EEG-based dementia diagnosis assistance method and device, and EEG-based dementia diagnosis assistance system comprising same

Abstract

An EEG-based dementia diagnosis assistance device according to the present embodiment comprises: an EEG data preprocessing unit for preprocessing a collected EEG data set of a test subject, the collected EEG data set being divided into time intervals by setting a window length and a shift length; an Hjorth parameter extraction unit for extracting, for each time interval, Hjorth parameters including an activity index, a mobility index, and a complexity index; and a dementia classification unit for classifying the test subject into one from among a Alzheimer's disease (AD) group, a frontotemporal dementia (FTD) group, and a cognitive normal (CN) group on the basis of the extracted Hjorth parameters. Therefore, AD, FTD, and CN subjects can be classified using Hjorth parameters composed of three major indicators of activity, mobility, and complexity through time domain analysis on EEG datasets.

Description

EEG-based dementia diagnosis assistance method and device and EEG-based dementia diagnosis assistance system including the same

[0001] The present invention relates to an EEG-based dementia diagnosis assistance method and device and an EEG-based dementia diagnosis assistance system including the same. More specifically, it relates to an EEG-based dementia diagnosis assistance method and device that classifies subjects with Alzheimer's disease (AD), frontotemporal dementia (FTD), and cognitive normal (CN) using Hjorth parameters consisting of three major indicators of activity, mobility, and complexity through time domain analysis of an electroencephalography (EEG) dataset, and an EEG-based dementia diagnosis assistance system including the same.

[0002]

[0003] Generally, dementia is a neurodegenerative disease characterized by a progressive decline in cognitive function, which can occur when specific clusters of brain cells fail to function properly. The most common types of dementia are Alzheimer's disease (AD) and frontotemporal dementia (FTD). Alzheimer's disease (AD) is characterized by memory loss, aphasia, and visuospatial problems, while frontotemporal dementia (FTD) is characterized by personality and behavioral changes.

[0004] In the early stages of dementia, diagnostic uncertainty frequently arises, and the broad scope of differential diagnosis makes it difficult to obtain a specific and timely diagnosis. Furthermore, cognitive decline may not be noticeable in the early stages, and prolonged asymptomatic periods hinder early detection. However, if patients are not properly managed during the early stages, the risk of continuous cognitive impairment increases, and the effectiveness of treatment decreases as diagnosis is delayed. Therefore, the early detection of dementia is a critical task and constitutes one of the most significant challenges in the field.

[0005] Existing methods for the early diagnosis of dementia include the patient's medical history, clinical observation, and neuropsychological tests such as the Mini-Mental State Examination (MMSE). However, these methods face difficulties when there are no obvious signs of cognitive decline.

[0006] In addition, imaging technologies such as Positron Emission Tomography (PET), Single Photon Emission Computed Tomography (SPECT), Magnetic Resonance Imaging (MRI), and Computed Tomography (CT) are also used. These methods can be used to observe structural changes in the brains of dementia patients and to identify affected areas and types of atrophy. While these technologies offer relatively high accuracy, they cost thousands of euros or dollars per case and require a long time for a final diagnosis. Furthermore, although imaging technologies possess high spatial resolution, they have low temporal resolution; consequently, they lack sensitivity to detect early risk factors or changes, making it difficult to detect the early stages of dementia. Consequently, many dementia patients are diagnosed only after significant neurodegeneration has already occurred, making early detection of the disease challenging.

[0007]

[0008] Accordingly, the technical problem of the present invention is based on this point, and the objective of the present invention is to provide an EEG-based dementia diagnostic assistance method that classifies subjects with Alzheimer's disease (AD), frontotemporal dementia (FTD), and cognitive normal (CN) using Hjorth parameters, which consist of three major indicators of activity, mobility, and complexity, through time domain analysis of an electroencephalogram (EEG) dataset.

[0009] Another objective of the present invention is to provide an EEG-based dementia diagnosis aid device for performing the above-described EEG-based dementia diagnosis aid method.

[0010] Another objective of the present invention is to provide an EEG-based dementia diagnosis assistance system comprising the above-described EEG-based dementia diagnosis assistance device.

[0011]

[0012] To realize the purpose of the present invention as described above, an EEG-based dementia diagnosis assistance method according to one embodiment comprises: (i) preprocessing a collected EEG dataset of a test subject, wherein the collected EEG dataset is divided into time intervals by setting a window length and a movement length; (ii) extracting Hjorth parameters including an activity index, a mobility index, and a complexity index for each time interval; and (iii) classifying the test subject into one of an Alzheimer's disease (AD) group, a frontotemporal dementia (FTD) group, and a cognitive normal (CN) group based on the extracted Hjorth parameters.

