Method and device for processing cerebral oxygen signal based on endogenous low-frequency oscillation

Abstract

This invention provides a brain oxygen signal processing method and device based on endogenous low-frequency oscillations, belonging to the field of computer-aided design technology. The method includes: S1, preprocessing and concentration conversion of the original light intensity signal, performing sliding windowing and low-frequency sub-band analysis on the hemoglobin concentration change signal to obtain the low-frequency sub-band signal; S2, extracting the instantaneous phase information and amplitude envelope information of the low-frequency sub-band signal, and determining the amplitude-phase coordination quantity and amplitude envelope coordination quantity; S3, using consistency gating parameters for joint constraints, constructing a low-frequency coupling structure matrix, and generating low-frequency coupling feature parameters; S4, performing quantitative mapping, hierarchical characterization, and confidence output of brain functional states to achieve quantitative characterization of the brain oxygen signal characteristics of the current analysis window. This invention models the coordinated change characteristics of multi-channel brain oxygen signal data in the low-frequency range, determines the coordination and stability of the spatial distribution of brain oxygen signals, and realizes the output of brain oxygen signal feature labels.

Description

Technical Field

[0001] This invention relates to the field of computer-aided design technology, and more particularly to the field of brain oxygen signal processing method, specifically to a brain oxygen signal processing method and processing device based on endogenous low-frequency oscillation. Background Technology

[0002] Objective quantification of brain function status is a crucial issue in clinical unconscious monitoring, intensive care, and consciousness assessment. Among existing brain function monitoring methods, electroencephalography (EEG) is susceptible to environmental noise and artifacts, while functional magnetic resonance imaging (fMRI) struggles to meet the needs of continuous bedside monitoring. Functional near-infrared spectroscopy (fNIRS) offers advantages such as being non-invasive, portable, and allowing for continuous monitoring, making it a promising candidate for brain function monitoring. However, current fNIRS-based methods primarily focus on local hemoglobin concentration changes, single statistics, or single-band energy characteristics, failing to adequately utilize the synergistic structure between low-frequency oscillations across multiple measurement channels. In particular, there is a lack of dedicated state quantification parameters that can be constructed based on the structural characteristics connecting these low-frequency oscillations across multiple measurement channels. In reality, the endogenous low-frequency oscillations in multi-channel brain oxygenation signals not only contain information on local intensity variations but also the synergistic relationships and structural features between different measurement channels in low-frequency activity. This information is significant for characterizing the overall structural extent of brain function activity under different states. Therefore, there is an urgent need to provide a state quantification method based on the co-structure of endogenous low-frequency oscillations in multiple measurement channels, so as to construct a dedicated index that can characterize the connection structure of low-frequency brain oxygen oscillations in multiple measurement channels within the current analysis window, and output brain functional state quantification parameters for monitoring devices to use. Summary of the Invention

[0003] To address the shortcomings of the prior art, the present invention aims to provide a brain oxygen signal processing method and device based on endogenous low-frequency oscillations. By modeling the coordinated variation characteristics of multi-channel brain oxygen signal data in the low-frequency range, it can comprehensively reflect the coordination and stability of the spatial distribution of low-frequency brain oxygen signals from multiple measurement channels. Gating constraints are applied to the consistency between different types of coordinated information to reduce the interference on the coupling weight construction results when there are abnormal fluctuations in a single coupling quantity or significant inconsistencies between two types of coordinated information, thereby improving the robustness of the low-frequency coupling structure representation. The processing device can quickly calculate and output the quantitative results of the current state of consciousness after the arrival of new analysis window data. Based on the overall coordinated structure of low-frequency brain oxygen activity from multiple measurement channels, it can provide a more stable and detailed quantitative representation of the current brain functional state.

[0004] Specifically, on the one hand, the present invention provides a brain oxygen signal processing method based on endogenous low-frequency oscillations, which includes the following steps: S1: Acquire the raw light intensity signal of the functional near-infrared spectrum, perform preprocessing and concentration conversion to obtain the hemoglobin concentration change signal of the measurement channel; perform sliding windowing and low-frequency subband analysis on the hemoglobin concentration change signal to obtain the low-frequency subband signal of the measurement channel; S2: Transform the low-frequency subband signal output in step S1, extract instantaneous phase information and amplitude envelope information, and determine the amplitude-phase coordination amount and amplitude envelope coordination amount between measurement channels; S3: Based on the amplitude-phase coordination and amplitude-envelope coordination quantities between measurement channels obtained in step S2, set consistency gating parameters to jointly constrain the amplitude-phase coordination and amplitude-envelope coordination quantities, obtain the coupling edge weights between measurement channels, and construct a low-frequency coupling structure matrix; perform eigenvalue decomposition on the low-frequency coupling structure matrix; extract the overall coupling strength, dominant coupling strength parameters, and node interference ordered parameters based on the low-frequency coupling structure matrix, and synthesize the low-frequency coupling characteristic parameters of each low-frequency sub-band according to a weighted geometric method; S4: Use the low-frequency coupling feature parameters output from step S3 to quantize, map, and classify the low-frequency coupling features of the brain oxygen signal corresponding to the current analysis window; obtain the quantized value of the low-frequency coupling feature. Low-frequency coupling feature hierarchical labels and corresponding hierarchical confidence levels To achieve a quantitative characterization of the brain oxygen signal characteristics of the current analysis window.

