WO2023163444A1

WO2023163444A1 - Method and device for generating precise ndt brain image

Info

Publication number: WO2023163444A1
Application number: PCT/KR2023/002186
Authority: WO
Inventors: 구용서; 민경문
Original assignee: 재단법인 아산사회복지재단
Priority date: 2022-02-25
Filing date: 2023-02-15
Publication date: 2023-08-31

Abstract

The present disclosure provides to a method and device for generating a precise nDT brain image. The method according to the present disclosure comprises the steps of: acquiring image data by a computer through magnetic resonance imaging (MRI); measuring, by the computer, electroencephalography data by using a navigated transcranial magnetic stimulation (nTMS) system; extracting, by the computer, noninvasive corticocortical evoked potential (nCCEP) data by using the electroencephalography data; determining, by the computer, a noninvasive dynamic tractography (nDT) by performing multimodal imaging on the basis of the image data and the nCCEP data; visualizing a precise nDT brain image by the computer on the basis of the noninvasive dynamic tractography; and by the computer, comparing the noninvasive dynamic tractography and the visualized precise nDT brain image with the data stored in a database, and analyzing the noninvasive dynamic tractography and the visualized precise nDT brain image to determine whether the brain is abnormal.

Description

NDT precise brain image generation method and apparatus

The present disclosure relates to a method and apparatus for generating nDT precise brain images.

Brain connectivity has been implicated in a variety of neurological and psychiatric disorders.

Among brain connectivity, effective connectivity provides information on the directionality of the connection and the transmission/reception pattern of neural signals, thereby enabling the construction of a brain connectivity model that understands the flow of nerve signals.

Effective connectivity also describes the flow of neural signals based on structural connectivity and functional connectivity based on specific hypotheses, but assumptions are always necessary and are subject to error. Therefore, in order to confirm effective connectivity, it is necessary to measure the directionality and propagation speed of large bundles of nerves through direct electrical stimulation.

For example, a method for confirming effective connectivity using invasive dynamic tractography (IDT) can determine which white matter actual nerve information is delivered by combining effective connectivity information and structural connectivity information. At this time, if information such as the strength and speed of connectivity and electrophysiological abnormalities are additionally included in the invasive dynamic nerve fiber map (IDT), it can be used for basic research on the pathophysiology and cognitive function of brain diseases.

As another example, effective connectivity confirmation methods using invasive dynamic nerve fiber maps (IDT) and invasive corticocortical evoked potentials (ICCEP) can be measured only by invasive brain waves, so they are effective only in the part where invasive brain waves are inserted Since connectivity was evaluated, effective connectivity of the whole brain could not be evaluated, and since it could only be measured in patients with epilepsy, the characteristics of normal people could not be reflected, errors could occur in the prediction model, and individual evaluation was impossible.

Therefore, research on improved methods and devices capable of directly evaluating the effective connectivity of entire brain regions for each individual in various subjects has been continuously conducted.

An object of the embodiments disclosed in the present disclosure is to provide a method and apparatus for generating nDT precise brain images using EEG data measured using a navigated transcranial magnetic stimulation (nTMS) system.

The problems to be solved by the present disclosure are not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below.

An nDT precise brain image generating apparatus according to the present disclosure for achieving the above technical problem includes a database; and a processor, wherein the processor measures EEG data using a navigated transcranial magnetic stimulation (nTMS) system when image data is acquired through Magnetic Resonance Imaging (MRI) imaging, and uses the EEG data to obtain noninvasive nCCEP (noninvasive) data. extract corticocortical evoked potential) data, determine a noninvasive dynamic tractography (nDT) by performing multimodal imaging based on the image data and the nCCEP data, and nDT precision based on the noninvasive dynamic path It is possible to determine whether there is an abnormality in the brain by visualizing the brain image and comparing and analyzing the non-invasive dynamic pathway and the visualized nDT precise brain image with data stored in a database.

In this case, the image data includes a T1-weighted image and a diffusion-weighted image, and when the processor acquires the image data, the computer removes noise from the diffusion-weighted image and removes the skull to obtain the diffusion-weighted image. Pre-processing is performed, and an FA map may be extracted using the pre-processed diffusion-enhanced image.

In addition, when measuring the EEG data, the processor divides the brain region using a cortical segmentation system, divides and displays the divided brain region on the T1-weighted image, and measures the exercise evoked potential to measure the intensity of stimulation. determining, stimulating the divided brain region with the determined intensity of stimulation using the nTMS system, and measuring and storing EEG data generated as the divided brain region is stimulated with the determined intensity of stimulation; The nTMS system may include, as a parameter, at least one of the frequency of stimulation, the strength of stimulation, the number of stimulations for each brain region, and the direction of stimulation.

In addition, when determining the non-invasive dynamic path, the processor extracts a first component in the form of a sharp wave appearing in an initial period of 10 ms to 50 ms of the nCCEP data, and based on the first component, the Euclidean path between end points of the path (Euclidean) Statistical analysis is performed by calculating the distance and statistically inferring the length of the path, and a non-invasive dynamic path is extracted by performing multimodal imaging based on the statistical analysis result, and the initial 50ms to 300ms of the nCCEP data A second component in the form of a slow wave appearing in the interval up to is extracted, effective connectivity data is extracted using the first component and the second component, a brain region is defined based on the effective connectivity data, and an electrical source image By performing the analysis, the defined brain region is visualized, the non-invasive dynamic path has a positive correlation between the length of the inferred path and the latency of the first component, and the length of the inferred path and the first component have a positive correlation. It may conform to a dynamic path model in which the amplitude of a component has a negative correlation and the FA map and the velocity of the first component have a positive correlation.

In addition, when defining the brain region, the processor defines a brain region in which the connectivity index of the effective connectivity data is equal to or greater than a preset threshold voltage as an epileptogenic zone, and A brain region in which the connectivity index of the effective connectivity data has connectivity but is lower than a preset threshold voltage is defined as a propagation zone, and a brain region in which the connectivity index of the effective connectivity data has no connectivity is defined as a non-relevant zone (Non- Involved Zone) can be defined.

