CN-121971034-A - Sleep slow wave detection method based on pseudo-linear enhancement phase-locked loop

CN121971034ACN 121971034 ACN121971034 ACN 121971034ACN-121971034-A

Abstract

The invention belongs to the technical field of biomedical signal processing, and particularly relates to a sleep slow wave detection method based on a pseudo-linear enhanced phase-locked loop. The method comprises the steps of obtaining and preprocessing NREM 3-phase EEG data, building a sleep slow oscillation phase tracking algorithm of a pseudo-linear enhanced phase-locked loop with mutually coupled phase loops and amplitude loops, tracking the phase and amplitude of the EEG data in real time, sending out an effective stimulation instruction according to detected real-time phase information, carrying out statistical verification on a closed-loop stimulation result, and evaluating algorithm performance. The method has the advantages of high phase tracking accuracy, small error of stimulation time, strong amplitude self-adaption capability, wide individual adaption range, remarkably improved clinical application range of the technology, low calculation complexity, real-time satisfaction of closed-loop requirements, less time consumption of single operation, simplified parameter calibration flow and convenience for clinical quick deployment, and can greatly improve the effectiveness of stimulation to enhance SO rhythms.

Inventors

LI QIUYU
CHEN CHEN
TANG ZHENNING
WANG ZAIHAO

Assignees

复旦大学

Dates

Publication Date: 20260505
Application Date: 20251215

Claims (7)

1. The sleep slow wave detection method based on the pseudo-linear enhancement phase-locked loop is characterized by comprising the following steps of: (1) Acquiring and preprocessing NREM 3-phase EEG data; (2) Setting up a sleep slow oscillation phase tracking algorithm of a pseudo-linear enhancement phase-locked loop with mutually coupled phase loops and amplitude loops, and tracking the phase and amplitude of EEG data in real time; (3) According to the detected real-time phase information, an effective stimulation instruction is sent out; (4) And carrying out statistical verification on the closed-loop stimulation result, and evaluating the algorithm performance.
2. The slow wave detection method based on pseudo-linear enhancement phase-locked loop according to claim 1, wherein in step (1), sleep data of N3 stage is screened out according to AASM standard (W/N1/N2/N3/REM), and bandpass filtering of 0.3-35Hz is performed to suppress baseline drift and power frequency noise interference.
3. The slow wave detection method based on the pseudo-linear enhancement phase-locked loop according to claim 2, wherein in the step (2), the phase loop is composed of a basic component phase discriminator and a numerical control oscillator, and a scaling factor Kp is added to realize the phase tracking of the EEG, the amplitude loop is composed of a basic component difference device and a amplitude discriminator to realize the amplitude tracking of the EEG, the PL-EPLL structure comprises a first-order proportional controller which builds the phase loop and the amplitude loop to be coupled with each other, and respectively adjusts the phase error and the amplitude error to zero to enable the system to reach a locking state and realize the phase and amplitude tracking; The PL-EPLL structure takes a reference frequency f 0 near a SO main frequency as a center frequency to generate an initial synchronous signal s (t) =asin (ωt+phi), wherein A is a synchronous signal amplitude, phi is a synchronous signal phase, omega is a frequency, a difference value between an input actual EEG signal x (t) and a synchronous signal s (t) is used as a control signal e (t) =x (t) -s (t), a synchronous signal amplitude A is combined to perform standardization processing on a phase error, the standardized phase error signal is regulated by a proportional controller with a proportionality coefficient Kp and then is input into a voltage-controlled oscillator of a phase ring, and meanwhile, the amplitude ring takes the control signal e (t) as an input and outputs a synchronous amplitude A after proportional control with a gain of 1, SO that synchronous regulation of the phase error and the amplitude error is realized, and a synchronous signal after real-time correction is generated, SO that the system achieves a locking state.
4. A slow wave detection method based on a pseudo-linear enhanced phase-locked loop according to claim 3, wherein in step (3), when the detected current phase reaches the phase space of the UP state, namely, the 0-90 ° phase interval, a stimulus instruction is issued.
5. The pseudo-linear enhanced phase locked loop based slow wave detection method of claim 4 wherein the stimulus is at least one of acoustic, optical, electrical or mechanical.
6. The pseudo-linear enhanced phase locked loop based slow wave detection method of claim 5, wherein in step (4), algorithm performance is measured by three indicators, namely, deviation of average phase angle to 45 °, ratio of number of ineffective stimuli in SO segments to total number of stimuli, ratio of number of SO segments not subjected to effective stimuli to total number of SO segments.
7. The slow wave detection method based on a pseudo-linear enhanced phase locked loop according to claim 6, wherein in step (4): the deviation of the average phase angle to 45 deg., noted CMAE, is used to measure the stimulation accuracy; the ratio of the number of ineffective stimuli in the SO segment to the total number of stimuli is marked as PNAS and used for measuring the number of redundant stimuli; the ratio of the number of SO fragments which are not stimulated effectively in the total number of SO fragments is recorded as LOSS and used for measuring the number of missed stimulation; the euclidean distance ED formed by the normalization process for the three evaluation indexes is expressed as: 。

