general-purpose computer such as a PC, or a dedicated computer such as an electroencephalograph or a dedicated apparatus capable of communicating with the electroencephalograph. The hardware configuration of the diagnosis support apparatus in the second embodiment is not limited to the example shown in FIG. 2 and may include other hardware elements not shown. For example, when the diagnosis support apparatus is realized as an electroencephalograph, the diagnosis support apparatus may include an electrode or the like for measuring an electroencephalogram. The number of hardware elements is not limited to the example in FIG. For example, the diagnosis support apparatus may have a plurality of CPUs 1.

  A diagnosis support program 7 is stored in the memory 2 of the diagnosis support apparatus according to the second embodiment. The diagnosis support program 7 may be stored in the ROM (memory 2) in advance, or may be installed via a communication unit 4 from a portable recording medium such as a CD or a memory card or another computer on the network. It may be stored in the memory 2. The diagnosis support program 7 may be sold in a state of being stored in a recording medium that is readable by a computer, or may be sold in a state of being stored in a server device on the Internet.

[Processing configuration]
FIG. 14 is a diagram conceptually illustrating a processing configuration example of the diagnosis support apparatus 20 in the second embodiment. The diagnosis support apparatus 20 has software elements as shown in FIG. Specifically, the diagnosis support apparatus 20 includes a reception unit 21, a data acquisition unit 22, a selection unit 23, a calculation unit 24, a distribution acquisition unit 25, a generation unit 26, and the like. Each of these software elements is realized, for example, by executing a diagnosis support program 7 stored in the memory 2 by the CPU 1.

  The data acquisition unit 21 acquires time-series data of electroencephalograms measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion. That is, the data acquisition unit 21 is the same as the data acquisition unit 11 of the first embodiment. As described above, the number of electroencephalogram measurement sites is not limited, but in this embodiment, the data acquisition unit 21 acquires electroencephalographic time series data measured at each of a plurality of predetermined sites.

  The data acquisition unit 21 may acquire the entire time-series data of the measured electroencephalogram as a target for extracting the feature value, or may acquire a part of it as a target. That is, the data acquisition unit 21 performs time corresponding to the extraction time point indicated by the user input from the time series data of the electroencephalogram measured at one or more predetermined sites based on the user input received by the receiving unit 22 described later. You may extract the time series data of an area. Hereinafter, the time-series data of the feature quantity extraction target acquired by the data acquisition unit 21 may be referred to as target electroencephalogram data.

The accepting unit 22 accepts a user input related to the extraction time point of the electroencephalogram time-series data.
Here, the “extraction time point” is information indicating time timing for specifying local data acquired as a target for extracting a feature quantity from time-series data of electroencephalograms. As long as the local data can be specified, the “extraction time point” may be the start time of the local data, the center time, or the end time. Further, the “extraction time point” may be information indicating the time interval (from the start time point to the end time point) of the local data.
For example, the reception unit 21 displays a screen for allowing the user to input the extraction time point on the display device connected via the input / output I / F unit 3, and performs the operation according to the input operation by the user on the screen. Accept extraction time.
This extraction time point is preferably a time timing at which a stable time zone with less noise in the electroencephalogram can be identified. As noise, there may be AC noise mixed from an external device, electromyogram generated by body movement (including eye movement) of the subject, and the like. Therefore, the reception unit 21 may cause the user to visually select a time zone with relatively little noise from the data by displaying time series data of the measured brain waves on a display device.

Based on the user input received by the receiving unit 22, the data acquisition unit 21 is based on the time series data of the electroencephalogram measured at one or more predetermined sites, and the time series of the time interval corresponding to the extraction time point indicated by the user input Data (target EEG data) is extracted. The longer the time length of the target electroencephalogram data is, the more accurate electroencephalogram feature quantity can be obtained. However, it is desirable that the time length of the target electroencephalogram data is set to an optimum time length in consideration of minimizing the amount of noise.
In this way, an input of an extraction time point that can specify a time zone with less noise is accepted, and time-series data of a time interval corresponding to the extraction time point is extracted as target electroencephalogram data, which contributes to AESD discrimination with high accuracy Can be obtained.

FIG. 15 is a diagram illustrating an example of accepted extraction time points and target electroencephalogram data. In the example of FIG. 15, the central time point of the target electroencephalogram data is specified as the extraction time point, and the time series data of the section (feature extraction interval) before and after T1 time (2 × T1) around the extraction time point is the target electroencephalogram. It is acquired as data.
In the example of FIG. 15, the same time interval is set as the feature extraction interval for all measurement sites, but a different time interval for each measurement site may be set as the feature extraction interval. Moreover, the time series data of the electroencephalogram relating to a part of the measurement sites among all the measurement sites may be the target electroencephalogram data.

  The selection unit 23 selects a predetermined number of pairs of single time sections at different times or different measurement sites from the time series data (target electroencephalogram data) acquired by the data acquisition unit 21. When the frequency information of the time correlation coefficient is used for the feature amount, the selection unit 23 selects a predetermined number of pairs of single time sections at different times, and when the spatial correlation coefficient is used for the feature amount, the selection unit 23 Selects a predetermined number of pairs of single time intervals at different measurement sites. Hereinafter, a pair of single time intervals at different times may be referred to as interval pairs within the measurement region, and a pair of single time intervals at different measurement regions may be referred to as interval pairs between the measurement regions.

When target electroencephalogram data relating to a plurality of measurement sites is obtained as in the present embodiment, the selection unit 23 selects a predetermined number of interval pairs in the measurement site for each measurement site, and between the measurement sites for each measurement site pair A predetermined number of segment pairs are selected. Here, when the number of measurement sites is represented by N (an integer greater than or equal to 2) and the number of selected pairs (predetermined number) is represented by S (an integer greater than or equal to 2), the interval pair within the measurement site is ( N × S) are selected, and [{N (N−1) / 2} × S] segment pairs between measurement sites are selected. {N (N-1) / 2} is the number of all combinations of two measurement sites out of N measurement sites.
In the present embodiment, the selection unit 23 selects (N × S) segment pairs in the measurement region, and selects [{N (N−1) / 2} × S] segment pairs between the measurement regions. . In this embodiment, S pairs of single time intervals are selected for each measurement site and each measurement site pair, but a different number of pairs may be selected.