In addition, when determining whether the brain is abnormal, the processor categorizes the seizure prognosis after epileptic surgery into true positive, false positive, true negative, and false negative, and evaluates the final accuracy of localization of the epilepsy-inducing site.

In addition, when visualizing the nDT precise brain image, the processor displays at least one of a matrix, a circle map, a brain volume, and a surface of the brain, when visualizing the nDT precise brain image. can be visualized using

In addition, the nDT precision brain image generation method according to the present disclosure for achieving the above technical problem includes acquiring image data by the computer through Magnetic Resonance Imaging (MRI) imaging; measuring EEG data by the computer using a navigated transcranial magnetic stimulation (nTMS) system; extracting, by the computer, noninvasive corticocortical evoked potential (nCCEP) data using the EEG data; determining, by the computer, a noninvasive dynamic tractography (nDT) by performing multimodal imaging based on the image data and the nCCEP data; Visualizing, by the computer, an nDT precise brain image based on the non-invasive dynamic path; and comparing and analyzing, by the computer, the non-invasive dynamic path and the visualized nDT precise brain image with data stored in a database, and determining whether there is an abnormality in the brain.

In addition to this, a computer program stored in a computer readable recording medium for execution to implement the present disclosure may be further provided.

In addition to this, a computer readable recording medium storing a computer program for executing a method for implementing the present disclosure may be further provided.

According to the above-described problem solving means of the present disclosure, it is possible to actively measure the epileptic induction site at a desired time point, the time required to measure the epileptic induction site and confirm effective connectivity suggesting the connection direction of the brain, and cost can be reduced. In addition, it is possible to evaluate brain connectivity for each patient, so that personalized treatment can be developed, and it can be applied to various brain diseases such as dementia, language disorder, autism, depression, and stroke.

The effects of the present disclosure 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 description below.

1 is a flowchart of a method for generating an nDT precise brain image according to an embodiment of the present disclosure.

FIG. 2 is a flowchart illustrating an embodiment of step S110 of FIG. 1 .

FIG. 3 is a flowchart illustrating an embodiment of step S120 of FIG. 1 .

4 is a diagram for explaining a divided brain region according to an embodiment of the present disclosure.

5 is a flowchart illustrating an embodiment of step S140 of FIG. 1 .

6 is a flowchart illustrating another embodiment of step S140 of FIG. 1 .

7 is a diagram for explaining an nDT precision brain image visualization method according to an embodiment of the present disclosure.

Advantages and features of the present invention, and methods of achieving them, will become clear with reference to the detailed description of the following embodiments taken in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, only these embodiments are intended to complete the disclosure of the present invention, and are common in the art to which the present invention belongs. It is provided to fully inform the person skilled in the art of the scope of the invention, and the invention is only defined by the scope of the claims.

Terminology used herein is for describing the embodiments and is not intended to limit the present invention. In this specification, singular forms also include plural forms unless specifically stated otherwise in a phrase. As used herein, "comprises" and/or "comprising" does not exclude the presence or addition of one or more other elements other than the recited elements. Like reference numerals throughout the specification refer to like elements, and “and/or” includes each and every combination of one or more of the recited elements. Although "first", "second", etc. are used to describe various components, these components are not limited by these terms, of course. These terms are only used to distinguish one component from another. Accordingly, it goes without saying that the first element mentioned below may also be the second element within the technical spirit of the present invention.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used with meanings commonly understood by those skilled in the art to which the present invention belongs. In addition, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless explicitly specifically defined.

In this specification, a 'computer' includes all of various devices capable of visually presenting a result to a user by performing calculation processing. For example, a computer includes not only a desktop PC and a notebook (Note Book) but also a smart phone, a tablet PC, a cellular phone, a PCS phone (Personal Communication Service phone), synchronous/asynchronous A mobile terminal of IMT-2000 (International Mobile Telecommunication-2000), a Palm Personal Computer (Palm PC), and a Personal Digital Assistant (PDA) may also be applicable. In addition, the computer may also correspond to medical equipment for obtaining or observing medical images. Also, the computer may correspond to a server computer connected to various client computers. Also, a computer may consist of one or more devices.

Operations in this specification may be operated through a processor and memory in the computer. That is, the processor may be composed of one or a plurality of processors. In this case, the one or more processors may be a general-purpose processor such as a CPU, an AP, or a digital signal processor (DSP), a graphics-only processor such as a GPU or a vision processing unit (VPU), or an artificial intelligence-only processor such as an NPU. The one or more processors control input data to be processed according to predefined operating rules or artificial intelligence models stored in a memory. Alternatively, when one or more processors are processors dedicated to artificial intelligence, the processors dedicated to artificial intelligence may be designed with a hardware structure specialized for processing a specific artificial intelligence model.

A predefined action rule or an artificial intelligence model is characterized in that it is created through learning. Here, being made through learning means that a basic artificial intelligence model is learned using a plurality of learning data by a learning algorithm, so that a predefined action rule or artificial intelligence model set to perform a desired characteristic (or purpose) is created. means burden. Such learning may be performed in the device itself in which artificial intelligence according to the present disclosure is performed, or through a separate server and/or system. Examples of the learning algorithm include supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but are not limited to the above examples.

An artificial intelligence model may be composed of a plurality of neural network layers. Each of the plurality of neural network layers has a plurality of weight values, and a neural network operation is performed through an operation between an operation result of a previous layer and a plurality of weight values. A plurality of weights possessed by a plurality of neural network layers may be optimized by a learning result of an artificial intelligence model. For example, a plurality of weights may be updated so that a loss value or a cost value obtained from an artificial intelligence model is reduced or minimized during a learning process. The artificial neural network may include a deep neural network (DNN), for example, a Convolutional Neural Network (CNN), a Deep Neural Network (DNN), a Recurrent Neural Network (RNN), a Restricted Boltzmann Machine (RBM), A deep belief network (DBN), a bidirectional recurrent deep neural network (BRDNN), or deep Q-networks, but is not limited to the above examples.

