Method and system for evaluating cognitive function based on sleep eeg phase amplitude coupling
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- TIANJIN UNIV
- Filing Date
- 2026-04-29
- Publication Date
- 2026-08-07
AI Technical Summary
[0004]针对现有技术多关注单一频段特征，缺乏跨频段耦合关系的定量建模，导致无法实现精准、高效实现认知功能评估的不足，本发明提出一种基于睡眠脑电相位波幅耦合的认知功能评估方法及系统，从而解决了现有技术存在的问题
本发明通过首创定量化跨频段耦合指标，揭示深层机制，将慢波振荡视为一个完整的相位周期，并精确定义纺锤波振幅峰值在该周期中的相位角度，这使得原本模糊的“耦合关系”变成了一个可测量、可量化的连续变量，“上升支为负，下降支为正”的符号定义与生理时序一致，使得分析结果具有明确的生理意义；根据相位角前移与认知障碍的特异性关联，建立明确的判别标准，该方法能揭示认知退化的深层神经机制，无创且成本较低，易于被接受，适合大规模筛查和长期动态监测，且突破传统单一频段分析的局限性，实现精准、高效的认知功能评估。
Smart Images
Figure CN122515696A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the interdisciplinary fields of signal and information processing and neurobiology, specifically to a cognitive function assessment method and system based on sleep EEG phase amplitude coupling. Background Technology
[0002] Currently, the assessment of cognitive impairment relies on neuropsychological scales, cerebrospinal fluid biomarkers, or imaging examinations, which have drawbacks such as high subjectivity, high invasiveness, and high cost.
[0003] Existing research indicates that microstructural abnormalities in sleep electroencephalography (sEEG) (such as disruption of the phase coupling between sleep spindles and slow-wave oscillations) are closely related to cognitive decline. However, current techniques for extracting sleep slow waves and sleep spindles primarily rely on frequency characteristics, typically employing Fourier transforms. This method, however, can only extract frequency features and is applicable to linear signal analysis, while EEG signals are complex nonlinear signals. This leads to an inability to accurately extract time-varying features and capture instantaneous phase relationships. Furthermore, traditional EEG analysis methods often focus on single-frequency band features, lacking quantitative modeling of cross-frequency band coupling relationships, making it difficult to reveal the deep mechanisms of cognitive decline and hindering accurate and efficient cognitive function assessment. Summary of the Invention
[0004] To address the shortcomings of existing technologies that focus primarily on single-frequency band characteristics and lack quantitative modeling of cross-frequency band coupling relationships, thus hindering accurate and efficient cognitive function assessment, this invention proposes a cognitive function assessment method and system based on sleep EEG phase amplitude coupling, thereby solving the problems existing in the prior art.
[0005] A cognitive function assessment method based on sleep EEG phase amplitude coupling includes the following steps: Collect the EEG signals of the subjects throughout the night and extract the EEG data of the non-rapid eye movement sleep period from the EEG signals of the subjects throughout the night; Multiscale decomposition was performed on EEG data during non-rapid eye movement sleep to extract slow wave oscillation frequency band and sleep spindle frequency band; Locate the peak B, the start point A of the rising branch, and the end point C of the falling branch in the slow wave oscillation frequency band, and determine the time point x of the maximum positive amplitude in the sleep spindle frequency band; calculate the phase coupling angle of the time point x of the maximum amplitude of the sleep spindle within the slow wave oscillation period. θ The phase coupling angle is determined by the relative position. θ The numerical range and distribution characteristics, where: if x is located in segment AB of the slow wave rising branch, then the phase coupling angle θ It is negative; if x is located in the BC segment of the slow wave descending branch, then the phase coupling angle is negative. θ It is a positive value; According to the phase coupling angle θ The numerical range and distribution characteristics are used to assess the cognitive function status of the subjects.
[0006] Furthermore, Gauss continuous wavelet transform was used to perform multi-scale decomposition of EEG data during non-rapid eye movement sleep, extracting the slow wave oscillation frequency band and the sleep spindle frequency band.
