WO2017193595A1 - 催眠状态脑电信号提取方法与系统 - Google Patents

催眠状态脑电信号提取方法与系统 Download PDF

Info

Publication number
WO2017193595A1
WO2017193595A1 PCT/CN2016/113118 CN2016113118W WO2017193595A1 WO 2017193595 A1 WO2017193595 A1 WO 2017193595A1 CN 2016113118 W CN2016113118 W CN 2016113118W WO 2017193595 A1 WO2017193595 A1 WO 2017193595A1
Authority
WO
WIPO (PCT)
Prior art keywords
wavelet
hypnotic state
eeg signal
wavelet function
eeg
Prior art date
Application number
PCT/CN2016/113118
Other languages
English (en)
French (fr)
Inventor
胡静
Original Assignee
广州视源电子科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 广州视源电子科技股份有限公司 filed Critical 广州视源电子科技股份有限公司
Publication of WO2017193595A1 publication Critical patent/WO2017193595A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/369—Electroencephalography [EEG]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/48—Other medical applications
    • A61B5/4806—Sleep evaluation

Definitions

  • the invention relates to the field of signal extraction technology, in particular to a hypnotic state EEG signal extraction method and system.
  • Hypnosis has become an extremely important means of clinical treatment of insomnia and sleep disorders, alone or in combination with other psychological treatments, can alleviate many physical and mental symptoms. Many studies have found that when the individual is under hypnosis, the Theta wave increases, the amplitude increases, and the high hypnotic susceptibility individual is more obvious than the low susceptibility individual.
  • a hypnotic state EEG signal extraction method comprising the steps of:
  • the initial signal of the hypnotic state EEG is collected, and the initial frequency-doubling notch processing is performed on the initial signal of the hypnotic state EEG, and the pre-processed EEG signal is obtained;
  • a wavelet function for extracting the hypnotic state EEG signal is selected;
  • the hypnotic EEG signal is extracted from the preprocessed EEG signal by the optimized wavelet function.
  • a hypnotic state EEG signal extraction system comprising:
  • a pre-processing module for collecting an initial signal of the hypnotic state EEG and presetting the initial signal of the hypnotic state EEG Double frequency notch processing to obtain pre-processed EEG signals;
  • a wavelet function selection module is configured to select a wavelet function for extracting a hypnotic state EEG signal according to a correlation between the wavelet function and the preprocessed EEG signal;
  • An optimization module for adaptively optimizing a center frequency-bandwidth ratio of the selected wavelet function to obtain an optimized wavelet function
  • the signal extraction module is configured to extract a hypnotic state EEG signal from the preprocessed EEG signal through the optimized wavelet function.
  • the method and system for extracting the hypnotic state EEG signal of the present invention collects the initial signal of the hypnotic state EEG, and performs preset frequency doubling notch processing, and selects the hypnotic state according to the correlation between the wavelet function and the post-brain electrical signal.
  • the wavelet function of the EEG signal adaptively optimizes the center-frequency-bandwidth ratio of the selected wavelet function, and extracts the hypnotic EEG signal from the pre-processed EEG signal through the optimized wavelet function.
  • the center frequency-bandwidth ratio of the wavelet function is adaptively optimized, the wavelet coefficient is adaptively adjusted, the resolution of the wavelet transform is improved, and the hypnotic state EEG signal can be extracted efficiently and accurately.
  • FIG. 1 is a schematic flow chart of a first embodiment of a method for extracting a hypnotic state EEG signal according to the present invention
  • FIG. 2 is a schematic flow chart of a second embodiment of a method for extracting a hypnotic state EEG signal according to the present invention
  • FIG. 3 is a schematic diagram showing a relationship between a frequency-bandwidth ratio and a wavelet entropy of a Morlet wavelet center
  • FIG. 4 is a schematic structural view of a first embodiment of a hypnotic state EEG signal extraction system according to the present invention.
  • FIG. 5 is a schematic structural view of a second embodiment of a hypnotic state EEG signal extraction system according to the present invention.
  • a hypnotic state EEG signal extraction method includes the following steps:
  • S100 collecting the initial signal of the hypnotic state EEG, and performing preset octave notch processing on the initial signal of the hypnotic state EEG, and obtaining the pre-processed EEG signal.
  • the hypnotic initial EEG signal can be acquired by existing instruments, for example, by an EEG machine.
  • the hypnotic state EEG signals collected by the EEG usually contain power frequency noise interference of a certain frequency, and there is no obvious fluctuation of the EEG signal.
  • the initial signal of the hypnotic state EEG has not only power frequency noise, but also a certain frequency. Harmonic Noise, by pre-set octave notch processing on the initial signals of these hypnotic EEGs, the pre-processed EEG signals with obvious EEG waveforms are obtained.
  • S200 Select a wavelet function for extracting a hypnotic state EEG signal according to a correlation between the wavelet function and the preprocessed EEG signal.
  • wavelet functions such as Haar wavelet, Coif wavelet, Meyer wavelet, Mexican wavelet, and Morlet wavelet.
  • Haar wavelet Coif wavelet
  • Meyer wavelet Meyer wavelet
  • Mexican wavelet Mexican wavelet
  • Morlet wavelet The correlation between different wavelet functions and pre-processed EEG signals is different.
  • Non-essential we can calculate the correlation coefficient between different wavelet functions and pre-processed EEG signals, and select the wavelet function with the largest correlation coefficient as the wavelet function for extracting the hypnotic EEG signals.
  • S300 adaptively optimizes the center-to-bandwidth ratio of the selected wavelet function to obtain an optimized wavelet function.
  • the center frequency and bandwidth are the key factors affecting the time-frequency resolution of the wavelet function. Changing the center frequency-bandwidth ratio changes the time-frequency resolution of the wavelet transform. When the center frequency-bandwidth ratio is optimal, the time-frequency resolution of the wavelet transform is the highest, which can more accurately extract the hypnotic brain telecom.
  • the center-frequency-bandwidth ratio of the wavelet function is optimized adaptively.
  • the adaptive optimization method can select relative wavelet entropy adaptive optimization, wavelet singular entropy adaptive optimization and wavelet entropy adaptive optimization.
  • the time-frequency resolution of the wavelet transform is significantly improved, and the high-sensitivity component in the hypnotic EEG signal can be extracted more effectively, and the pre-processed EEG is obtained through the optimized wavelet function.
  • the hypnotic EEG signal is accurately extracted from the signal.
  • the extracted high-sensitivity hypnotic EEG signal can be displayed on the one hand on the EEG monitor, the electronic sleep instrument, the biofeedback sleep device including the hypnotic EEG module, as a basis for diagnosis; Hypnosis analysis and sleep information matching provide an accurate and reliable data foundation.
  • the method for extracting the electroencephalogram signal of the hypnotic state collecting the initial signal of the hypnotic state EEG, and performing the preset frequency doubling notch processing, and selecting the hypnotic state EEG according to the correlation between the wavelet function and the post-brain electrical signal
  • the wavelet function of the signal adaptively optimizes the center-frequency-bandwidth ratio of the selected wavelet function, and extracts the hypnotic EEG signal from the pre-processed EEG signal through the optimized wavelet function.