FIG. 16 is a diagram illustrating an example of a pair of single time intervals at different times and different measurement sites. In FIG. 16, a pair of single time intervals at different times (interval pairs within a measurement region) is indicated by reference signs S1 and S2, and a pair of single time intervals at different measurement regions (interval pair between measurement regions). Are indicated by reference numerals S3 and S4.
In the present embodiment, as shown in FIG. 16, the section pairs between the measurement sites are set in the same time zone, but may be set in different time zones.

As the number of selected single time interval pairs increases, the processing load increases and the processing time becomes longer, but a highly accurate electroencephalogram feature quantity can be obtained. The number of the pairs is set to an appropriate number in consideration of the time length of the target electroencephalogram data and the time length of a single time interval, for example.
The time length of a single time interval is related to the wavelength length (frequency) of the electroencephalogram to be examined for correlation. That is, by changing the time length of a single time section, it becomes possible to investigate different frequency bands of the electroencephalogram, and a highly accurate electroencephalogram feature quantity can be obtained. The time length of the single time interval is preferably set to about 0.1 to 2 seconds, for example.
The highly accurate electroencephalogram feature amount is feature information that includes the features of the electroencephalogram contributing to AESD discrimination, so that the AESD discrimination accuracy can be increased.

The target electroencephalogram data is selected from a section with a relatively small amount of noise, but it is difficult to completely remove the noise. Therefore, the selection unit 23 may select a single time interval as follows. Specifically, the selection unit 23 randomly selects a predetermined number of single time interval pairs, and selects the upper limit threshold value for the time series data of the single time interval from among the selected single time intervals. A single time interval including an amplitude value that exceeds or falls below the lower threshold is excluded, and the single time interval is reselected until a predetermined number of single time interval pairs are obtained.
Here, the “upper limit threshold” and the “lower limit threshold” are set to an upper limit value and a lower limit value of an amplitude value that can be identified as a noise component such as an electromyogram. As a result, it is possible to select only a single time interval that does not include a noise component by excluding a single time interval that includes a noise component. As a result, a highly accurate electroencephalogram feature quantity can be obtained.

As described in the overview, the calculation unit 24 calculates a plurality of correlation coefficients of at least one of the time correlation coefficient and the spatial correlation coefficient in the time series data acquired by the data acquisition unit 21.
In the present embodiment, the calculation unit 24 uses, as the time correlation coefficient or the spatial correlation coefficient, the correlation coefficient of the time series data of each single time interval for each pair of single time intervals selected by the selection unit 23. Calculate each. According to the notation example described above, the calculation unit 24 calculates (N × S) time correlation coefficients and [{N (N−1) / 2} × S] spatial correlation coefficients. .

As a correlation coefficient calculation method, a known method may be used. For example, the correlation coefficient can be calculated by the following equation. In the following equation, R (x, y) represents a correlation coefficient of a pair (x, y) of time series data (vector) in a single time interval, and C (x, y) represents a single time interval. The covariance of a pair (x, y) of time-series data (vector) is shown. T ₂ indicates the number of amplitude values included in a single time interval, and x _i and y _i indicate one amplitude value included in the single time interval. m _x represents the average value of the time series data of a single time interval x, m _y denotes an average value of the time series data of a single time interval y. The range of the correlation coefficient R (x, y) is −1 or more and 1 or less.

The distribution acquisition unit 25 acquires the correlation coefficient frequency information calculated by the calculation unit 24 as a feature quantity of the subject's brain waves. The “correlation coefficient frequency information” is as described above.
In the present embodiment, the distribution acquisition unit 25 counts the occurrence frequency for each correlation coefficient based on the correlation coefficient calculated by the calculation unit 24 for each pair selected by the selection unit 23. The frequency information is acquired.
Here, the “occurrence frequency for each correlation coefficient” is the number of occurrences or the occurrence ratio for each time correlation coefficient or each spatial correlation coefficient. The granularity of “every time correlation coefficient” or “every spatial correlation coefficient” is arbitrary, and may be, for example, a predetermined value range or a value after a predetermined fraction processing. Good.

FIG. 17 is a diagram illustrating an example of the occurrence frequency for each time correlation coefficient acquired with respect to one measurement site. In the example of FIG. 17, the number of time correlation coefficient values calculated by the calculation unit 24 is counted for each value range in increments of 0.1. For example, the distribution acquisition unit 25 may count the number for each coefficient value with respect to the time correlation coefficient up to the S first decimal place calculated for each measurement site.
The occurrence frequency for each spatial correlation coefficient is similarly counted. For example, the distribution acquisition unit 25 may count the number of each coefficient value for the spatial correlation coefficient up to the S first decimal place calculated for each measurement site pair.

The distribution acquisition unit 25 acquires the frequency information by adding the occurrence frequencies for each correlation coefficient counted for each measurement site or each measurement site pair. The frequency information acquired in the present embodiment can be expressed as a correlation coefficient histogram.
FIG. 18 is a diagram illustrating an example of creating frequency information (histogram) of the time correlation coefficient. In the example of FIG. 18, the distribution acquisition unit 25 adds up the number of occurrences for each time correlation coefficient counted for each measurement site, and normalizes the norm to be 1, so that the histogram of the time correlation coefficient is obtained. Is generated as the frequency information. The histogram shown in FIG. 18 is formed of 20 bins (correlation coefficient values in increments of 0.1). At this time, the distribution acquisition unit 25 adds the number of occurrences for each spatial correlation coefficient counted for each measurement site pair and performs the same normalization, thereby further using the spatial correlation coefficient histogram as the frequency information. Generate.
In this manner, the distribution acquisition unit 25 can generate the histogram of the temporal correlation coefficient and the histogram of the spatial correlation coefficient separately, and use these two histograms as the feature quantity of the subject's brain wave.

FIG. 19 is a diagram illustrating an example of generating the correlation coefficient frequency information by combining the frequency information of the time correlation coefficient and the frequency information of the spatial correlation coefficient. As shown in FIG. 19, the distribution acquisition unit 25 connects the frequency information of the time correlation coefficient and the frequency information of the spatial correlation coefficient in the horizontal direction and normalizes the frequency information of the correlation coefficient. It can also be. In this case, the distribution acquisition unit 25 can use one connected and normalized histogram as the feature quantity of the subject's brain wave.
Further, the distribution acquisition unit 25 can also add (combine) the frequency information of the temporal correlation coefficient and the frequency information of the spatial correlation coefficient as the correlation coefficient frequency information. Specifically, a combined histogram can be generated by adding the frequency of each bin of the histogram of the temporal correlation coefficient and the histogram of the spatial correlation coefficient.