[0005] Preferably, step S3 specifically includes: S31: Map the amplitude envelope covariance to Interval; set consistency gating parameters to characterize the consistency between amplitude-phase coherence and amplitude-envelope coherence, set low-frequency coupling edge weights, and construct a low-frequency coupling structure matrix; S32: Perform eigenvalue decomposition on the low-frequency coupled structure matrix, assuming the eigenvalues ​​satisfy: ; S33: Set coupling interference parameters to characterize the interference distribution law of the measurement channel in the current low-frequency coupling structure, and obtain the node interference ordered parameters; S34: Set the low-frequency coupling structure strength, calculate the energy weight of the low-frequency sub-band, and synthesize the low-frequency coupling structure strength of each low-frequency sub-band according to the weighted geometric method to obtain the low-frequency coupling characteristic parameters.

[0006] Preferably, the consistency gating parameter in step S31 is: ; in, For the first The measurement channel and the first The measurement channel in the first Consistency gating parameters for each low-frequency subband; To find the minimum value of the function; For the first The measurement channel and the first The measurement channel in the first Amplitude and phase coordination of each low-frequency sub-band; For the first The first analysis window, the first Low-frequency subband measurement channel With measurement channel The normalized amplitude envelope co-variance between them; The parameter is the minimum value. ; ; This is the index number for the low-frequency subband.

[0007] Preferably, the low-frequency coupling structure matrix in step S31 is: ; ; in, For the first Low-frequency coupling structure matrix under each low-frequency sub-band; For the first The first analysis window, the first Low-frequency subband measurement channel With measurement channel The low-frequency composite coupling strength between them after being constrained by consistency gating; This is to measure the total number of channels.

[0008] Preferably, the dominant coupling strength parameter in step S32 is: ; in, For the first The first analysis window Dominant coupling strength parameters in low-frequency subbands; For the first The first analysis window The first eigenvalue of the low-frequency coupling structure matrix under each low-frequency sub-band arranged in descending order of numerical value. It is the sum of all eigenvalues; Index for measurement channel numbering.

[0009] Preferably, the node interference ordering parameters in step S33 are: ; ; in, For the first The first analysis window Ordered parameters of node interference under a low-frequency sub-band; This is the normalized term for the total number of measurement channels, expressed as the natural logarithm. For the first The first analysis window The first low-frequency subband The coupling interference parameters of each measurement channel; For the first The first analysis window The first low-frequency subband The normalized interference distribution of each measurement channel is used to characterize the relative interference ratio of the measurement channels in the current low-frequency coupling structure; This is the sum of the coupling interference parameters of all measurement channels.

[0010] Preferably, step S34 specifically includes: ; ; in, For the first Low-frequency coupling characteristic parameters of each analysis window; These are the coupling structure coefficients; For the first The first analysis window Low-frequency coupling structure strength under a low-frequency sub-band; The overall coupling strength; The dominant coupling strength parameter; For the ordered parameters of node interference; For the first Energy weights of each low-frequency subband within each analysis window; This represents the total number of low-frequency sub-bands. for total; For the first The measurement channel of the first A low-frequency sub-band signal.

[0011] Preferably, step S4 specifically includes: S41: Low-frequency coupling characteristic parameters obtained in step S3 Used as the first The original brain functional state quantification parameters corresponding to each analysis window are used to realize the quantitative mapping of brain functional state. S42: Determine the reference prototype values ​​of awake characteristics based on brain functional state calibration samples. and unconscious feature reference prototype value , obtain the state quantization value of the current analysis window and the feature distance between the two state prototypes; S43: Based on the characteristic distance relationship between the current analysis window and the two types of state prototypes, determine whether the brain functional state corresponding to the current analysis window is the conscious reference prototype or the unconscious reference prototype.

[0012] On the other hand, the present invention provides a processing device for a brain oxygen signal processing method based on endogenous low-frequency oscillation, which includes: a light source driving unit, a photoelectric detection unit, a signal acquisition unit, a main control processing unit, and a probe assembly; The light source driving unit is connected to the emitting probe in the probe assembly, and controls each near-infrared emitting probe to emit near-infrared light according to a preset driving sequence. The photoelectric detection unit is connected to the receiving probe in the probe assembly to receive near-infrared light signals and convert the received light signals into corresponding analog electrical signals for output.

[0013] The signal acquisition unit is connected to the light source driving unit and the photoelectric detection unit respectively. It processes the analog electrical signals output by each receiving probe according to the light source driving timing to obtain the original light intensity signal corresponding to the measurement channel. The main control processing unit is connected to the signal acquisition unit to receive the raw light intensity signal of the measurement channel output by the signal acquisition unit, and to execute the brain oxygen signal processing method based on endogenous low-frequency oscillation to generate low-frequency coupling feature parameters and state recognition results corresponding to the brain functional state. The probe assembly includes a light-shielding part and a buffer contact part. The light-shielding part is used to reduce ambient light interference, and the buffer contact part is used to improve contact stability.