  • the center frequency-bandwidth ratio of the wavelet function is adaptively optimized, the wavelet coefficient is adaptively adjusted, the resolution of the wavelet transform is improved, and the hypnotic state EEG signal can be extracted efficiently and accurately.
  • step S100 includes:
  • S120 collecting an initial signal of hypnotic state EEG through an electroencephalograph, an EEG biofeedback sleeper, or an EEG electrode.
  • EEG machine Through the accurate scientific equipment such as EEG machine, EEG biofeedback sleeper or EEG electrode, the initial signal of hypnotic state EEG is accurately collected.
  • S140 Perform a 50 Hz octave notch on the initial signal of the hypnotic state EEG to obtain a pre-processed EEG signal.
  • the hypnotic state EEG initial signal collected in step S120 contains 50 Hz power frequency noise interference and 50 Hz harmonic noise, such as 100 Hz harmonic noise, 150 Hz harmonic noise, and 200 Hz harmonic noise. Since there is no obvious EEG signal fluctuation law and there is a large noise in the hypnotic state EEG initial signal, it needs to be subjected to 50Hz octave notch processing. After the frequency doubling notch, the curve shows obvious EEG fluctuation. shape. It is not necessary to pre-process the signal by using a 50Hz octave trap (50Hz notch, 100Hz notch, 150Hz notch, 200Hz notch, and 250Hz notch).
  • 50Hz octave trap 50Hz notch, 100Hz notch, 150Hz notch, 200Hz notch, and 250Hz notch.
  • step S200 includes:
  • S220 Calculate correlation coefficients between different types of wavelet functions and pre-processed EEG signals, respectively.
  • the wavelet transform is based on the wavelet basis function, and a series of sub-wavelets are obtained by transforming to approximate the EEG signal under hypnosis.
  • the wavelet basis function is extremely important, which is related to the accuracy of wavelet approximation.
  • the basis function ⁇ (t) ⁇ L 2 (R) must be satisfied. Requirements.
  • ⁇ (t) is a wavelet basis function, and a series of sub-wavelets are obtained by time-scale transformation, such as
  • the correlation coefficient between the wavelet signal and the EEG signal under hypnosis is selected as the evaluation index to select the appropriate mother wavelet function.
  • the calculation of the correlation coefficient is as follows:
  • ⁇ xy is their correlation coefficient
  • ⁇ 1, ⁇ xy reflects the degree of similarity between x(t) and y(t).
  • the mother wavelets most commonly used for hypnotic EEG signal extraction include Haar wavelet function, Coif wavelet function, Meyer wavelet function, Mexican Hat wavelet function and Morlet wavelet function. The correlation between hypnotic EEG signals and wavelet signals The calculation results are shown in Table 1:
  • S240 Select a wavelet function corresponding to the maximum correlation coefficient as a wavelet function for extracting a hypnotic state EEG signal.
  • the wavelet function corresponding to the largest correlation coefficient is selected as the wavelet function for extracting the hypnotic EEG signal.
  • the waveform of the Morlet wavelet is in the form of oscillation attenuation, which is closest to the hypnotic state EEG signal, and the correlation coefficient between the two is the largest among all the wavelet waveforms. Therefore, Morlet wavelet is selected as a wavelet analysis tool for hypnotic EEG signal extraction.
  • step S300 includes:
  • S320 Acquire wavelet coefficients of the selected wavelet function, obtain a probability distribution sequence p i according to the wavelet coefficient conversion, and calculate a value of the probability distribution sequence p i .
  • S340 Draw a relationship between a center frequency-band ratio of the selected wavelet function and wavelet entropy according to the value of the probability distribution sequence p i .
  • S380 adaptively optimize the center frequency-bandwidth ratio of the selected wavelet function according to the optimal value to obtain an optimized wavelet function.
  • the wavelet entropy is selected to optimize the center-frequency-bandwidth ratio of the wavelet function, and the center frequency f c and the bandwidth ⁇ f are key factors affecting the wavelet time-frequency resolution. Changing the center frequency-bandwidth ratio changes the time-frequency resolution of the wavelet transform. When the center frequency-bandwidth ratio is optimal, the time-frequency resolution of the wavelet transform is the highest.
  • the CMOR wavelet of the Morlet wavelet is explained by the mother wavelet function.
  • the mother wavelet expression of the CMOR wavelet is as follows:
  • f c represents the characteristic frequency of the mother wave ⁇ (t) and is also the center frequency
  • ⁇ t is the standard deviation of the Gaussian window
  • ⁇ f is the bandwidth
  • ⁇ f 1/2 ⁇ t .
  • Analysis of the mother wavelet of the CMOR wavelet shows that the speed of the CMOR wavelet waveform oscillation is determined by the bandwidth ⁇ f , and the oscillation frequency of the waveform is determined by the center frequency f c . According to the above formula, the frequency resolution of CMOR wavelet can be calculated.
  • f s is the sampling frequency
  • f c is the center frequency
  • ⁇ f is the bandwidth
  • f i is the signal analysis frequency.
  • p i is a probability distribution sequence
  • the wavelet coefficients obtained by the conversion is uncertain.
  • the conversion formula is as follows:
  • X(f i ,t) is the wavelet coefficient.
  • the relationship between the center frequency-bandwidth ratio f c / ⁇ f and the Shannon wavelet entropy is shown in Figure 3.
  • the center frequency-bandwidth ratio f c / ⁇ f 4.43
  • the Shannon wavelet entropy probability optimal theory it can be seen that when the Shannon wavelet entropy reaches the minimum value, the Morlet wavelet center frequency-bandwidth ratio parameter is optimal, and the corresponding base wavelet is obtained. It is the wavelet that best matches the feature component.
  • the step of extracting the hypnotic EEG signal from the pre-processed EEG signal by the optimized wavelet function comprises:
  • Step 1 Obtain the mother wavelet of the optimized wavelet function, and stretch and translate the mother wavelet to obtain the wavelet.
  • the wavelet is generated by the mother wave expansion and translation, and the expression is as follows:
  • a is the frequency scaling factor
  • b is the time shifting factor
  • * indicates that the function is a complex function.
  • Step 2 Scale-convert the parameters in the wavelet, and obtain the weight coefficients of the sub-wavelets.
  • the scale conversion includes converting the scaling factor into frequency and converting the translation factor into time.
  • the scale conversion is performed on the parameters in the wavelet w(a, b), the frequency expansion factor a is represented by the frequency f, and the translation factor b is represented by the time t, and the weighting coefficient X(f, t) is obtained.
  • Step 3 Construct a three-dimensional plane of the weighting coefficient wavelet curve, wherein the space X axis is the time axis, the space Y axis is the frequency axis, and the space Z axis is the weighting coefficient axis.
  • the x-axis is the time axis (t)
  • the y-axis is the frequency axis (f)
  • the z-axis is the wavelet coefficient (in dB).
  • Step 4 Obtain the optimized wavelet function period, and find the maximum frequency in the three-dimensional plane of the weighting coefficient wavelet curve, and calculate the continuous threshold value.
  • Step 5 Find the time interval value corresponding to the weighting coefficient in the three-dimensional plane of the weighting coefficient wavelet curve that is greater than the frequency maximum portion.
  • the frequency f' represents the frequency value of the highest point of the three-dimensional plane of the wavelet curve.