The generation unit 26 uses the feature amount acquired by the distribution acquisition unit 25 to generate discrimination information related to the convulsive status encephalopathy (AESD) of the subject. The discrimination information is as described above, and the specific contents of the subject are not limited as long as the doctor who has seen the information can discriminate and predict something about AESD.
For example, the generation unit 26 generates discrimination information including the subject's brain wave feature amount acquired by the distribution acquisition unit 25 and a brain wave feature amount sample acquired in advance from a population of AESD patients in a comparable state. can do. In this case, the generation unit 26 may generate discrimination information in which the correlation coefficient histogram obtained from the subject's brain wave and the correlation coefficient histogram obtained in advance from the population of the AESD patient are displayed side by side.
Further, the generation unit 26 may use the characteristic amount of the subject's brain wave acquired by the distribution acquisition unit 25 as the discrimination information as it is. In this case, the feature quantity of the electroencephalogram obtained in advance from the population of the AESD patient may be provided so that it can be referred to the doctor separately by printed matter or display.

The generation unit 26 compares the brain wave feature quantity sample measured in advance from the population with the subject's brain wave feature quantity, thereby determining whether the subject is AESD or the subject's AESD has sequelae. It is also possible to generate discrimination information.
In this case, the electroencephalogram feature amount sample only needs to be stored in the electroencephalogram database (DB) 27. The electroencephalogram DB 27 may be realized on the memory 2 of the diagnosis support apparatus 20 or may be realized on another computer (such as a server). When the electroencephalogram DB 27 exists on another computer, the generation unit 26 can refer to the electroencephalogram DB 27 through communication with the computer via the communication unit 4.

Further, for example, the generation unit 26 inputs the feature quantity of the subject's electroencephalogram to the discriminator learned from the feature quantity sample of the electroencephalogram measured in advance from a population of patients other than the AESD patient or ASED. It is also possible to automatically determine whether or not AESD. The discriminator can be generated using a known pattern recognition method such as SVM (Support Vector Machine). This discriminator may be realized on the diagnosis support apparatus 20 or may be realized on another computer.
It is also possible to automatically determine whether or not the subject has AESD with sequelae by making the population an AESD patient with sequelae. At this time, the production | generation part 26 can produce | generate the discrimination information which shows the discrimination | determination result with a character string or a numerical value. For example, the generating unit 26 includes discrimination information including a character string such as “the subject has a high possibility of convulsion-type acute encephalopathy” or “the subject has a high possibility of convulsion-type acute encephalopathy with a sequela”. Can be generated.

[Operation example / Information processing flow]
The diagnosis support apparatus 20 according to the second embodiment executes the information processing flow illustrated in FIGS. 20 and 21 when the diagnosis support program 7 is executed by the CPU 1 as described above. In other words, the CPU 1 loads and executes the diagnosis support program 7 from the memory 2 to realize an information processing flow as shown in FIGS. 20 and 21 in cooperation with other hardware elements.
FIG. 20 is a flowchart showing the overall operation of the diagnosis support apparatus 20 in the second embodiment. FIG. 21 is a flowchart showing detailed operations of (S54) and (S55) shown in FIG.

The CPU 1 (diagnosis support apparatus 10) acquires time-series data of electroencephalograms measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion (initial stage of disease) (S51). CPU1 may acquire site | part ID which is the identification number of the site | part where the brain wave was measured with the time series data of the brain wave. The specific content of (S51) is the same as the processing content of the data acquisition unit 21. For example, the CPU 1 can acquire time-series data of an electroencephalogram measured by an electroencephalograph from the electroencephalograph, another computer, or a portable recording medium via the communication unit 4. When the diagnosis support apparatus 20 is an electroencephalograph, the CPU 1 acquires time series data of electroencephalograms by processing a potential difference signal obtained from the electrodes.
As described above, the number of electroencephalogram measurement sites is not limited, but in this embodiment, the CPU 1 acquires time-series data of electroencephalograms measured at each of a plurality of predetermined sites.

  Subsequently, the CPU 1 accepts a user input relating to the extraction time point of the electroencephalogram time-series data (S52). The specific content of (S52) is the same as the processing content of the reception part 22. FIG. For example, the CPU 1 displays the time series data of the brain waves of each of the plurality of predetermined sites acquired in (S51) on the display device, and the extraction time point in each time series data according to the user operation for the display Receive input information about.

  Based on the user input received in (S52), the CPU 1 extracts time-series data of a time interval corresponding to the extraction time point indicated by the user input from the time-series data of brain waves measured at a plurality of predetermined sites. (S53). By this processing, the target electroencephalogram data is specified for each measurement site. The method for extracting the target electroencephalogram data is as described for the data acquisition unit 21.

  CPU1 selects the section pair in a measurement part, and the section pair between measurement parts regarding the target electroencephalogram data of a plurality of measurement parts extracted (S54) and (S55). In the present embodiment, as described above, the CPU 1 selects (N × S) section pairs in the measurement site and selects [{N (N−1) / 2} × S] section pairs between the measurement sites. To do. Specific contents of (S54) and (S55) will be described later with reference to FIG. The specific contents of (S54) and (S55) are the same as the processing contents of the selection unit 23.

The CPU 1 calculates the correlation coefficient of the time series data of each single time section as the time correlation coefficient for each of the section pairs in the measurement region selected in (S54) (S56).
Further, the CPU 1 calculates the correlation coefficient of the time series data of each single time section as the spatial correlation coefficient for each of the section pairs between the measurement sites selected in (S55) (S57).
The specific contents of (S56) and (S57) are the same as the processing contents of the calculation unit 24. In this embodiment, (N × S) time correlation coefficients and [{N (N-1) / 2} × S] spatial correlation coefficients.