[0014] Preferably, the main control processing unit and signal acquisition unit include: a data preprocessing and concentration conversion module, a sliding window module, a low-frequency oscillation analysis module, a coupled structure feature calculation module, a state index generation module, and a state output module; The data preprocessing and concentration conversion module processes the original light intensity signal and converts the preprocessed light intensity signal into a hemoglobin concentration change signal; The sliding window module performs sliding window processing on the hemoglobin concentration change signal; The low-frequency oscillation analysis module performs low-frequency sub-band decomposition on the measurement channel window signal within each analysis window to obtain the sub-band signal on the low-frequency sub-band, performs transformation, and extracts instantaneous phase information and amplitude envelope information to obtain the analytical signal, instantaneous phase, and amplitude envelope. The coupled structure feature calculation module calculates the cooperative relationship between measurement channels based on the instantaneous phase information and amplitude envelope information between the measurement channels; The state index generation module generates low-frequency coupling characteristic parameters corresponding to the current analysis window based on the overall coupling strength, the dominant coupling strength parameter node interference ordered parameter, and the weights of each low-frequency sub-band. The state output module is used to perform state quantization mapping, state prototype distance discrimination, and discrimination confidence output on the low-frequency coupled feature parameters output by the state index generation module.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) This invention models the coordinated change characteristics of multi-measurement channel brain oxygen signal data in the low-frequency range. Instead of relying solely on the amplitude of a single measurement channel, a single statistical parameter, or a simple threshold, it starts from the coordinated relationship between low-frequency oscillations of multiple measurement channels and constructs low-frequency coupling characteristic parameters that can characterize the overall signal state within the current analysis window. Compared with existing methods, this invention can more comprehensively reflect the coordination and stability of low-frequency brain oxygen signals in the spatial distribution of multiple measurement channels, thereby improving the ability to distinguish different brain functional states.

[0016] (2) This invention constructs a dedicated quantitative index based on the overall structure of low-frequency oscillations of multiple measurement channels within the current analysis window; it performs gating constraints on the consistency between different types of cooperative information to reduce the interference on the construction results of coupling edge weights when the single coupling quantity fluctuates abnormally or the two types of cooperative information are significantly inconsistent, thereby improving the robustness of the low-frequency coupling structure characterization; on this basis, the constructed low-frequency coupling characteristic parameters can simultaneously reflect the cooperative strength between measurement channels, the dominant coupling strength parameters, and the interference distribution law, thus having good structural significance and physiological signal basis, and can more stably characterize the difference between the conscious reference prototype and the unconscious reference prototype.

[0017] (3) The present invention proposes a processing device that can quickly complete the calculation and output of the quantitative characteristics of the current state after the arrival of new analysis window data. Compared with the existing monitoring methods that rely on a single statistical feature or fixed empirical parameters, the present invention can perform a more stable and detailed quantitative characterization of the current brain function state based on the overall synergistic structure of low-frequency brain oxygen activity in multiple measurement channels. Attached Figure Description

[0018] Figure 1 This is a control block diagram of the brain oxygen signal processing method based on endogenous low-frequency oscillations of the present invention. Figure 2 This is a low-frequency coupling matrix average diagram of the sober reference prototype sample under different states in the embodiments of the present invention; Figure 3 This is a low-frequency coupling matrix average diagram of the unconscious reference prototype in different states in the embodiments of the present invention; Figure 4 These are low-frequency coupling feature parameter diagrams of two sets of samples under different states in this embodiment of the invention; Figure 5 This is a graph showing the state quantization values ​​of samples under different states in embodiments of the present invention; Figure 6 This is a physical image of the processing device worn in an embodiment of the present invention. Detailed Implementation

[0019] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

[0020] This invention proposes a brain oxygen signal processing method based on endogenous low-frequency oscillations, such as... Figure 1 As shown, the original light intensity signal is preprocessed and concentration converted. The hemoglobin concentration change signal is then subjected to sliding windowing and low-frequency sub-band analysis to obtain the low-frequency sub-band signal. Instantaneous phase and amplitude envelope information of the low-frequency sub-band signal are extracted to determine amplitude-phase co-variance and amplitude-envelope co-variance. Consistency gating parameters are used for joint constraints to construct a low-frequency coupling structure matrix and generate low-frequency coupling feature parameters. The brain functional state is then quantitatively mapped, hierarchically characterized, and confidence-outputted to achieve a quantitative characterization of the brain oxygen signal characteristics of the current analysis window. Specifically, the process includes the following steps: Step S1: Acquire the raw light intensity signal of the functional near-infrared spectrum, perform preprocessing and concentration conversion to obtain the hemoglobin concentration change signal of the measurement channel; perform sliding windowing and low-frequency subband analysis on the hemoglobin concentration change signal to obtain the low-frequency subband signal of the measurement channel.