[0007] Furthermore, the location of the wave peak B, the starting point A of the rising branch, and the ending point C of the falling branch in the slow wave oscillation frequency band specifically includes the following steps: The sleep spindle wave was found in the sleep spindle wave frequency band, and the slow wave oscillation waveform coupled with the sleep spindle wave was found in the slow wave oscillation frequency band. The local maximum detection algorithm is used to identify all local peaks in the slow wave oscillation waveform coupled with sleep spindles, and the main peak of the slow wave oscillation that conforms to physiological characteristics is selected and defined as peak B. Starting from point B, backtrack and search for the rising segment from the local minimum to point B in the slow wave oscillation waveform, and take the lowest point as point A. Tracing backward from point B, search for the endpoint of the slow wave oscillation waveform from the local maximum value at point B to the minimum value of the falling segment, and mark it as the endpoint of the falling branch, point C.
[0008] Furthermore, determining the time point x of the maximum positive amplitude in the sleep spindle wave frequency band specifically includes the following steps: The sleep spindle wave signals in the sleep spindle wave frequency band were observed and analyzed to identify the start and end times of each spindle wave event. Within each spindle wave event window, locate the point of maximum positive amplitude of the sleep spindle wave signal, denoted as time point x.
[0009] Furthermore, the phase coupling angle of the time point x with the maximum positive amplitude in the sleep spindle wave frequency band within the slow wave oscillation period. θ Specifically, it is expressed as: If x is located in the rising branch of a slow wave: ; like Located in the lower rising branch of the slow wave: .
[0010] Furthermore, the calculation process for the phase position of the time point (x) of the maximum positive amplitude of the sleep spindle wave within the slow wave oscillation period specifically includes the following steps: Map time point x to the corresponding slow-wave oscillation period; where a complete period is defined by adjacent points A and C. If x is located in the rising branch AB segment of the slow wave, then calculate the time delay of x relative to point B Δt1=x–B, and the total duration of the rising branch T1=B–A; If x is located in the descent branch BC segment of the slow wave, then calculate the time delay of x relative to point B: Δt2 = x – B; total duration of the descent branch: T2 = C – B.
[0011] This invention also includes a cognitive function assessment system based on sleep EEG phase amplitude coupling, comprising: The signal acquisition module is used to collect the subject's EEG signals throughout the night and extract the EEG data from the non-rapid eye movement sleep period. The signal processing module is used to perform multi-scale decomposition of EEG data during non-rapid eye movement sleep and extract the slow wave oscillation frequency band and the sleep spindle frequency band. The phase coupling analysis module is used to locate the peak B, the starting point A of the rising branch, and the ending point C of the falling branch in the slow wave oscillation frequency band, and to determine the time point x of the maximum positive amplitude in the sleep spindle frequency band; it also calculates the phase coupling angle of the time point x of the maximum amplitude of the sleep spindle within the slow wave oscillation period. θ The phase coupling angle is determined by the relative position. θ The numerical range and distribution characteristics, where: if x is located in segment AB of the slow wave rising branch, then the phase coupling angle θ It is negative; if x is located in the BC segment of the slow wave descending branch, then the phase coupling angle is negative. θ It is a positive value; The evaluation module is used to evaluate the phase coupling angle. θ The numerical range and distribution characteristics are used to assess the cognitive function status of the subjects.