[0013] In one embodiment of the present invention, the moving length may be a first length in the classification between the dementia patient group including the AD group and the FTD group and the CN group and between the AD group and the CN group, and a second length longer than the first length in the classification between the FTD group and the CN group.

[0014] In one embodiment of the present invention, step (i) comprises: (i-1) a step of checking whether to classify a dementia patient (AD&FTD) group including the AD group and the FTD group and a cognitively normal (CN) group; (i-2) a step of, if checked to classify the dementia patient (AD&FTD) group and the cognitively normal (CN) group, setting a movement length to a first length and setting a window length to a second length longer than the first length; (i-3) a step of checking whether to classify the dementia patient (AD&FTD) group and the cognitively normal (CN) group if not checked to classify the dementia patient (AD&FTD) group and the cognitively normal (CN) group; (i-4) a step of feeding back to step (ii) if checked to classify the Alzheimer's disease (AD) group and the cognitively normal (CN) group; (i-5) a step of checking whether to classify the frontotemporal dementia (FTD) group and the cognitive normal (CN) group if it is not checked that the Alzheimer's disease (AD) group and the cognitive normal (CN) group are classified; (i-6) a step of terminating if it is not checked that the frontotemporal dementia (FTD) group and the cognitive normal (CN) group are classified; and (i-7) a step of setting the movement length to the first length and setting the window length to the second length if it is checked that the frontotemporal dementia (FTD) group and the cognitive normal (CN) group are classified.

[0015] To realize another objective of the present invention as described above, an EEG-based dementia diagnostic aid device according to one embodiment comprises: an EEG data preprocessing unit that preprocesses an electroencephalogram (EEG) dataset of a subject collected for examination, and divides the collected EEG dataset into time intervals by setting a window length and a movement length for the data; a Hjorth parameter extraction unit that extracts Hjorth parameters including an activity index, a mobility index, and a complexity index for each time interval; and a dementia classification unit that classifies the subject into one of an Alzheimer's disease (AD) group, a frontotemporal dementia (FTD) group, and a cognitive normal (CN) group based on the extracted Hjorth parameters.

[0016] In one embodiment of the present invention, the preprocessing unit may divide the EEG data set into window length and movement length according to time intervals.

[0017] In one embodiment of the present invention, the EEG data preprocessing unit can remove artifacts from the EEG data set using Artifact Subspace Reconstruction (ASR).

[0018] In one embodiment of the present invention, the EEG data preprocessing unit can separate eye blinking and jaw movement artifacts from the EEG data set through Independent Component Analysis (ICA).

[0019] In one embodiment of the present invention, the EEG data preprocessing unit can remove low-frequency drift and high-frequency noise by passing only a certain frequency band through a Butterworth bandpass filter for the EEG data set.

[0020] In one embodiment of the present invention, the dementia classification unit may include a first classifier for classifying a dementia patient (AD & FTD) group and a cognitively normal (CN) group; a second classifier for classifying an Alzheimer's disease (AD) group and a cognitively normal (CN) group; and a second classifier for classifying a frontotemporal dementia (FTD) group and a cognitively normal (CN) group.

[0021] In one embodiment of the present invention, the first classifier and the second classifier may include a Linear Discriminant Analysis (LDA) classification model, and the third classifier may include a Support Vector Machine (SVM) classification model.

[0022] To realize another objective of the present invention as described above, an EEG-based dementia diagnosis assistance system according to one embodiment comprises: a test probe attached to a pre-set area of ​​a test subject to collect an electroencephalogram (EEG) data set of the test subject; an EEG data preprocessing unit that preprocesses the collected EEG data set of the test subject, wherein the collected EEG data set is divided into time intervals by setting a window length and a movement length for the collected EEG data set; a Hjorth parameter extraction unit that extracts Hjorth parameters including an activity index, a mobility index, and a complexity index for each time interval; and a dementia classification unit that classifies the test subject into one of an Alzheimer's disease (AD) group, a frontotemporal dementia (FTD) group, and a cognitive normal (CN) group based on the extracted Hjorth parameters.

[0023]

[0024] According to this EEG-based dementia diagnostic aid method and device and the EEG-based dementia diagnostic aid system including it, it is possible to classify subjects with Alzheimer's disease (AD), frontotemporal dementia (FTD), and cognitive normal (CN) by using Hjorth parameters, which consist of three major indicators of activity, mobility, and complexity, through time domain analysis of an EEG dataset.

[0025]

[0026] FIG. 1 is a diagram illustrating an EEG-based dementia diagnosis assistance system according to one embodiment of the present invention.

