Method, device and computer program for generating standardized electroencephalogram images for learning artificial intelligence models

Abstract

A method, device, and computer program for generating a standardized electroencephalogram image for artificial intelligence model training according to various embodiments of the present invention are provided. The method for generating a standardized electroencephalogram image for artificial intelligence model training according to various embodiments of the present invention is performed by a computing device and includes the steps of collecting a plurality of electroencephalogram signals from a user, processing the collected plurality of electroencephalogram signals, and generating an electroencephalogram image using the processed plurality of electroencephalogram signals.

Description

[Technical field]

[0001] Various embodiments of the present invention relate to a method, apparatus, and computer program for generating a standardized electroencephalogram image for training an artificial intelligence model, and more particularly, to a method for generating a standardized electroencephalogram image as training data for improving the accuracy, sensitivity, and specificity of an artificial intelligence model that analyzes an imaged electroencephalogram signal. [Background technology]

[0002] Brain waves are electrical currents that occur when signals are transmitted between brain nerves in the nervous system, and appear differently depending on the state of the mind and body. By measuring and analyzing the brain's activity in the form of signals, a person's condition can be determined. As brain waves can be used to determine not only a person's physical and mental state, but also whether there is a brain disorder, technology to analyze and interpret brain waves is being actively developed.

[0003] A typical EEG test involves measuring spontaneous electrical activity generated by neuronal activity in the brain over a period of time in a non-invasive manner through electrodes attached to various parts of the scalp, and in special cases, invasive methods (e.g. electrocorticography) may also be used. In general, when applying EEG diagnostically, the focus is on the spectral information of EEG, and in this case, the frequency type of EEG signals is used.

[0004] However, with such typical EEG tests, there is a problem in that it is difficult to distinguish even the subtle differences in EEG. To improve this, there has been a trend toward further improving and developing software that can analyze EEG, supported by recent developments in computer technology.

[0005] Conventionally, a method has been developed and utilized in which a machine learning-based artificial intelligence model that is pre-trained using learning data generated by imaging brainwaves is used to analyze a user's brainwaves. In particular, a topographic map (Topomap) of brainwaves generated by imaging brainwaves (e.g., Figure 1) is used as learning data for training the artificial intelligence model.

[0006] Such EEG topographical maps are suitable for doctors and other specialists to estimate and diagnose a patient's condition. However, as shown in Figure 1, the data on the edges of the image is unclear, making it unsuitable for training artificial intelligence models. In fact, it has been confirmed that artificial intelligence models trained using such EEG topographical maps as training data have poor performance, including reduced accuracy.

[0007] In order to solve the above problems, prior art document 1 (Korean Patent Registration No. 10-2151497) discloses a configuration for generating an image map for training an artificial intelligence model based on electroencephalogram information, and prior art document 2 (Korean Patent Registration No. 10-1748731) discloses a configuration for generating an electroencephalogram image using an electroencephalogram signal.

[0008] However, the configuration disclosed in Prior Art 1 has the problem that an image map is generated by selectively taking into account only EEG signals in a specific frequency band in order to improve the performance of the operation of identifying specific brain diseases, making it difficult to analyze EEG signals in all frequency bands using only one image map, and each axis of the image map is simply composed of the positions of the left and right channels that measure EEG signals, making it difficult to observe changes in EEG signals due to changes in frequency bands.

[0009] In addition, the configuration disclosed in Prior Art 2 is a configuration that generates an EEG image in a form to which Eigenface technology can be applied, for the purpose of classifying EEGs using Eigenface technology rather than training an artificial intelligence model, and even if an artificial intelligence model is trained using this, there is a problem in that it is difficult to take into account the importance of each frequency band. Summary of the Invention [Problem to be solved by the invention]

[0010] The problem to be solved by the present invention is to provide a method, device, and computer program for generating a standardized EEG image for learning an artificial intelligence model that analyzes EEG by generating a standardized EEG image using a plurality of EEG signals collected from a user, and using the standardized EEG image to learn a machine learning-based artificial intelligence model to improve the performance (e.g., modeling accuracy, sensitivity, specificity, etc.) of the machine learning-based artificial intelligence model for analyzing EEG, thereby enabling not only comparison of EEG signals collected through the left and right channels with just one standardized EEG image, but also analysis of EEG signals in all frequency bands, and analyzing EEG signals taking into account the importance of each frequency band.

[0011] The problems to be solved by the present invention are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the following description. [Means for solving the problem]

[0012] A standardized EEG image generating method for artificial intelligence model learning according to one embodiment of the present invention for solving the above-mentioned problems may include, in a method performed by a computing device, a step of collecting a plurality of EEG signals from a user, a step of processing the collected plurality of EEG signals, and a step of generating an EEG image using the processed plurality of EEG signals.

[0013] In various embodiments, the step of collecting the plurality of electroencephalogram signals may include collecting a plurality of electroencephalogram signals having different frequency bands through a plurality of electroencephalogram measurement channels attached to different positions on the user's head.

[0014] In various embodiments, the step of processing the collected EEG signals may include a step of calculating an index value for each of the collected EEG signals, a step of pre-processing the EEG signals for which the index values ​​have been calculated, and filtering EEG signals of a specific frequency band from the EEG signals for which the index values ​​have been calculated, and a step of determining pixel values ​​corresponding to each of the pre-processed EEG signals using the calculated index values, and the step of generating the EEG image may include a step of generating the EEG image using the determined pixel values.

[0015] In various embodiments, the step of calculating the index value may include a step of calculating at least one of absolute power, relative power, a standard value (Z-score), complexity, and entropy for each of the pre-processed EEG signals as the index value.

[0016] In various embodiments, the step of generating the EEG image using the determined pixel values ​​includes the step of arranging the determined pixel values ​​for each of the pre-processed EEG signals on a predetermined template to generate the EEG image, wherein the predetermined template has pixel values ​​according to frequency changes arranged on a first axis and pixel values ​​according to position changes of the point of interest of the user's brain arranged on a second axis perpendicular to the first axis, and a preset reference frequency value is set as the central axis of the first axis, and pixel values ​​corresponding to EEG signals collected from the left region of the brain are arranged in a left region based on the central axis, and pixel values ​​corresponding to EEG signals collected from the right region of the brain are arranged in a right region based on the central axis.

[0017] In various embodiments, the step of generating the EEG image by placing the determined pixel values ​​on a preset template may include dividing each of the left and right sides of the first axis based on the central axis into frequency bands according to the type of EEG signal to generate a plurality of unit sections, and dividing the first axis so that the plurality of unit sections have the same length regardless of the range of the frequency band according to the type of EEG signal.

[0018] In various embodiments, the step of dividing the first axis may include, when a request is received from the user to extend a length of a first unit section among the generated plurality of unit sections, extending the length of the first unit section and evenly shortening lengths of the remaining unit sections by the length by which the first unit section is extended, and, when a request is received from the user to shorten a length of a second unit section among the generated plurality of unit sections, shortening the length of the second unit section and evenly extending lengths of the remaining unit sections by the length by which the second unit section is shortened.

[0019] In various embodiments, the step of generating the EEG image by arranging the determined pixel values ​​on a preset template may include a step of dividing the first axis into frequency bands according to the type of EEG signal to generate a plurality of unit intervals, and determining the length of each of the plurality of unit intervals based on the importance of each frequency band input by the user.