[0012] This invention provides a cognitive function assessment method based on sleep EEG phase amplitude coupling, which has the following beneficial effects: This invention reveals deep mechanisms by pioneering a quantitative cross-frequency coupling index. It treats slow-wave oscillations as a complete phase cycle and precisely defines the phase angle of the spindle wave amplitude peak within this cycle. This transforms the previously ambiguous "coupling relationship" into a measurable and quantifiable continuous variable. The symbolic definition of "negative for the rising branch and positive for the falling branch" aligns with physiological timing, giving the analysis results clear physiological significance. Based on the specific correlation between phase angle shift and cognitive impairment, a clear discrimination criterion is established. This method can reveal the deep neural mechanisms of cognitive decline. It is non-invasive, low-cost, easily accepted, suitable for large-scale screening and long-term dynamic monitoring, and breaks through the limitations of traditional single-frequency analysis, achieving accurate and efficient cognitive function assessment. Attached Figure Description
[0013] Figure 1 This is a schematic diagram showing the correspondence between the scale and frequency of electroencephalogram (EEG) signals in the implementation of this invention. Figure 2This is a schematic diagram illustrating the time-frequency analysis of EEG signals at different scales and the cross-scale phase-amplitude coupling relationship in an embodiment of the present invention; Figure 2 (a) is a scale decomposition graph showing the changes of EEG signals at different scales over time in a normal person; Figure 2 (b) is a diagram showing the cross-scale phase-amplitude coupling relationship in normal individuals; Figure 3 This is a diagram showing the average phase coupling angle relationship of slow wave oscillations-sleep spindle EEG signals in NC, aMCI, and AD subjects in this embodiment of the invention. Figure 3 (a) is the graph of the function; Figure 3 (b) is a circular angle diagram with the same meaning as (a) in 3; Figure 4 This is a schematic diagram of the cosine function of the slow-wave oscillation model constructed in this embodiment of the invention; Figure 5 This is a flowchart of a cognitive function assessment method based on the phase-amplitude coupling relationship of sleep EEG signals in an embodiment of the present invention. Detailed Implementation
[0014] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.
[0015] This invention proposes a cognitive function assessment method based on the phase-amplitude coupling relationship of sleep EEG signals, such as... Figure 5 As shown, the specific steps include: S1. Collect the subject's EEG signals throughout the night using a multi-lead EEG device and extract the EEG data during non-rapid eye movement sleep (NREM).
[0016] S2. The Gaussian continuous wavelet transform (CWT) is used to perform multi-scale decomposition of EEG data during non-REM sleep. This converts one-dimensional non-stationary EEG signals into two-dimensional time-frequency signals, progressively refining the signal across multiple scales to achieve precise synchronization analysis in the time and frequency domains. The calculation process is expressed as follows:
[0017] (1) The wavelet mother function is translated β Units, scaling α After a unit transformation, the family of wavelet functions is obtained. ,in t Indicates time, α Indicates the scaling factor. β For corresponding time variablest The translation parameters and the translation scaling process are shown in formula (2). The wavelet function formula is expressed as:
[0018] (2) like Figure 1 The diagram shows the correspondence between scale and frequency: slow wave oscillation band (0.5-1.25 Hz, corresponding to scale 15) and sleep spindle wave band (12-16 Hz, corresponding to scale 8).
[0019] S3. Locate the peak (point B), the starting point of the rising branch (point A), and the ending point of the falling branch (point C) of the slow wave oscillation, and calculate the phase position of the time point (x) of the maximum positive amplitude of the sleep spindle wave within the slow wave oscillation period.
[0020] (1) Locating the characteristic points of slow wave oscillation (points A, B, and C) specifically includes the following steps: ① Peak (Point B) detection: First, search for sleep spindle waves on the 8th scale EEG signal band extracted in step S2. Since a slow wave oscillation is always coupled on the sleep spindle wave, the slow wave oscillation waveform coupled with the sleep spindle wave on the 15th scale can be found.
[0021] A local maximum detection algorithm was used to identify all local peaks in the slow-wave oscillation signal coupled with sleep spindles, and the main peak of the slow-wave oscillation that conforms to physiological characteristics was selected and defined as point B.
[0022] ② Determining the starting point of the ascending branch (point A) and the ending point of the descending branch (point C): Rising support point (point A): Tracing back from point B, search for the rising segment of the signal from the local minimum to point B, and take the lowest point as point A.