  The CPU 1 obtains the frequency information of the correlation coefficient as the feature quantity of the subject's brain wave by counting the occurrence frequency for each time correlation coefficient or each spatial correlation coefficient calculated in (S56) (S58). In the present embodiment, the CPU 1 acquires the frequency information of the time correlation coefficient by adding up the occurrence frequencies for each time correlation coefficient counted for each measurement site, and the spatial phase counted for each measurement site pair. The frequency information of the spatial correlation coefficient is acquired by adding the occurrence frequency for each number of relationships. In this way, the CPU 1 acquires the frequency information of the time correlation coefficient and the frequency information of the spatial correlation coefficient separately, and the two frequency information may be used as the feature amount, or one frequency obtained by integrating them. Information may be acquired as the feature amount. The specific content of (S58) is the same as the processing content of the distribution acquisition unit 25.

  CPU1 produces | generates the discrimination | determination information regarding a test subject's AESD using the feature-value acquired by (S58) (S59). The discrimination information is as described above, and the specific content of (S59) is the same as the processing content of the generation unit 26. In (S59), the CPU 1 may further include a step of comparing the electroencephalogram feature quantity sample measured in advance from the population with the subject's electroencephalogram feature quantity obtained in (S58). By this comparison, it is possible to automatically determine whether or not the subject has AESD or whether or not the subject has AESD with a sequelae.

Next, detailed operations of (S54) and (S55) will be described with reference to FIG.
Selection of the section pair in the measurement site (S54) is executed as follows.
CPU1 specifies the target electroencephalogram data of one measurement site | part from the target electroencephalogram data of the some measurement site | part extracted by (S53) (S60).
Subsequently, the CPU 1 randomly selects a predetermined number (S) of pairs of single time intervals in the target electroencephalogram data specified in (S60) (S61).

The CPU 1 includes, in the (2 × S) single time intervals selected in (S61), the time series data in the single time interval includes a single amplitude value that exceeds the upper threshold or falls below the lower threshold. Existence of one hour section is confirmed (S62). That is, the CPU 1 confirms the existence of a single time interval including noise (S62). The upper threshold and the lower threshold are as described above.
When there is no single time interval including noise (S62; NO), the CPU 1 executes (S65) without executing (S63) and (S64). When there is a single time interval including noise (S62; YES), the CPU 1 eliminates the single time interval including the noise (S63), and instead selects a new single time interval (S64). ).

  The CPU 1 does not have a single time interval including noise (S62; NO), that is, if S pairs of single time intervals not including noise are selected, the CPU 1 has not yet been specified in (S60) ( The presence of an unprocessed measurement site is confirmed (S65). When the CPU 1 finishes selecting S single time interval pairs for the target electroencephalogram data of all the measurement sites (S65; NO), it ends the processing of FIG. On the other hand, if there is a measurement part that has not yet been specified (unprocessed) in (S60) (S65; YES), CPU 1 specifies the target electroencephalogram data of the unprocessed measurement part (S60), (S61) ) And subsequent steps.

  In this way, by selecting a predetermined number of pairs in a single time interval that does not contain noise and calculating the correlation coefficient for each such pair, the characteristic amount of the subject's brain wave acquired in (S58) is calculated. The accuracy can be increased, and consequently the accuracy of the discrimination information generated in (S59) can be improved.

[Operation and effect of the second embodiment]
In the second embodiment, time series data of electroencephalograms measured at a plurality of predetermined parts are acquired, and a predetermined number of pairs of single time intervals at different times or different measurement parts are selected from the time series data. Then, for each pair of single time intervals, a correlation coefficient between time series data is calculated as a time correlation coefficient or a spatial correlation coefficient, and for each measurement site or each measurement site pair, for each time correlation coefficient or for each space The occurrence frequency for each correlation coefficient is counted. Further, the occurrence frequencies counted for each measurement site or each measurement site pair are added together, and the frequency information of the calculated correlation coefficient is used as a feature quantity of the subject's brain wave, thereby generating discrimination information regarding the subject's AESD. The
It has been confirmed by the present inventors that it is possible to determine whether or not the AESD is based on the characteristic amount of the electroencephalogram acquired in this way, and it is also possible to determine whether or not the AESD has residual sequelae. ing. Since the feature amount is frequency information of the correlation coefficient, according to the second embodiment, it is possible to quantitatively determine AESD. As a result, as in the first embodiment, according to the second embodiment, it is possible to perform quantitative discrimination prediction related to AESD, which is difficult in other examinations and visual diagnosis of the electroencephalogram, which is an early stage of disease. Can support early diagnosis of AESD.

[Modification of Second Embodiment]
The diagnosis support apparatus 20 described above may not include the reception unit 22. In this case, (S52) may not be executed in the flowchart of FIG. For example, a section with relatively low noise may be automatically specified by a computer, or target electroencephalogram data may be arbitrarily extracted from measured electroencephalogram time-series data regardless of the presence or absence of noise. .
Moreover, the above-described diagnosis support apparatus 20 does not need to exclude and reselect a single time interval including noise. In this case, (S62), (S63), and (S64) may not be executed in the flowchart of FIG.

  In the second embodiment described above, discrimination information related to AESD is generated, but discrimination information related to other encephalopathy such as FS, MERS, and EP (Epilepsy) can also be generated. In this case, a brain wave feature amount sample obtained in advance from a patient with the encephalopathy type is stored in the electroencephalogram DB 27, and the subject's brain wave feature amount and the brain wave feature amount of each encephalopathy type are stored. By comparing each of them and obtaining a similarity, information indicating a candidate for encephalopathy can be presented.

  FIG. 22 is a flowchart showing a schematic operation of the diagnosis support apparatus 20 in the modification of the second embodiment. In this modification, for each type of encephalopathy, an electroencephalogram DB 271 storing an electroencephalogram feature amount sample acquired in advance from a patient with the encephalopathy type, and an electroencephalogram feature amount sample acquired in advance from an AESD patient in which the sequelae remain. The stored brain wave DB 272 is used. The electroencephalogram DBs 271 and 272 may be provided by the diagnosis support apparatus 20 or may be provided by another computer.

The generation unit 26 of the diagnosis support apparatus 20 automatically determines the type of encephalopathy by comparing the feature quantity of the subject's brain wave with the feature quantity sample stored in the brain wave DB 271 (S71). A known pattern recognition technique can also be used for discrimination of this encephalopathy type. The production | generation part 26 produces | generates and shows the discriminating information which shows the candidate of the name of the encephalopathy type | mold corresponding to a test subject's symptom (S72).
The doctor determines whether or not the subject is AESD with reference to the discrimination information.
When the subject determines that AESD, the generation unit 26 automatically determines the sequelae by comparing the feature quantity of the subject's brain wave with the feature quantity sample stored in the brain wave DB 272 (S75). The generation unit 26 generates and presents discrimination information indicating whether or not the sequelae remain in the subject or the predicted level of the sequelae based on the discrimination result (S76).
Thus, by using the electroencephalogram feature amount acquired by the method of the second embodiment described above, encephalopathy discrimination and sequelae discrimination can be performed.