[0020] In various embodiments, the step of generating the EEG image by placing the determined pixel values ​​on a preset template may include the steps of: generating a pixel value matrix using a plurality of pixel values ​​determined corresponding to each of the pre-processed EEG signals, in which pixel values ​​according to frequency changes of the pre-processed EEG signals are arranged in a row direction in the generated pixel value matrix, and pixel values ​​according to position changes of a point of interest of the pre-processed EEG signals are arranged in a column direction in the generated pixel value matrix; converting the generated pixel value matrix into a square matrix based on rows or columns; and placing the pixel value matrix converted into the square matrix on the preset template to generate a standardized EEG image.

[0021] In various embodiments, the step of generating the EEG image by arranging the determined pixel values ​​on a preset template may include a step of arranging a plurality of pixel values ​​determined corresponding to each of the plurality of preprocessed EEG signals in an area on the preset template corresponding to the frequency and the position of a point of interest of each of the plurality of preprocessed EEG signals, performing image smoothing on each of the plurality of pixel values ​​with pixel values ​​arranged at adjacent positions, and performing image resizing to change the image-smoothed EEG image into a square of a preset size to generate a standardized EEG image.

[0022] In various embodiments, the method may further include a step of generating training data for training an artificial intelligence model using the generated EEG image, the generated training data including one standardized EEG image or including a plurality of standardized EEG images combined in the form of a regular rectangle.

[0023] A standardized EEG image generating device for learning an artificial intelligence model according to another embodiment of the present invention for solving the above-mentioned problems includes a processor, a network interface, a memory, and a computer program loaded into the memory and executed by the processor, the computer program including instructions for collecting a plurality of EEG signals from a user, instructions for processing the collected plurality of EEG signals, and instructions for generating an EEG image using the processed plurality of EEG signals.

[0024] A computer program recorded on a computer-readable recording medium according to yet another embodiment of the present invention for solving the above-mentioned problems can be stored on the computer-readable recording medium in order to be combined with a computing device and execute the steps of collecting a plurality of electroencephalogram signals from a user, processing the collected plurality of electroencephalogram signals, and generating an electroencephalogram image using the processed plurality of electroencephalogram signals.

[0025] Further details of the invention are contained in the detailed description and the drawings. Effect of the Invention

[0026] According to various embodiments of the present invention, a standardized EEG image is generated using multiple EEG signals collected from a user, and a machine learning-based artificial intelligence model is trained using the standardized EEG image to analyze EEGs, thereby dramatically improving the performance (e.g., modeling accuracy, sensitivity, specificity, etc.) of the machine learning-based artificial intelligence model. This has the advantage that not only can EEG signals collected through the left and right channels be compared using just one standardized EEG image, but also EEG signals in all frequency bands can be analyzed, and EEG signals can be analyzed taking into account the importance of each frequency band.

[0027] The 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.

Brief Description of the Drawings

[0028] [Figure 1] It is a drawing showing an electroencephalogram topographic map used in conventional artificial intelligence model learning.

[0029] [Diagram 2] It is a drawing showing a standardized electroencephalogram image generation system for artificial intelligence model learning according to an embodiment of the present invention.

[0030] [Diagram 3] It is a hardware configuration diagram of a standardized electroencephalogram image generation device for artificial intelligence model learning according to another embodiment of the present invention.

[0031] [Figure 4] It is a flowchart of a standardized electroencephalogram image generation method for artificial intelligence model learning according to still another embodiment of the present invention.

[0032] [Diagram 5] In various embodiments, it is a flowchart of a method for processing a plurality of electroencephalogram signals.

[0033] [Figure 6] In various embodiments, it is a flowchart of a method for generating a standardized electroencephalogram image using a plurality of processed electroencephalogram signals.

[0034] [Figure 7] It is a drawing showing a standardized electroencephalogram image generation template applicable to various embodiments. [Figure 8] It is a drawing showing a standardized electroencephalogram image generation template applicable to various embodiments.

[0035] [Figure 9] 11 is a diagram illustrating a process of dividing a first axis according to a frequency band in various embodiments.

[0036] [Figure 10] 4 is a diagram illustrating a process of performing image smoothing between pixel values ​​in various embodiments.

[0037] [Figure 11] 1 is a diagram illustrating an example of a standardized electroencephalogram image according to various embodiments.

[0038] [Figure 12] 1 is a diagram illustrating an example of learning data for learning an artificial intelligence model in various embodiments.

[0039] [Figure 13] 1 is a diagram illustrating a standardized electroencephalogram image series that can be applied to various embodiments. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0040] The advantages and features of the present invention, as well as the methods for achieving them, will become apparent from the following detailed description of the embodiments in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, and may be embodied in various different forms, and the embodiments are provided to complete the disclosure of the present invention and to fully inform those skilled in the art of the present invention of the scope of the present invention, and the present invention is defined only by the scope of the claims.

[0041] The terms used in the present specification are for the purpose of describing the embodiments and are not intended to limit the present invention. In the present specification, the singular includes the plural unless otherwise specified in the text. The terms "comprises" and / or "comprising" used in the specification do not exclude the presence or addition of one or more other elements in addition to the elements mentioned. The same reference numerals refer to the same elements throughout the specification, and "and / or" includes each and every combination of one or more of the elements mentioned. Although "first", "second", etc. are used to describe various elements, it is understood that these elements are not limited by these terms. These terms are used merely to distinguish one element from another element. Therefore, it is understood that the first element mentioned below may be the second element within the technical concept of the present invention.

[0042] Unless otherwise defined, all terms (including technical and scientific terms) used herein may be used in the sense commonly understood by a person of ordinary skill in the art to which the present invention belongs. Furthermore, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless specifically and explicitly defined.

[0043] The term "module" or "module" as used herein means a software or hardware component such as an FPGA or ASIC, and the "module" or "module" performs a certain function. However, the term "module" or "module" is not limited to software or hardware. A "module" or "module" may be configured to reside on an addressable storage medium and to execute on one or more processors. Thus, by way of example, a "module" or "module" includes components such as software components, object-oriented software components, class components, and task components, as well as processes, functions, attributes, procedures, subroutines, program code segments, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The components and functions provided within a "module" or "module" may be combined into a smaller number of components and "modules" or "modules" or may be further separated into additional components and "modules" or "modules".

[0044] Spatially relative terms such as "below," "beneath," "lower," "above," "upper," and the like, may be used to facilitate describing the relationship of one component to another as depicted in the drawings. Spatially relative terms should be understood to include different orientations of the components in use or operation in addition to the orientation depicted in the drawings. For example, if a component depicted in the drawings is turned over, a component described as "below" or "beneath" another component may be placed "above" the other component. Thus, the exemplary term "below" can include both an orientation of below and above. Components may be oriented in other directions, and the spatially relative terms may be interpreted accordingly.

[0045] In this specification, the term "computer" refers to any type of hardware device including at least one processor, and may also include software configurations that operate on the hardware device in accordance with the embodiment. For example, the term "computer" refers to any type of device, including, but not limited to, a smartphone, tablet PC, desktop, notebook computer, and user clients and applications that run on each device.