  • the energy threshold is indicated by K. If X(f', t) > K(f'), the parameter ⁇ can adjust the threshold size. Let [t',t"] denote the portion of the curve flat line X(f',t)>K(f').
  • Step 6 When the time interval value is greater than the continuous threshold value, determine that the EEG signal corresponding to the portion of the three-dimensional plane of the weighting coefficient wavelet curve that is greater than the frequency maximum is an effective hypnotic state EEG signal.
  • Step 7 Through the optimized wavelet function, the time-frequency information of the effective hypnotic EEG signal is extracted, and the hypnotic state EEG signal is obtained.
  • the wavelet-entropy adaptively optimized CMOR wavelet is used to extract the time-frequency information of the EEG signal under hypnosis.
  • Shannon wavelet entropy adaptively optimizes the center-to-bandwidth ratio parameter of the CMOR wavelet, the time-frequency resolution of the wavelet transform is obvious. The improvement can more effectively extract high-sensitivity components in hypnotic EEG signals.
  • a hypnotic state EEG signal extraction system includes:
  • the pre-processing module 100 is configured to collect an initial signal of the electroencephalogram of the hypnotic state, and perform a preset frequency doubling notch processing on the initial signal of the hypnotic state EEG to obtain a pre-processed electroencephalogram signal;
  • the wavelet function selection module 200 is configured to select a wavelet function for extracting a hypnotic state EEG signal according to a correlation between the wavelet function and the preprocessed EEG signal;
  • the optimization module 300 is configured to adaptively optimize the center frequency-bandwidth ratio of the selected wavelet function to obtain an optimized wavelet function
  • the signal extraction module 400 is configured to extract a hypnotic state EEG signal from the preprocessed EEG signal through the optimized wavelet function.
  • the hypnotic state EEG signal extraction system of the present invention collects the initial signal of the hypnotic state EEG, and performs preset frequency doubling notch processing, and selects the hypnotic state EEG according to the correlation between the wavelet function and the post-brain electrical signal.
  • the wavelet function of the signal adaptively optimizes the center-frequency-bandwidth ratio of the selected wavelet function, and extracts the hypnotic EEG signal from the pre-processed EEG signal through the optimized wavelet function.
  • the center frequency-bandwidth ratio of the wavelet function is adaptively optimized, the wavelet coefficient is adaptively adjusted, the resolution of the wavelet transform is improved, and the hypnotic state EEG signal can be extracted efficiently and accurately.
  • the pre-processing module 100 includes:
  • the initial signal acquisition unit 120 is configured to collect an initial signal of the hypnotic state EEG through an electroencephalograph, an EEG biofeedback sleeper, or an EEG electrode.
  • the pre-processing unit 140 is configured to perform a 50 Hz octave notch processing on the hypnotic state EEG initial signal to obtain a pre-processed EEG signal.
  • the hypnotic state EEG initial signal collected by the instrument and equipment contains 50Hz power frequency noise interference and 50Hz harmonic noise, such as 100Hz harmonic noise, 150Hz harmonic noise and 200Hz harmonic noise. Since there is no obvious EEG signal fluctuation law and there is a large noise in the hypnotic state EEG initial signal, it needs to be subjected to 50Hz octave notch processing. After the frequency doubling notch, the curve shows obvious EEG fluctuation. shape.
  • the pre-processing unit 140 may use a 50 Hz multiplier trap (50 Hz notch, 100 Hz notch, 150 Hz notch, 200 Hz notch, 250 Hz notch, etc.) to preprocess the signal.
  • the wavelet function selection module 200 includes:
  • the correlation coefficient calculation unit 220 is configured to separately calculate correlation coefficients of different types of wavelet functions and pre-processed EEG signals.
  • the selecting unit 240 is configured to select a wavelet function corresponding to the maximum correlation coefficient as a wavelet function for extracting a hypnotic state EEG signal.
  • the wavelet function corresponding to the largest correlation coefficient is selected as the wavelet function for extracting the hypnotic EEG signal.
  • the optimization module 300 includes:
  • the calculating unit 320 is configured to obtain wavelet coefficients of the selected wavelet function, obtain a probability distribution sequence p i according to the wavelet coefficient conversion, and calculate a value of the probability distribution sequence p i .
  • the curve drawing unit 340 is configured to draw a relationship between the center frequency-band ratio of the selected wavelet function and the wavelet entropy according to the value of the probability distribution sequence p i .
  • the searching unit 360 is configured to search for an optimal value of the selection center frequency-bandwidth ratio according to the relationship curve.
  • the optimizing unit 380 is configured to adaptively optimize the center frequency-bandwidth ratio of the selected wavelet function according to the optimal value to obtain an optimized wavelet function.
  • the wavelet entropy is selected to optimize the center-frequency-bandwidth ratio of the wavelet function, and the center frequency f c and the bandwidth ⁇ f are key factors affecting the wavelet time-frequency resolution. Changing the center frequency-bandwidth ratio changes the time-frequency resolution of the wavelet transform. When the center frequency-bandwidth ratio is optimal, the time-frequency resolution of the wavelet transform is the highest.
  • the signal extraction module 400 includes:
  • the subwavelet acquisition unit is configured to obtain the mother wavelet of the optimized wavelet function, and expand and translate the mother wavelet to obtain the subwavelet.
  • the conversion unit is configured to perform scale conversion on the parameters in the sub-wavelets to obtain weighting coefficients of the sub-wavelets, wherein the scale conversion includes converting the scaling factor into a frequency and converting the translation factor into time.
  • the building unit is configured to construct a three-dimensional plane of the weighting coefficient wavelet curve, wherein the space X axis is a time axis, the space Y axis is a frequency axis, and the space Z axis is a weighting coefficient axis.
  • the threshold limit calculation unit is configured to obtain the optimized wavelet function period, and find the frequency maximum value in the three-dimensional plane of the weight coefficient wavelet curve, and calculate the continuous threshold value.
  • the time interval value calculation unit is configured to search for a time interval value corresponding to the weighting coefficient in the three-dimensional plane of the wavelet curve of the weighting coefficient greater than the frequency maximum portion.
  • the comparing unit is configured to determine, when the time interval value is greater than the continuous threshold value, the EEG signal corresponding to the portion of the three-dimensional plane of the weighting coefficient wavelet curve that is greater than the frequency maximum value as the effective hypnotic state EEG signal.
  • the extracting unit is configured to extract time-frequency information of the effective hypnotic state EEG signal through the optimized wavelet function, and obtain a hypnotic state EEG signal.
  • CMOR wavelet optimized by wavelet entropy is used to optimize the EEG signal under hypnosis.
  • Time-frequency information extraction using Shannon wavelet entropy adaptive optimization of the center-to-bandwidth ratio parameter of CMOR wavelet, the time-frequency resolution of wavelet transform is significantly improved, and the high-sensitivity component in hypnotic EEG signal can be extracted more effectively. .