According to an exemplary embodiment of the present disclosure, a processor may implement artificial intelligence. Artificial intelligence refers to a machine learning method based on an artificial neural network in which a machine learns by mimicking a human's biological neuron. The methodology of artificial intelligence includes supervised learning in which input data and output data are provided together as training data according to the learning method, so that the answer (output data) of the problem (input data) is determined, and only input data is provided without output data. In unsupervised learning, where the answer (output data) of the problem (input data) is not determined, and whenever an action is taken in the current state, a reward is given in the external environment. , it can be classified as reinforcement learning in which learning proceeds in the direction of maximizing this reward. In addition, the methodology of artificial intelligence may be classified according to the architecture, which is the structure of the learning model. The architecture of widely used deep learning technology is Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN). , Transformers, and generative adversarial networks (GANs).

The devices and systems may include artificial intelligence models. The artificial intelligence model may be one artificial intelligence model or may be implemented as a plurality of artificial intelligence models. Artificial intelligence models may be composed of neural networks (or artificial neural networks), and may include statistical learning algorithms that mimic biological neurons in machine learning and cognitive science. A neural network may refer to an overall model having a problem-solving ability by changing synaptic coupling strength through learning of artificial neurons (nodes) formed in a network by synaptic coupling. Neurons in a neural network may contain a combination of weights or biases. A neural network may include one or more layers composed of one or more neurons or nodes. Illustratively, the device may include an input layer, a hidden layer, and an output layer. A neural network constituting the device can infer a result (output) to be predicted from an arbitrary input (input) by changing the weight of a neuron through learning.

The processor generates a neural network, trains or learns the neural network, performs an operation based on received input data, generates an information signal based on a result of the execution, or generates a neural network. Neural network models include GoogleNet, AlexNet, VGG Network, etc., CNN (Convolution Neural Network), R-CNN (Region with Convolution Neural Network), RPN (Region Proposal Network), RNN (Recurrent Neural Network), S-DNN (Stacking-based deep Neural Network), S-SDNN (State-Space Dynamic Neural Network), Deconvolution Network, DBN (Deep Belief Network), RBM (Restrcted Boltzman Machine), Fully Convolutional Network . For example, the neural network may include a deep neural network.

Neural networks include CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), perceptron, multilayer perceptron, FF (Feed Forward), RBF (Radial Basis Network), DFF (Deep Feed Forward), LSTM (Long Short Term Memory), GRU (Gated Recurrent Unit), AE (Auto Encoder), VAE (Variational Auto Encoder), DAE (Denoising Auto Encoder), SAE (Sparse Auto Encoder), MC (Markov Chain), HN (Hopfield Network), BM(Boltzmann Machine), RBM(Restricted Boltzmann Machine), DBN(Depp Belief Network), DCN(Deep Convolutional Network), DN(Deconvolutional Network), DCIGN(Deep Convolutional Inverse Graphics Network), GAN(Generative Adversarial Network) ), LSM (Liquid State Machine), ELM (Extreme Learning Machine), ESN (Echo State Network), DRN (Deep Residual Network), DNC (Differentiable Neural Computer), NTM (Neural Turning Machine), CN (Capsule Network), It will be appreciated by those skilled in the art that it may include any neural network, including but not limited to Kohonen Network (KN) and Attention Network (AN).

According to an exemplary embodiment of the present disclosure, the processor may include a Convolution Neural Network (CNN), a Region with Convolution Neural Network (R-CNN), a Region Proposal Network (RPN), a Recurrent Neural Network (RNN), such as GoogleNet, AlexNet, VGG Network, and the like. ), S-DNN (Stacking-based deep neural network), S-SDNN (State-Space Dynamic Neural Network), Deconvolution Network, DBN (Deep Belief Network), RBM (Restrcted Boltzman Machine), Fully Convolutional Network, LSTM (Long Short-Term Memory) Network, Classification Network, Generative Modeling, eXplainable AI, Continual AI, Representation Learning, AI for Material Design, BERT for natural language processing, SP-BERT, MRC/QA, Text Analysis, Dialog System, GPT-3 , GPT-4, Visual Analytics for vision processing, Visual Understanding, Video Synthesis, ResNet Data intelligence for Anomaly Detection, Prediction, Time-Series Forecasting, Optimization, Recommendation, and Data Creation. , but not limited thereto.

A processor in the present specification may control an operation of generating an nDT precise brain image using EEG data measured based on a nTMS (navigated transcranial magnetic stimulation) system described in FIGS. 1 to 7 below.

Specifically, when image data is acquired through Magnetic Resonance Imaging (MRI) imaging, the processor measures EEG data using an nTMS system, extracts noninvasive corticocortical evoked potential (nCCEP) data using the EEG data, , determining a noninvasive dynamic tractography (nDT) by performing multimodal imaging based on the image data and the nCCEP data, visualizing an nDT precise brain image based on the noninvasive dynamic tractography, and By comparing and analyzing the invasive dynamic pathway and the visualized nDT precision brain image with data stored in a database (memory), it is possible to determine whether there is an abnormality in the brain.

In this case, the image data includes a T1-weighted image and a diffusion-enhanced image, and the processor, when obtaining the image data through the MRI imaging, removes noise from the diffusion-enhanced image and removes the skull to enhance the diffusion-enhanced image. An image may be pre-processed, and an FA map may be extracted using the pre-processed diffusion-enhanced image.

In addition, when measuring EEG data using the nTMS system, the processor divides brain regions using a cortical segmentation system, divides and displays the divided brain regions on the T1-weighted image, and calculates motion evoked potentials. The intensity of stimulation is determined by measuring, the divided brain regions are stimulated with the determined intensity of stimulation using the nTMS system, and the EEG data generated by stimulating the divided brain regions with the determined intensity of stimulation can be measured and stored. In this case, the nTMS system may include at least one of the frequency of stimulation, the intensity of stimulation, the number of stimulation for each brain region, and the direction of stimulation as a parameter.

