KR20220146124A - Apparatus for diagnosis and prediction of tinnitus based on electroencephalography, method for processing electroencephalography signal and computer program for the same - Google Patents

Abstract

An apparatus for diagnosing and predicting tinnitus based on brainwaves comprises: a stimulus presentation unit configured to output an auditory stimulus signal including a standard stimulus and a target stimulus; a measuring unit configured to measure a brainwave signal of a subject in a resting state or in an information processing state of listening to the auditory stimulation signal; and an analysis unit configured to extract one or more features of the measured brainwave signal and classify the brainwave signal by multivariate pattern analysis (MVPA) using the one or more features to determine whether the subject has a tinnitus symptom. The apparatus for diagnosing and predicting tinnitus based on brainwaves, if used, can accurately diagnose and/or predict the tinnitus only by measuring the brainwaves in a short period of time through any one among an approach method adopting a top-down information processing abnormality of the patient's auditory center, an approach method adopting a bottom-up information processing abnormality of patient's auditory peripheral nerves, or a resting state, depending on the cause of the tinnitus, and can diagnose the tinnitus quickly and accurately even if a skilled experimenter does not read the brainwave signal by applying MVPA technology to various methods based on a time axis, a frequency axis, an event-related potential (ERP), a time-frequency map, a cross-frequency coupling (CFC), etc. to determine the tinnitus patient.

Description

EEG-based tinnitus diagnosis and prediction device, EEG signal processing method, and computer program therefor

Embodiments relate to an EEG-based tinnitus diagnosis and prediction apparatus, an EEG signal processing method, and a computer program therefor. In more detail, the embodiments measure EEG signals during a resting state, bottom-up information processing, or top-down information processing paradigm tasks, and perform multivariate pattern analysis (MVPA). ) technology to analyze the EEG characteristics of patients with tinnitus compared to normal people by determining the activation pattern and timing of EEG with different states.

Tinnitus refers to a symptom in which sound is sensed in the ear or in the head even when there is no external sound stimulus. In the United States, 25.3% of adults have experienced tinnitus, and it is known that 7.9% of them frequently experience tinnitus. In particular, it is predicted that the proportion of the population suffering from tinnitus disease due to high-frequency auditory signals is expected to increase worldwide as the environment in which people wear earphones increases in recent years.

Meanwhile, with the development of EEG signal processing technology, technologies for diagnosing and predicting various types of diseases using EEG are being developed. Among them, as an example in which EEG signal processing is applied to tinnitus, International Publication No. WO2012/121450 discloses an apparatus for detecting tinnitus using EEG and a method for detecting tinnitus using the same.

However, the prior art disclosed in International Publication No. WO2012/121450 is a univariate method for diagnosing tinnitus by acquiring an Auditory Evoked Potential (AEP) EEG of a subject according to a stimulus sound and analyzing only the time axis analysis component of the EEG. ) method, so it has the disadvantage of being very vulnerable to EEG noise. In addition, the prior art has a limitation in that it is not possible to predict the tinnitus before the onset of only the diagnosis of tinnitus that has already occurred.

International Publication No. WO2012/121450

According to one aspect of the present invention, it is possible to diagnose a tinnitus patient with a high recognition rate using only EEG measurement, and by using Multivariate Pattern Analysis (MVPA) technology to determine the activation pattern and timing of EEG with different states. It is possible to provide an EEG-based tinnitus diagnosis and prediction device, an EEG signal processing method, and a computer program for extracting EEG signal characteristics of a patient with tinnitus that can be distinguished from a normal person and effectively predicting the occurrence of tinnitus.

The technical problems of the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the following description.

According to an aspect of the present invention, an EEG-based tinnitus diagnosis and prediction apparatus includes: a stimulus presentation unit configured to output an auditory stimulus signal including a standard stimulus and a target stimulus; a measuring unit configured to measure an EEG signal of a subject in a resting state or in an information processing state listening to the auditory stimulus signal; and an analysis unit configured to extract one or more features of the measured EEG signal and determine the presence or absence of tinnitus symptoms in the subject by classifying the EEG signal by multivariate pattern recognition using the one or more features.