  An example is given to the following and the content regarding the above-mentioned second embodiment is explained in detail. The present invention is not limited by the following examples.

  The present inventors can discriminate AESD by using either one or both of the frequency information of the temporal correlation coefficient and the frequency information of the spatial correlation coefficient acquired by the above-described method as the feature quantity of the electroencephalogram. It was proved as follows.

  First, time series data of past EEGs measured from 26 pediatric patients sent to the hospital by convulsion were used. The breakdown of pediatric patients is 14 boys and 12 girls, 12 patients diagnosed with AESD, 9 patients diagnosed with FS, 3 patients diagnosed with MERS, and EP (Epileptic ( Two patients are diagnosed with Epilepsy)). Of the 12 patients diagnosed with AESD, 4 patients have sequelae and 8 patients have sequelae.

  By applying the above-described method to the above-described time-series data of the electroencephalogram of each pediatric patient, the feature values of each pediatric patient were acquired. In this embodiment, a temporal correlation coefficient histogram, a spatial correlation coefficient histogram, and an integrated histogram obtained by integrating them are acquired as feature quantities of brain waves.

Specifically, the extraction time point (central time point) is designated for the time series data of the brain waves of each child patient (S52), and the time series data for a total of 2 minutes before and after the extraction time is the target electroencephalogram data. (S53). In this example, the target electroencephalogram data was extracted in the same time interval for the electroencephalogram time-series data of all measurement sites.
Then, the time length of the single time interval was set to 1 second, and 10,000 unit time interval pairs (S = 10000) were selected for each measurement site or each measurement site pair (S54) and (S55). Correlation coefficients were calculated for each pair of the selected unit time intervals (S56) and (S57).
In addition, the number of occurrences for each time correlation coefficient or each spatial correlation coefficient is counted for each measurement site or each measurement site pair, and these are added together to obtain a histogram of the time correlation coefficient and the spatial correlation coefficient. A histogram and a histogram integrating them were acquired (S58). Each histogram is normalized so that the number of bins is 20, the bin width is 0.1, and the norm is 1. That is, the frequency of occurrence for each correlation coefficient in increments of 0.1 was counted.
In the present embodiment, the processes (S62), (S63), and (S64) for reselecting a single time interval including noise were not executed.

Verification of discrimination regarding AESD was performed using each histogram acquired in this manner as a feature amount. In this verification, pattern recognition was performed by a discriminator using linear SVM for the feature values of the electroencephalograms of 26 pediatric patients. Specifically, feature values of one patient out of 26 patients are extracted as test examples, and feature values of the remaining patients are used as training examples. The verification was performed by repeating this so that the feature values of all patients become test cases once (the leave one-out method).
As a result, it was possible to obtain a high recognition rate (all 100%) by two-class discrimination between AESD and others (FS, MERS, EP). That is, according to the present example, it was proved that the subject's AESD can be discriminated with high accuracy by using the electroencephalogram feature amount acquired by the above-described method.

FIG. 23 is a diagram illustrating a histogram of temporal correlation coefficients, FIG. 24 is a diagram illustrating a histogram of spatial correlation coefficients, and FIG. 25 illustrates a histogram of temporal correlation coefficients and a histogram of spatial correlation coefficients. It is a figure which shows the integrated histogram. 23, 24 and 25, histograms of three pediatric patients diagnosed with AESD (ID3, ID7 and ID9) and three pediatric patients diagnosed with other (EP, MERS, FS) ( ID42, ID47, ID53) histograms are shown respectively.
23, 24, and 25, the temporal correlation coefficient histogram shape (temporal correlation coefficient frequency information), the spatial correlation coefficient histogram shape (spatial correlation coefficient frequency information), It has been demonstrated that any of the integrated histogram shapes (frequency information for both temporal and spatial correlation coefficients) can be used to distinguish AESD or other.
In addition, these figures demonstrate that it is possible to determine the subject's AESD using the frequency information of the time correlation coefficient or the spatial correlation coefficient itself as the determination information.

In the plurality of flowcharts used in the above description, a plurality of information processing steps are described in order, but the information processing execution order in each embodiment is not limited to the description order. In each embodiment, the illustrated order can be changed within a range that does not hinder the contents. For example, (S54) and (S55) shown in FIG. 20 may be executed in parallel, or the execution order may be switched. Similarly, (S56) and (S57) may be executed in parallel, or the execution order may be switched.
Moreover, each above-mentioned embodiment and each modification can be combined in the range with which the content does not conflict.

  Part or all of the above contents can also be specified as follows. However, the above-mentioned content is not limited to the following description.