[0023] Descent endpoint (point C): Tracing backward from point B, the endpoint of the search signal from the local maximum value at point B to the minimum value of the descent segment is marked as the descent endpoint point C.
[0024] (2) Location and phase mapping of the time point (x) of the maximum positive amplitude of the sleep spindle wave: ① Maximum positive amplitude detection: The sleep spindle wave signals (12-16 Hz) extracted in step S2 are observed and analyzed to identify the start and end times of each spindle wave.
[0025] Within each spindle wave window, the point of maximum positive amplitude of the positioning signal is denoted as time point x.
[0026] ② Phase position calculation: Map time point x to the corresponding slow wave oscillation period (with adjacent points A and C as a complete period).
[0027] If x is located in the rising branch of the slow wave (segment AB): calculate the time of x relative to point B: Δt = x – B; total duration of the rising branch: T = B – A.
[0028] If x is located in the descent branch of the slow wave (segment BC): calculate the time of x relative to point B: Δt = x – B; total duration of the descent branch: T = C – B.
[0029] S4. Symbol definition: The slow wave oscillation waveform is regarded as a function curve. The phase angle of the rising branch (AB segment) is negative (-180° to 0°) and the falling branch (BC segment) is positive (0° to +180°), which is consistent with the physiological timing characteristics of slow wave oscillation.
[0030] Calculate the phase coupling angle using the formula. θ : If x is in the ascending branch: ; If x is located in the lower ascending branch: .
[0031] S5. Establish an evaluation model, with the following judgment criteria: Normal control (NC): θ For positive values (0° to +180°), the coupling angle is concentrated in the slow wave descending branch.
[0032] aMCI stage: θ When the value is negative (-180° to 0°), the coupling angle shifts forward to the rising branch of the slow wave.
[0033] AD stage: θ Significant negative offset (-180° to 0°), coupling angle is relatively NC , aMCI The stage shifts significantly forward, and the coupling dispersion increases.
[0034] S6. Output a cognitive function assessment report, including a phase coupling angle distribution map, damage severity grading, and intervention recommendations.
[0035] This invention analyzes the dynamic changes in the phase-amplitude coupling angle between slow-wave oscillations and sleep spindles: it is the first to propose using the forward shift (negation) of the slow-wave oscillation-sleep spindle phase angle as a specific marker of aMCI, breaking through the limitations of traditional single-band analysis; it focuses on the signal of the left parietal lobe (lead P3), improving spatial sensitivity and clinical operability; it provides non-invasive real-time monitoring: suitable for hospital, home or community scenarios, supporting long-term dynamic tracking of cognitive function.
[0036] Figure 2A schematic diagram illustrating the time-frequency analysis of EEG signals at different scales and the cross-scale phase-amplitude coupling relationship; Figure 2 (a) is a scale decomposition graph of the changes of EEG signals at different scales over time in normal individuals. The EEG signals were decomposed at different scales using Gauss continuous wavelet transform. It was observed that small-scale signals are often enveloped by large-scale signals, that is, there is a special coupling relationship between oscillating signals at different scales, which provides a basis for further analysis. Figure 2 (b) is a cross-scale phase amplitude coupling relationship diagram of normal people. The waveforms of the 12-16 Hz (corresponding to the 8th scale) frequency band and the slow wave frequency band (corresponding to the 14th scale and above) of the EEG signal were extracted. It was found that the EEG signal of the 15th scale was most closely coupled with the EEG signal of the 8th scale.
[0037] Figure 3 Plotting the positional relationship of the average phase coupling angle between slow-wave oscillations and sleep spindle EEG signals in subjects with NC, aMCI, and AD. (Calculate the slow-wave oscillation-sleep spindle coupling angle between different cognitive levels.) Figure 3 (a) is a function graph showing that the maximum positive phase amplitude of the sleep spindle in the NC group is located on the descending branch after the peak of the slow wave oscillation, while the maximum positive phase amplitude of the sleep spindle in the aMCI and AD groups is located on the ascending branch before the peak of the slow wave oscillation, and the aMCI group is located between the NC and AD groups. Figure 3 (b) is a circular angle diagram with the same meaning as 3(a), which more intuitively shows the angular relationship between the three.