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Informatics (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Physics & Mathematics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Psychiatry (AREA)
  • Psychology (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

一种催眠状态脑电信号提取方法与系统，采集催眠状态脑电初始信号，并进行预设倍频陷波处理（S100），根据小波函数与理脑后电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数（S200），自适应优化选取的所述小波函数的中心频率-带宽比（S300），通过所述优化后的小波函数，从所述预处理脑电信号中提取催眠状态脑电信号（S400）。整个过程中，自适应优化小波函数的中心频率-带宽比，实现小波系数自适应调节，提高小波变换的分辨率，能有效且准确提取催眠状态脑电信号。

Description

催眠状态脑电信号提取方法与系统 技术领域
本发明涉及信号提取技术领域，特别是涉及催眠状态脑电信号提取方法与系统。
背景技术
睡眠是人类非常重要且不可或缺的生理需要。近些年来随着人们生活节奏的加快，工作压力的增加，运动量的缺乏或其他原因引起的烦躁、身心不安，导致失眠患者越来越多，已严重影响到人们的身心健康，使工作效率与生活质量下降。
催眠已成为临床治疗失眠和睡眠障碍的一种极为重要的手段，单独或联合其它的心理疗法，可以缓解诸多身心症状。许多研究发现，个体在催眠状态下脑电Theta波增加，振幅增加，高催眠感受性个体比低感受性个体更明显。
为了实现催眠状态下高催眠感受性脑电的自动检测和正确地进行催眠感受性的分类判别，必须从催眠状态下脑电信号中提取最有判别力的信息。因此，寻求有效的催眠状态脑电信号提取方法迫在眉睫。
发明内容
基于此，有必要针对目前不能有效的提取催眠状态脑电信号的问题，提供一种有效且准确的催眠状态脑电信号提取方法与系统。
一种催眠状态脑电信号提取方法，包括步骤：
采集催眠状态脑电初始信号，并对催眠状态脑电初始信号进行预设倍频陷波处理，获得预处理脑电信号；
根据小波函数与预处理脑电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数；
自适应优化选取的小波函数的中心频率-带宽比，获得优化后的小波函数；
通过优化后的小波函数，从预处理脑电信号中提取催眠状态脑电信号。
一种催眠状态脑电信号提取系统，包括：
预处理模块，用于采集催眠状态脑电初始信号，并对催眠状态脑电初始信号进行预设 倍频陷波处理，获得预处理脑电信号；
小波函数选择模块，用于根据小波函数与预处理脑电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数；
优化模块，用于自适应优化选取的小波函数的中心频率-带宽比，获得优化后的小波函数；
信号提取模块，用于通过优化后的小波函数，从预处理脑电信号中提取催眠状态脑电信号。
本发明催眠状态脑电信号提取方法与系统，采集催眠状态脑电初始信号，并进行预设倍频陷波处理，根据小波函数与理脑后电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数，自适应优化选取的小波函数的中心频率-带宽比，通过优化后的小波函数，从预处理脑电信号中提取催眠状态脑电信号。整个过程中，自适应优化小波函数的中心频率-带宽比，实现小波系数自适应调节，提高小波变换的分辨率，能有效且准确提取催眠状态脑电信号。
附图说明
图1为本发明催眠状态脑电信号提取方法第一个实施例的流程示意图；
图2为本发明催眠状态脑电信号提取方法第二个实施例的流程示意图；
图3为Morlet小波中心频率-带宽比和小波熵之间的曲线关系示意图；
图4为本发明催眠状态脑电信号提取系统第一个实施例的结构示意图；
图5为本发明催眠状态脑电信号提取系统第二个实施例的结构示意图。
具体实施方式
如图1所示，一种催眠状态脑电信号提取方法，包括步骤：
S100：采集催眠状态脑电初始信号，并对催眠状态脑电初始信号进行预设倍频陷波处理，获得预处理脑电信号。
催眠状态脑电初始信号可以通过现有的仪器设备采集，例如可以通过脑电图机采集。脑电图机采集到的催眠状态脑电信号通常含有一定频率的工频噪声干扰，没有明显的脑电信号波动规律，经过分析，催眠状态脑电初始信号不仅有工频噪声，还有一定频率的谐波 噪声，通过对这些催眠状态脑电初始信号进行预设倍频陷波处理，获得信号的曲线呈现出明显的脑电波动形状的预处理脑电信号。
S200：根据小波函数与预处理脑电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数。
小波函数有多种类型，例如Haar小波、Coif小波、Meyer小波、Mexican小波以及Morlet小波，不同的小波函数与预处理脑电信号之间相关性不同。非必要的，我们可以计算不同小波函数与预处理脑电信号之间相关性系数，选取相关性系数最大的小波函数作为用于提取催眠状态脑电信号的小波函数。
S300：自适应优化选取的小波函数的中心频率-带宽比，获得优化后的小波函数。
中心频率和带宽是影响小波函数时频分辨率关键因素。改变中心频率-带宽比就可以改变小波变换的时频分辨率。当中心频率-带宽比达到最优时，小波变换的时频分辨率最高，其能够更加准确提取催眠状态脑电信。在这里，采用自适应方式优化小波函数的中心频率-带宽比，自适应优化方式可以选择相对小波熵自适应优化，小波奇异熵自适应优化以及小波熵自适应优化等。
S400：通过优化后的小波函数，从预处理脑电信号中提取催眠状态脑电信号。
优化后小波函数的中心频率-带宽后，小波变换的时频分辨率得到明显的改善，能够更有效地提取催眠脑电信号中的高感受性成分，通过优化后的小波函数，从预处理脑电信号中准确提取催眠状态脑电信号。提取得到的高感受性催眠脑电信号，一方面，可在包含催眠脑电模块的脑电监护仪、电子睡眠仪、生物反馈睡眠仪设备上显示出来，作为诊断的基础；另一方面，为接下来的催眠分析、睡眠信息匹配提供准确可靠的数据基础。