In an embodiment, the stimulus presentation unit is further configured to generate the auditory stimulus signal by randomly mixing a standard stimulus having a first frequency and a target stimulus having a second frequency different from the first frequency.

In an embodiment, the analysis unit may include: a feature extraction unit configured to generate a pattern vector corresponding to the one or more characteristics using the brain wave signal; a pattern generating unit configured to generate a pattern signal including a training set and a testing set by using the plurality of pattern vectors corresponding to the symptoms of the subject; and a classifier configured to perform learning using the learning set of the pattern signal, and to determine a symptom of the subject by evaluating a learning result using the evaluation set.

In an embodiment, the pattern vector is a time axis analysis method, a frequency axis analysis method, a time-frequency map analysis method, an Event-Related Potential (ERP) analysis method, or cross-frequency tuning (Cross-Frequency Coupling; CFC) includes feature values extracted by one or more of the analysis methods.

In an embodiment, the pattern generator is further configured to generate the pattern signal using an average value of the pattern vectors generated from the plurality of EEG signals.

The EEG-based tinnitus diagnosis and prediction apparatus according to an embodiment further includes an input unit configured to receive an input signal input by a subject listening to the auditory stimulation signal in response to the stimulation signal.

According to an aspect of the present invention, there is provided an EEG signal processing method, comprising: measuring, by a signal processing device, an EEG signal of a subject in a resting state or an information processing state listening to an auditory stimulus signal including a standard stimulus and a target stimulus; extracting, by the signal processing device, one or more features of the measured EEG signal; and classifying, by the signal processing device, the EEG signal by multivariate pattern recognition using the one or more features to determine whether the subject has tinnitus symptoms.

The EEG signal processing method according to an embodiment further includes, before the step of measuring the EEG signal, the step of outputting the auditory stimulation signal by the signal processing device.

In one embodiment, the step of outputting the auditory stimulus signal includes, by the signal processing device, a standard stimulus having a first frequency and a target stimulus having a second frequency different from the first frequency by randomly mixing the auditory stimulus generating a signal.

In an embodiment, the classifying the EEG signal may include: generating, by the signal processing device, a pattern vector corresponding to the one or more features by using the EEG signal; generating, by the signal processing device, a pattern signal including a learning set and an evaluation set using a plurality of the pattern vectors corresponding to the symptoms of the subject; performing, by the signal processing device, learning of the classifier using the training set of the pattern signal; and determining, by the signal processing device, a symptom of the subject by using the learning result of the classifier.

In an embodiment, the generating of the pattern signal includes generating, by the signal processing apparatus, the pattern signal using an average value of the pattern vectors generated from the plurality of EEG signals.

The EEG signal processing method according to an embodiment further includes, by the signal processing apparatus, receiving an input signal input by a subject listening to the auditory stimulation signal in response to the stimulation signal.

The computer program according to an aspect of the present invention is combined with hardware to execute the EEG signal processing method according to the above-described embodiments, and may be stored in a computer-readable recording medium.

According to the EEG-based tinnitus diagnosis and prediction apparatus and EEG signal processing method according to an aspect of the present invention, depending on the cause of tinnitus, an approach method due to a top-down information processing abnormality of the patient's auditory center and the patient's hearing There is an advantage in that it is possible to accurately diagnose and predict tinnitus only by measuring EEG in a short time through either the approach due to the bottom-up information processing abnormality of the peripheral nerve or the resting state.

The EEG-based tinnitus diagnosis and prediction device and EEG signal processing method according to an aspect of the present invention uses a technology that combines cognitive neuroscience and brain engineering, and multivariate pattern analysis (MVPA) technology is applied on the time axis, frequency axis, and By applying various methods such as Time-Frequency Map, Event-Related Potential (ERP), and Cross-Frequency Coupling (CFC) to identify patients with tinnitus, experienced experimenters can use EEG There is an advantage in that accurate diagnosis is possible even if the signal is not read.