[0027] Figure 2 is a block diagram illustrating an EEG-based dementia diagnostic aid device illustrated in Figure 1.

[0028] Figure 3 is a graph illustrating the window length and shift length in brainwave signal segmentation for normalization.

[0029] FIG. 4 is a flowchart illustrating an EEG-based dementia diagnosis assistance method according to one embodiment of the present invention.

[0030] Figure 5 is a flowchart illustrating the step of dividing the EEG data set shown in Figure 4 according to time intervals.

[0031] Figure 6 shows a topographic plot illustrating the ability of the Hjorth parameter to distinguish between dementia (AD and FTD) groups and cognitively normal (CN) groups.

[0032] Figure 7 shows a topographic plot illustrating the ability of the Hjorth parameter to distinguish between the Alzheimer's disease (AD) group and the frontotemporal dementia (FTD) group.

[0033] Figures 8a, 8b, and 8c show heatmaps showing the performance over time intervals between Alzheimer's disease (AD), frontotemporal dementia (FTD), and cognitive normal (CN).

[0034] FIG. 9 is a block diagram illustrating an exemplary computer suitable for use in an embodiment of the present invention.

[0035]

[0036] Embodiments of the present invention are described below with reference to the attached drawings so that those skilled in the art can easily implement the invention. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. Furthermore, in order to clearly explain the present invention in the drawings, parts unrelated to the explanation have been omitted, and similar parts throughout the specification are denoted by similar reference numerals.

[0037] Throughout the specification, when a part is described as being "connected" to another part, this includes not only cases where they are "directly connected" but also cases where they are "electrically connected" with other elements interposed between them. Furthermore, when a part is described as "including" a component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components, and it should be understood that this does not preclude the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0038] Terms such as “about,” “substantially,” etc., used throughout the specification, are used to mean at or near the stated value when inherent manufacturing and material tolerances are presented in the stated meaning, and are used to prevent unscrupulous infringers from unfairly exploiting the disclosure in which precise or absolute values ​​are mentioned to aid in understanding the invention. Terms such as “step” or “step of” used throughout the specification of the invention do not mean “step for”.

[0039] In this specification, the term "part" includes a unit realized by hardware, a unit realized by software, and a unit realized using both. Additionally, one unit may be realized using two or more pieces of hardware, and two or more units may be realized by one piece of hardware. Meanwhile, the term "part" is not limited to software or hardware; the "part" may be configured to reside in an addressable storage medium or configured to run on one or more processors. Accordingly, as an example, the "part" includes components such as software components, object-oriented software components, class components, and task components, as well as processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided within the components and "parts" may be combined into a smaller number of components and "parts" or further separated into additional components and "parts." In addition, the components and '~parts' may be implemented to play one or more CPUs within the device or secure multimedia card.

[0040] Some of the operations or functions described herein as being performed by a terminal, device, or device may instead be performed by a server connected to said terminal, device, or device. Likewise, some of the operations or functions described as being performed by a server may also be performed by a terminal, device, or device connected to said server.

[0041] In this specification, some of the operations or functions described as mapping or matching with a terminal may be interpreted as meaning mapping or matching the terminal's unique number or personal identification information, which is the terminal's identifying data.

[0042] The present invention will be described in detail below with reference to the attached drawings.

[0043] FIG. 1 is a diagram illustrating an EEG-based dementia diagnosis assistance system according to one embodiment of the present invention.

[0044] Referring to FIG. 1, an EEG-based dementia diagnosis assistance system (100) according to one embodiment of the present invention includes a test probe (110) and an EEG-based dementia diagnosis assistance device (130).

[0045] The test probe (110) includes brainwave measuring electrodes (117a, 117b) and a reference electrode (119), and is attached to a pre-set area of ​​the test subject to acquire the brainwave signal of the test subject to test the activity of the nerve cell itself.

[0046] The brainwave measuring electrodes (117a, 117b) are attached to the forehead or head of the subject to measure the subject's brainwaves. By measuring potential values ​​(brainwaves) at different locations, the brainwave measuring electrodes (117a, 117b) enable the EEG-based dementia diagnosis aid device (130) to identify potential differences. The reference electrode (119) operates as a ground electrode.

[0047] The test probe (110) includes components that perform the aforementioned operations and transmits data measured by the brainwave measurement electrode (117) to the EEG-based dementia diagnosis aid device (130). The test probe (110) may include a connector (120) connected to the EEG-based dementia diagnosis aid device (130) at one end. The test probe (110) can be attached to and detached from the EEG-based dementia diagnosis aid device (130) using the connector (120), thereby determining whether the two components (110, 130) are electrically connected. Since the test probe (110) is attached to a test subject and used, there is a need for replacement periodically or after each use. Accordingly, to facilitate the replacement of only the test probe (110), the test probe (110) may include a connector (120) at one end.