[0021] Step S11: For The raw intensity signal of the functional near-infrared spectrum of the measurement channel is preprocessed and converted to obtain the hemoglobin concentration change signal of the measurement channel. Let the first... Each measurement channel at wavelength The original light intensity signal collected at the location is ,in, , Indicates the sampling time. In the embodiments of the present invention, the following are used: =20-channel functional near-infrared brain oxygenation monitoring device. (Regarding the raw light intensity signal) Outlier removal, motion artifact correction, baseline drift removal, and bandpass filtering are performed to obtain the preprocessed light intensity signal. Based on the modified Beer-Lambert law, the preprocessed light intensity signal is converted into a hemoglobin concentration change signal, thus obtaining the oxyhemoglobin concentration change signal of the measurement channel. and signals of changes in deoxyhemoglobin concentration Detect the signal of changes in oxyhemoglobin concentration. As the input signal for subsequent low-frequency coupled structure analysis, the output of step S11 was ultimately determined to be the signal of change in hemoglobin concentration in the measurement channel.

[0022] Step S12: Perform sliding windowing and low-frequency subband analysis on the hemoglobin concentration change signal output from step S11 to obtain the low-frequency subband signal of the measurement channel. Set the sliding analysis window length to... The window movement step length is Then in the first Within the analysis window, the first... The window signal for each measurement channel is represented as follows: ,in, This refers to the sampling point number within the current analysis window; Indicates the analysis window number; Number the first measurement channel. , To measure the total number of channels, the window signal is divided into... There are several low-frequency sub-bands, and the set of low-frequency sub-bands is represented as: ; for the first The first analysis window The window signal of the measurement channel is subjected to low-frequency subband filtering to obtain the first measurement channel. The subband signal on each low-frequency subband is: ; in, For the first The measurement channel of the first A low-frequency sub-band signal; For the first Window signals for each measurement channel; For the first Low-frequency subband Bandpass filter operator; This refers to the sampling point number within the current analysis window; For the first One low-frequency sub-band; For low-frequency subband index number, .

[0023] The output of step S12 is the low-frequency subband signal of each measurement channel within each analysis window. .

[0024] Step S2: Transform the low-frequency subband signal output in step S12, extract instantaneous phase information and amplitude envelope information, and determine the amplitude-phase coordination and amplitude envelope coordination between measurement channels; for the... The first analysis window, the first The first low-frequency subband Subband signal of each measurement channel After transformation, the corresponding analytic signal is obtained as follows: ; in, For the first The measurement channel in the first A low-frequency sub-band analytical signal; For transformation operators; The imaginary unit; This is the number for the second measurement channel.

[0025] The low-frequency subband analytical signal is represented in polar coordinates as follows: ; in, For the first The measurement channel in the first The amplitude envelope of each low-frequency sub-band; The base is the natural number; For the first The measurement channel in the first The instantaneous phase of a low-frequency sub-band.

[0026] For any two measurement channels and Set it in the first The phase difference under each low-frequency sub-band is: ; in, For the first The measurement channel and the first The measurement channel in the first The instantaneous phase difference of each low-frequency sub-band.

[0027] Based on this, the amplitude-phase coordination between the two measurement channels is obtained as follows: ; ; in, For the first The measurement channel and the first The measurement channel in the first Amplitude and phase coordination of each low-frequency sub-band; For the first The measurement channel and the first The measurement channel in the first Analytical signal complex cross-multiplication terms in each low-frequency sub-band; It is a symbolic function; To extract the imaginary part; To calculate the expectation of the sample points within the current analysis window; The parameter is the minimum value. To analyze the signal The conjugate of complex numbers.

[0028] Obtain the amplitude envelope coordination between the two measurement channels for: ; in, For the first The measurement channel and the first The measurement channel in the first The amplitude envelope coordination quantity of each low-frequency sub-band; For the first The average amplitude envelope of each measurement channel within the current analysis window; For the first The average amplitude envelope of each corresponding measurement channel within the current analysis window; This is the length of the sliding analysis window.

[0029] The output of step S2 is the amplitude and phase coordination between the measurement channels in each analysis window and each low-frequency subband. and amplitude envelope coherence .

[0030] Step S3: Based on the amplitude-phase coordination and amplitude-envelope coordination quantities between measurement channels obtained in Step S2, joint constraints are applied to the coordination quantities using consistency gating parameters to obtain the coupling edge weights between measurement channels and construct a low-frequency coupling structure matrix; subsequently, based on the low-frequency coupling structure matrix, the overall coupling strength, dominant coupling strength parameters, and nodal interference order parameters are extracted to generate the low-frequency coupling characteristic parameter LCOI, short for Low-Frequency Coupling Organization Index. Figure 4 The diagram shows the low-frequency coupling characteristic parameters of two sets of samples under different states in an embodiment of the present invention. It should be noted that the differences in brain oxygen signals from multiple measurement channels in the target low-frequency subband under different brain functional states do not necessarily manifest as an absolute increase or decrease in a single amplitude feature, a single phase coupling index, or a single measurement channel energy feature, but rather as an overall change in the synergistic relationship and structure between multiple measurement channels. Therefore, this invention constructs low-frequency coupling characteristic parameters that reflect the low-frequency coupling structural pattern within the current analysis window by jointly extracting amplitude-phase synergy, amplitude-envelope synergy, and consistency gating results, and achieves state quantification and discrimination based on brain functional state calibration samples.