Using the EEG-based tinnitus diagnosis and prediction device and EEG signal processing method according to an aspect of the present invention, the treatment success rate is increased before the onset of the patient through diagnosis and prediction of tinnitus, and the proportion of the population suffering from tinnitus disease due to the use of earphones, etc. It is expected to contribute to increasing social productivity and reducing medical costs in an environment that is expected to increase.

Effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a schematic block diagram of an EEG-based tinnitus diagnosis and prediction apparatus according to an exemplary embodiment.
2 is a conceptual diagram illustrating an examination environment using an EEG-based tinnitus diagnosis and prediction apparatus according to an exemplary embodiment.
3 is a flowchart illustrating each step of the EEG signal processing method according to an embodiment.
4 is a conceptual diagram illustrating an auditory stimulation signal applied by an EEG signal processing method according to an exemplary embodiment.
5 is a conceptual diagram for explaining an EEG signal analysis process by the EEG signal processing method according to an embodiment.
6A and 6B are results of analyzing the response of the subject to an oddball task for a top-down information processing test and a passive task for a bottom-up information processing test, respectively, by the EEG signal processing method according to an embodiment. is a graph representing
7A and 7B are graphs showing the results of analyzing the response of the subject to the target stimulus and the standard stimulus, respectively, by the EEG signal processing method according to the embodiment.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

1 is a schematic block diagram of an EEG-based tinnitus diagnosis and prediction apparatus according to an exemplary embodiment.

Referring to FIG. 1 , an EEG-based tinnitus diagnosis and prediction apparatus 1 includes a stimulus presentation unit 10 , a measurement unit 20 , and an analysis unit 40 . In an embodiment, the EEG-based tinnitus diagnosis and prediction apparatus 1 includes an input unit 30 for receiving an input of a subject. Also, in an embodiment, the analysis unit 40 includes a feature extraction unit 401 , a pattern generation unit 402 , and a classifier 403 .

The apparatus described herein may be entirely hardware, or may have aspects that are partly hardware and partly software. For example, the EEG-based tinnitus diagnosis and prediction device 1 of the present specification and each unit (10, 20, 30, 40) included therein processes data in a specific format and content or/or uses an electronic communication method. Hardware for sending and receiving and software related thereto may be collectively referred to. As used herein, terms such as “unit”, “module”, “device”, “terminal”, “server” or “system” are intended to refer to a combination of hardware and software driven by the hardware. For example, the hardware may be a data processing device including a CPU or other processor. In addition, software driven by hardware may refer to a running process, an object, an executable file, a thread of execution, a program, and the like.

In addition, each element constituting the EEG-based tinnitus diagnosis and prediction apparatus 1 according to the present embodiment is not necessarily intended to refer to separate devices physically separated from each other. That is, the stimulus presentation unit 10, the measurement unit 20, the input unit 30, and the analysis unit 40 of FIG. 1 use hardware constituting the EEG-based tinnitus diagnosis and prediction device 1 by the corresponding hardware. It is only functionally classified according to the operation, and each part does not necessarily have to be provided independently of each other. Of course, depending on the embodiment, it is also possible to implement one or more of each part of the EEG-based tinnitus diagnosis and prediction device 1 as a separate device physically separated from each other.

The stimulus presentation unit 10 is configured to output an auditory stimulus signal including a standard stimulus and a target stimulus. To this end, the stimulus presentation unit 10 may include a device such as a speaker for reproducing sound. The standard stimulus and the target stimulus may have different frequencies, which will be described in detail later. Also, the stimulus presentation unit 10 may include a display device for reproducing an image or an image provided along with the auditory stimulus.

The measurement unit 20 is configured to measure an EEG signal of a subject who listens to the auditory stimulation signal reproduced by the stimulation presentation unit 10 . To this end, the measurement unit 20 may include one or more sensors disposed to be worn on the subject's head or to be fixed adjacent to the subject's head. In the present specification, the EEG signal includes information measured from the brain by any brain imaging method. In an embodiment, the EEG signal may mean an EEG (Electroencephalogram) signal that is invasively specified using a device attached to the head surface of the subject.