本发明催眠状态脑电信号提取方法，采集催眠状态脑电初始信号，并进行预设倍频陷波处理，根据小波函数与理脑后电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数，自适应优化选取的小波函数的中心频率-带宽比，通过优化后的小波函数，从预处理脑电信号中提取催眠状态脑电信号。整个过程中，自适应优化小波函数的中心频率-带宽比，实现小波系数自适应调节，提高小波变换的分辨率，能有效且准确提取催眠状态脑电信号。
如图2所示，在其中一个实施例中，步骤S100包括：
S120：通过脑电图机、脑电生物反馈睡眠仪或脑电电极，采集催眠状态脑电初始信号。
通过脑电图机、脑电生物反馈睡眠仪或脑电电极这些精准的科学仪器设备，准确采集催眠状态脑电初始信号。
S140：对催眠状态脑电初始信号进行50Hz倍频陷波处理，获得预处理脑电信号。
步骤S120中采集到的催眠状态脑电初始信号含有50Hz的工频噪声干扰和还有50Hz的谐波噪声，例如100Hz的谐波噪声、150Hz的谐波噪声以及200Hz的谐波噪声等。由于未这些催眠状态脑电初始信号没有明显的脑电信号波动规律且存在较大噪声，需要对其进行50Hz倍频陷波处理，经倍频陷波之后，曲线才呈现出明显的脑电波动形状。非必要的可以选用50Hz倍频陷波器(50Hz陷波器、100Hz陷波器、150Hz陷波器、200Hz陷波器以及250Hz陷波器等)对信号进行预处理。
如图2所示，在其中一个实施例中，步骤S200包括：
S220：分别计算不同类型小波函数与预处理脑电信号的相关系数。
假设预处理脑电信号为x(k)，样本总数为N。小波变换以小波基函数为基础，通过变换得到一系列子小波，用于逼近催眠状态下脑电信号。小波基函数极为重要，关系着子波逼近的准确度，基函数ψ(t)∈L2(R)必须满足
Figure PCTCN2016113118-appb-000001
的要求。ψ(t)为小波基函数，通过时间-尺度变换得到一系列子小波，如
Figure PCTCN2016113118-appb-000002
其中，a,b∈R,a≠0。
针对催眠状态下脑电信号，选取小波信号和催眠状态下脑电信号之间的互相关系数作为评价指标，来选取合适的母小波函数。互相关系数计算式如下：
Figure PCTCN2016113118-appb-000003
y(t)＝[y0(t),y1(t),...,yn(t)]T为t时刻两个能量有限的确定信号,ρxy为它们的相关系数,且|ρxy|≤1，ρxy反映了x(t)和y(t)之间的相似程度。目前最常用于催眠状态脑电信号提取的母小波有Haar小波函数、Coif小波函数、Meyer小波函数、Mexican Hat小波函数和Morlet小波函数等，催眠状态脑电信号和小波信号之间的互相关系数计算结果如表1所示：
表1催眠状态脑电信号和小波信号之间的互相关系数
Figure PCTCN2016113118-appb-000004
S240：选取相关系数最大所对应的小波函数为用于提取催眠状态脑电信号的小波函数。
相关系数越大表明小波函数与预处理脑电信号相关性越好，选取相关系数最大所对应的小波函数为用于提取催眠状态脑电信号的小波函数。具体来说，通过上述表1可知Morlet小波的波形为振荡衰减形式，与催眠状态脑电信号最为接近，两者的互相关系数是所有小波波形中最大的。因此，选用Morlet小波作为催眠状态脑电信号提取的小波分析工具。
如图2所示，在其中一个实施例中，步骤S300包括：
S320：获取选取的所述小波函数的小波系数，根据所述小波系数转换获得概率分布序列pi，并计算所述概率分布序列pi的值。
S340：根据所述概率分布序列pi的值，绘制选取的小波函数的中心频率-带宽比与小波熵之间的关系曲线。
S360：根据关系曲线，查找选中心频率-带宽比的最优值。
S380：根据最优值，自适应优化选取的小波函数的中心频率-带宽比，获得优化后的小波函数。
在本实施例中，选用小波熵来优化小波函数的中心频率-带宽比，中心频率fc和带宽σf是影响小波时频分辨率关键因素。改变中心频率-带宽比就可以改变小波变换的时频分辨率。当中心频率-带宽比达到最优时，小波变换的时频分辨率最高。下面以母小波函数为Morlet小波的CMOR小波进行解释说明。CMOR小波的母小波表达式如下所示：
Figure PCTCN2016113118-appb-000005
其中，fc表示母波ψ(t)的特征频率，也是中心频率，σt为高斯窗的标准差，通常取值 为1，σf为带宽，通常σf＝1/2π·σt。分析CMOR小波的母小波可知，CMOR小波波形振荡衰减的快慢由带宽σf决定，波形的振荡频率由中心频率fc决定。根据上述公式可以计算CMOR小波的频率分辨率
Figure PCTCN2016113118-appb-000006
和时间分辨率
Figure PCTCN2016113118-appb-000007
其中，fs为采样频率，fc为中心频率，σf为带宽，fi为信号分析频率。利用Shannon熵优化小波变换中心频率-带宽比的核心思想，就是用概率分布序列pi来表示小波系数，然后计算pi的值，表达式如下所示：
Figure PCTCN2016113118-appb-000008
其中，pi是一个概率分布序列，通过小波系数转换得到，具有不确定性。其转换公式如下所示：
Figure PCTCN2016113118-appb-000009
X(fi,t)为小波系数。中心频率-带宽比fc/σf和Shannon小波熵之间的曲线关系，如图3所示。当中心频率-带宽比fc/σf＝4.43时，基于Shannon小波熵概率最优理论，可知当Shannon小波熵达到最小值时，Morlet小波中心频率-带宽比参数达到最优，对应的基小波就是与特征成分最匹配的小波。
在其中一个实施例中，通过优化后的小波函数，从预处理脑电信号中提取催眠状态脑电信号的步骤包括：
步骤一：获取优化后的小波函数的母小波，对母小波进行伸缩和平移得到子小波。
子波通过母波伸缩和平移生成，表达式如下示：
Figure PCTCN2016113118-appb-000010
a为频率伸缩因子，b为时间平移因子，*表征该函数是一个复函数。
步骤二：对子小波中的参数进行尺度转换，获得子小波的加权系数，其中，尺度转换包括伸缩因子转换为频率以及平移因子转换为时间。