[0031] Step S31: To unify phase coupling information and amplitude coupling information, first map the amplitude envelope co-coordinated quantity to... The interval is: ; in, For the first The first analysis window, the first Low-frequency subband measurement channel With measurement channel The normalized amplitude envelope of the co-variable.

[0032] To characterize the degree of consistency between amplitude-phase coherence and amplitude-envelope coherence, the consistency gating parameter is set as follows: ; in, For the first The measurement channel and the first The measurement channel in the first Consistency gating parameters for each low-frequency subband; This is a function that takes the minimum value.

[0033] Consistency gating parameters are used to characterize amplitude and phase coordination. Co-occurrence with amplitude envelope The degree of consistency between them; when and When the values ​​are close to and at a high level, The value is relatively large; when the difference between the two increases or either synergistic quantity is low. The value is reduced; the consistency gating parameter is not used to presuppose that the amplitude-phase co-occurrence or amplitude-envelope co-occurrence must be higher in a certain brain functional state, but rather to emphasize the consistency and common support of the two types of co-occurrence information in the current analysis window and the current low-frequency subband, thereby improving the robustness of the coupling edge weight construction results. Then set the first The first analysis window, the first Low-frequency subband measurement channel With measurement channel Low-frequency coupling edge weights between And based on this, construct the first The low-frequency coupling structure matrix under each low-frequency sub-band is: ; ; in, For the first The low-frequency coupling structure matrix under each low-frequency subband has its diagonal elements set to zero. It is a symmetric matrix. For the first The first analysis window, the first Low-frequency subband measurement channel With measurement channel The low-frequency integrated coupling strength between the two channels after being constrained by consistency gating is the edge weight of the corresponding measurement channel pair in the low-frequency coupling structure matrix.

[0034] Setting the first The overall coupling strength under each low-frequency subband is: ; in, For the first The first analysis window The overall coupling strength under each low-frequency subband is used to characterize the overall level of the coupling weights between measurement channels in the current low-frequency coupling structure matrix; This is to measure the total number of channels.

[0035] Step S32: For the low-frequency coupling structure matrix Perform eigenvalue decomposition, and assume that the eigenvalues ​​satisfy: ;in, Low-frequency coupling structure matrix eigenvalues; the dominant coupling strength parameter is set as: ; in, For the first The first analysis window The dominant coupling strength parameter under each low-frequency subband is used to characterize the proportion of the dominant coupling mode in the current low-frequency coupling structure; For the first The first analysis window The first eigenvalue of the low-frequency coupling structure matrix under each low-frequency subband, arranged in descending order of numerical value, is used to characterize the dominant coupling strength in the current low-frequency coupling structure. It is the sum of all eigenvalues; Index for measurement channel numbering.

[0036] Step S33: To characterize the interference distribution of the measurement channel in the current low-frequency coupling structure, the first step is set... The coupling interference parameters for each measurement channel are: ; in, For the first The first analysis window The first low-frequency subband The coupling interference parameters of each measurement channel are used to characterize the interference intensity of that measurement channel in the current low-frequency coupling structure; For the first Measurement channels within the analysis window With measurement channel In the Coupled side weights under a low-frequency subband.

[0037] The normalized interference distribution is defined as follows: ; in, For the first The first analysis window The first low-frequency subband The normalized interference distribution of each measurement channel is used to characterize the relative interference ratio of the measurement channels in the current low-frequency coupling structure; This is the sum of the coupling interference parameters of all measurement channels.

[0038] Based on this, the node interference order parameters are set as follows: ; in, For the first The first analysis window The node interference ordered parameters under each low-frequency subband are used to characterize the interference distribution law of the measurement channel in the current low-frequency coupling structure; This is the normalized term for the total number of measurement channels, expressed in natural logarithmic form.

[0039] Step S34: Set the first The low-frequency coupling structure strength under each low-frequency subband is: ; in, For the first The first analysis window The low-frequency coupling structure strength under each low-frequency subband is used to comprehensively characterize the overall coordination degree, dominant coupling strength parameters, and node interference order parameters of brain oxygen oscillations in multiple measurement channels under the current low-frequency subband. The overall coupling strength; The dominant coupling strength parameter; These are the ordered parameters for node interference.

[0040] Calculate the first The energy weights of each low-frequency sub-band within each analysis window are: ; in, For the first The energy weights of each low-frequency subband within each analysis window.

[0041] Finally, the low-frequency coupling structure strengths of each low-frequency sub-band are synthesized using a weighted geometric method to obtain the first... Low-frequency coupling characteristic parameters of each analysis window: ; in, For the first Low-frequency coupling characteristic parameters of each analysis window; This is the coupling structure coefficient, which is set to 100 in this example.