[0046] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0047] Although each step described in this specification is described as being performed by a computer, the subject matter of each end system is not limited to this, and depending on the embodiment, at least a portion of each step may be performed by different devices.

[0048]

[0049] FIG. 2 is a diagram illustrating a standardized electroencephalogram image generating system for artificial intelligence model learning according to an embodiment of the present invention.

[0050] Referring to FIG. 2, a standardized electroencephalogram image generating system for artificial intelligence model learning according to one embodiment of the present invention may include a standardized electroencephalogram image generating device 100, a user terminal 200 and an external server 300.

[0051] Here, the standardized EEG image generating system for artificial intelligence model learning illustrated in FIG. 1 is according to one embodiment, and its components are not limited to the embodiment illustrated in FIG. 1, and may be added, modified or deleted as necessary.

[0052] In one embodiment, the standardized electroencephalogram image generating device 100 can generate standardized electroencephalogram images as learning data for training an artificial intelligence model (e.g., an image analysis model that uses electroencephalogram images generated by imaging electroencephalogram signals as input data and outputs result data related to the user's state information) using a plurality of electroencephalogram signals collected from a user.

[0053] For example, the standardized electroencephalogram image generating device 100 can generate an electroencephalogram image by processing a plurality of electroencephalogram signals collected from a user, converting the processed plurality of electroencephalogram signals into pixel values, and placing the converted pixel values ​​on a pre-set template.

[0054] In addition, the standardized EEG image generating device 100 can process the EEG image generated in the above process (e.g., image smoothing and image resizing) to generate a standardized EEG image for training an artificial intelligence model.

[0055] In various embodiments, the standardized electroencephalogram image generating device 100 can be connected to an electroencephalogram measuring device (not shown) via a network 400 and can collect a plurality of electroencephalogram signals collected through the electroencephalogram measuring device.

[0056] Here, the network 400 may refer to a connection structure that allows information exchange between each of a plurality of nodes such as terminals and servers, etc. For example, the network 400 may include a local area network (LAN), a wide area network (WAN), the Internet (WWW), wired and wireless data communication networks, telephone networks, wired and wireless television communication networks, etc.

[0057] In addition, the wireless data communication network includes, but is not limited to, 3G, 4G, 5G, 3GPP (3rd Generation Partnership Project), 5GPP (5th Generation Partnership Project), LTE (Long Term Evolution), WIMAX (World Interoperability for Microwave Access), Wi-Fi, the Internet, LAN (Local Area Network), Wireless LAN (Wireless Local Area Network), WAN (Wide Area Network), PAN (Personal Area Network), RF (Radio Frequency), Bluetooth network, NFC (Near-Field Communication) network, satellite broadcasting network, analog broadcasting network, DMB (Digital Multimedia Broadcasting) network, etc.

[0058] In various embodiments, the standardized EEG image generating device 100 can generate training data using the standardized EEG image generated by the above-mentioned method, and can train an artificial intelligence model using the generated training data.

[0059] Here, the artificial intelligence model is composed of one or more network functions, which may be composed of a collection of interconnected computational units that may generally be referred to as "nodes." Such "nodes" may be referred to as "neurons." One or more network functions are composed of at least one or more nodes. The nodes (or neurons) that make up one or more network functions may be interconnected by one or more "links."

[0060] In an artificial intelligence model, one or more nodes connected through links can form a relationship between input nodes and output nodes relative to each other. The concepts of input nodes and output nodes are relative, and any node that has an output node relationship with one node can also have an input node relationship with another node, and vice versa. As mentioned above, the input node to output node relationship can be generated around links. One or more output nodes can be connected to one input node through links, and vice versa.

[0061] In the relationship between an input node and an output node connected through a link, the value of the output node may be determined based on data input to the input node. Here, the node connecting the input node and the output node may have a weight. The weight may be variable and may be changed by a user or an algorithm so that the artificial intelligence model performs a desired function. For example, when one or more input nodes are connected to one output node through respective links, the output node may determine an output node value based on the value input to the input node connected to the output node and the weight set for the link corresponding to each input node.

[0062] As described above, in an AI model, one or more nodes are connected to each other through one or more links to form a relationship between input nodes and output nodes within the AI ​​model. The characteristics of the AI ​​model can be determined by the number of nodes and links, the correlation between the nodes and links, and the weighted values ​​assigned to each link within the AI ​​model. For example, if there are two AI models that have the same number of nodes and links but different weighted values ​​between the links, the two AI models can be recognized as different from each other.

[0063] Some of the nodes constituting the artificial intelligence model may constitute a layer based on the distance from the first input node. For example, a set of nodes whose distance from the first input node is n may constitute an n-th layer. The distance from the first input node may be defined by the minimum number of links that must be traversed to reach the corresponding node from the first input node. However, such a definition of a layer is arbitrary for the purpose of explanation, and the degree of a layer in an artificial intelligence model may be defined in a manner different from the above. For example, the layer of a node may be defined by the distance from the final output node.

[0064] The first input node may refer to one or more nodes to which data is directly input without passing through a link in relation to other nodes in the artificial intelligence model. Or, in the artificial intelligence model network, it may refer to a node that does not have other input nodes connected to a link in the node-to-node relationship based on the link. Similarly, the final output node may refer to one or more nodes that do not have an output node in relation to other nodes in the artificial intelligence model. Also, the hidden node may refer to a node that constitutes the artificial intelligence model, rather than the first input node and the last output node. The artificial intelligence model according to an embodiment of the present invention may be an artificial intelligence model in which the nodes of the input layer may be more than the nodes of the hidden layer close to the output layer, and the number of nodes decreases as the input layer progresses to the hidden layer.

[0065] The artificial intelligence model may include one or more hidden layers. The hidden nodes of the hidden layers may receive the output of the previous layer and the output of the surrounding hidden nodes as input. The number of hidden nodes in each hidden layer may be the same or different. The number of nodes in the input layer may be determined based on the number of data fields of the input data, and may be the same or different from the number of hidden nodes. Input data input to the input layer may be operated by the hidden nodes of the hidden layer and may be output by a fully connected layer (FCL), which is an output layer.

[0066] In addition, the standardized electroencephalogram image generating device 100 can train an artificial intelligence model using a standardized electroencephalogram image. Specifically, the standardized electroencephalogram image generating device 100 can perform training on one or more network functions constituting an artificial intelligence model using a labeled training data set. For example, the standardized electroencephalogram image generating device 100 can input each of the training input data sets to one or more network functions, and derive an error by comparing each of the output data calculated by the one or more network functions with each of the training output data sets corresponding to the labels of each of the training input data sets. That is, in training of the artificial intelligence model, the training input data can be input to an input layer of one or more network functions, and the training output data can be compared with the output of one or more network functions. The standardized electroencephalogram image generating device 100 can train an artificial intelligence model based on an error between the calculation result of one or more network functions for the training input data and the training output data (label).

[0067] In addition, the standardized electroencephalogram image generating device 100 can adjust weights of one or more network functions based on an error in a backpropagation manner. That is, the standardized electroencephalogram image generating device 100 can adjust weights so that the output of one or more network functions approaches the learning output data based on an error between the operation result of one or more network functions on the learning input data and the learning output data.