[0048] The EEG-based dementia diagnostic aid (130) diagnoses the type of dementia and the likelihood of dementia in the subject based on the test results from the test probe (110).

[0049] Figure 2 is a block diagram illustrating an EEG-based dementia diagnostic aid device illustrated in Figure 1.

[0050] Referring to FIGS. 1 and 2, the EEG-based dementia diagnosis aid device (130) includes an EEG data preprocessing unit (132), a Hjorth parameter extraction unit (134), and a dementia classification unit (136). In this embodiment, the EEG-based dementia diagnosis aid device (130) is shown to be composed of an EEG data preprocessing unit (132), a Hjorth parameter extraction unit (134), and a dementia classification unit (136), but this is a logical distinction made for convenience of explanation and is not a hardware distinction.

[0051] The EEG data preprocessing unit (132) preprocesses the collected EEG data set, and divides the collected EEG data into time intervals by setting a window length and a moving length.

[0052] The EEG data preprocessing unit (132) processes the EEG dataset using the Artifact Subspace Reconstruction (ASR) routine in the EEGLAB Matlab software. Specifically, the ASR processes artifacts in the EEG dataset. In particular, it can effectively remove electrical noise, eye blinking, and motion artifacts that may affect brain waves. It performs the task of extracting and removing artifacts using the ASR routine provided by the EEGLAB software. This improves the quality of the data and enhances the accuracy of the analysis.

[0053] Additionally, the EEG data preprocessing unit (132) removes "eye blink artifacts" and "jaw movement artifacts" through Independent Component Analysis (ICA). Specifically, ICA is an algorithm that separates mixed signals into independent components. In an EEG dataset, artifacts such as eye blinks, jaw movements, and heartbeats may exist, and ICA can separate these artifacts into independent components and remove them. In this process, eye blink artifacts and jaw movement artifacts are removed, leaving only the actual brainwave signals.

[0054] The EEG data preprocessing unit (132) applies a Butterworth bandpass filter to the EEG data set to filter it into a frequency band in the range of 0.5 to 45 Hz and removes low-frequency drift and high-frequency noise. Specifically, the EEG data set contains various frequency bands. However, the main frequency band required for analysis is usually between 0.5 Hz and 45 Hz. To this end, a Butterworth bandpass filter is used to pass only signals within the range of 0.5 to 45 Hz, and remove other low-frequency (drift) and high-frequency (noise) components. This filtering process improves the quality of the signal and provides data suitable for analysis.

[0055] The EEG data preprocessing unit (132) converts noise-removed brainwave data into an electroencephalogram (EEG) signal. For example, the EEG data preprocessing unit (132) can remove additional noise by performing a Fast Fourier Transform (FFT) on the EEG data set, which is noise-removed brainwave data. After removing the additional noise, the EEG data preprocessing unit (132) can obtain an EEG data set, which is a pure electroencephalogram signal, by performing an Inverse Fast Fourier Transform (IFFT) on the Fast Fourier Transformed data.

[0056] The EEG data preprocessing unit (132) re-referencing the signal filtered by the average value of two reference electrodes (A1 and A2) and from which low-frequency drift and high-frequency noise have been removed. Specifically, the EEG data set is generally re-referencing using two reference electrodes. In this case, the EEG data set is re-referencing using the average value of two reference electrodes, A1 and A2. This allows the signal recorded at each electrode to be compared with the average of the reference electrodes, thereby more accurately reflecting the difference between the electrodes.

[0057] The EEG data preprocessing unit (132) normalizes the data length to provide consistency in analysis because the EEG recording time of each test subject is different.

[0058] Figure 3 is a graph illustrating, exemplarily, the window length and shift length in brainwave signal segmentation for normalization by the EEG data preprocessing unit shown in Figure 2.

[0059] Referring to Fig. 3, each data point is divided into a window length, for example, from 20 seconds to 100 seconds. Each window is divided by a movement length, moving at intervals of, for example, 10 seconds to 100 seconds. Here, the window length refers to a specific time interval, and the movement length refers to the time interval between the starting point of the previous window and the starting point of the next window. In this process, 90 time interval configurations for analysis are generated.

[0060] Extracting features using overlapping time intervals can improve the signal-to-noise ratio (SNR) and enhance classification accuracy. Specifically, using overlapping time intervals allows for a better identification of signal patterns through features extracted from each interval. Furthermore, this overlapping improves the SNR, enabling the extraction of more accurate features. The features extracted through this process are advantageous for model training and can improve the performance of the classifier.