对子波w(a,b)中的参数进行尺度转换，用频率f表示频率伸缩因子a，用时间t表示平移因子b，就可以得到加权系数X(f,t)。
步骤三：构建加权系数小波曲线三维平面，其中，空间X轴为时间轴，空间Y轴为频率轴，空间Z轴为加权系数轴。
构建加权系数小波曲线三维平面，在X(f,t)小波曲线三维平面中，x轴为时间轴(t)，y轴为频率轴(f)，z轴为小波系数(单位为dB)。
步骤四：获取优化后的小波函数周期，并查找加权系数小波曲线三维平面中频率最大值，计算连续阈限值。
定义“连续”阈限DTmax，DTmax(f)＝c/f，其中，c表示特征小波周期，f为频率。
步骤五：查找加权系数小波曲线三维平面中加权系数大于频率最大值部分所对应的时间区间值。
在局部最大值点，频率f'表示小波曲线三维平面最高点的频率取值。能量阈值用K表示。如果X(f',t)>K(f')，那么参数β可以调整阈值大小。令[t',t″]表示曲线平线上X(f',t)>K(f')的部分。
步骤六：当时间区间值大于连续阈限值时，判定加权系数小波曲线三维平面中加权系数大于频率最大值的部分对应的脑电信号为有效催眠状态脑电信号。
如果时间宽度t″-t'大于“连续”阈限DTmax，那么该区域就认为是催眠感受性高的脑电信号。
步骤七：通过优化后的小波函数，对有效催眠状态脑电信号进行时频信息提取，获得催眠状态脑电信号。
利用小波熵自适应优化后的CMOR小波对催眠状态下脑电信号进行时频信息提取，利用Shannon小波熵自适应优化CMOR小波的中心频率-带宽比参数后，小波变换的时频分辨率得到明显的改善，能够更有效地提取催眠脑电信号中的高感受性成分。
如图4所示，一种催眠状态脑电信号提取系统，包括：
预处理模块100，用于采集催眠状态脑电初始信号，并对催眠状态脑电初始信号进行预设倍频陷波处理，获得预处理脑电信号；
小波函数选择模块200，用于根据小波函数与预处理脑电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数；
优化模块300，用于自适应优化选取的小波函数的中心频率-带宽比，获得优化后的小波函数；
信号提取模块400，用于通过优化后的小波函数，从预处理脑电信号中提取催眠状态脑电信号。
本发明催眠状态脑电信号提取系统，采集催眠状态脑电初始信号，并进行预设倍频陷波处理，根据小波函数与理脑后电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数，自适应优化选取的小波函数的中心频率-带宽比，通过优化后的小波函数，从预处理脑电信号中提取催眠状态脑电信号。整个过程中，自适应优化小波函数的中心频率-带宽比，实现小波系数自适应调节，提高小波变换的分辨率，能有效且准确提取催眠状态脑电信号。
如图5所示，在其中一个实施例中，预处理模块100包括：
初始信号采集单元120，用于通过脑电图机、脑电生物反馈睡眠仪或脑电电极，采集催眠状态脑电初始信号。
预处理单元140，用于对催眠状态脑电初始信号进行50Hz倍频陷波处理，获得预处理脑电信号。
仪器设备采集到的催眠状态脑电初始信号含有50Hz的工频噪声干扰和还有50Hz的谐波噪声，例如100Hz的谐波噪声、150Hz的谐波噪声以及200Hz的谐波噪声等。由于未这些催眠状态脑电初始信号没有明显的脑电信号波动规律且存在较大噪声，需要对其进行50Hz倍频陷波处理，经倍频陷波之后，曲线才呈现出明显的脑电波动形状。非必要的，预处理单元140可以选用50Hz倍频陷波器(50Hz陷波器、100Hz陷波器、150Hz陷波器、200Hz陷波器以及250Hz陷波器等)对信号进行预处理。
如图5所示，在其中一个实施例中，小波函数选择模块200包括：
相关系数计算单元220，用于分别计算不同类型小波函数与预处理脑电信号的相关系数。
选取单元240，用于选取相关系数最大所对应的小波函数为用于提取催眠状态脑电信号的小波函数。
相关系数越大表明小波函数与预处理脑电信号相关性越好，选取相关系数最大所对应的小波函数为用于提取催眠状态脑电信号的小波函数。
如图5所示，在其中一个实施例中，优化模块300包括：
计算单元320，用于获取选取的所述小波函数的小波系数，根据所述小波系数转换获得概率分布序列pi，并计算所述概率分布序列pi的值。曲线绘制单元340，用于根据所述概率分布序列pi的值，绘制选取的小波函数的中心频率-带宽比与小波熵之间的关系曲线。
查找单元360，用于根据关系曲线，查找选中心频率-带宽比的最优值。
优化单元380，用于根据最优值，自适应优化选取的小波函数的中心频率-带宽比，获得优化后的小波函数。
在本实施例中，选用小波熵来优化小波函数的中心频率-带宽比，中心频率fc和带宽σf是影响小波时频分辨率关键因素。改变中心频率-带宽比就可以改变小波变换的时频分辨率。当中心频率-带宽比达到最优时，小波变换的时频分辨率最高。
在其中一个实施例中，信号提取模块400包括：
子小波获取单元，用于获取优化后的小波函数的母小波，对母小波进行伸缩和平移得到子小波。
转换单元，用于对子小波中的参数进行尺度转换，获得子小波的加权系数，其中，尺度转换包括伸缩因子转换为频率以及平移因子转换为时间。
构建单元，用于构建加权系数小波曲线三维平面，其中，空间X轴为时间轴，空间Y轴为频率轴，空间Z轴为加权系数轴。
阈限值计算单元，用于获取优化后的小波函数周期，并查找加权系数小波曲线三维平面中频率最大值，计算连续阈限值。
时间区间值计算单元，用于查找加权系数小波曲线三维平面中加权系数大于频率最大值部分所对应的时间区间值。
比较单元，用于当时间区间值大于连续阈限值时，判定加权系数小波曲线三维平面中加权系数大于频率最大值的部分对应的脑电信号为有效催眠状态脑电信号。
提取单元，用于通过优化后的小波函数，对有效催眠状态脑电信号进行时频信息提取，获得催眠状态脑电信号。
以CMOR小波为例，利用小波熵自适应优化后的CMOR小波对催眠状态下脑电信号 进行时频信息提取，利用Shannon小波熵自适应优化CMOR小波的中心频率-带宽比参数后，小波变换的时频分辨率得到明显的改善，能够更有效地提取催眠脑电信号中的高感受性成分。
以上实施例仅表达了本发明的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本发明构思的前提下，还可以做出若干变形和改进，这些都属于本发明的保护范围。因此，本发明专利的保护范围应以所附权利要求为准。