(Appendix 1) Data acquisition means for acquiring time-series data of brain waves measured at a predetermined site on the subject's head within a predetermined time after convulsion,
Calculating means for calculating a power value of a predetermined frequency band for the predetermined portion by performing frequency analysis on the acquired time-series data;
Using the calculated power value, generating means for generating discrimination information on convulsive status acute encephalopathy or febrile seizure regarding the subject,
A diagnostic support device for convulsion type acute encephalopathy comprising:
(Supplementary note 2) The calculation means calculates a power value for each of a plurality of frequency bands predetermined for the predetermined part,
The generating means generates the discrimination information using a plurality of power values calculated for the plurality of frequency bands.
The diagnosis support apparatus according to appendix 1.
(Supplementary Note 3) The data acquisition unit acquires time series data of brain waves at a plurality of predetermined sites on the subject's head,
The calculating means calculates the power value of one or a plurality of frequency bands determined in advance for each predetermined part in each of the electroencephalograms of the plurality of predetermined parts,
The generating means generates the discrimination information using a plurality of power values calculated for the plurality of predetermined portions and the one or a plurality of frequency bands.
The diagnosis support apparatus according to appendix 1.
(Supplementary Note 4) Different frequency bands are predetermined for the first predetermined portion and the second predetermined portion included in the plurality of predetermined portions.
The diagnosis support apparatus according to appendix 3.
(Supplementary Note 5) The generating means includes
By comparing the learning data of convulsive intussusception type acute encephalopathy and the learning data of febrile seizure with the calculated power value, the power value is statistically classified into acute encephalopathy or febrile seizure.
Generating the discrimination information including the result of the statistical classification or the basis data of the statistical classification;
The diagnosis support apparatus according to any one of appendices 1 to 4.
(Appendix 6) The generating means includes:
By the comparison, the degree of similarity to the learning data of convulsion intussusception type acute encephalopathy and the degree of similarity to the learning data of febrile seizure are calculated,
Generating the discrimination information including the calculated similarity,
The diagnosis support apparatus according to appendix 5.
(Additional remark 7) Based on the electrode arrangement | positioning prescribed | regulated by the international 10-20 method, the said predetermined site | part is between F3 and C3, between F4 and C4, between C3 and P3, or between C4 and P4. Between
The diagnosis support apparatus according to any one of appendices 1 to 6.
(Appendix 8) The data acquisition means acquires a plurality of time series data of brain waves measured at different times within the predetermined time,
The calculation means calculates the power value by analyzing the frequency of each of the plurality of time series data,
The generating means generates the discrimination information by further using information on the time change of the calculated power value.
The diagnosis support apparatus according to any one of appendices 1 to 7.
(Supplementary note 9) Data acquisition means for acquiring time series data of brain waves measured at a predetermined site on the subject's head at different times within a predetermined time after convulsion,
Calculating means for calculating a power value of a predetermined frequency band for each of the predetermined portions by performing frequency analysis on each of the plurality of acquired time series data;
Using the time variation of the calculated power value, generating means for generating discrimination information on convulsive status encephalopathy or febrile convulsions for the subject,
A diagnostic support device for convulsion type acute encephalopathy comprising:
(Supplementary Note 10) Acquire time series data of brain waves measured at a predetermined site on the subject's head within a predetermined time after convulsion,
By performing frequency analysis on the acquired time-series data, a power value in a predetermined frequency band for the predetermined part is calculated,
Using the calculated power value, generating discriminating information on convulsive status acute encephalopathy or febrile seizure regarding the subject,
Diagnosis support program for convulsion-type acute encephalopathy that causes a computer to do this.
(Supplementary note 11) The calculation of the power value is to calculate a power value for each of a plurality of frequency bands predetermined for the predetermined part,
The generation of the discrimination information generates the discrimination information using a plurality of power values calculated for the plurality of frequency bands.
The diagnostic support program according to attachment 10.
(Additional remark 12) The acquisition of the time series data respectively acquires time series data of electroencephalograms at a plurality of predetermined sites on the subject's head,
The calculation of the power value is to calculate a power value of one or a plurality of frequency bands determined in advance for each predetermined part in each electroencephalogram of the plurality of predetermined parts,
The generation of the discrimination information generates the discrimination information using a plurality of power values calculated for the plurality of predetermined portions and the one or a plurality of frequency bands.
The diagnostic support program according to attachment 10.
(Supplementary Note 13) Different frequency bands are predetermined for the first predetermined portion and the second predetermined portion included in the plurality of predetermined portions.
The diagnostic support program according to attachment 12.
(Supplementary Note 14) By comparing the learning data of convulsive intussusception type acute encephalopathy and the learning data of febrile convulsion with the calculated power value, the power value is statistically classified into convulsive intussusception type acute encephalopathy or febrile seizure,
The computer further executes:
The generation of the discrimination information generates the discrimination information including a result of the statistical classification or ground data of the statistical classification.
The diagnostic support program according to any one of appendices 10 to 13.
(Additional remark 15) By the said comparison, the similarity with the learning data of the convulsion intussusception type acute encephalopathy and the learning data of the febrile convulsions are respectively calculated.
The computer further executes:
The generation of the discrimination information generates the discrimination information including the calculated similarity.
The diagnostic support program according to appendix 14.
(Additional remark 16) The said predetermined part is based on the electrode arrangement | positioning prescribed | regulated by the international 10-20 method, between F3 and C3, between F4 and C4, between C3 and P3, or between C4 and P4, Between
The diagnostic support program according to any one of appendices 10 to 15.
(Supplementary Note 17) The acquisition of the time series data is obtained by acquiring a plurality of time series data of brain waves measured at different times within the predetermined time,
The calculation of the power value calculates the power value by performing frequency analysis on each of the plurality of time series data,
The generation of the discrimination information generates the discrimination information by further using information on the time change of the calculated power value.
The diagnostic support program according to any one of appendices 10 to 16.
(Supplementary Note 18) Acquire time series data of brain waves measured at a predetermined site on the subject's head at different times within a predetermined time after convulsion,
By performing frequency analysis on each of the obtained time series data, each power value of a predetermined frequency band for the predetermined part is calculated,
Using the time variation of the calculated power value, to generate discrimination information for convulsive acute encephalopathy or febrile seizure for the subject,
Diagnosis support program for convulsion-type acute encephalopathy that causes a computer to do this.
(Supplementary note 19) Acquire time series data of brain waves measured at a predetermined site on the subject's head within a predetermined time after convulsion,
By performing frequency analysis on the acquired time series data, a power value in a frequency band predetermined for the predetermined part is calculated.
Diagnosis support method for convulsive status-type acute encephalopathy.
(Supplementary note 20) Each time-series data of brain waves measured at a predetermined site on the subject's head at different times within a predetermined time after convulsion is obtained,
By performing frequency analysis on each of the obtained time series data, each power value of a predetermined frequency band for the predetermined part is calculated,
Obtaining information on time variation of the calculated power value;
Diagnosis support method for convulsive status-type acute encephalopathy.