[0068] When the learning of one or more network functions is performed for a predetermined epoch or more, the standardized electroencephalogram image generating device 100 may determine whether or not to halt the learning using the verification data. The predetermined epoch may be a part of the entire learning target epoch. The verification data may be composed of at least a part of the labeled learning data set. That is, the standardized electroencephalogram image generating device 100 may perform learning of an artificial intelligence model through a learning data set, and after the learning of the artificial intelligence model is repeated for a predetermined epoch or more, it may determine whether or not the learning effect of the artificial intelligence model is equal to or higher than a predetermined level using the verification data. For example, when performing learning with a target number of iterative learning times of 10 times using 100 pieces of learning data, the standardized electroencephalogram image generating device 100 may perform 10 iterative learning, which is a predetermined epoch, and then perform 3 iterative learning using 10 pieces of verification data. If the change in the output of the artificial intelligence model during the 3 iterative learning times is below a predetermined level, it may determine that further learning is meaningless and terminate the learning. That is, the validation data may be used to determine the completion of learning based on whether the effect of the learning per epoch is above or below a certain level in the iterative learning of the artificial intelligence model. The above-mentioned numbers of learning data, validation data, and the number of iterations are merely examples and are not limited thereto.

[0069] The standardized electroencephalogram image generating device 100 can generate an artificial intelligence model by testing the performance of one or more network functions using a test data set and determining whether or not one or more network functions are activated. The test data can be used to verify the performance of the artificial intelligence model and can be composed of at least a portion of the training data set. For example, 70% of the training data set can be used for training the artificial intelligence model (i.e., training to adjust weights to output result values ​​similar to the labels), and 30% can be used as test data for verifying the performance of the artificial intelligence model.

[0070] The standardized electroencephalogram image generating device 100 can input a test data set to an artificial intelligence model that has completed training, measure an error, and determine whether or not to activate the artificial intelligence model depending on whether the performance is above a predetermined standard. The standardized electroencephalogram image generating device 100 can verify the performance of the artificial intelligence model that has completed training using test data for the artificial intelligence model that has completed training, and can activate the artificial intelligence model for use in other applications if the performance of the artificial intelligence model that has completed training is above a predetermined standard.

[0071] In addition, the standardized electroencephalogram image generating device 100 may deactivate and discard the AI ​​model if the performance of the AI ​​model that has completed learning is below a predetermined standard. For example, the standardized electroencephalogram image generating device 100 may determine the performance of the AI ​​model generated based on factors such as accuracy, precision, and recall. The above-mentioned performance evaluation criteria are merely examples and are not limited thereto. In addition, the standardized electroencephalogram image generating device 100 according to various embodiments of the present invention may generate a plurality of AI models by independently learning each AI model, and may evaluate the performance and use only AI models with a certain level of performance or higher for electroencephalogram image analysis.

[0072] In one embodiment, the user terminal 200 can be connected to the standardized EEG image generating device 100 via the network 400, and can receive standardized EEG images generated by the standardized EEG image generating device 100 performing a standardized EEG image generation process for learning an artificial intelligence model, or can receive the results of analyzing multiple EEG signals through an artificial intelligence model trained using the standardized EEG images.

[0073] Here, the user terminal 200 is a wireless communication device that ensures portability and mobility, and may include, but is not limited to, all kinds of handheld-based wireless communication devices such as navigation, PCS (Personal Communication System), GSM (Global System for Mobile communications), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), Wibro (Wireless Broadband Internet) terminals, smartphones, smartpads, tablet PCs, etc.

[0074] In one embodiment, the external server 300 may be connected to the standardized electroencephalogram image generating device 100 through the network 400, and may store and provide various information and data required for the standardized electroencephalogram image generating device 100 to perform a standardized electroencephalogram image generating process for artificial intelligence model learning, or may collect and store various data generated by performing the standardized electroencephalogram image generating process for artificial intelligence model learning. For example, the external server 300 may be, but is not limited to, a storage server separately provided outside the standardized electroencephalogram image generating device 100. Hereinafter, a hardware configuration of the standardized electroencephalogram image generating device 100 performing a standardized electroencephalogram image generating process for artificial intelligence model learning will be described with reference to FIG. 3.

[0075]

[0076] FIG. 3 is a hardware configuration diagram of a standardized electroencephalogram image generating device for learning an artificial intelligence model according to another embodiment of the present invention.

[0077] Referring to Fig. 3, a standardized electroencephalogram image generating device 100 (hereinafter, "computing device 100") according to another embodiment of the present invention may include one or more processors 110, a memory 120 for loading a computer program 151 executed by the processor 110, a bus 130, a communication interface 140, and a storage 150 for storing the computer program 151. Here, Fig. 3 illustrates only components related to the embodiment of the present invention. Therefore, a person skilled in the art to which the present invention belongs will understand that other general components may be included in addition to the components illustrated in Fig. 3.

[0078] The processor 110 controls the overall operation of each component of the computing device 100. The processor 110 may include a central processing unit (CPU), a microprocessor unit (MPU), a microcontroller unit (MCU), a graphic processing unit (GPU), or any other type of processor commonly known in the technical field of the present invention.

[0079] Additionally, the processor 110 may execute operations for at least one application or program for implementing a method according to an embodiment of the present invention, and the computing device 100 may include one or more processors.

[0080] In various embodiments, the processor 110 may further include a random access memory (RAM, not shown) and a read-only memory (ROM, not shown) for temporarily and / or permanently storing signals (or data) processed within the processor 110. The processor 110 may also be implemented in the form of a system on chip (SoC) including at least one of a graphics processing unit, a RAM, and a ROM.

[0081] The memory 120 stores various data, instructions and / or information. The memory 120 can load a computer program 151 from the storage 150 to perform the methods / operations according to various embodiments of the present invention. When the computer program 151 is loaded into the memory 120, the processor 110 can perform the methods / operations by executing one or more instructions constituting the computer program 151. The memory 120 may be embodied as a volatile memory such as a RAM, although the scope of the present disclosure is not limited in this respect.

[0082] The bus 130 provides a communication function between the components of the computing device 100. The bus 130 may be implemented as various types of buses such as an address bus, a data bus, and a control bus.

[0083] The communication interface 140 supports wired and wireless Internet communication of the computing device 100. The communication interface 140 may also support various communication methods other than Internet communication. To this end, the communication interface 140 may be configured to include a communication module that is well known in the art of the present invention. In some embodiments, the communication interface 140 may be omitted.

[0084] The storage 150 may non-temporarily store a computer program 151. When a standardized electroencephalogram image generation process for artificial intelligence model training is performed through the computing device 100, the storage 150 may store various information required to provide the standardized electroencephalogram image generation process for artificial intelligence model training.

[0085] Storage 150 may be configured to include non-volatile memory such as ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), flash memory, etc., a hard disk, a removable disk, or any form of computer-readable recording medium widely known in the technical field to which the present invention belongs.

[0086] The computer program 151, when loaded into the memory 120, may include one or more instructions that cause the processor 110 to perform the methods / operations according to various embodiments of the present invention. That is, the processor 110 may execute the one or more instructions to perform the methods / operations according to various embodiments of the present invention.

[0087] In one embodiment, the computer program 151 may include one or more instructions for performing a standardized electroencephalogram image generating method for artificial intelligence model training, including the steps of collecting a plurality of electroencephalogram signals from a user, processing the collected plurality of electroencephalogram signals, and generating an electroencephalogram image using the processed plurality of electroencephalogram signals.