[0042] The output of step S3 is the low-frequency coupling characteristic parameter of the current analysis window. Low-Frequency Coupling Organization Index (LFCO) is a low-frequency coupling characteristic parameter.

[0043] Step S4: Use the low-frequency coupling feature parameters output in step S3 to perform quantitative mapping, hierarchical characterization, and confidence output of the brain functional state corresponding to the current analysis window.

[0044] Step S41: Quantitative mapping of brain functional states; Low-frequency coupling feature parameters obtained in step S3 Used as the first The original brain functional state quantification parameters correspond to each analysis window; to facilitate unified comparison between different samples, low-frequency coupling feature parameters are further mapped to a preset quantization interval to obtain the brain functional state quantification parameters. for: ; in, The lower bound of the low-frequency coupling characteristic parameter is set based on the statistical results of the brain functional state calibration sample; The upper bound of the low-frequency coupling characteristic parameters is set based on the statistical results of brain functional state calibration samples; This is a quantitative value for brain functional status, used to characterize the overall structural degree and state-related differences of low-frequency brain oxygen oscillations in multiple measurement channels within the current analysis window.

[0045] like Figure 5 The figure shown is a graph of state quantization values ​​of samples under different states in an embodiment of the present invention; the magnitude of the state quantization values ​​is determined by the brain functional state calibration samples, and is uniformly quantified based on the distribution differences of low-frequency coupling feature parameters in samples under different states.

[0046] Step S42: Obtain the distance of brain functional state features; to improve the stability of hierarchical representation, determine the reference prototype value of awake features based on the brain functional state calibration samples. and unconscious feature reference prototype value Among them, the reference prototype values ​​for conscious and unconscious features were obtained by statistically analyzing the state quantification values ​​of the corresponding state samples, specifically as follows: ; ; in, The reference prototype value for the sobriety characteristic; This serves as a reference prototype value for unconscious characteristics. Number of conscious reference prototype samples; Number of unconscious reference prototype samples; The state quantification values ​​corresponding to the conscious reference prototype and the unconscious reference prototype samples; The state quantification values ​​corresponding to the conscious reference prototype and the unconscious reference prototype samples; For reference to the prototype sample index; Index of unconscious reference prototype samples.

[0047] The feature distance between the current state quantization value of the analysis window and the two state prototypes is obtained as follows: ; ; in, The feature distance between the current analysis window and the sober reference prototype; This represents the feature distance between the current analysis window and the unconscious reference prototype.

[0048] Step S43: Hierarchical representation and confidence output; Based on the feature distance relationship between the current analysis window and the two state prototypes, the brain functional state is discriminated; when At that time, the brain functional state corresponding to the current analysis window is determined to be the awake reference prototype, such as... Figure 2 The image shows the low-frequency coupling matrix average of the sober reference prototype sample in different states in an embodiment of the present invention; when At that time, the brain functional state corresponding to the current analysis window is determined to be the unconscious reference prototype, such as... Figure 3 The image shown is a low-frequency coupling matrix average diagram of the unconscious reference prototype in different states in an embodiment of the present invention; to characterize the accuracy of the current discrimination result, a confidence level for brain function grading is set. for: ; in, To characterize the confidence level of brain function grading. The larger the value, the more obviously the current analysis window is biased towards a certain state prototype, and the higher the accuracy of the corresponding judgment result; the smaller the value, the more the current analysis window is located between two state prototypes, and the uncertainty of the corresponding judgment result is relatively high.

[0049] The output of step S4 includes: quantification of brain functional state. Confidence of brain functional state labels and brain functional grading representations This enables the quantitative characterization and hierarchical output of the low-frequency coupling characteristics of brain oxygen signals.

[0050] On the other hand, the present invention also provides a processing device based on a brain oxygen signal processing method using endogenous low-frequency oscillations, specifically a 20-channel functional near-infrared brain oxygen monitoring device, such as... Figure 6The diagram shows the actual device being worn in an embodiment of the present invention; specifically, it includes: a light source driving unit, a photoelectric detection unit, a signal acquisition unit, a main control processing unit, and a probe assembly. Multiple near-infrared emitting probes and multiple near-infrared receiving probes are arranged alternately to form multiple measurement channels with different or identical source-detection distances between adjacent emitting and receiving probes; these measurement channels are used to acquire near-infrared light intensity changes corresponding to changes in hemoglobin concentration. The probe fixing base adopts a cap-like structure to accommodate different objects under test and to reduce measurement errors caused by body movement, probe offset, or unstable contact during detection. The probe assembly includes a light-shielding part and a buffer contact part; the light-shielding part is used to reduce ambient light interference, and the buffer contact part is used to improve contact stability.

[0051] The light source driving unit is connected to the emitting probe in the probe assembly. It is used to control each near-infrared emitting probe to emit near-infrared light of different wavelengths according to the preset driving sequence, and to adjust the driving current of each emitting probe so that the incident light of the measurement channel is stably output.

[0052] The photoelectric detection unit is connected to the receiving probe in the probe assembly to receive the scattered and absorbed near-infrared light signal and convert the received light signal into a corresponding analog electrical signal output.