Description

Sleep slow wave detection method based on pseudo-linear enhancement phase-locked loop Technical Field The invention belongs to the technical field of biomedical signal processing, and particularly relates to a sleep slow wave detection method. Background Sleep is an important physiological process of the human body, and the quality of the sleep directly influences the neural plasticity, memory consolidation and cognitive function. Slow vibration (SO) in deep sleep stage (N3 stage) is a core characterization of sleep quality, representing low frequency, high amplitude electroencephalogram (EEG) rhythms with frequency <1Hz, which is a key mechanism for achieving memory coding and nerve repair, and phase coupling action with sleep spindle waves. Therefore, enhancing SO rhythm by targeted stimulation has become a core technical direction for improving sleep disorder and improving sleep quality. At present, a closed-loop stimulation paradigm based on real-time EEG feedback is widely applied to sleep regulation, and the paradigm realizes accurate regulation and control of SO rhythms through a closed-loop link of EEG signal acquisition, neurophysiologic analysis and synchronous stimulation output. The accuracy of the stimulation timing is the technical core, namely, only when the stimulation occurs in the UP state (0-90 DEG phase interval) of SO, the neural oscillation can be effectively enhanced, and the sleep stability can be disturbed when the stimulation acts in the DOWN state. This characteristic places extremely high demands on the real-time phase tracking of the EEG signal. The common practice in offline analysis is to remove high-frequency interference by a digital filter, then perform hilbert transform, and estimate the real phase of the signal. However, as a causal system, the FIR of the higher order narrowband filtering introduces higher group delay, while the IIR generates nonlinear phase delay, resulting in phase distortion, which is not suitable for a real-time system. The existing SO Phase tracking and stimulation technology is mainly divided into two types, namely a self-adaptive amplitude threshold detection (Adaptive Threshold, AT) method, a Phase-Locked Loop (PLL) method, a Phase difference elimination method and a continuous Phase tracking method, wherein after SO candidate signals are extracted through band-pass filtering, the UP state is judged according to whether the signal amplitude exceeds a dynamic threshold value and stimulation is triggered, but the SO amplitude has obvious individual difference and dynamic fluctuation, stimulation omission easily occurs to low-amplitude SO groups such as the elderly, the accuracy is difficult to guarantee, and the Phase-Locked Loop (PLL) method is adopted, and Phase difference is eliminated through a negative feedback mechanism by generating reference sine wave fitting SO signals. PLL's perform well in stable SO sequences, but are highly sensitive to signal amplitude variations, poorly fitting individuals, and prone to losing lock when EEG signals mutate in opposite phase, resulting in a deviation in stimulation timing. Disclosure of Invention The invention aims to provide a sleep slow wave detection method based on a pseudo-linear enhanced phase-locked loop (PL-EPLL) with high accuracy, strong real-time performance and wide adaptability, SO as to solve the problems of low identification rate of low-amplitude SO signals, easy locking losing under reverse mutation and poor individual adaptability of the existing phase tracking algorithm. The invention provides a method for detecting sleep slow waves (sleep slow oscillation phases) based on a pseudo-linear enhanced phase-locked loop (PL-EPLL), which comprises the following steps: (1) Acquiring and preprocessing NREM 3-phase EEG data; (2) Setting up a sleep slow oscillation phase tracking algorithm of a PL-EPLL structure with mutually coupled phase rings and amplitude rings, and tracking the phase and amplitude of EEG data in real time; (3) According to the detected real-time phase information, an effective stimulation instruction is sent out; (4) And carrying out statistical verification on the closed-loop stimulation result, and evaluating the algorithm performance. In the invention, in the step (1), sleep data of the N3 stage is screened out according to AASM standard (W/N1/N2/N3/REM), and band-pass filtering of 0.3-35Hz is carried out to inhibit interference such as baseline drift, power frequency noise and the like. In the invention, in the step (2), the phase loop consists of a basic component phase discriminator and a numerical control oscillator, and a scaling factor Kp is added to realize the phase tracking of the EEG, the amplitude loop consists of a basic component differentiator and a amplitude discriminator to realize the amplitude tracking of the EEG, and the PL-EPLL structure comprises a first-order proportional controller which builds the phase loop and the amplitude loop to be coupled wi