[0088] The steps of a method or algorithm described in connection with the embodiments of the present invention may be embodied directly in hardware, or in a software module executed by hardware, or in a combination thereof. The software module may reside in a Random Access Memory (RAM), Read Only Memory (ROM), Erasable Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), Flash Memory, a hard disk, a removable disk, a CD-ROM, or any other form of computer readable recording medium commonly known in the art to which the present invention pertains.

[0089] The components of the present invention may be embodied as a program (or application) and stored on a medium to be executed in combination with a computer, which is hardware. The components of the present invention may be implemented as software programming or software elements, and similarly, embodiments may include various algorithms embodied as a combination of data structures, processes, routines or other programming constructs, and may be embodied in programming or scripting languages ​​such as C, C++, Java, assembler, etc. Functional aspects may be embodied as algorithms executed by one or more processors. Hereinafter, a standardized EEG image generation process for artificial intelligence model learning performed by a computing device 100 will be described with reference to FIGS. 4 to 11.

[0090]

[0091] FIG. 4 is a flowchart of a standardized electroencephalogram image generating method for learning an artificial intelligence model according to yet another embodiment of the present invention.

[0092] 4, in step S110, the computing device 100 may collect a plurality of electroencephalogram signals from a first user (e.g., a person whose electroencephalogram signal is to be analyzed, such as a patient). For example, the computing device 100 may collect a plurality of electroencephalogram signals having different frequency bands through an electroencephalogram measuring device (not shown) including a plurality of electroencephalogram measuring channels attached to different positions on the head of the first user.

[0093] Here, the EEG measuring device may be a device that includes multiple channels (2, 4, 8, 16, 19, 24, 68, 128 or 256 channels, with caps or individual electrodes attached) attached to different positions on the head of the first user, and measures independent EEG signals through each channel according to the EEG measuring 10-20 system, and can collect multiple EEG signals through each of the multiple channels. For example, the EEG measuring device may include 19 channels (e.g., Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5, T6, Fz, Cz, Pz) (here, Fz, Cz, Pz are common channels), and can measure 19 independent EEG signals through the 19 channels. However, it is not limited thereto.

[0094] In various embodiments, the computing device 100 may collect, through the EEG measuring device, a number of EEG signals measured in a normal state when the first user is not performing any particular action, but is not limited thereto, and may collect a number of EEG signals measured while the first user is performing various actions or performing various tests.

[0095] In step S120, the computing device 100 may process a plurality of electroencephalogram signals of the first user collected through the electroencephalogram measuring device. Hereinafter, a method for processing a plurality of electroencephalogram signals performed by the computing device 100 will be described with reference to FIG.

[0096]

[0097] FIG. 5 is a flow diagram of a method for processing a plurality of electroencephalogram signals, according to various embodiments.

[0098] 5, in step S210, the computing device 100 may pre-process the plurality of EEG signals to filter EEG signals of a particular frequency band. For example, the computing device 100 may filter out unnecessary frequency bands or frequency bands that may be vulnerable to noise (e.g., EEG signals corresponding to a frequency band less than 4 Hz, EEG signals corresponding to a frequency band exceeding 45 Hz) from the plurality of EEG signals to extract only EEG signals corresponding to a frequency band in the range of 1 to 45 Hz.

[0099] In various embodiments, when the computing device 100 receives a selection of an EEG signal of a specific frequency band (e.g., an EEG signal corresponding to any one of delta waves, alpha waves, beta waves, and gamma waves) from a second user (e.g., a person who wishes to obtain the results of analyzing an EEG signal, such as a doctor, or an administrator of a standardized EEG image generation process for learning an artificial intelligence model), the computing device 100 can filter (or null) the EEG signals corresponding to the remaining frequency bands except for the EEG signal of the specific frequency band selected by the second user.

[0100] In step S220, the computing device 100 can calculate an index value for each of the preprocessed EEG signals using the preprocessed EEG signals (the result of filtering the EEG signals corresponding to a specific frequency band) in step S210.

[0101] Here, the index value for each of the pre-processed EEG signals may include, but is not limited to, at least one of absolute power, relative power, Z-score, complexity, and entropy.

[0102] First, the computing device 100 may calculate absolute power for each of the plurality of electroencephalogram signals as an index value. For example, the computing device 100 may calculate the absolute power for each of the plurality of pre-processed electroencephalogram signals by calculating the sum of the power values ​​(e.g., the degree to which each of the plurality of electroencephalogram signals appears) of each of the plurality of pre-processed electroencephalogram signals. However, the present invention is not limited to this, and various methods for calculating the absolute power for the electroencephalogram signals may be applied.

[0103] Furthermore, the computing device 100 may calculate the relative power for each of the plurality of EEG signals as an index value. For example, the computing device 100 may calculate the ratio (%) of the power value for each frequency of the preprocessed plurality of EEG signals to the sum of the power values ​​of the preprocessed plurality of EEG signals to calculate the relative power for each of the preprocessed plurality of EEG signals. For example, the computing device 100 may calculate the power value (PSD) for a specific frequency band and calculate the relative power based on the largest power value of the calculated values. However, the present invention is not limited to this, and various methods for calculating the relative power for the EEG signals may be applied.

[0104] Furthermore, the computing device 100 may calculate a standard value (Z-score) for each of the plurality of EEG signals as an index value. For example, the computing device 100 may normally distribute the preprocessed plurality of EEG signals, and may calculate a standard value using the normally distributed plurality of EEG signals (e.g., (intensity of EEG signal-standard deviation value of intensity of the plurality of preprocessed EEG signals) / (average value of intensity of the plurality of preprocessed EEG signals)). However, the present invention is not limited thereto, and various methods for calculating a standard value for an EEG signal may be applied.

[0105] In addition, the computing device 100 may calculate the complexity and entropy of each of the plurality of EEG signals as index values. For example, the computing device 100 may calculate the complexity of each of the pre-processed plurality of EEG signals using a complexity calculation method using approximate entropy (e.g., a method of strategically calculating whether the characteristics of the EEG signal are maintained similarly in other m+1 sample periods for an m sample period). However, the present invention is not limited thereto, and various methods for calculating the complexity of the EEG signal may be applied.

[0106] In step S230, the computing device 100 may determine pixel values ​​corresponding to each of the pre-processed electroencephalogram signals using the index values ​​calculated by the above method. Here, the method of determining pixel values ​​may be based on color data according to the magnitude of index values ​​(e.g., data in which pixel values ​​of colors (e.g., red, orange, yellow, green, blue, indigo, purple) according to the magnitude of index values ​​are matched and stored), but is not limited thereto.

[0107]

[0108] Referring again to FIG. 3, in step S130, the computing device 100 may generate an electroencephalogram image using the pixel values ​​determined in the above manner.

[0109] In various embodiments, the computing device 100 may generate an electroencephalogram image by arranging pixel values ​​determined for each of the pre-processed electroencephalogram signals on a preset template. Hereinafter, a method for generating an electroencephalogram image performed by the computing device 100 will be described with reference to FIGS. 6 to 11.

[0110]

[0111] FIG. 6 is a flow chart of a method for generating a standardized EEG image using a plurality of processed EEG signals, according to various embodiments.