The analyzer 40 may extract one or more features from the EEG signal measured by the measurement unit 20 and analyze the EEG signal through multivariate pattern analysis (MVPA) using one or more features. To this end, the analyzer 40 may include a feature extractor 401 for extracting a pattern vector from an EEG signal and a pattern generator 402 for generating a pattern signal using the pattern vectors. In addition, the analysis unit 40 uses a part of the pattern vectors of the pattern signal as a training set and uses the other part as a testing set to learn the classifier 403 to analyze the EEG signal. can

The analysis unit 40 may determine whether the subject has tinnitus symptoms based on the classification result of the classifier 403 for the EEG signal. For example, the analysis unit 40 may classify the pattern vector corresponding to the evaluation set using the learning result using the pattern vectors corresponding to the learning set, which determines whether the subject has tinnitus symptoms corresponding to the evaluation set. it means. The analysis process of the EEG signal by the analysis unit 40 will be described later in detail with reference to FIG. 5 .

2 is a conceptual diagram illustrating an examination environment using an EEG-based tinnitus diagnosis and prediction apparatus according to an exemplary embodiment.

Referring to FIG. 2 , the subject 2 may listen to the auditory stimulus signal reproduced through the stimulus presentation unit 10 while wearing the measuring unit 20 including the sensor device for EEG measurement on the head. Alternatively, the stimulus presentation unit 10 may include sound reproducing means such as earphones worn on both ears of the subject 2 . While the subject 2 listens to the auditory stimulus signal, the EEG signal of the subject 2 is measured by the measurement unit 20 . However, it will be easily understood that the shape of each device shown in FIG. 2 is an example for convenience of description, and does not limit the configuration of the EEG-based tinnitus diagnosis and prediction device according to the embodiments.

In embodiments, the auditory stimulus signal provided to the subject 2 may be divided into an oddball task and a passive task. In the odd ball task, the subject 2 must distinguish the standard stimulus from the target stimulus, and provide a situation in which the target stimulus is distinguished from the standard stimulus by pressing a button of the input unit 30 . On the other hand, the passive task corresponds to a simple listening task, and the subject 2 only needs to passively listen to the provided auditory stimulus.

3 is a flowchart illustrating each step of the EEG signal processing method according to an embodiment.

The brainwave signal processing method according to the embodiments includes a memory in which one or more program instructions for performing an operation for processing the brainwave signal are stored and a processor for executing the instructions stored in the memory It may be performed by a computing device, and this device is referred to as a signal processing device in this specification. The EEG-based tinnitus diagnosis and prediction apparatus 1 according to the embodiments described above with reference to FIG. 1 may correspond to an example of the signal processing apparatus.

Referring to FIG. 3 , in order to diagnose whether the subject has tinnitus symptoms, an auditory stimulus signal to be provided to the examinee may be first generated and output to be heard by the examinee ( S1 ).

4 is a conceptual diagram illustrating an auditory stimulation signal applied by an EEG signal processing method according to an exemplary embodiment. Referring to FIG. 4 , the auditory stimulus signal may be configured by mixing two sound stimuli, a standard stimulus (S) and a target stimulus (T). For example, the standard stimulus S and the target stimulus T may be arranged to have a preset inter-stimulus interval d. In this case, the interval d between the stimuli may be, for example, 1300 to 1700 ms, but is not limited thereto. Also, in an embodiment, the frequency of occurrence of the standard stimulus S and the target stimulus T may be set so that the frequency of occurrence of the standard stimulus S is higher. For example, the standard stimulus S may be presented with an occurrence frequency of 80% and the target stimulus T may be presented with an occurrence frequency of 20%, but is not limited thereto.

The standard stimulus (S) and the target stimulus (T) are for diagnosing the occurrence of tinnitus or/or predicting the possibility of occurrence of tinnitus through the response of the subject to distinguish the target stimulus (T), and each stimulus has different frequencies has For example, in one embodiment, the standard stimulus S may be a sound signal having a frequency of 500 Hz. Also, in one embodiment, the target stimulus T may be a sound signal having a frequency of 8 kHz in the case of a healthy subject, and a frequency matching the individual tinnitus pitch of the subject in the case of a subject having tinnitus symptoms.