[0038] Figure 4 This is a schematic diagram of the cosine function for the constructed slow-wave oscillation model. The slow-wave oscillation is considered as a standard cosine function waveform. The troughs A and C represent the lowest points of the slow-wave oscillation amplitude, and the crest B represents the highest point. The curve between A and B represents the rising branch of the slow wave, and the curve between B and C represents the falling branch. The total phase from A to C is 360°. The phase of crest B is defined as 0°, the phase from B to A is from 0° to -180°, and the phase from B to C is from 0° to +180°.
[0039] Example 1: The subjects' EEG signals throughout the night were collected via lead P3, and signals at scale 15 (slow wave oscillation) and scale 8 (sleep spindle) were extracted.
[0040] Analysis shows that the maximum amplitude time point of the sleep spindle wave is located in the rising branch of the slow wave. The phase coupling angle of the aMCI group was calculated. θ =-18.4°.
[0041] Output an assessment report to evaluate the risk of cognitive impairment.
[0042] Example 2: In the AD group, the phase coupling angle θ in lead P3 was -58.7°, and the dispersion of coupling events (standard deviation > 25°) was significantly higher than the normal threshold.
[0043] Output an assessment report to evaluate the risk of cognitive impairment.
[0044] Based on the same inventive concept, this invention also proposes a cognitive function assessment system based on sleep EEG phase amplitude coupling, comprising: The signal acquisition module is used to collect the subject's EEG signals throughout the night and extract the EEG data from the non-rapid eye movement (NREM) sleep period.
[0045] The signal processing module is used to perform multi-scale decomposition of EEG data during non-rapid eye movement sleep to extract the slow wave oscillation frequency band and the sleep spindle frequency band.
[0046] The phase coupling analysis module is used to locate the peak B, the starting point A of the rising branch, and the ending point C of the falling branch in the slow wave oscillation frequency band, and to determine the time point x of the maximum positive amplitude in the sleep spindle frequency band; it also calculates the phase coupling angle of the time point x of the maximum amplitude of the sleep spindle within the slow wave oscillation period. θ The phase coupling angle is determined by the relative position. θ The numerical range and distribution characteristics, where: if x is located in segment AB of the slow wave rising branch, then the phase coupling angle θ It is negative; if x is located in the BC segment of the slow wave descending branch, then the phase coupling angle is negative. θ It is a positive value.
[0047] The evaluation module is used to evaluate the phase coupling angle. θ The numerical range and distribution characteristics are used to assess the cognitive function status of the subjects.
[0048] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.
Claims
1. A cognitive function assessment method based on sleep EEG phase amplitude coupling, characterized in that, Includes the following steps: Collect the EEG signals of the subjects throughout the night and extract the EEG data of the non-rapid eye movement sleep period from the EEG signals of the subjects throughout the night; Multiscale decomposition was performed on EEG data during non-rapid eye movement sleep to extract slow wave oscillation frequency band and sleep spindle frequency band; Locate the peak B, the starting point A of the rising branch, and the ending point C of the falling branch in the slow wave oscillation frequency band, and determine the time point x of the maximum positive amplitude in the sleep spindle wave frequency band; Calculate the phase coupling angle of the time point x of the maximum amplitude of the sleep spindle wave within the slow wave oscillation period. θ The phase coupling angle is determined by the relative position. θ The numerical range and distribution characteristics, where: if x is located in segment AB of the slow wave rising branch, then the phase coupling angle θ It is negative; if x is located in the BC segment of the slow wave descending branch, then the phase coupling angle is negative. θ It is a positive value; According to the phase coupling angle θ The numerical range and distribution characteristics are used to assess the cognitive function status of the subjects.