[0053] The signal acquisition unit is connected to the light source driving unit and the photoelectric detection unit, respectively. It is used to synchronously sample, amplify, filter, and convert the analog electrical signals output by each receiving probe according to the light source driving timing to obtain the original light intensity signal corresponding to the measurement channel. ,in, Indicates the measurement channel number. Indicates the wavelength of the emitted light. Indicates the sampling time.

[0054] The main control processing unit is connected to the signal acquisition unit and is used to receive the raw light intensity signal from the measurement channel output by the signal acquisition unit. It then executes a brain oxygen signal processing method based on endogenous low-frequency oscillations to generate low-frequency coupling characteristic parameters and state recognition results corresponding to brain functional states. The main control processing unit consists of a processor, memory, and program instructions. The memory stores data processing programs for implementing each functional module. After the processor calls the program instructions, it sequentially completes data preprocessing and concentration conversion, sliding windowing, low-frequency oscillation analysis, coupling structure feature calculation, state index generation, and state output. The main control processing unit includes a data preprocessing and concentration conversion module, a sliding windowing module, a low-frequency oscillation analysis module, a coupling structure feature calculation module, a state index generation module, and a state output module.

[0055] The data preprocessing and concentration conversion module is used to process the raw light intensity signal output by the signal acquisition unit. Outlier removal, motion artifact correction, baseline drift removal, and filtering are performed. Based on the improved Beer-Lambert law, the preprocessed light intensity signal is converted into a hemoglobin concentration change signal to obtain the oxyhemoglobin concentration change signal of the measurement channel. .

[0056] The sliding window module is used to perform sliding window processing on the hemoglobin concentration change signal; the length of the sliding analysis window is set to... The window movement step length is Then the first The first analysis window The window signal of each measurement channel is denoted as ,in, This indicates the sampling point number within the current analysis window. Indicates the analysis window number.

[0057] The low-frequency oscillation analysis module is used to perform low-frequency sub-band decomposition on the measurement channel window signals within each analysis window to obtain the first... Subband signal on a low-frequency subband The sub-band signal is then transformed to extract the corresponding instantaneous phase information and amplitude envelope information, resulting in an analytical signal. Instantaneous phase and amplitude envelope .

[0058] The coupled structure feature calculation module is used to calculate the coordination relationship between measurement channels based on the instantaneous phase information and amplitude envelope information between the measurement channels; the coupled structure feature calculation module determines the amplitude and phase coordination amount respectively. and amplitude envelope coherence And the amplitude envelope covariance is mapped to the normalized amplitude envelope covariance. Use consistency gating factor Constructing coupled edge weights And form a low-frequency coupling structure matrix. Based on this, the overall coupling strength is extracted. Dominant coupling strength parameters and node interference ordered parameters .

[0059] The state index generation module is used to generate the state index based on the overall coupling strength. Dominant coupling strength parameters Node interference ordered parameters And the weights of each low-frequency subband, generating the low-frequency coupling characteristic parameters corresponding to the current analysis window. , This is the Low-Frequency Coupling Organization Index, also known as the low-frequency coupling characteristic parameter.

[0060] The state output module is used to process the low-frequency coupling characteristic parameters output by the state exponent generation module. The state quantization mapping, state prototype distance discrimination, and discrimination confidence output are performed; the state output module first performs state quantization mapping, state prototype distance discrimination, and discrimination confidence output based on low-frequency coupling feature parameters. Generate the state quantization value corresponding to the current analysis window. This facilitates unified comparisons between different samples; then, based on pre-established reference prototype values ​​of sobriety features... and unconscious feature reference prototype value Calculate the feature distance between the current state quantization value and the two types of state prototypes respectively. and The system outputs corresponding status labels based on distance relationships; the status output module also outputs the confidence level of the current hierarchical representation result. This characterizes the degree to which the current analysis window is biased towards a particular state prototype and the accuracy of the discrimination results. The feature distance between the current state quantification value and the awake feature reference prototype value... The feature distance is less than or equal to the reference prototype value of the unconscious feature. When the current analysis window's corresponding status label is determined to be the conscious reference prototype; when Greater than At that time, the state label corresponding to the current analysis window is determined to be the unconscious reference prototype. Confidence level is then determined. The larger the value, the clearer the state attribution corresponding to the current analysis window, and the higher the accuracy of the discrimination result; discrimination confidence. The smaller the value, the more likely the current analysis window is located between two state prototypes, indicating a relatively high uncertainty in its state transition characteristics or discrimination. The state output module can also directly output continuous state quantization values. or low-frequency coupling characteristic parameters It can be called by display and alarm units, external monitoring equipment or closed-loop control modules to realize status trend display, continuous monitoring result output or subsequent alarm analysis.