Referring again to FIG. 3 , the brain wave signal of the subject may be measured while the auditory stimulation signal as described above is heard to the subject ( S2 ). In the case of an odd ball task corresponding to a top-down information processing approach, when a target stimulus is distinguished from standard stimuli, the subject can provide a response through a button or the like. On the other hand, in the case of a manual task corresponding to a bottom-up information processing approach, the subject may listen to an auditory stimulus signal while watching an image or video provided without a special response. In this case, the image or image is for diverting the subject's attention from the auditory stimulus signal to the visual stimulus signal, and may have no sound.

In addition, according to an embodiment, an EEG signal in a resting state may also be used as information for tinnitus diagnosis and prediction. In this embodiment, the auditory stimulation signal is intermittently heard to the subject, but the EEG signal of the subject is measured in a section where there is no auditory stimulation signal, or the step (S1) of presenting the auditory stimulation signal is omitted and the examinee does not present the auditory stimulation signal It is also possible to measure EEG signals of

In the EEG signal processing method according to the embodiments, a time axis analysis method, a frequency axis analysis method, a time-frequency map analysis method, and an event-induced potential (Event-) in the EEG signal of the subject measured as described above By extracting features in various ways, such as a Related Potential (ERP) analysis method or a Cross-Frequency Coupling (CFC) analysis method, a time point and activation pattern that can distinguish a normal state from a tinnitus state can be derived. In addition, by using a classifier at every time point of the measured EEG, it is possible to derive the accuracy for all time points by classifying the normal state and the tinnitus state.

For example, using the EEG signal processing method according to the embodiments, depending on the cause of tinnitus, an approach due to a top-down information processing abnormality of the patient's auditory center, or a bottom-up method of the patient's auditory peripheral nerves (bottom-up) By using the approach due to information processing abnormality, the normal state and the tinnitus state can be distinguished by using the EEG signals in the information processing state of the subject. In addition, according to an embodiment, the normal state and the tinnitus state may be distinguished by cross-frequency tuning (CFC) analysis using features extracted from EEG signals in the subject's resting state.

5 is a conceptual diagram for explaining an EEG signal analysis process by the EEG signal processing method according to an embodiment.

Referring to FIGS. 3 and 5 , first, an EEG signal 501 of a subject is measured over time as an example of an EEG signal using an electrode attached to the subject's head, and a feature for classifying a state from the EEG signal can be extracted (S3). At this time, each feature value is obtained by analyzing EEG signals in various ways, and is obtained by methods such as time axis, frequency axis, time-frequency map, event-induced potential (ERP), and cross-frequency tuning (CFC). It may be extracted, but is not limited thereto.

The extracted feature values may be combined into pattern vectors 511, ..., 51N (S4). That is, each of the pattern vectors 511 , ..., 51N is a vector having a plurality of components corresponding to feature values. In addition, the components included in each of the pattern vectors 511, ..., 51N, that is, the feature values, are values extracted by different methods from the same EEG signal for analysis based on multivariate pattern recognition (MVPA). can That is, the feature values extracted from the original EEG signal by various methodologies may constitute the pattern vectors 511, ..., 51N corresponding to the corresponding EEG signal. In addition, it is possible to obtain a plurality of pattern vectors 511, ..., 51N respectively corresponding to a plurality of repeated trials through a plurality of measurements.

Next, pattern signals 521 and 522 may be constructed using pattern vectors 511 , ..., 51N for pattern classification ( S5 ). In this case, each of the pattern signals 521 and 522 corresponds to different symptoms (eg, different subjects), and includes a pattern vector corresponding to a learning set and a pattern vector corresponding to an evaluation set, respectively.