Here, the processor may determine the minimum TMS stimulation intensity. That is, the processor may determine the stimulation intensity by measuring a motor evoked potential (MEP) before stimulation with TMS. In this case, the processor may perform 10 consecutive stimulations at an optimal position where the MEP is best induced, and determine a minimum stimulation intensity at which 50 μV or more is induced more than 5 times as a resting motor threshold (RMT).

In addition, the processor may stimulate the brain region using the TANS method. Here, the stimulation intensity can be stimulated up to 80 times every 5-6 seconds with a single pulse of 120% of the RMT intensity.

In addition, the processor selects nerve fibers that can be stimulated by TMS through Atlas, identifies the location of the gyrus of the largest network cluster using at least one of diffusion-weighted imaging (DWI) and functional magnetic resonance imaging (fMRI), and each Create more than 3,000 search grids in the gyrus range that can stimulate nerve fibers, randomly select about 600 coil positions, conduct E-field modeling including various coil directions for each selected target cluster, and The stimulation location can be determined by selecting the hotspot that can stimulate the nerve fibers to the maximum.

In addition, when determining the non-invasive dynamic path, the processor extracts a first component in the form of a sharp wave appearing in an initial period of 10 ms to 50 ms of the nCCEP data, and based on the first component, the Euclidean path between end points of the path (Euclidean) Statistical analysis is performed by calculating the distance and statistically inferring the length of the path, and a non-invasive dynamic path can be extracted by performing multimodal imaging based on the statistical analysis result, wherein the non-invasive dynamic path is The inferred path length and the latency of the first component have a positive correlation, the inferred path length and the amplitude of the first component have a negative correlation, and the FA map and the first component have a negative correlation. can fit a dynamic path model in which the velocity of is positively correlated. In this case, the processor may validate the dynamic pathway based on the CIPI analysis method, diffusion-weighted imaging (DWI) signal, functional magnetic resonance imaging (fMRI) signal, and seizure prognostic information after epilepsy surgery.

Here, the processor may use the CIPI analysis method in the process of confirming the nCCEP. In this case, the processor may separate or combine independent elements using CIPI analysis. In other words, the CIPI (common independent process identification) analysis method is a method of confirming whether different brain signals are statistically the same. It is possible to identify nCCEPs originating from different nerve fibers by selecting the same elements and separating or combining independent elements. there is. In addition, the processor can identify independent nCCEPs for each nerve fiber using the CIPI analysis method, and can confirm similarity of nCCEPs for the same nerve fiber for each individual.

In addition, since TMS stimulation generates very large artifacts, and it may be difficult to evaluate early brain signals, the processor removes artifacts from transcranial magnetic stimulation-EEG (TMS-EEG) for non-invasive brain signal can be recorded.

In addition, when determining the non-invasive dynamic path, the processor extracts a second component in the form of a slow wave appearing in the interval from the initial 50 ms to 300 ms of the nCCEP data, and using the first component and the second component An operation of extracting effective connectivity data, defining a brain region based on the effective connectivity data, and visualizing the defined brain region by performing electrical source image analysis by the computer may be further performed. In this case, when the processor defines a brain region based on the effective connectivity data, a brain region in which a connectivity index of the effective connectivity data is equal to or greater than a preset threshold voltage is an epileptogenic zone. ), and a brain region in which the connectivity index of the effective connectivity data has connectivity but is lower than a preset threshold voltage is defined as a propagation zone, and a brain region in which the connectivity index of the effective connectivity data has no connectivity is defined as a propagation zone. It can be defined as a non-involved zone.

In addition, when visualizing the nDT precise brain image based on the non-invasive dynamic path, the processor converts the nDT precise brain image into a matrix, a circle map, a brain volume, and a surface of the brain. It can be visualized using at least one of (Surface).

In addition, the processor compares and analyzes the non-invasive dynamic pathway and the visualized nDT precision brain image with data stored in a database to determine whether there is an abnormality in the brain. It can be categorized as true negative or false negative, and the final accuracy with which the epilepsy-induced site is localized can be evaluated.

In the present specification, 'brain wave data' is data recorded by deriving and amplifying current generated according to brain activity. 'EEG data' is obtained in the form of a waveform graph by electroencephalography (EEG), which records the electrical activity of the brain by attaching electrodes to the scalp.

In the present specification, a 'navigated transcranial magnetic stimulation (nTMS) system' recognizes the location of stimulation and the motion of a patient when stimulation is applied to the surface of a patient's head with a transducer using two or more cameras. It may refer to a system that maps and provides a recognized result to an MR image. Hereinafter, the navigational transcranial magnetic stimulation system may be referred to as an 'nTMS system'.

Hereinafter, a method and program for generating an nDT precise brain image according to embodiments of the present invention will be described with reference to the drawings.

1 is a flowchart of a method for generating an nDT precise brain image according to an embodiment of the present invention.

Referring to FIG. 1 , a method for generating an nDT precise brain image (S100) according to an embodiment of the present invention may include steps S110 to S160.

In step S110, the processor of the computer may obtain image data of the patient's brain provided through magnetic resonance imaging (hereinafter referred to as 'MRI'). Image data of the patient's brain may include T1 weighted imaging and diffusion weighted imaging (DWI). Imaging data of the patient's brain can be used to generate nDT precise brain images in subsequent steps.

In step S120, the processor of the computer may measure and store EEG data using the nTMS system. According to the nTMS system, the processor of the computer recognizes the location of the stimulation and the motion of the patient when the stimulation is applied to the surface of the patient's head, and at the same time measures and stores EEG data provided through electroencephalography (EEG).

In one embodiment, the processor of the computer may acquire EEG data of the patient for a predetermined period of time, and the EEG data may be divided into predetermined intervals. For example, the predetermined period may be a minimum unit time length for determining the state of the brain, or may be a period arbitrarily designated by the user. The pre-determined period may refer to all times during which EEG data is measured using the nTMS system.