(Supplementary Note 21) Data acquisition means for acquiring time series data of brain waves measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion,
Calculation means for calculating a plurality of correlation coefficients of at least one of the time correlation coefficient and the spatial correlation coefficient in the acquired time series data;
A distribution acquisition means for acquiring frequency information of the calculated correlation coefficient as a feature quantity of the brain wave of the subject;
Using the acquired feature amount, generating means for generating discrimination information related to the convulsive status acute encephalopathy of the subject,
A diagnostic support device for convulsion type acute encephalopathy comprising:
(Supplementary Note 22) Selection means for selecting a predetermined number of pairs of single time intervals at different times or different measurement sites from the acquired time series data,
Further comprising
The calculating means calculates, for each of the selected pairs, a correlation coefficient of time series data of each single time interval as the temporal correlation coefficient or the spatial correlation coefficient,
The distribution acquisition means acquires the frequency information by counting the occurrence frequency for each correlation coefficient based on the correlation coefficient calculated for each selected pair.
The diagnostic support apparatus for convulsive status acute encephalopathy according to appendix 21.
(Supplementary Note 23) The data acquisition means acquires time series data of brain waves measured at each of a plurality of predetermined sites,
The selection means selects each pair of the predetermined number of single time intervals for each measurement site or each measurement site pair,
The distribution acquisition means acquires the frequency information by adding the occurrence frequency for each correlation coefficient counted for each measurement site or each measurement site pair,
The diagnostic support device for convulsive status acute encephalopathy according to appendix 22.
(Supplementary Note 24) The selection means includes:
Randomly select the predetermined number of single time interval pairs;
From the selected single time intervals, the single time interval in which the time series data of the single time interval includes an amplitude value that exceeds the upper threshold or falls below the lower threshold,
Reselecting the single time interval until the number of pairs of single time intervals is the predetermined number;
The diagnostic support apparatus for convulsive status acute encephalopathy according to appendix 22 or 23.
(Supplementary Note 25) Accepting means for accepting user input related to extraction time of time series data of electroencephalogram,
The data acquisition means is based on the accepted user input, and from the time series data of the electroencephalogram measured at the one or more predetermined sites, at a time interval corresponding to the extraction time point indicated by the user input Extract series data,
The diagnostic support device for convulsive status acute encephalopathy according to any one of appendices 21 to 24.
(Additional remark 26) The said production | generation means discriminate | determines whether the said test subject is a convulsion type | formula acute encephalopathy by comparing the feature-value sample of the electroencephalogram previously acquired from the population with the feature-value of the said test subject's electroencephalogram. Generating information or discrimination information as to whether or not the subject has convulsive status encephalopathy with sequelae,
The diagnostic support device for convulsive status acute encephalopathy according to any one of appendices 21 to 25.
(Supplementary note 27) Obtain time-series data of brain waves measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion,
Calculating a plurality of correlation coefficients of at least one of the time correlation coefficient and the spatial correlation coefficient in the acquired time series data;
Obtaining the frequency information of the calculated correlation coefficient as a feature quantity of the subject's brain wave,
Using the acquired feature amount, to generate discrimination information regarding the convulsive status acute encephalopathy of the subject,
Diagnosis support program for convulsion-type acute encephalopathy that causes a computer to do this.
(Supplementary Note 28) Select a predetermined number of pairs of single time intervals at different times or different measurement sites from the acquired time series data.
The computer further executes:
The calculation of the correlation coefficient, for each of the selected pair, to calculate the correlation coefficient of the time series data of each single time interval as the time correlation coefficient or the spatial correlation coefficient, respectively,
The acquisition of the frequency information acquires the frequency information by counting the occurrence frequency for each correlation coefficient based on the correlation coefficient calculated for each selected pair.
The diagnostic support program for convulsion type acute encephalopathy according to appendix 27.
(Supplementary note 29) The acquisition of the time series data is obtained by acquiring time series data of brain waves measured at each of a plurality of predetermined sites,
The selection of the pair of single time intervals is to select the predetermined number of single time interval pairs for each measurement site or each measurement site pair,
The frequency information is acquired by adding the occurrence frequency for each correlation coefficient counted for each measurement site or each measurement site pair,
The diagnostic support program for convulsion type acute encephalopathy according to appendix 28.
(Supplementary Note 30) The selection of the pair of single time intervals is as follows.
Randomly select the predetermined number of single time interval pairs;
Removing a single time interval in which the time series data of the single time interval includes an amplitude value that exceeds the upper threshold or falls below the lower threshold from the plurality of selected single time intervals;
Reselecting the single time interval until the number of pairs of single time intervals is the predetermined number;
The diagnostic support program for the convulsive status encephalopathy according to appendix 28 or 29.
(Supplementary Note 31) Accepting user input regarding the time of extraction of time series data of brain waves,
The computer further executes:
The acquisition of the time series data is based on the received user input, from the time series data of the electroencephalogram measured at the one or more predetermined sites, the time series of the time interval corresponding to the extraction time point indicated by the user input Extract data,
The diagnostic support program for convulsive status acute encephalopathy according to any one of appendices 27 to 30.
(Additional remark 32) The feature-value sample of the electroencephalogram measured beforehand from the population and the feature-value of the subject's electroencephalogram are compared.
The computer further executes:
The discriminating information is generated based on the comparison result to generate discriminating information as to whether or not the subject has convulsive acute encephalopathy or discriminating information as to whether or not the subject has seizure-type acute encephalopathy. To
The diagnostic support program for convulsive status acute encephalopathy according to any one of appendices 27 to 31.
(Supplementary Note 33) Acquire time-series data of brain waves measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion,
Calculating a plurality of correlation coefficients of at least one of the time correlation coefficient and the spatial correlation coefficient in the acquired time series data;
Obtaining frequency information of the calculated correlation coefficient as a feature quantity of the subject's brain wave,
Diagnosis support method for convulsive status-type acute encephalopathy.

1 CPU
2 Memory 3 Input / output I / F unit 4 Communication unit 7 Diagnosis support program 10, 20 Diagnosis support device 11, 21 Data acquisition unit 12 Calculation unit 13, 26 Generation unit 22 Reception unit 23 Selection unit 24 Calculation unit 25 Distribution acquisition unit 27 EEG database (DB)

Claims (19)