[0112] Referring to FIG. 6, in step S310, the computing device 100 may place pixel values ​​determined for each of the pre-processed electroencephalogram signals on a preset template.

[0113] Here, as shown in Fig. 7(A), the preset template may have pixel values ​​according to frequency changes on a first axis (X-axis) and pixel values ​​according to position changes of a region of interest (ROI) of the first user's brain on a second axis (Y-axis) perpendicular to the first axis. However, without being limited thereto, the preset template may have pixel values ​​according to frequency changes on a first axis and pixel values ​​according to the positions of each of a plurality of EEG measurement channels attached to the first user's head on a second axis.

[0114] In addition, the preset template may have a preset frequency value (e.g., 1 Hz, which is the lowest frequency value among the multiple preprocessed EEG signals) as the central axis of the first axis, and pixel values ​​corresponding to EEG signals collected from the left side of the brain may be arranged in the left region based on the central axis of 1 Hz, and pixel values ​​corresponding to EEG signals collected from the right side of the brain may be arranged in the right region based on the central axis of 1 Hz.

[0115] Here, the first axis may be divided by a preset unit frequency (e.g., 0.25 Hz), and the second axis may be divided according to the number of points of interest (e.g., Desikan-Killiany atlas A total of 34 points of interest based on regions (e.g. banks-STS, caudal-anterior-cingulate, caudal-middle-fronta, cuneus, entorhinal, frontal-pole, fusiform, inferior-parietal, inferior ior-tempora, insula, isthmus-cingulate, lateral-occipital, lateral-orbitofrontal, lingual, medial-orbitofrontal, middle-temporal, paracentral, parahipp ocampal, pars-opercularis, pars-orbitalis, pars-triangularis, pericalcarine, post-central, posterior-cingulate, precentral, precuneus, rostral-anterior-cingulate, rostral-middle-frontal, superior-frontal, superior-parietal, superior-temporal, supramarginal, temporal-pole and transverse-temporal).

[0116] That is, the preset template may include an area in which a total of 11,968 pixel values ​​are arranged by dividing the first axis into 352 and the second axis into 34, and the computing device 100 may calculate a total of 11,968 index values ​​by calculating index values ​​for the pre-processed EEG signal in 0.25 Hz units for 34 points of interest, and may determine 11,968 pixel values ​​corresponding to the calculated 11,968 index values. However, the number of divisions of the first axis and the second axis may be set in various ways without being limited thereto.

[0117] In addition, the second axis of the preset template is described here as being divided according to the positions of 34 points of interest based on the Desikan-Killiany atlas regions, but is not limited to this and may be divided according to the positions of 11 points of interest based on the positions of multiple EEG measurement channels attached to the head of the first user, as shown in FIG. 8(A) (e.g., 11 positions in each of the left and right regions based on the first axis (F3, C3, C4, P3, O1, T3, T4, T5, Fz, Cz, Pz, of which Fz, Cz, Pz are positions where the left and right sides overlap)).

[0118] In various embodiments, the computing device 100 may generate a plurality of unit sections by dividing each of the left and right sides of the first axis based on a central axis (1 Hz) set on the first axis into frequency bands according to the type of electroencephalogram signal. For example, the computing device 100 may divide each of the left 1 Hz to 45 Hz section and the right 1 Hz to 45 Hz section based on the central axis of the first axis into a total of eight sections (e.g., delta section (1 to 4 Hz section), theta section (4 to 8 Hz section), alpha 1 section (8 to 10 Hz section), alpha 2 section (10 to 12 Hz section), beta 1 section (12 to 15 Hz section), beta 2 section (15 to 20 Hz section), beta 3 section (20 to 30 Hz section), and gamma section (30 to 45 Hz section)) according to the type of electroencephalogram signal.

[0119] In this case, the computing device 100 may divide the first axis so that the length of each unit interval divided according to the type of electroencephalogram signal has the same length regardless of the range of the frequency band.

[0120] For example, as shown in Fig. 9, the delta section and theta section have sections in the range of 3 Hz and 4 Hz, respectively, while the alpha 1 section and the alpha 2 section have sections in the range of 2 Hz, so that pixel values ​​may be allocated in a relatively narrower section compared to delta waves and theta waves. In this way, that is, when the areas in which pixel values ​​corresponding to each EEG signal are allocated are uneven, there is a problem that the learning and analysis effect may be reduced for EEG signals allocated in a relatively narrower section such as the alpha 1 section and the alpha 2 section.

[0121] Considering this, the computing device 100 may divide the first axis to have sections of the same length for each frequency band regardless of the range of each frequency band. Through this, pixel values ​​corresponding to EEG signals having a relatively narrow frequency band such as alpha waves and delta waves (e.g., EEG signals having a frequency band within a 4 Hz range) may be arranged in a relatively wide region, and pixel values ​​corresponding to EEG signals having a wide frequency band such as beta waves and gamma waves may be arranged in a relatively narrow region. For example, in the case of alpha 1 waves, eight pixel values ​​may be arranged in the alpha 1 section having a first length in the first axis direction, whereas in the case of gamma waves, 60 pixel values ​​may be arranged in the gamma section having the same first length as the alpha 1 section in the first axis direction.

[0122] In various embodiments, the computing device 100 evenly divides each of the left and right sides of the first axis into frequency bands according to the type of EEG signal to generate a plurality of unit sections having the same length, but when a request is received from the second user to adjust the length of a specific section, the computing device 100 can evenly adjust the lengths of the remaining unit sections by the amount of change caused by adjusting the length of the specific section.

[0123] For example, when the computing device 100 receives a request from the second user to extend the length of the first unit section in order to more precisely observe the EEG signal for the first unit section, the computing device 100 can extend the length of the first unit section by the amount requested by the second user, and can evenly shorten the lengths of the remaining unit sections by the amount by which the first unit section is extended (e.g., when the length of the first unit section is extended by 1, each of the remaining seven unit sections can be shortened by 1 / 7).

[0124] In addition, when the computing device 100 receives a request from the second user to shorten the length of the second unit interval because the importance of the second unit interval is low, the computing device 100 can shorten the length of the second unit interval by the amount requested by the second user, and can evenly extend the lengths of the remaining unit intervals by the length by which the second unit interval was shortened (for example, if the length of the second unit interval is shortened by 1, each of the remaining seven unit intervals can be extended by 1 / 7).

[0125] In various embodiments, the computing device 100 generates a plurality of unit sections by dividing each of the left and right sides of the first axis by frequency bands according to the type of electroencephalogram signal based on a central axis (1 Hz) set on the first axis, and may determine the length of each of the plurality of unit sections based on the importance of each frequency band input by the second user, and may divide the first axis according to the length of each of the plurality of unit sections determined. For example, the computing device 100 may receive a priority (e.g., 1 to 8 ranks) for each of a total of eight types of electroencephalogram signals (e.g., delta waves, theta waves, alpha 1 waves, alpha 2 waves, beta 1 waves, beta 2 waves, beta 3 waves, and gamma waves) based on the type of electroencephalogram signal, and may divide the first axis so that a section with a higher priority has a relatively longer length than a section with a lower priority based on the set priority.

[0126] In step S320, the computing device 100 places a plurality of pixel values ​​determined corresponding to each of the pre-processed EEG signals on a preset template, and may perform image smoothing for each of the plurality of pixel values ​​with pixel values ​​located at adjacent positions.