In step S130, the processor of the computer may extract noninvasive corticocortical evoked potential (hereinafter referred to as 'nCCEP') data based on the stored EEG data. The nCEEP data is data capable of grasping brain connectivity by recording brain waves at a location apart from a location where electrical stimulation is applied, and may represent data representing effective connectivity of the brain.

In this case, the processor may validate the dynamic pathway based on the CIPI analysis method, diffusion-weighted imaging (DWI) signal, functional magnetic resonance imaging (fMRI) signal, and seizure prognostic information after epilepsy surgery.

In step S140, the processor of the computer performs multi-modal imaging based on the image data and nCCEP data obtained through MRI imaging to obtain noninvasive dynamic tractography (hereinafter referred to as 'nDT'). ) can be extracted. The nDT may be a result of realizing an accurate brain connectivity location by combining structural connectivity of the brain represented by image data obtained through MRI imaging through multimodal imaging and effective connectivity of the brain represented by nCCEP data.

Multimodal imaging may refer to the operation of a deep-learning model trained to extract significant information from image data and nCCEP data obtained through MRI imaging. The deep learning model can find spatial information such as the position, direction, and size of an object in real time, and can learn to process the image based on this.

In step S150, the processor of the computer may visualize the nDT precise brain image based on the determined nDT. The nDT precise brain image may be data expressed on a brain map (atlas). However, the present invention is not limited thereto, and nDT precision brain imaging can be implemented in various ways. Various methods of implementing nDT precision brain imaging will be described in detail with reference to FIG. 7 described later.

In step S160, the processor of the computer may determine whether the patient's brain is abnormal by comparing the nDT and the visualized nDT precise brain image with the brain image stored in the database. Computers can analyze abnormal areas of the brain and localize epilepsy-causing areas. In addition, the computer may classify and analyze the prognosis of seizures after epilepsy surgery. For example, the processor of the computer may classify the prognosis following epilepsy surgery according to Engel and International League Against Epilepsy (ILAE) scale criteria. In addition, the computer can evaluate the final accuracy of localization after categorizing the seizure prognosis after epilepsy surgery into true positive, false positive, true negative, and false negative.

According to an embodiment of the present disclosure, since it is possible to actively check brain connectivity, it is possible to reduce time and cost consumed in examination. In addition, it is possible to non-invasively evaluate the connectivity of the entire brain of each individual, and individualized treatment can be performed by localizing the epileptic region using nDT precise brain images generated for each individual. In addition, there is an effect that can confirm the brain connectivity of the unconscious patient.

Effective connectivity and structural connectivity information for language functions can be visualized using embodiments according to the present disclosure. For example, embodiments according to the present disclosure may be applied not only to epilepsy, but also to various brain diseases such as dementia, language disorder, autism, depression, and stroke. Specifically, it can be used as a diagnostic and monitoring biomarker by identifying connectivity abnormalities in patients with language dysfunction using the embodiments according to the present disclosure. In addition, by combining the navigation software used during brain tumor surgery with the embodiments according to the present disclosure, surgical treatment is performed by avoiding specific structures, thereby preventing postoperative complications such as language dysfunction, and performing neurological monitoring during surgery. Embodiments according to the present disclosure may be used.

Embodiments according to the present disclosure may visualize effective connectivity and structural connectivity information for cognitive functions. For example, embodiments according to the present disclosure can be used as diagnostic and monitoring biomarkers by identifying connectivity abnormalities in patients with cognitive dysfunction. For example, through a prospective study to extract nDT precision brain images according to the present disclosure for patients with lesions and to identify abnormalities in specific cognitive function areas in patients with abnormalities in specific tracts, specific cognition You can check the tract responsible for the function. In addition, the embodiments according to the present disclosure can be used as monitoring indicators in the evaluation of stopping epilepsy drugs and whether driving is possible.

FIG. 2 is a flowchart for explaining step S110 of FIG. 1 . Hereinafter, description will be made with reference to FIG. 1, and overlapping descriptions will be omitted.

Referring to FIG. 2 , step S110 may include steps S111, S112, S113, and S114.

In step S111, MRI imaging may be performed.

In step S112, the processor of the computer may acquire T1 weighted imaging and diffusion weighted imaging (DWI) provided through MRI imaging.

A T1-weighted image is an image in which a difference in tissue relaxation time is reflected as a signal difference, and T1 may be a constant that causes image contrast by emphasizing different tissue components. For example, in a T1-weighted image, fat may have high contrast, water may have low contrast, and air and dense bone may have the lowest contrast. The T1-weighted image may be implemented in 3D.

The diffusion-weighted image (DWI) may be an image reflecting a difference in intensity of a signal caused by a phenomenon in which material molecules move from a side having a high molecular concentration to a side having a low molecular concentration.

In step S113, the processor of the computer may pre-process the diffusion-enhanced image (DWI). For example, the computer can remove noise in the diffusion-weighted image (DWI) and adjust the inhomogeneity and gradient eddy currents of the B1 field. The processor of the computer can bias-correct by removing the skull of the diffusion-weighted image (DWI) and correcting EPI (Echo Planar Imaging) distortion.

In step S113, the processor of the computer may extract a fractional anisotropy (hereinafter referred to as 'FA') map using the preprocessed diffusion-enhanced image (DWI). The FA map shows that FA decreases as the diffusion of material molecules is free in various directions and isotropic, and FA increases as the diffusion of material molecules is biased in one direction and becomes anisotropic. there is.

FIG. 3 is a flowchart for explaining step S120 of FIG. 1, and FIG. 4 is a diagram for explaining the brain region divided in step S121. Hereinafter, description will be made with reference to FIGS. 1 and 2, and overlapping descriptions will be omitted.

Step S120 may include steps S121, S122, S123, and S124.

In step S121, the processor of the computer may segment the brain region based on the T1-weighted image. For example, as shown in FIG. 4 , the processor of the computer may segment the patient's brain region using a cortical parcellation system (CPS), and may include a pre-photographed T1-weighted image of the patient. Brain regions can be distinguished and displayed.