[0061] Referring again to FIG. 2, the Hjorth parameter extraction unit (134) extracts Hjorth parameters including an activity index, a mobility index, and a complexity index for each time interval. Hjorth parameters are statistical characteristic indicators used when extracting signal features in the time domain.

[0062] The activity coordinate represents the power of the signal and is defined as the variance of the time signal. Activity is expressed by the following equation (1). Here, y(t) represents the signal.

[0063] [Formula 1]

[0064]

[0065] The mobility index refers to the value of the mean frequency. Mobility is defined as the square root of the value obtained by dividing the variance of the first derivative of the signal dy(t) / dt by the variance of the original signal y(t). Mobility is expressed by the following equation (2).

[0066] [Equation 2]

[0067]

[0068] The complexity index indicates the degree of signal frequency variation and compares how close the signal is to a pure sine wave. The closer the value is to 1, the more similar the signal is to a sine wave. Complexity is expressed by the following formula (3).

[0069] [Equation 3]

[0070]

[0071] These activity, mobility, and complexity indices are calculated for each channel and each time segment.

[0072] The dementia classification unit (136) includes a first classifier (210), a second classifier (220), and a third classifier (230), and classifies the subject to examination into one of the following groups based on the extracted Hjorth parameters: Alzheimer's disease (AD), frontotemporal dementia (FTD), and cognitive normal (CN).

[0073] The first classifier (210) classifies the test subjects into a dementia patient (AD&FTD) group and a cognitive normal group based on the extracted Hjorth parameters. In this embodiment, the test subjects may be classified into a dementia patient (AD&FTD) group or a cognitive normal group, or may not be classified.

[0074] The second classifier (220) classifies the test subject into an Alzheimer's disease (AD) group and a cognitive normal group based on the extracted Hjorth parameters. In this embodiment, the test subject may be classified into an Alzheimer's disease (AD) group and a cognitive normal group, or may not be classified.

[0075] The third classifier (230) classifies the test subjects into a frontotemporal dementia (FTD) group and a cognitively normal group based on the extracted Hjorth parameters. In this embodiment, the test subjects may be classified into a frontotemporal dementia (FTD) group and a cognitively normal group, or may not be classified.

[0076] In this embodiment, the first classifier (210) and the second classifier (220) may include a Linear Discriminant Analysis (LDA) classification model, and the third classifier (230) may include a Support Vector Machine (SVM) classification model.

[0077] FIG. 4 is a flowchart illustrating an EEG-based dementia diagnosis assistance method according to one embodiment of the present invention.

[0078] Referring to FIG. 4, an EEG data set of the subject is collected (step S110). Step S110 can be performed by the examination probe (110) described in FIG. 1.

[0079] Next, the EEG data set collected in step S110 is preprocessed, and the collected EEG data set is divided into time intervals by setting a window length and a movement length (step S120). Step S120 can be performed by the EEG data preprocessing unit (132) described in FIG. 2.

[0080] Next, for each time interval of the EEG dataset, Hjorth parameters including activity indicators, mobility indicators, and complexity indicators are extracted (step S130). Step S130 can be performed by the Hjorth parameter extraction unit (134) described in FIG. 2.

[0081] Next, based on the Hjorth parameters extracted in step S130, the subjects are classified into one of the following groups: Alzheimer's disease (AD), frontotemporal dementia (FTD), and cognitive normal (CN) (step S140). Step S140 can be performed by the dementia classification unit (136) described in FIG. 2.

[0082] Figure 5 is a flowchart illustrating the step of dividing the EEG data set shown in Figure 4 according to time intervals.

[0083] Referring to Fig. 5, check whether to classify the dementia patient (AD&FTD) group and the cognitively normal (CN) group (step S121).

[0084] If it is checked that the dementia patient (AD&FTD) group and the cognitively normal (CN) group are classified in step S121, the movement length is set to a first length, e.g., 10 seconds, and the window length is set to a second length, e.g., 80 seconds, which is longer than the first length, to split the EEG data set (step S122), and then feed back to step S130.

[0085] If it is not checked that the dementia patient (AD&FTD) group and the cognitively normal (CN) group are classified in step S121, check whether the Alzheimer's disease (AD) group and the cognitively normal (CN) group are classified (step S123).

[0086] If it is checked that the Alzheimer's disease (AD) group and the cognitive normal (CN) group are classified in step S123, feedback is given to step S122.

[0087] If it is not checked that the Alzheimer's disease (AD) group and the cognitive normal (CN) group are classified in step S123, check whether the frontotemporal dementia (FTD) group and the cognitive normal (CN) group are classified (step S124).

[0088] If the frontotemporal dementia (FTD) group and cognitive normal (CN) group are not checked in step S124, terminate.