[0127] For example, as shown in FIG. 10, the computing device 100 may perform image smoothing between pixel values ​​arranged in a first region 10 and pixel values ​​arranged in a second region 20 that is adjacent to the pixel values ​​arranged in the first region 10, so that the pixel values ​​arranged in the first region 10 and the pixel values ​​arranged in the second region 20 become continuous.

[0128] Here, a method of performing image smoothing between pixel values ​​arranged in the first region 10 and pixel values ​​arranged in the second region 20 may involve, but is not limited to, a method of arranging an average value of pixel values ​​arranged in the first region 10 and pixel values ​​arranged in the second region 20 in a third region 30, which is a region where pixel values ​​arranged in the first region 10 and pixel values ​​arranged in the second region 20 intersect.

[0129] In addition, in order to derive more continuous results, in addition to the method of placing the average value of the pixel values ​​placed in the first region 10 and the pixel values ​​placed in the second region 20 in the third region 30, the third region 30 can be subdivided into a plurality of third regions, and the average value can be reflected sequentially in the plurality of subdivided regions.

[0130] For example, after dividing the third region 30, which is a region where the pixel values ​​arranged in the first region 10 and the pixel values ​​arranged in the second region 20 are adjacent, into three third regions (e.g., region 3-1, region 3-2, and region 3-3), a first average value which is the average of the pixel values ​​arranged in the first region 10 and the pixel values ​​arranged in the second region 20 can be arranged in region 3-2. Also, a second average value which is the average of the pixel values ​​arranged in the first region 10 and the first average value can be arranged in region 3-1, and a third average value which is the average of the first average value and the pixel values ​​arranged in the second region 20 can be arranged in region 3-3.

[0131] In step S330, the computing device 100 performs image resizing to change the EEG image generated by arranging a plurality of pixel values ​​on a preset template through step S320 into a regular square of a preset size, thereby generating a standardized EEG image (e.g., FIG. 11) (here, FIG. 11(A) is a standardized EEG image when the index value is absolute power, and FIG. 11(B) is a standardized EEG image when the index value is relative power).

[0132] Here, the preset size may be a value set by the second user, but is not limited thereto. Also, the standardized EEG image may be generated in a regular square shape to improve the performance of the artificial intelligence model, but is not limited thereto, and various shapes may be applied.

[0133] Also, here, the image smoothing operation (step S320) and the image resizing operation (step S330) performed by the computing device 100 are described as being performed sequentially, but this is not limited thereto, and the EEG image can be changed into a regular square through image resizing, and image smoothing can also be performed.

[0134] In various embodiments, the computing device 100 may generate a pixel value matrix using a plurality of pixel values ​​determined corresponding to each of the pre-processed EEG signals, and use the pixel value matrix to generate a standardized EEG image.

[0135] First, the computing device 100 may generate a pixel value matrix using a plurality of pixel values ​​determined corresponding to each of the preprocessed EEG signals. For example, the computing device 100 may generate a 352×34 pixel value matrix by arranging pixel values ​​(352) according to frequency changes of the preprocessed EEG signals in the row direction and arranging pixel values ​​(34) according to position changes of the points of interest of the preprocessed EEG signals in the column direction, considering that the first axis is divided by 0.25 Hz, which is a preset unit frequency, to form 352 regions and the second axis is divided by the number of points of interest to form 34 regions.

[0136] Thereafter, the computing device 100 may convert the pixel value matrix into a square matrix (e.g., a matrix having the same number of rows and columns) based on the rows or columns of the 352×34 pixel value matrix in order to generate a standardized electroencephalogram image in a square shape. At this time, it is preferable to convert based on an axis on which most pixel values ​​are arranged among the rows or columns of the pixel value matrix. For example, the computing device 100 may convert the 352×34 pixel value matrix into a 352×352 pixel value matrix based on the number of rows. At this time, the computing device 100 may perform smoothing on pixel values ​​during the process of converting the 352×34 pixel value matrix into the 352×352 pixel value matrix (e.g., step S320).

[0137] Thereafter, the computing device 100 can generate a standardized electroencephalogram image in a square shape by placing the 352×352 pixel matrix on the square template.

[0138] In various embodiments, the computing device 100 can generate a series of standardized EEG images based on the time at which the EEG signal was measured using a pre-defined template in which the first axis is pixel values ​​according to frequency change and the second axis is pixel values ​​according to the position of the point of interest or pixel values ​​according to the position of the EEG measurement channel, as shown in Figures 7(B) and 8(B).

[0139] In step S340, the computing device 100 may generate training data for training an artificial intelligence model using the standardized electroencephalogram image generated in step S330.

[0140] In various embodiments, the computing device 100 may generate a single training data set using a single standardized electroencephalogram image, as shown in FIG. 12(A).

[0141] In various embodiments, the computing device 100 may generate a single training data set by combining a plurality of standardized EEG images in a square shape, as shown in FIG. 12(B).

[0142] In this case, the multiple standardized EEG images included in one learning data may be multiple standardized EEG images generated based on the same index value, such as combining multiple standardized EEG images when the index value is absolute power or combining multiple standardized EEG images when the index value is relative power, but is not limited to this. For example, a standardized EEG image when the index value is absolute power, a standardized EEG image when the index value is relative power, a standardized EEG image when the index value is a standard value, and a standardized EEG image when the index value is complexity or entropy may be combined, i.e., multiple standardized EEG images generated using different index values ​​may be combined to generate one learning data.

[0143]

[0144] The above-mentioned standardized electroencephalogram image generating method for training an artificial intelligence model has been described with reference to the flowchart shown in the drawings. For ease of explanation, the standardized electroencephalogram image generating method for training an artificial intelligence model has been illustrated and described as a series of blocks, but the present invention is not limited to the order of the blocks, and some blocks may be performed in a different order or simultaneously than those illustrated and described herein. In addition, new blocks not described in this specification and drawings may be added, or some blocks may be deleted or changed.

[0145] In addition, in various embodiments of the present invention, the method for generating a standardized EEG image for training an artificial intelligence model has been described as generating a standardized EEG image for training an artificial intelligence model by arranging pixel values ​​according to frequency changes on the first axis and pixel values ​​according to position changes of the user's brain's point of interest (or the position of the EEG measurement channel) on the second axis, but is not limited to this, and the standardized EEG image for training an artificial intelligence model may be generated in series according to the frequency of the EEG signal as shown in FIG. 13, but may be generated by identically arranging pixel values ​​(or pixel values ​​according to the position of the EEG measurement channel) according to position changes of the user's brain's point of interest on the first axis and the second axis.

[0146]

[0147] Although the embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art to which the present invention pertains will understand that the present invention can be embodied in other specific forms without changing the technical spirit or essential features of the present invention. Therefore, the embodiments described above should be understood to be illustrative in all respects and not restrictive.