In step S122, the processor of the computer may determine the intensity of stimulation to be used in the nTMS system by measuring a Motor Evoked Potential (MEP). Hereinafter, the strength of the stimulus determined in step S122 may be referred to as 'RMT (Resting Motor Threshold)'.

For example, the processor of the computer may provide a single pulse to a coil disposed perpendicular to the motor cortical region to be stimulated, and continuously apply 10 pulses to a location where a motor evoked potential (MEP) is best induced. stimulus can be provided. The processor of the computer may determine, as the RMT, the minimum stimulus intensity at which a motor evoked potential (MEP) of 50 μV or more is induced 5 times or more out of 10 times. The coil may be an 8-character coil.

In step S123, the processor of the computer may stimulate the divided brain regions with RMT using the nTMS system. For example, the processor of the computer may stimulate each divided brain region with the stimulus intensity RMT determined in step S122 through coils disposed in each brain region divided through step S121. . The coil may be an 8-character coil.

For example, the nTMS system may include parameters such as frequency of the stimulation, stimulation intensity, number of stimulations, and stimulation orientation. For example, Frequency of the stimulation is 0.2 Hz, Number of stimulations is 30 times on average for each brain region, and Stimulation Orientation is by rotating the angle of the coil clockwise by 45 °. It is set to change, and the stimulation intensity (Stimulation intensity) is set to 100% of the RMT, but can be set to be adjusted by 10% based on the patient's pain level. According to an embodiment, the processor of the computer may stimulate each divided brain region up to 80 times every 5 to 6 seconds using a single pulse having an intensity of 120% of the RMT.

In step S124, the processor of the computer may measure and store EEG data generated by stimulating the brain region. For example, the processor of a computer may measure brain waves using 64 or more channels of electrodes while stimulating a brain region. At this time, the processor of the computer may maintain the impedance of each channel to 5KΩ or less, and may set the sampling rate of EEG data to 2000Hz. The processor of the computer may filter the measured EEG data using a signal bandpass filter having a frequency of 1 Hz or more and 100 Hz or less. However, embodiments according to the present disclosure are not limited thereto, and the processor of the computer may filter the measured EEG data using a signal band communication filter having various frequencies.

In addition, the processor of the computer maintains the input of the amplifier from 100 ms before stimulation to 2 ms after stimulation in order to avoid amplifier saturation due to stimulation of the brain region (amplifier input constant) (sample-and-hold circuit) -hold circuit) can be used. However, embodiments according to the present disclosure are not limited thereto, and the sample hold circuit may be omitted. For example, a sample hold circuit may be omitted when a DC amplifier is used.

5 is a flowchart for explaining step S140 of FIG. 1, and FIG. 6 is a flowchart for explaining another embodiment of step S140 of FIG. Hereinafter, all descriptions will be made with reference to FIGS. 1 to 4, and overlapping descriptions will be omitted.

Referring to FIG. 5 , step S140 may include steps S141, S142, and S143.

In step S141, the processor of the computer may extract a component from nCCEP data. The component may include a first component N1 in the form of a sharp wave appearing in an initial period of 10 ms to 50 ms. The first component N1 may be a waveform that appears when nerve cells are excited by electrical stimulation. The processor of the computer may extract the first component N1 by checking electrical stimulation by the nTMS system and negative deflection appearing in the initial period from 10 ms to 50 ms. The computer can calculate the amplitude, latency and velocity of the first component N1 for each tract. For example, the processor of the computer may calculate the speed of the first component N1 by dividing the minimum path length between the two electrodes by the latency of the first component N1 between the two electrodes.

Depending on the embodiment, the component may further include a second component N2 in the form of a slow wave appearing in an initial period of 50 ms to 300 ms. The second component N2 may be a waveform that appears when the brain goes through various inhibition processes. The second component N2 is an abnormal brain wave, and may be used as an indicator suggesting an epileptic induction site. The computer may extract the second component N2 by calculating the absolute value of the standard deviation (Z score) appearing in the interval from the initial 50 ms to 300 ms using the standard deviation of the baseline. For example, the computer may calculate the standard deviation for each electrode and determine that the data is significant when the peak of the second component (N2) is greater than a threshold value of ±6 standard deviations from the amplitude of the baseline. there is.

In step S142, the processor of the computer may perform statistical analysis based on the extracted components. The computer can calculate the Euclidean distance between the endpoints of the path and statistically infer the length of the path in order to minimize the biologically implausible path. The processor of the computer may perform statistical analysis using a deep learning model that has been trained to minimize biologically implausible paths.

The deep learning model can be trained to compute a group-level linear regression model related to distance and length. For example, the deep learning model may be trained to perform regression analysis for each interval in which the peak amplitude of the first component N1 and the latency of the first component N1 are averaged for each 10 mm interval of the path length. The deep learning model can calculate the average velocity for 15 FA bins of the same size in order to create a model that can compare the propagation velocity of the first component N1, and use the calculated result value in the regression model. can be learned to do. Depending on the embodiment, the deep learning model may be trained to remove paths with model errors greater than 3 standard deviations of the mean. According to the embodiment, the deep learning model uses only the amplitude, latency, and speed of the first significant component (N1), the length of the connection path extracted through diffusion-weighted image (DWI), and electrode pairs having an average FA value greater than 0.2. and can be trained to perform statistical analysis.

In step S143, the processor of the computer may extract a non-invasive dynamic path (nDT) by performing multi-modal imaging based on statistical analysis. The processor of the computer may determine whether the first component N1 conforms to the nDT model, and may select a part conforming to the nDT model and determine it as the nDT. nDT has a positive correlation between the length of the connection path inferred in step S142 and the latency of the first component N1, and nDT is the length of the connection path inferred in step S142 and the first component N1. It may be data conforming to an nDT model in which the amplitude of β has a negative correlation and the FA map and the velocity of the first component N1 have a positive correlation.