[0089] If it is checked that the frontotemporal dementia (FTD) group and the cognitive normal (CN) group are classified in step S124, the window length is set to a second length, e.g. 80 seconds, and the movement length is set to a second length, e.g. 80 seconds, to split the EEG dataset and then feed back to step S130.

[0090] Below, the differences in Hjorth parameters among the AD, FTD, and CN groups are explained.

[0091] Figure 6 shows a topographic plot illustrating the ability of Hjorth parameters to distinguish between dementia (AD and FTD) groups and cognitively normal (CN) groups. In particular, the differences in Hjorth parameters (activity, mobility, complexity) between the dementia (AD and FTD) groups and the cognitively normal (CN) groups are illustrated. Hjorth parameters were extracted from the EEG dataset of each subject, and the topographic plot represents the average Hjorth parameters of each channel.

[0092] Referring to Figure 6, in terms of activity indicators, the CN group showed larger values ​​in the occipital lobe than the dementia (AD and FTD) groups.

[0093] In terms of mobility and complexity indices, AD / FTD showed larger values ​​in the frontal and temporal lobes than the dementia (AD and FTD) group.

[0094] Figure 7 shows a topographic plot illustrating the ability of the Hjorth parameter to distinguish between the Alzheimer's disease (AD) group and the frontotemporal dementia (FTD) group.

[0095] Referring to Figure 7, differences were observed within the dementia groups, namely the Alzheimer's disease (AD) group and the frontotemporal dementia (FTD) group.

[0096] In other words, regarding activity indicators, FTD showed larger values ​​in the frontal and temporal lobes than in the AD group.

[0097] AD and FTD showed similar patterns in mobility and complexity indices. However, FTD exhibited larger values ​​than AD in mobility and complexity indices.

[0098] As can be seen in Figures 6 and 7, although there was no perfect match in the location and value range, it can be confirmed that each Hjorth parameter can distinguish dementia.

[0099] Below, the selection of the optimal time interval for the three groups is explained.

[0100] Figures 8a, 8b, and 8c show heatmaps illustrating time-interval performance among Alzheimer's disease (AD), frontotemporal dementia (FTD), and cognitive normal (CN). Here, columns represent window lengths and rows represent moving lengths. The heatmaps represent the maximum Area Under the Curve (AUC) derived from various machine learning models for different window lengths and moving lengths. In the heatmaps for each group, red squares indicate the optimal time interval.

[0101] Referring to Figure 8a, in the heatmap showing the classification performance between the dementia (AD and FTD) group and the CN group, optimal performance was observed when the shift length was 10 seconds and the window length was 80 seconds. Also, when Linear Discriminant Analysis (LDA) was used as the classifier, the maximum AUC was found to be 0.874.

[0102] Referring to Figure 8b, in the heatmap showing the classification performance between the AD group and the CN group, optimal performance was observed when the movement length was 10 seconds and the window length was 80 seconds. Also, when LDA was used as the classifier, the maximum AUC was found to be 0.881.

[0103] Referring to Figure 8c, in the heatmap showing the classification performance between the FTD group and the CN group, optimal performance was observed when the movement length was 80 seconds and the window length was 80 seconds. Also, when a Support Vector Machine (SVM) was used as the classifier, the maximum AUC was found to be 0.908.

[0104] In all three classifications, the optimal window length was found to be 80 seconds, and the movement lengths varied from 10 seconds to 80 seconds.

[0105] Below, the dementia classification performance in the optimal time interval is described.

[0106] The results of comparing the performance of CatBoost, LightGBM, XGBoost, SVM, and LDA as classification models are presented in Table 1.

[0107] [Table 1]

[0108]

[0109] Referring to Table 1, the CatBoost classification model is based on the gradient boosting algorithm and has strengths in handling categorical variables. The LightGBM classification model is a Gradient Boosting framework developed by Microsoft and has strengths that are advantageous for processing very large datasets. The XGBoost classification model, short for Extreme Gradient Boosting, is based on the gradient boosting algorithm and is a widely used model known for its excellent performance. The Support Vector Machine (SVM) classification model finds a hyperplane that separates data in high dimensions and is primarily used in classification problems; it performs classification by maximizing the expansion of the decision boundary. The Linear Discriminant Analysis (LDA) classification model classifies data by maximizing the difference in variance between classes and minimizing the variance within classes, primarily separating data linearly while emphasizing the differences between classes.

[0110] Performance evaluations for each classification model were performed using various measurement methods (i.e., accuracy, sensitivity, specificity, PPV, NPV, AUC).