Claims

1. 1. A method performed by a computing device, comprising: collecting a plurality of electroencephalographic signals for a user; processing the collected electroencephalogram signals; and generating an electroencephalogram image using the processed electroencephalogram signals; The step of processing the collected electroencephalogram signals includes: pre-processing the collected electroencephalogram signals and filtering out electroencephalogram signals of a specific frequency band from the collected electroencephalogram signals; calculating an index value for each of the electroencephalogram signals obtained by filtering the electroencephalogram signals in the specific frequency bands; and determining pixel values ​​corresponding to each of the pre-processed electroencephalogram signals using the calculated index values; The step of generating an electroencephalogram image includes: generating the electroencephalogram image using the determined pixel values; generating the electroencephalogram image using the determined pixel values, placing the pixel values ​​determined for each of the pre-processed electroencephalogram signals on a preset template to generate the electroencephalogram image; The predefined template includes: A pixel value according to a change in frequency is arranged on a first axis, and a pixel value according to a change in position of the point of interest of the user's brain is arranged on a second axis perpendicular to the first axis, A preset reference frequency value is set as a central axis of the first axis, and pixel values ​​corresponding to electroencephalogram signals collected from the left side of the brain are arranged in a left region based on the central axis, and pixel values ​​corresponding to electroencephalogram signals collected from the right side of the brain are arranged in a right region based on the central axis. A standardized method for generating EEG images for training artificial intelligence models.

2. The step of acquiring the plurality of electroencephalogram signals comprises:

2. The method of claim 1, further comprising the step of collecting a plurality of electroencephalogram signals having different frequency bands through a plurality of electroencephalogram measurement channels attached to different positions on the user's head.

3. The step of calculating the index value comprises:

2. The method of claim 1, further comprising the step of calculating at least one of absolute power, relative power, a standard value (Z-score), complexity, and entropy as an index value for each of the pre-processed electroencephalogram signals.

4. The step of generating the electroencephalogram image by arranging the determined pixel values ​​on a preset template includes: generating a plurality of unit sections by dividing each of the left and right sides of the first axis based on the central axis into frequency bands according to the type of electroencephalogram signal; 2. The method of claim 1, wherein the step of generating the plurality of unit intervals includes a step of dividing the first axis so that the plurality of unit intervals have the same length regardless of a frequency band range according to the type of the EEG signal.

5. The step of dividing the first axis includes: extending the length of the first unit section and shortening the lengths of the remaining unit sections by the extended length of the first unit section when a request to extend the length of a first unit section among the generated unit sections is received from the user; and 5. The method for generating a standardized electroencephalogram image for training an artificial intelligence model according to claim 4, further comprising the step of shortening the length of the second unit interval and equally extending the lengths of the remaining unit intervals by the shortened length of the second unit interval when a request is received from the user to shorten the length of a second unit interval among the generated plurality of unit intervals.

6. The step of generating the electroencephalogram image by arranging the determined pixel values ​​on a preset template includes:

2. The method for generating a standardized electroencephalogram image for training an artificial intelligence model according to claim 1, further comprising: dividing the first axis into frequency bands according to the type of electroencephalogram signal to generate a plurality of unit intervals; and determining the length of each of the plurality of unit intervals based on the importance of each frequency band input by the user.

7. The step of generating the electroencephalogram image by arranging the determined pixel values ​​on a preset template includes: generating a pixel value matrix using a plurality of pixel values ​​determined corresponding to each of the pre-processed electroencephalogram signals, the generated pixel value matrix having pixel values ​​according to frequency changes of the pre-processed electroencephalogram signals arranged in a row direction and pixel values ​​according to position changes of the point of interest of the pre-processed electroencephalogram signals arranged in a column direction; converting the generated pixel value matrix into a square matrix on a row or column basis; and 2. The method of claim 1, further comprising the step of placing the pixel value matrix converted into a square matrix on the preset template to generate a standardized electroencephalogram image.

8. The step of generating the electroencephalogram image by arranging the determined pixel values ​​on a preset template includes: arranging a plurality of pixel values ​​determined corresponding to each of the pre-processed electroencephalogram signals in an area corresponding to the frequency and the location of the point of interest of each of the pre-processed electroencephalogram signals on the preset template, and performing image smoothing with pixel values ​​arranged in adjacent positions for each of the plurality of pixel values; and 2. The method of claim 1, further comprising the step of performing image resizing to change the image-smoothed EEG image into a square of a preset size to generate a standardized EEG image.

9. 9. The method of claim 8, further comprising the step of generating training data for training an artificial intelligence model using the generated EEG image, the generated training data including one standardized EEG image or including a plurality of standardized EEG images combined in the form of a regular rectangle.

10. Processor; Network interface; memory; and A computer program that is loaded into the memory and executed by the processor, The computer program comprises: instructions to a user to collect a plurality of electroencephalogram signals; instructions for processing the collected electroencephalographic signals; and instructions for generating an electroencephalogram image using the processed electroencephalogram signals; The instructions for processing the collected electroencephalogram signals include: instructions for pre-processing the collected electroencephalogram signals to filter electroencephalogram signals of a specific frequency band from the collected electroencephalogram signals; instructions for calculating an index value for each of the electroencephalogram signals obtained by filtering the electroencephalogram signals in the specific frequency bands; and instructions for determining pixel values ​​corresponding to each of the pre-processed electroencephalogram signals using the calculated index values; The instructions for generating the electroencephalogram image include: instructions for generating the electroencephalogram image using the determined pixel values; instructions for generating the electroencephalogram image using the determined pixel values, and instructions for placing the determined pixel values ​​for each of the pre-processed electroencephalogram signals on a pre-defined template to generate the electroencephalogram image. The predefined template includes: A pixel value according to a change in frequency is arranged on a first axis, and a pixel value according to a change in position of the point of interest of the user's brain is arranged on a second axis perpendicular to the first axis, A preset reference frequency value is set as a central axis of the first axis, and pixel values ​​corresponding to electroencephalogram signals collected from the left side of the brain are arranged in a left region based on the central axis, and pixel values ​​corresponding to electroencephalogram signals collected from the right side of the brain are arranged in a right region based on the central axis. A standardized electroencephalogram image generator for training artificial intelligence models.

11. In combination with a computing device, collecting a plurality of electroencephalographic signals for a user; processing the collected electroencephalogram signals; and A computer program stored on a computer-readable recording medium for executing a step of generating an electroencephalogram image using the processed electroencephalogram signals, The step of processing the collected electroencephalogram signals includes: pre-processing the collected electroencephalogram signals and filtering out electroencephalogram signals of a specific frequency band from the collected electroencephalogram signals; calculating an index value for each of the electroencephalogram signals obtained by filtering the electroencephalogram signals in the specific frequency bands; and determining pixel values ​​corresponding to each of the pre-processed electroencephalogram signals using the calculated index values; The step of generating an electroencephalogram image includes: generating the electroencephalogram image using the determined pixel values; generating the electroencephalogram image using the determined pixel values, placing the pixel values ​​determined for each of the pre-processed electroencephalogram signals on a preset template to generate the electroencephalogram image; The predefined template includes: A pixel value according to a change in frequency is arranged on a first axis, and a pixel value according to a change in position of the point of interest of the user's brain is arranged on a second axis perpendicular to the first axis, A preset reference frequency value is set as a central axis of the first axis, and pixel values ​​corresponding to electroencephalogram signals collected from the left side of the brain are arranged in a left region based on the central axis, and pixel values ​​corresponding to electroencephalogram signals collected from the right side of the brain are arranged in a right region based on the central axis. A computer program stored on a computer-readable recording medium.

Images