2. The cognitive function assessment method based on sleep EEG phase amplitude coupling according to claim 1, characterized in that, Gauss continuous wavelet transform was used to perform multi-scale decomposition on EEG data during non-rapid eye movement sleep, extracting the slow wave oscillation frequency band and the sleep spindle frequency band.
3. The cognitive function assessment method based on sleep EEG phase amplitude coupling according to claim 1, characterized in that, The specific steps for locating the peak B, the starting point A of the rising branch, and the ending point C of the falling branch in the slow wave oscillation frequency band are as follows: The sleep spindle wave was found in the sleep spindle wave frequency band, and the slow wave oscillation waveform coupled with the sleep spindle wave was found in the slow wave oscillation frequency band. The local maximum detection algorithm is used to identify all local peaks in the slow wave oscillation waveform coupled with sleep spindles, and the main peak of the slow wave oscillation that conforms to physiological characteristics is selected and defined as peak B. Starting from point B, backtrack and search for the rising segment from the local minimum to point B in the slow wave oscillation waveform, and take the lowest point as point A. Tracing backward from point B, search for the endpoint of the slow wave oscillation waveform from the local maximum value at point B to the minimum value of the falling segment, and mark it as the endpoint of the falling branch, point C.
4. The cognitive function assessment method based on sleep EEG phase amplitude coupling according to claim 1, characterized in that, The determination of the time point x of the maximum positive amplitude in the sleep spindle wave frequency band specifically includes the following steps: The sleep spindle wave signals in the sleep spindle wave frequency band were observed and analyzed to identify the start and end times of each spindle wave event. Within each spindle wave event window, locate the point of maximum positive amplitude of the sleep spindle wave signal, denoted as time point x.
5. The cognitive function assessment method based on sleep EEG phase amplitude coupling according to claim 1, characterized in that, The phase coupling angle of the time point x with the maximum positive amplitude in the sleep spindle wave band within the slow wave oscillation period. θ Specifically, it is expressed as: If x is located in the rising branch of a slow wave: ； like Located in the lower rising branch of the slow wave: 。 6. The cognitive function assessment method based on sleep EEG phase amplitude coupling according to claim 1, characterized in that, The calculation process for the phase position of the time point (x) of the maximum positive amplitude of the sleep spindle wave within the slow wave oscillation period includes the following steps: Map time point x to the corresponding slow-wave oscillation period; where a complete period is defined by adjacent points A and C. If x is located in the rising branch AB segment of the slow wave, then calculate the time delay of x relative to point B Δt1=x–B, and the total duration of the rising branch T1=B–A; If x is located in the descent branch BC segment of the slow wave, then calculate the time delay of x relative to point B: Δt2 = x – B; total duration of the descent branch: T2 = C – B.
7. A cognitive function assessment system based on sleep EEG phase amplitude coupling, characterized in that, include: The signal acquisition module is used to collect the subject's EEG signals throughout the night and extract the EEG data from the non-rapid eye movement sleep period. The signal processing module is used to perform multi-scale decomposition of EEG data during non-rapid eye movement sleep and extract the slow wave oscillation frequency band and the sleep spindle frequency band. The phase coupling analysis module is used to locate the peak B, the starting point A of the rising branch, and the ending point C of the falling branch in the slow wave oscillation frequency band, and to determine the time point x of the maximum positive amplitude in the sleep spindle frequency band. Calculate the phase coupling angle of the time point x of the maximum amplitude of the sleep spindle wave within the slow wave oscillation period. θ The phase coupling angle is determined by the relative position. θ The numerical range and distribution characteristics, where: if x is located in segment AB of the slow wave rising branch, then the phase coupling angle θ It is negative; if x is located in the BC segment of the slow wave descending branch, then the phase coupling angle is negative. θ It is a positive value; The evaluation module is used to evaluate the phase coupling angle. θ The numerical range and distribution characteristics are used to assess the cognitive function status of the subjects.