[0111] For AD&FTD / CN classification, LDA demonstrated the best performance, with an accuracy of 0.902 and an AUC of 0.874. This model showed the highest sensitivity of 0.965, indicating its ability to effectively identify dementia. Additionally, it exhibited high performance in CN prediction with an NPV of 0.913.

[0112] For AD / CN classification, LDA demonstrated the best performance, with an accuracy of 0.886 and an AUC of 0.881. With a sensitivity of 0.926 and a specificity of 0.835, it showed the ability to distinguish between AD and CN groups in a balanced manner. This suggests that its AD prediction performance is particularly excellent.

[0113] For FTD / CN classification, SVM demonstrated the best performance, with an accuracy of 0.919 and an AUC of 0.907. With a sensitivity of 0.859 and a specificity of 0.955, it shows performance capable of distinguishing between the FTD and CN groups in a balanced manner. This indicates particularly high performance in CN prediction.

[0114] This result shows that each classification model and metric can have different strengths in specific classification tasks.

[0115] FIG. 9 is a block diagram illustrating an exemplary computer suitable for use in an embodiment of the present invention.

[0116] Referring to FIG. 9, the computer (400) may be one or more components included in the EEG-based dementia diagnosis aid device (130, described in FIG. 2). In the illustrated embodiment, each component may have different functions and capabilities in addition to those described below, and may include additional components in addition to those not described below.

[0117] The computer (400) includes at least one processor (410), a computer-readable storage medium (420), and a communication bus (430). The processor (410) can enable the computer (400) to operate according to the exemplary embodiment described above. For example, the processor (410) can execute one or more programs stored in the computer-readable storage medium (420). The one or more programs may include one or more computer-executable instructions, and the computer-executable instructions may be configured to enable the computer (400) to perform operations according to the exemplary embodiment when executed by the processor (410).

[0118] A computer-readable storage medium (420) is configured to store computer-executable instructions or program code, program data and / or other suitable forms of information. A program (422) stored in the computer-readable storage medium (420) includes a set of instructions executable by a processor (410). In one embodiment, the computer-readable storage medium (420) may be memory (volatile memory such as random access memory, non-volatile memory, or a suitable combination thereof), one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, other forms of storage media that can be accessed by a computer (400) and store desired information, or a suitable combination thereof.

[0119] The communication bus (430) interconnects various other components of the computer (400), including the processor (410) and the computer-readable storage medium (420).

[0120] The computer (400) may further include one or more input / output interfaces (440) and one or more network communication interfaces (460) that provide interfaces for one or more input / output devices (450). The input / output interfaces (440) and network communication interfaces (460) are connected to a communication bus (430). The input / output devices (450) may be connected to other components of the computer (400) through the input / output interfaces (440). An exemplary input / output device (450) may include an input device such as a pointing device (such as a mouse or trackpad), a keyboard, a touch input device (such as a touchpad or touchscreen), a voice or sound input device, various types of sensor devices and / or imaging devices, and / or an output device such as a printer, a speaker, and / or a network card. An exemplary input / output device (450) may be included inside the computer (400) as a component constituting the computer (400), or it may be connected to the computer as a separate device distinct from the computer (400).

[0121] The device described above may be implemented as a hardware component, a software component, and / or a combination of a hardware component and a software component. For example, the device and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are also possible.

[0122] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or instruct the processing unit independently or collectively. Software and / or data may be embodied in any type of machine, component, physical device, virtual equipment, computer storage medium, or device so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and may be stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0123] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the embodiment, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

[0124] As described above, according to the present invention, through time domain analysis of an EEG dataset, subjects with Alzheimer's disease (AD), frontotemporal dementia (FTD), and cognitive normal (CN) can be classified using Hjorth parameters, which consist of three main indicators of activity, mobility, and complexity.

[0125] Although the invention has been described above with reference to embodiments, those skilled in the art will understand that various modifications and changes can be made to the invention without departing from the spirit and scope of the invention as set forth in the following claims.

[0126] [Explanation of the symbol]

[0127] 100: Dementia diagnostic assistance system 110: Test probe

[0128] 120 : Connector 130 : Dementia diagnostic assistive device

[0129] 132: EEG Data Preprocessing Unit 134: Hjorth Parameter Extraction Unit

[0130] 136 : Dementia Classification Unit 210 : Classifier 1

[0131] 220 : Classifier 2 230 : Classifier 3

[0132] 400 : Computer 410 : Processor

[0133] 420: Computer-readable storage medium 422: Program

[0134] 430: Communication bus 440: Input / output interface

[0135] 450: Input / Output Device 460: Network Communication Interface

Claims

1. (i) A step of preprocessing the collected electroencephalogram (EEG) dataset of the subject, wherein the collected EEG dataset is divided into time intervals by setting a window length and a shift length; (ii) a step of extracting Hjorth parameters including activity indicators, mobility indicators, and complexity indicators for each time interval; and (iii) A method for assisting in the diagnosis of dementia based on an EEG, characterized by including the step of classifying a subject into one of the following groups based on extracted Hjorth parameters: Alzheimer's disease (AD), frontotemporal dementia (FTD), and cognitive normal (CN).