In one embodiment, in order to reduce the computational load and increase the signal-to-noise ratio, the pattern vectors 511, ..., 51N are clustered into a certain number of groups and are averaged pattern vectors calculated as the average value of the pattern vectors in each group. can be converted The classifier may be trained using averaged pattern vectors, wherein a certain percentage of pattern vectors among the pattern vectors may be allocated to the training set and the remaining pattern vectors may be allocated to the evaluation set for evaluating the learning result. For example, of the k averaged pattern vectors, k-n pattern vectors may be allocated to the training set, and the remaining n pattern vectors may be allocated to the evaluation set to construct the pattern signals 521 and 522 (k and n). is any natural number).

Next, the EEG signal can be classified by classifying the pattern signals 521 and 522 generated as described above using a classifier (S6). In an embodiment, a support vector machine (SVM)-based classifier may be used as the classifier, but is not limited thereto. The classifier performs learning on parameters related to an analysis model that classifies normal people and tinnitus patients through machine learning or artificial intelligence using pattern vectors corresponding to the learning set among the pattern signals 521 and 522. can

The classification result for the pattern vectors means the classification result of the subject who provided the EEG signal corresponding to the corresponding pattern vector, and furthermore, the corresponding EEG signal. Accordingly, symptom classification for determining whether a subject has tinnitus symptoms or the possibility of occurrence of tinnitus may be performed through the classification of the pattern vector. For example, a time point at which a feature corresponding to tinnitus is extracted from the EEG signal and a feature pattern (ie, activation pattern) corresponding to tinnitus may be determined ( S7 ).

In one embodiment, a decoding signal 530 indicating the accuracy of the subject's response is generated using the feature vector extracted at all time points of the EEG signal, and the normal state and tinnitus symptom state are generated using a classifier at every time point. It is also possible to further increase the accuracy of symptom classification by classifying them.

6A and 6B are graphs showing the results of analyzing the EEG-based classification accuracy of the subject for the odd ball task and the manual task, respectively, by the EEG signal processing method according to an embodiment. Through ERP) analysis, the response of the subject is expressed with the decoding accuracy for each time. In addition, FIGS. 7A and 7B are graphs showing the results of analyzing the EEG-based classification accuracy of the subject with respect to the target stimulus and the standard stimulus, respectively, by the EEG signal processing method according to the embodiment, and similarly to FIGS. 6A and 6B, the event-evoked potential Through (ERP) analysis, the classification accuracy of the corresponding stimulus according to the different stimulus types of the subject is shown by time.

As shown, it can be confirmed that the tinnitus patient group has a statistically significant difference in classification accuracy compared to the normal control group. For example, in the odd ball task, both the normal group and the tinnitus patient group were able to discriminate the target stimulus, but the tinnitus patient showed reduced discrimination accuracy compared to the control group for a time interval of 180 to 680 ms after stimulation, thereby reducing hearing abnormalities. could check In addition, the tinnitus patient group showed lower decoding accuracy than the control group in the time interval of 180 to 610 ms after stimulation even in the case of the response to the standard stimulus in the odd ball task and the manual task. In addition, in FIGS. 6A, 6B, 7A, and 7B, the bars 601, 602, 701, and 702 displayed on the respective graphs mean confidence intervals of classification accuracy obtained by the statistical method.

Through the above analysis, it is possible to find a time point at which the normal state and the tinnitus state can be distinguished and an activation pattern for diagnosing and/or predicting the tinnitus state from the EEG signal.

The operation by the EEG signal processing method according to the embodiments described above may be at least partially implemented as a computer program and recorded in a computer-readable recording medium. A computer-readable recording medium in which a program for implementing an operation by the EEG signal processing method according to the embodiments is recorded and includes all types of recording devices in which computer-readable data is stored. Examples of the computer-readable recording medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, and optical data storage device. In addition, the computer-readable recording medium may be distributed in network-connected computer systems, and the computer-readable code may be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present embodiment may be easily understood by those skilled in the art to which the present embodiment belongs.