When the processor of the computer further extracts the second component N2, the processor of the computer implements a pattern corresponding to the epilepsy inducing region using the amplitude ratio of the first component N1 and the second component N2. can That is, according to an embodiment of the present disclosure, since nCCEP data can be measured for the entire brain, a pattern corresponding to an epilepsy-inducing region can be implemented through the amplitude ratio of the first component N1 and the second component N2. . Therefore, in this case, several more steps may be performed to extract the non-invasive dynamic pathway (nDT). This will be described in more detail with reference to FIG. 6 .

Referring to FIG. 6 , step S140′ may include steps S141, S143, S144, S145, S146, and S147, and steps S141 and S143 are steps S141 and 143 of FIG. ) may be the same step. Therefore, a description overlapping with that of FIG. 5 will be omitted.

In step S144, the processor of the computer may perform effective connectivity analysis using the amplitude ratio of the first component N1 and the second component N2. For example, the computer can extract the overall effective connectivity data of one subject, organize and visualize the effective connectivity data through a matrix.

In this case, the bipolar channel may constitute a node, and the root mean square (RMS) may constitute an edge. Root Mean Square (RMS) can be configured with the size of nCCEP data to reconstruct an effective connectivity network indicator. The computer can extract the effective connectivity by calculating the average value of the connection weights connecting all nodes in the same area and between areas.

In step S145, the processor of the computer may define brain regions using the available connectivity data. The computer can use the available connectivity data to define brain regions as Epileptogenic Zone (EZ), Propagation Zone (PZ), and Non-Involved Zone (NIZ).

For example, the processor of the computer may define the corresponding brain region as an epileptic zone (EZ) when the connectivity index of the brain region is equal to or greater than a preset threshold voltage. The preset threshold voltage may be a value indicating an epileptic zone (EZ). The processor of the computer may define the corresponding brain region as a proliferative zone (PZ) when the connectivity index of the brain region has connectivity but is less than a specific threshold voltage indicating an epileptic zone (EZ). The processor of the computer may define the brain region as a non-involved zone (NIZ) when the connectivity index of the brain region has no connectivity.

In step S146, the processor of the computer may perform electrical source imaging (ESI) analysis on the defined brain regions and visualize them using various algorithms. For example, the processor of a computer may use a fast spatiotemporal iteratively reweighted edge sparsity (FAST-IRES) algorithm to visualize data.

The FAST-IRES algorithm determines the range by measuring the connectivity of each node by imposing a penalty according to the amplitude of the EEG data, and converts the brain signal measured from the scalp into a time base function through component analysis. function) and solve the identified focal source using a convex optimization tool.

In step S147, the processor of the computer may determine the direction based on the extracted effective connectivity data. The processor of the computer has no discharge in the epileptic zone (EZ), an epileptic zone (EZ) adjacent zone anatomically adjacent to the epileptic zone (EZ), an interictal state, and no discharge Directionality can be determined from electrical stimulation to a zone adjacent to the zone NIZ.

In step S143, the processor of the computer may implement a pattern corresponding to the epileptic induction site by performing multimodal imaging. The processor of the computer may output a pattern corresponding to the epileptic region as data such as nDT.

Referring to FIG. 7 , various methods of implementing nDT precision brain imaging according to an embodiment of the present invention are illustrated.

According to the first display method v1, the nDT precision brain image may be implemented as a matrix. For example, the processor of the computer may use a two-dimensional heat map to represent the evoked connection strength (eg, RMS) of stimulating the pair of connections.

According to the second display method (v2), the nDT precision brain image may be implemented as a circle map. For example, the processor of a computer may visualize a connectivity matrix.

According to the third display method (v3), the nDT precision brain image can be implemented by visualizing the brain volume. For example, the processor of the computer may represent the response around the electrode location according to linear interpolation to estimate the activity of the unsampled surrounding area.

According to the fourth display method (v4), the nDT precision brain image can be implemented by visualizing data on the surface of the brain. For example, the processor of a computer may display a connectivity matrix on a three-dimensional brain surface along with connectivity locations to directly display connectivity and connectivity locations on the brain surface.

Meanwhile, the disclosed embodiments may be implemented in the form of a recording medium storing instructions executable by a computer. Instructions may be stored in the form of program codes, and when executed by a processor, create program modules to perform operations of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

Computer-readable recording media include all types of recording media in which instructions that can be decoded by a computer are stored. For example, there may be read only memory (ROM), random access memory (RAM), magnetic tape, magnetic disk, flash memory, optical data storage device, and the like.

As above, the disclosed embodiments have been described with reference to the accompanying drawings. Those skilled in the art to which the present disclosure pertains will understand that the present disclosure may be implemented in a form different from the disclosed embodiments without changing the technical spirit or essential features of the present disclosure. The disclosed embodiments are illustrative and should not be construed as limiting.

Claims

database; and

contains a processor;

the processor,

When image data is acquired through MRI (Magnetic Resonance Imaging) imaging, EEG data is measured using nTMS (navigated transcranial magnetic stimulation) system,

Extracting noninvasive corticocortical evoked potential (nCCEP) data using the brain wave data,

Determining a noninvasive dynamic tractography (nDT) by performing multimodal imaging based on the image data and the nCCEP data,

Based on the non-invasive dynamic pathway, nDT precise brain imaging is visualized,

An nDT precise brain image generating device that determines whether there is an abnormality in the brain by comparing and analyzing the non-invasive dynamic pathway and the visualized nDT precise brain image with data stored in a database.
According to claim 1,

The image data includes a T1-weighted image and a diffusion-weighted image,

the processor,

Upon obtaining the image data, the computer pre-processes the diffusion-enhanced image by removing noise from the diffusion-enhanced image and removing a skull;

An apparatus for generating an nDT precise brain image, which extracts an FA map using the preprocessed diffusion-weighted image.
According to claim 2,

the processor,

When measuring the brain wave data, the brain region is divided using a cortical segmentation system,

The divided brain regions are divided and displayed on the T1-weighted image;

determine the strength of the stimulus by measuring the motor evoked potential;

Stimulating the divided brain region with the determined intensity of stimulation using the nTMS system;