Claims (10)

  1. 一种催眠状态脑电信号提取方法，其特征在于，包括步骤：
    采集催眠状态脑电初始信号，并对所述催眠状态脑电初始信号进行预设倍频陷波处理，获得预处理脑电信号；
    根据小波函数与所述预处理脑电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数；
    自适应优化选取的所述小波函数的中心频率-带宽比，获得优化后的小波函数；
    通过所述优化后的小波函数，从所述预处理脑电信号中提取催眠状态脑电信号。
  2. 根据权利要求1所述的催眠状态脑电信号提取方法，其特征在于，所述采集催眠状态脑电初始信号，并对所述催眠状态脑电初始信号进行预设倍频陷波处理，获得预处理脑电信号的步骤包括：
    通过脑电图机、脑电生物反馈睡眠仪或脑电电极，采集催眠状态脑电初始信号；
    对所述催眠状态脑电初始信号进行50Hz倍频陷波处理，获得预处理脑电信号。
  3. 根据权利要求1或2所述的催眠状态脑电信号提取方法，其特征在于，所述根据小波函数与所述预处理脑电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数的步骤包括：
    分别计算不同类型小波函数与所述预处理脑电信号的相关系数；
    选取所述相关系数最大所对应的小波函数为用于提取催眠状态脑电信号的小波函数。
  4. 根据权利要求1或2所述的催眠状态脑电信号提取方法，其特征在于，所述自适应优化选取的所述小波函数的中心频率-带宽比，获得优化后的小波函数的步骤包括：
    获取选取的所述小波函数的小波系数，根据所述小波系数转换获得概率分布序列，并计算所述概率分布序列的值；
    根据所述概率分布序列的值，绘制选取的所述小波函数的中心频率-带宽比与小波熵之间的关系曲线；
    根据所述关系曲线，查找选中心频率-带宽比的最优值；
    根据所述最优值，自适应优化选取的所述小波函数的中心频率-带宽比，获得优化后的小波函数。
  5. 根据权利要求1或2所述的催眠状态脑电信号提取方法，其特征在于，所述通过 所述优化后的小波函数，从所述预处理脑电信号中提取催眠状态脑电信号的步骤包括：
    获取所述优化后的小波函数的母小波，对所述母小波进行伸缩和平移得到子小波；
    对所述子小波中的参数进行尺度转换，获得所述子小波的加权系数，其中，所述尺度转换包括伸缩因子转换为频率以及平移因子转换为时间；
    构建所述加权系数小波曲线三维平面，其中，空间X轴为时间轴，空间Y轴为频率轴，空间Z轴为所述加权系数轴；
    获取所述优化后的小波函数周期，并查找所述加权系数小波曲线三维平面中频率最大值，计算连续阈限值；
    查找所述加权系数小波曲线三维平面中所述加权系数大于所述频率最大值部分所对应的时间区间值；
    当所述时间区间值大于所述连续阈限值时，判定所述加权系数小波曲线三维平面中所述加权系数大于所述频率最大值的部分对应的脑电信号为有效催眠状态脑电信号；
    通过所述优化后的小波函数，对所述有效催眠状态脑电信号进行时频信息提取，获得催眠状态脑电信号。
  6. 一种催眠状态脑电信号提取系统，其特征在于，包括：
    预处理模块，用于采集催眠状态脑电初始信号，并对所述催眠状态脑电初始信号进行预设倍频陷波处理，获得预处理脑电信号；
    小波函数选择模块，用于根据小波函数与所述预处理脑电信号之间相关性，选取用于提取催眠状态脑电信号的小波函数；
    优化模块，用于自适应优化选取的所述小波函数的中心频率-带宽比，获得优化后的小波函数；
    信号提取模块，用于通过所述优化后的小波函数，从所述预处理脑电信号中提取催眠状态脑电信号。
  7. 根据权利要求6所述的催眠状态脑电信号提取系统，其特征在于，所述预处理模块包括：
    初始信号采集单元，用于通过脑电图机、脑电生物反馈睡眠仪或脑电电极，采集催眠状态脑电初始信号；
    预处理单元，用于对所述催眠状态脑电初始信号进行50Hz倍频陷波处理，获得预处 理脑电信号。
  8. 根据权利要求6或7所述的催眠状态脑电信号提取系统，其特征在于，所述小波函数选择模块包括：
    相关系数计算单元，用于分别计算不同类型小波函数与所述预处理脑电信号的相关系数；
    选取单元，用于选取所述相关系数最大所对应的小波函数为用于提取催眠状态脑电信号的小波函数。
  9. 根据权利要求6或7所述的催眠状态脑电信号提取系统，其特征在于，所述优化模块包括：
    计算单元，用于获取选取的所述小波函数的小波系数，根据所述小波系数转换获得概率分布序列，并计算所述概率分布序列的值；
    曲线绘制单元，用于根据所述概率分布序列的值，绘制选取的所述小波函数的中心频率-带宽比与小波熵之间的关系曲线；
    查找单元，用于根据所述关系曲线，查找选中心频率-带宽比的最优值；
    优化单元，用于根据所述最优值，自适应优化选取的所述小波函数的中心频率-带宽比，获得优化后的小波函数。
  10. 根据权利要求6或7所述的催眠状态脑电信号提取系统，其特征在于，所述信号提取模块包括：
    子小波获取单元，用于获取所述优化后的小波函数的母小波，对所述母小波进行伸缩和平移得到子小波；
    转换单元，用于对所述子小波中的参数进行尺度转换，获得所述子小波的加权系数，其中，所述尺度转换包括伸缩因子转换为频率以及平移因子转换为时间；
    构建单元，用于构建所述加权系数小波曲线三维平面，其中，空间X轴为时间轴，空间Y轴为频率轴，空间Z轴为所述加权系数轴；
    阈限值计算单元，用于获取所述优化后的小波函数周期，并查找所述加权系数小波曲线三维平面中频率最大值，计算连续阈限值；
    时间区间值计算单元，用于查找所述加权系数小波曲线三维平面中所述加权系数大于所述频率最大值部分所对应的时间区间值；
    比较单元，用于当所述时间区间值大于所述连续阈限值时，判定所述加权系数小波曲线三维平面中所述加权系数大于所述频率最大值的部分对应的脑电信号为有效催眠状态脑电信号；
    提取单元，用于通过所述优化后的小波函数，对所述有效催眠状态脑电信号进行时频信息提取，获得催眠状态脑电信号。
PCT/CN2016/113118 2016-05-13 2016-12-29 催眠状态脑电信号提取方法与系统 WO2017193595A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610322430.4A CN106037722B (zh) 2016-05-13 2016-05-13 催眠状态脑电信号提取方法与系统
CN201610322430.4 2016-05-13

Publications (1)

Publication Number Publication Date
WO2017193595A1 true WO2017193595A1 (zh) 2017-11-16

Family

ID=57177582

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/113118 WO2017193595A1 (zh) 2016-05-13 2016-12-29 催眠状态脑电信号提取方法与系统

Country Status (2)

Country Link
CN (1) CN106037722B (zh)
WO (1) WO2017193595A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113283062A (zh) * 2021-05-10 2021-08-20 杭州电子科技大学 一种运动想象脑电信号多变量建模方法
CN114504327A (zh) * 2021-12-28 2022-05-17 深圳大学 一种脑电噪声的处理方法、装置及计算机设备
CN116369866A (zh) * 2023-06-05 2023-07-04 安徽星辰智跃科技有限责任公司 基于小波变换的睡眠趋稳性量化及调节方法、系统和装置