2. In paragraph 1, the above-mentioned moving length is, In the classification between the dementia patient group including the AD group and the FTD group and the CN group, and the classification between the AD group and the CN group, the first length is, An EEG-based dementia diagnostic aid method characterized by a second length being longer than the first length in the classification between the above FTD group and the above CN group.

3. In paragraph 1, the above step (i) is, (i-1) A step of checking whether to classify a dementia patient (AD&FTD) group including the AD group and the FTD group and a cognitively normal (CN) group; (i-2) If the above-mentioned dementia patient (AD&FTD) group and the above-mentioned cognitively normal (CN) group are checked as being classified, the movement length is set to a first length, and the window length is set to a second length that is longer than the first length; (i-3) If it is not checked that the above dementia patient (AD&FTD) group and the above cognitive normal (CN) group are classified, a step of checking whether the above Alzheimer's disease (AD) group and the above cognitive normal (CN) group are classified; (i-4) A step of feeding back to step (ii) when it is checked that the above Alzheimer's disease (AD) group and the above cognitive normal (CN) group are classified; (i-5) A step of checking whether to classify the Alzheimer's disease (AD) group and the cognitive normal (CN) group if it is not checked to classify the frontotemporal dementia (FTD) group and the cognitive normal (CN) group; (i-6) A step of terminating if the above frontotemporal dementia (FTD) group and the above cognitive normal (CN) group are not checked; and (i-7) An EEG-based dementia diagnosis aid method characterized by including the step of, when checking that the frontotemporal dementia (FTD) group and the cognitive normal (CN) group are classified, setting the movement length to the first length and setting the window length to the second length.

4. An EEG data preprocessing unit that preprocesses the collected electroencephalogram (EEG) dataset of a test subject, wherein the collected EEG dataset is divided into time intervals by setting a window length and a shift length; A Hjorth parameter extraction unit that extracts Hjorth parameters including activity indicators, mobility indicators, and complexity indicators for each time interval; and An EEG-based dementia diagnostic aid characterized by including a dementia classification unit that classifies a test subject into one of three groups—Alzheimer's disease (AD), frontotemporal dementia (FTD), and cognitive normal (CN)—based on extracted Hjorth parameters.

5. An EEG-based dementia diagnosis aid device according to claim 4, wherein the preprocessing unit divides the EEG data set into window length and movement length according to time intervals.

6. A dementia diagnostic aid based on claim 4, wherein the EEG data preprocessing unit removes artifacts from the EEG data set using ASR (Artifact Subspace Reconstruction).

7. An EEG-based dementia diagnostic aid device according to claim 4, wherein the EEG data preprocessing unit separates eye blinking and jaw movement artifacts from the EEG data set through Independent Component Analysis (ICA).

8. An EEG-based dementia diagnosis aid device according to claim 4, wherein the EEG data preprocessing unit removes low-frequency drift and high-frequency noise by passing only a certain frequency band through a Butterworth bandpass filter for the EEG data set.

9. In paragraph 4, the above dementia classification unit, A first classifier for classifying dementia patients (AD & FTD) and cognitively normal (CN) groups; A second classifier for classifying Alzheimer's disease (AD) groups and cognitive normal (CN) groups; and An EEG-based dementia diagnostic aid characterized by including a second classifier that classifies a frontotemporal dementia (FTD) group and a cognitively normal (CN) group.

10. In claim 9, the first classifier and the second classifier include an LDA (Linear Discriminant Analysis) classification model, and An EEG-based dementia diagnostic aid characterized in that the above-mentioned third classifier includes a Support Vector Machine (SVM) classification model.

11. A test probe attached to a pre-set area of ​​the test subject to collect the test subject's electroencephalogram (EEG) data set; and An EEG-based dementia diagnosis assistance system characterized by comprising: an EEG data preprocessing unit that preprocesses an EEG dataset of a collected subject and divides the collected EEG dataset into time intervals by setting a window length and a shift length; a Hjorth parameter extraction unit that extracts Hjorth parameters including activity indicators, mobility indicators, and complexity indicators for each time interval; and a dementia classification unit that classifies the subject into one of an Alzheimer's disease (AD) group, a frontotemporal dementia (FTD) group, and a cognitive normal (CN) group based on the extracted Hjorth parameters.

Images