Measuring and storing EEG data generated as the divided brain regions are stimulated with the determined intensity of stimulation;

The nTMS system includes, as a parameter, at least one of the frequency of stimulation, the intensity of stimulation, the number of stimulation for each brain region, and the direction of stimulation, nDT precision brain image generating device.
According to claim 3,

the processor,

When determining the non-invasive dynamic path, extracting a first component in the form of a sharp wave appearing in an initial period of 10 ms to 50 ms of the nCCEP data,

Statistical analysis is performed by calculating a Euclidean distance between path endpoints based on the first component and statistically inferring a path length;

Extracting a non-invasive dynamic path by performing multimodal imaging based on the statistical analysis results,

Extracting a second component in the form of a slow wave appearing in the interval from the initial 50 ms to 300 ms of the nCCEP data,

Extracting effective connectivity data using the first component and the second component;

Defining a brain region based on the effective connectivity data;

Visualizing the defined brain region by performing electrical source image analysis;

In the non-invasive dynamic path, the length of the inferred path and the latency of the first component have a positive correlation, and the length of the inferred path and the amplitude of the first component have a negative correlation, An apparatus for generating nDT precise brain images, which conforms to a dynamic path model in which the FA map and the speed of the first component have a positive correlation.
According to claim 4,

the processor,

When defining the brain region, a brain region in which the connectivity index of the effective connectivity data is equal to or greater than a preset threshold voltage is defined as an epileptogenic zone,

A brain region where the connectivity index of the effective connectivity data has connectivity but is lower than a preset threshold voltage is defined as a propagation zone,

A brain region in which the connectivity index of the effective connectivity data has no connectivity is defined as a non-involved zone.
According to claim 5,

the processor,

When determining whether the brain is abnormal, the prognosis of seizures after epilepsy surgery is categorized into true positive, false positive, true negative, and false negative,

An nDT precise brain image generating device that evaluates the final accuracy of localization of an epilepsy-induced region.
According to claim 6,

the processor,

Visualizing the nDT precise brain image using at least one of a matrix, a circle map, a brain volume, and a surface of the brain when visualizing the nDT precise brain image , nDT precision brain image generating device.
In the nDT precision brain image generation method performed by a computer,

obtaining, by the computer, image data through Magnetic Resonance Imaging (MRI) imaging;

measuring EEG data by the computer using a navigated transcranial magnetic stimulation (nTMS) system;

extracting, by the computer, noninvasive corticocortical evoked potential (nCCEP) data using the EEG data;

determining, by the computer, a noninvasive dynamic tractography (nDT) by performing multimodal imaging based on the image data and the nCCEP data;

Visualizing, by the computer, an nDT precise brain image based on the non-invasive dynamic path; and

The nDT precise brain image generation method comprising the step of determining, by the computer, whether there is an abnormality in the brain by comparing and analyzing the non-invasive dynamic path and the visualized nDT precise brain image with data stored in a database.
According to claim 8,

The image data includes a T1-weighted image and a diffusion-weighted image,

Obtaining the image data,

pre-processing, by the computer, the diffusion-enhanced image by removing noise from the diffusion-enhanced image and removing a skull; and

Further comprising the step of extracting, by the computer, an FA map using the preprocessed diffusion-weighted image.
According to claim 9,

The step of measuring the brain wave data,

dividing, by the computer, brain regions using a cortical segmentation system, and displaying the divided brain regions in the T1-weighted image;

determining, by the computer, the strength of the stimulus by measuring the motor evoked potential;

stimulating, by the computer, the divided brain regions with the determined intensity of stimulation using the nTMS system; and

Measuring and storing EEG data generated as the computer stimulates the divided brain regions with the determined intensity of stimulation;

The nTMS system includes, as a parameter, at least one of the frequency of stimulation, the intensity of stimulation, the number of stimulation for each brain region, and the direction of stimulation, nDT precise brain image generation method.
According to claim 10,

The step of determining the non-invasive dynamic path,

extracting, by the computer, a first component in the form of a sharp wave appearing in an initial period of 10 ms to 50 ms of the nCCEP data;

performing statistical analysis by the computer by calculating a Euclidean distance between path endpoints based on the first component and statistically inferring a path length;

extracting a non-invasive dynamic path by performing multimodal imaging based on the statistical analysis result;

extracting, by the computer, a second component in the form of a slow wave appearing in a section from an initial period of 50 ms to 300 ms of the nCCEP data;

extracting, by the computer, valid connectivity data using the first component and the second component;

defining, by the computer, a brain region based on the valid connectivity data; and

further comprising the computer visualizing the defined brain region by performing electrical source image analysis;

In the non-invasive dynamic path, the length of the inferred path and the latency of the first component have a positive correlation, and the length of the inferred path and the amplitude of the first component have a negative correlation, A method for generating nDT precise brain images, wherein the FA map and the velocity of the first component conform to a dynamic path model having a positive correlation.
According to claim 11,

Defining the brain region,

defining, by the computer, a brain region in which a connectivity index of the effective connectivity data is equal to or greater than a preset threshold voltage as an epileptogenic zone;

defining a brain region where the connectivity index of the effective connectivity data has connectivity but is lower than a preset threshold voltage as a propagation zone; and

And defining a brain region in which the connectivity index of the effective connectivity data has no connectivity as a non-involved zone.
According to claim 12,

The step of determining whether the brain is abnormal,

categorizing, by the computer, the prognosis of seizures after epilepsy surgery into true positive, false positive, true negative, and false negative; and

The nDT precise brain image generation method, further comprising the step of allowing the computer to evaluate the final accuracy of localization of the epilepsy-induced region.
According to claim 13,

Visualizing the nDT precise brain image,

Visualizing, by the computer, the nDT precise brain image using at least one of a matrix, a circle map, a brain volume, and a surface of the brain. Methods for generating brain images.
A computer-readable recording medium in which a program for performing the method for generating an nDT precise brain image according to claim 8 is stored in combination with a computer, which is hardware.