Although the present invention as described above has been described with reference to the embodiments shown in the drawings, it will be understood that these are merely exemplary, and that various modifications and variations of the embodiments are possible therefrom by those of ordinary skill in the art. However, such modifications should be considered to be within the technical protection scope of the present invention. Accordingly, the true technical protection scope of the present invention should be defined by the technical spirit of the appended claims.

Claims (14)

a stimulus presentation unit configured to output an auditory stimulus signal including a standard stimulus and a target stimulus;
a measuring unit configured to measure an EEG signal of a subject in a resting state or in an information processing state listening to the auditory stimulus signal; and
EEG-based tinnitus diagnosis and prediction device comprising an analysis unit configured to extract one or more features of the measured EEG signal and classify the EEG signal by multivariate pattern recognition using the one or more features to determine the presence or absence of tinnitus symptoms in a subject .
According to claim 1,
The stimulus presentation unit is further configured to generate the auditory stimulus signal by randomly mixing a standard stimulus having a first frequency and a target stimulus having a second frequency different from the first frequency.
According to claim 1,
The analysis unit,
a feature extraction unit configured to generate a pattern vector corresponding to the one or more features by using the EEG signal;
a pattern generating unit configured to generate a pattern signal including a learning set and an evaluation set by using the plurality of pattern vectors corresponding to the symptoms of the subject; and
and a classifier configured to perform learning using the learning set of the pattern signal and to determine a symptom of a subject by evaluating a learning result using the evaluation set.
4. The method of claim 3,
The pattern vector is EEG-based tinnitus diagnosis including feature values extracted by at least one of a time axis analysis method, a frequency axis analysis method, a time-frequency map analysis method, an event evoked potential analysis method, and a cross frequency tuning analysis method and predictive devices.
4. The method of claim 3,
The pattern generator, EEG-based tinnitus diagnosis and prediction apparatus further configured to generate the pattern signal using an average value of the pattern vectors generated from the plurality of EEG signals.
According to claim 1,
EEG-based tinnitus diagnosis and prediction apparatus further comprising an input unit configured to receive an input signal input by a subject listening to the auditory stimulation signal in response to the stimulation signal.
measuring, by the signal processing device, an EEG signal of a subject in a resting state or an information processing state listening to an auditory stimulus signal including a standard stimulus and a target stimulus;
extracting, by the signal processing device, one or more features of the measured EEG signal; and
and classifying, by the signal processing device, the EEG signal by multivariate pattern recognition using the one or more features to determine whether the subject has tinnitus symptoms.
8. The method of claim 7,
Before measuring the EEG signal, the EEG signal processing method further comprising the step of outputting the auditory stimulus signal by the signal processing device.
9. The method of claim 8,
The step of outputting the auditory stimulus signal may include: generating, by the signal processing device, a standard stimulus having a first frequency and a target stimulus having a second frequency different from the first frequency to generate the auditory stimulus signal EEG signal processing method comprising a.
8. The method of claim 7,
Classifying the EEG signal includes:
generating, by the signal processing device, a pattern vector corresponding to the one or more features by using the EEG signal;
generating, by the signal processing device, a pattern signal including a learning set and an evaluation set using a plurality of the pattern vectors corresponding to the symptoms of the subject;
performing, by the signal processing device, learning of the classifier using the training set of the pattern signal; and
and determining, by the signal processing device, a symptom of the subject by using the learning result of the classifier.
11. The method of claim 10,
The pattern vector is an EEG signal processing method including feature values extracted by at least one of a time axis analysis method, a frequency axis analysis method, a time-frequency map analysis method, an event evoked potential analysis method, and a cross-frequency tuning analysis method .
11. The method of claim 10,
The generating of the pattern signal includes, by the signal processing device, generating the pattern signal by using an average value of the pattern vectors generated from a plurality of the EEG signals.
8. The method of claim 7,
The EEG signal processing method further comprising the step of receiving, by the signal processing device, an input signal input by a subject listening to the auditory stimulation signal in response to the stimulation signal.
A computer program stored in a computer-readable recording medium in combination with hardware to execute the EEG signal processing method according to any one of claims 7 to 13.

Images