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106037722B (zh) * 2016-05-13 2019-02-01 广州视源电子科技股份有限公司 催眠状态脑电信号提取方法与系统
CN106580319A (zh) * 2016-12-20 2017-04-26 广州视源电子科技股份有限公司 一种基于小波变换的脑电放松度识别方法及装置
CN106725487A (zh) * 2016-12-20 2017-05-31 广州视源电子科技股份有限公司 一种呼吸率提取方法及装置
CN111905229A (zh) * 2020-07-13 2020-11-10 湖南科技大学 一种基于5g的钢琴音乐催眠治疗控制系统及方法
CN118296306B (zh) * 2024-05-28 2024-09-06 小舟科技有限公司 基于分形维数增强的脑电信号处理方法、装置及设备

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101632587A (zh) * 2009-08-05 2010-01-27 南京大学 一种基于小波变换的步态信号提取方法
CN101732047A (zh) * 2009-12-16 2010-06-16 天津大学 复合下肢想象动作脑电的能量特征提取方法
CN103815901A (zh) * 2013-11-22 2014-05-28 刘志勇 一种应用于单导联便携式脑电设备的频域特征提取算法
US20150038804A1 (en) * 2013-07-30 2015-02-05 Magdy Younes Method, non-transitory computer readable medium and apparatus for arousal intensity scoring
US20150257700A1 (en) * 2010-08-02 2015-09-17 Chi Yung Fu Apparatus for treating a patient
CN105559777A (zh) * 2016-03-17 2016-05-11 北京工业大学 基于小波包和lstm型rnn神经网络的脑电识别方法
CN106037722A (zh) * 2016-05-13 2016-10-26 广州视源电子科技股份有限公司 催眠状态脑电信号提取方法与系统

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7373198B2 (en) * 2002-07-12 2008-05-13 Bionova Technologies Inc. Method and apparatus for the estimation of anesthetic depth using wavelet analysis of the electroencephalogram
US7809433B2 (en) * 2005-08-09 2010-10-05 Adidas Ag Method and system for limiting interference in electroencephalographic signals
CN103793593B (zh) * 2013-11-15 2018-02-13 吴一兵 一种获取大脑状态客观定量指标的方法
CN103989471B (zh) * 2014-05-08 2015-11-04 东北大学 一种基于脑电图识别的疲劳驾驶检测方法
CN104545901B (zh) * 2015-01-29 2016-09-21 中国科学院电子学研究所 脑电检测系统

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101632587A (zh) * 2009-08-05 2010-01-27 南京大学 一种基于小波变换的步态信号提取方法
CN101732047A (zh) * 2009-12-16 2010-06-16 天津大学 复合下肢想象动作脑电的能量特征提取方法
US20150257700A1 (en) * 2010-08-02 2015-09-17 Chi Yung Fu Apparatus for treating a patient
US20150038804A1 (en) * 2013-07-30 2015-02-05 Magdy Younes Method, non-transitory computer readable medium and apparatus for arousal intensity scoring
CN103815901A (zh) * 2013-11-22 2014-05-28 刘志勇 一种应用于单导联便携式脑电设备的频域特征提取算法
CN105559777A (zh) * 2016-03-17 2016-05-11 北京工业大学 基于小波包和lstm型rnn神经网络的脑电识别方法
CN106037722A (zh) * 2016-05-13 2016-10-26 广州视源电子科技股份有限公司 催眠状态脑电信号提取方法与系统

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113283062A (zh) * 2021-05-10 2021-08-20 杭州电子科技大学 一种运动想象脑电信号多变量建模方法
CN114504327A (zh) * 2021-12-28 2022-05-17 深圳大学 一种脑电噪声的处理方法、装置及计算机设备
CN114504327B (zh) * 2021-12-28 2024-05-17 深圳大学 一种脑电噪声的处理方法、装置及计算机设备
CN116369866A (zh) * 2023-06-05 2023-07-04 安徽星辰智跃科技有限责任公司 基于小波变换的睡眠趋稳性量化及调节方法、系统和装置
CN116369866B (zh) * 2023-06-05 2023-09-01 安徽星辰智跃科技有限责任公司 基于小波变换的睡眠趋稳性量化及调节方法、系统和装置

Also Published As

Publication number Publication date
CN106037722B (zh) 2019-02-01
CN106037722A (zh) 2016-10-26

Similar Documents

Publication Publication Date Title
WO2017193595A1 (zh) 催眠状态脑电信号提取方法与系统
CN105411565B (zh) 基于广义尺度小波熵的心率变异性特征分类方法
CN107361766B (zh) 一种基于emd域多维信息的情绪脑电信号识别方法
Aboalayon et al. Efficient sleep stage classification based on EEG signals
CN108416367B (zh) 基于多传感器数据决策级融合的睡眠分期方法
Samieinasab et al. Fetal phonocardiogram extraction using single channel blind source separation
Zakaria et al. Three resnet deep learning architectures applied in pulmonary pathologies classification
CN111329475B (zh) 基于集成模型的单通道脑电信号睡眠分期方法及系统
Dan et al. An identification system based on portable EEG acquisition equipment
CN110942103A (zh) 分类器的训练方法和计算机可读存储介质
Guo et al. DS-CNN: Dual-stream convolutional neural networks-based heart sound classification for wearable devices
Kuila et al. Feature extraction of electrocardiogram signal using machine learning classification
Kiran et al. TQWT based features for classification of ALS and healthy EMG signals
Ibrahim et al. ECG signals for human identification based on fiducial and non-fiducial approaches
CN111783715A (zh) 基于脉搏信号特征提取的身份识别方法
CN108470182A (zh) 一种用于非对称脑电特征增强与识别的脑-机接口方法
JP2004248714A (ja) 生体信号を用いた認証方法および認証装置
CN114469041A (zh) 一种运动过程中心率变化数据特征分析方法
KR102630840B1 (ko) 근전도 신호 기반 인식 정보 추출 시스템과 이를 이용한 근전도 신호 기반 인식 정보 추출 방법
Bassiouni et al. A study on the Intelligent Techniques of the ECG-based Biometric Systems
Zehir et al. Support vector machine for human identification based on non-fiducial features of the ecg
Sen et al. THE ROLE OF MACHINE LEARNING IN ENHANCING SLEEP STAGE DETECTION ACCURACY WITH SINGLE-CHANNEL EEG
Zhang et al. Research on feature extraction algorithm commonly used in brain-computer interface technology
Nguyen et al. MPCNN: A Novel Matrix Profile Approach for CNN-based Single Lead Sleep Apnea In Classification Problem
CN115105042B (zh) 自主神经系统稳定性的评估方法、评估系统以及评估设备

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16901550

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 16901550

Country of ref document: EP

Kind code of ref document: A1