Data acquisition means for acquiring time-series data of electroencephalograms measured at a predetermined site on the subject's head within a predetermined time after convulsion;
Calculating means for calculating a power value of a predetermined frequency band for the predetermined portion by performing frequency analysis on the acquired time-series data;
Using the calculated power value, generating means for generating discrimination information on convulsive status acute encephalopathy or febrile seizure regarding the subject,
A diagnostic support device for convulsion type acute encephalopathy comprising: The calculation means calculates a power value for each of a plurality of frequency bands predetermined for the predetermined part,
The generating means generates the discrimination information using a plurality of power values calculated for the plurality of frequency bands.
The diagnosis support apparatus according to claim 1. The data acquisition means respectively acquires time series data of brain waves at a plurality of predetermined sites on the subject's head,
The calculating means calculates the power value of one or a plurality of frequency bands determined in advance for each predetermined part in each of the electroencephalograms of the plurality of predetermined parts,
The generating means generates the discrimination information using a plurality of power values calculated for the plurality of predetermined portions and the one or a plurality of frequency bands.
The diagnosis support apparatus according to claim 1. Different frequency bands are predetermined for the first predetermined portion and the second predetermined portion included in the plurality of predetermined portions,
The diagnosis support apparatus according to claim 3. The generating means includes
By comparing the learning data of convulsive intussusception type acute encephalopathy and the learning data of febrile seizure with the calculated power value, the power value is statistically classified into acute encephalopathy or febrile seizure.
Generating the discrimination information including the result of the statistical classification or the basis data of the statistical classification;
The diagnosis support apparatus according to any one of claims 1 to 4. The generating means includes
By the comparison, the degree of similarity to the learning data of convulsion intussusception type acute encephalopathy and the degree of similarity to the learning data of febrile seizure are calculated,
Generating the discrimination information including the calculated similarity,
The diagnosis support apparatus according to claim 5. The predetermined portion is between F3 and C3, between F4 and C4, between C3 and P3, or between C4 and P4, based on the electrode arrangement defined in the International 10-20 Act. ,
The diagnosis support apparatus according to any one of claims 1 to 6. The data acquisition means acquires a plurality of time series data of brain waves measured at different times within the predetermined time,
The calculation means calculates the power value by analyzing the frequency of each of the plurality of time series data,
The generating means generates the discrimination information by further using information on the time change of the calculated power value.
The diagnosis support apparatus according to any one of claims 1 to 7. Data acquisition means for respectively acquiring time series data of electroencephalograms measured at a predetermined site on the subject's head at different times within a predetermined time after convulsions;
Calculating means for calculating a power value of a predetermined frequency band for each of the predetermined portions by performing frequency analysis on each of the plurality of acquired time series data;
Using the time variation of the calculated power value, generating means for generating discrimination information on convulsive status encephalopathy or febrile convulsions for the subject,
A diagnostic support device for convulsion type acute encephalopathy comprising: Obtain time-series data of brain waves measured at a predetermined site on the subject's head within a predetermined time after convulsion,
By performing frequency analysis on the acquired time-series data, a power value in a predetermined frequency band for the predetermined part is calculated,
Using the calculated power value, generating discriminating information on convulsive status acute encephalopathy or febrile seizure regarding the subject,
Diagnosis support program for convulsion-type acute encephalopathy that causes a computer to do this. Obtain time-series data of brain waves measured at a predetermined site on the subject's head within a predetermined time after convulsion,
By performing frequency analysis on the acquired time series data, a power value in a frequency band predetermined for the predetermined part is calculated.
Diagnosis support method for convulsive status-type acute encephalopathy. Data acquisition means for acquiring time-series data of electroencephalograms measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion;
Calculation means for calculating a plurality of correlation coefficients of at least one of the time correlation coefficient and the spatial correlation coefficient in the acquired time series data;
A distribution acquisition means for acquiring frequency information of the calculated correlation coefficient as a feature quantity of the brain wave of the subject;
Using the acquired feature amount, generating means for generating discrimination information related to the convulsive status acute encephalopathy of the subject,
A diagnostic support device for convulsion type acute encephalopathy comprising: Selection means for selecting a predetermined number of pairs of single time intervals at different times or different measurement sites from the acquired time series data,
Further comprising
The calculating means calculates, for each of the selected pairs, a correlation coefficient of time series data of each single time interval as the temporal correlation coefficient or the spatial correlation coefficient,
The distribution acquisition means acquires the frequency information by counting the occurrence frequency for each correlation coefficient based on the correlation coefficient calculated for each selected pair.
The diagnostic support apparatus for the convulsive stack type acute encephalopathy according to claim 12. The data acquisition means acquires time series data of brain waves measured at each of a plurality of predetermined sites,
The selection means selects each pair of the predetermined number of single time intervals for each measurement site or each measurement site pair,
The distribution acquisition means acquires the frequency information by adding the occurrence frequency for each correlation coefficient counted for each measurement site or each measurement site pair,
The diagnostic support apparatus of the convulsive status-type acute encephalopathy of Claim 13. The selection means includes
Randomly select the predetermined number of single time interval pairs;
From the selected single time intervals, the single time interval in which the time series data of the single time interval includes an amplitude value that exceeds the upper threshold or falls below the lower threshold,
Reselecting the single time interval until the number of pairs of single time intervals is the predetermined number;
The diagnostic support apparatus of the convulsive status type acute encephalopathy according to claim 13 or 14. Accepting means for accepting user input regarding the time of extraction of time series data of brain waves;
The data acquisition means is based on the accepted user input, and from the time series data of the electroencephalogram measured at the one or more predetermined sites, at a time interval corresponding to the extraction time point indicated by the user input Extract series data,
The convulsive stack type acute encephalopathy diagnosis support apparatus according to any one of claims 12 to 15. The generation means compares the brain wave feature quantity sample acquired in advance from a population with the subject's brain wave feature quantity, thereby determining whether or not the subject has convulsive acute encephalopathy or the subject. Generate discriminating information on whether or not convulsion-type acute encephalopathy with sequelae,
The convulsive stack type acute encephalopathy diagnosis support apparatus according to any one of claims 12 to 16. Obtain EEG time-series data measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion,
Calculating a plurality of correlation coefficients of at least one of the time correlation coefficient and the spatial correlation coefficient in the acquired time series data;
Obtaining the frequency information of the calculated correlation coefficient as a feature quantity of the subject's brain wave,
Using the acquired feature amount, to generate discrimination information regarding the convulsive status acute encephalopathy of the subject,
Diagnosis support program for convulsion-type acute encephalopathy that causes a computer to do this. Obtain EEG time-series data measured at one or more predetermined sites on the subject's head within a predetermined time after convulsion,
Calculating a plurality of correlation coefficients of at least one of the time correlation coefficient and the spatial correlation coefficient in the acquired time series data;
Obtaining frequency information of the calculated correlation coefficient as a feature quantity of the subject's brain wave,
Diagnosis support method for convulsive status-type acute encephalopathy.

Images