[0061] The beneficial effects of this invention are as follows: This invention proposes a brain oxygen signal processing method and device based on endogenous low-frequency oscillations; it models the coordinated change characteristics of multi-channel brain oxygen signal data in the low-frequency range, constructing low-frequency coupling characteristic parameters that can characterize the overall signal state within the current analysis window, thus more comprehensively reflecting the coordination and stability of the spatial distribution of low-frequency brain oxygen signals from multiple measurement channels, thereby improving the ability to distinguish different brain functional states; it constructs dedicated quantitative indicators around the overall structure of low-frequency oscillations from multiple measurement channels within the current analysis window, possessing good structural significance and physiological signal basis, and can more stably characterize the differences between conscious and unconscious reference prototypes; the processing device can quickly complete the calculation and output of the quantitative results of the current state of consciousness after the arrival of new analysis window data; compared with existing monitoring methods that rely on single statistical features or fixed empirical parameters, this invention can provide a more stable and detailed quantitative characterization of the current brain functional state based on the overall coordinated structure of low-frequency brain oxygen activity from multiple measurement channels.

[0062] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A brain oxygen signal processing method based on endogenous low-frequency oscillations, characterized in that, It includes: S1: Acquire the raw light intensity signal of the functional near-infrared spectrum, perform preprocessing and concentration conversion to obtain the hemoglobin concentration change signal of the measurement channel; perform sliding windowing and low-frequency subband analysis on the hemoglobin concentration change signal to obtain the low-frequency subband signal of the measurement channel; S2: Transform the low-frequency subband signal output in step S1, extract instantaneous phase information and amplitude envelope information, and determine the amplitude-phase coordination amount and amplitude envelope coordination amount between measurement channels; S3: Based on the amplitude-phase coordination quantity and amplitude-envelope coordination quantity between the measurement channels obtained in step S2, set consistency gating parameters to jointly constrain the amplitude-phase coordination quantity and amplitude-envelope coordination quantity, obtain the coupling edge weights between the measurement channels, and construct the low-frequency coupling structure matrix; Eigenvalue decomposition is performed on the low-frequency coupling structure matrix; based on the low-frequency coupling structure matrix, the overall coupling strength, dominant coupling strength parameters, and node interference ordered parameters are extracted, and the low-frequency coupling structure strength of each low-frequency sub-band is synthesized into low-frequency coupling characteristic parameters in a weighted geometric manner; S4: Use the low-frequency coupling feature parameters output in step S3 to quantify, map, and hierarchically characterize the low-frequency coupling features of the brain oxygen signal corresponding to the current analysis window; Obtain the quantization value of low-frequency coupling characteristics Low-frequency coupling feature hierarchical labels and corresponding hierarchical confidence levels To achieve a quantitative characterization of the brain oxygen signal characteristics of the current analysis window.

2. The brain oxygen signal processing method based on endogenous low-frequency oscillations according to claim 1, characterized in that: Step S3 is as follows: S31: Map the amplitude envelope covariance to interval; A consistency gating parameter is set to characterize the degree of consistency between the amplitude-phase coherence quantity and the amplitude-envelope coherence quantity. Low-frequency coupling edge weights are set to construct a low-frequency coupling structure matrix. S32: Perform eigenvalue decomposition on the low-frequency coupled structure matrix, assuming the eigenvalues ​​satisfy: ; S33: Set coupling interference parameters to characterize the interference distribution law of the measurement channel in the current low-frequency coupling structure, and obtain the node interference ordered parameters; S34: Set the low-frequency coupling structure strength, calculate the energy weight of the low-frequency sub-band, and synthesize the low-frequency coupling structure strength of each low-frequency sub-band according to the weighted geometric method to obtain the low-frequency coupling characteristic parameters.

3. The brain oxygen signal processing method based on endogenous low-frequency oscillations according to claim 2, characterized in that: The consistency gating parameters in step S31 are: ； in, For the first The measurement channel and the first The measurement channel in the first Consistency gating parameters for each low-frequency subband; To find the minimum value of the function; For the first The measurement channel and the first The measurement channel in the first Amplitude and phase coordination of each low-frequency sub-band; For the first The first analysis window, the first Low-frequency subband measurement channel With measurement channel The normalized amplitude envelope co-variance between them; The parameter is the minimum value. ; ; This is the index number for the low-frequency subband.

4. The brain oxygen signal processing method based on endogenous low-frequency oscillations according to claim 2, characterized in that: The low-frequency coupling structure matrix in step S31 is: ; ; in, For the first Low-frequency coupling structure matrix under each low-frequency sub-band; For the first The first analysis window, the first Low-frequency subband measurement channel With measurement channel The low-frequency composite coupling strength between them after being constrained by consistency gating; This is to measure the total number of channels.

5. The brain oxygen signal processing method based on endogenous low-frequency oscillations according to claim 2, characterized in that: The dominant coupling strength parameter in step S32 is: ； in, For the first The first analysis window Dominant coupling strength parameters in low-frequency subbands; For the first The first analysis window The first eigenvalue of the low-frequency coupling structure matrix under each low-frequency sub-band arranged in descending order of numerical value. It is the sum of all eigenvalues; Index for measurement channel numbering.

6. The brain oxygen signal processing method based on endogenous low-frequency oscillations according to claim 2, characterized in that: The node interference ordering parameters in step S33 are: ； ； in, For the first The first analysis window Ordered parameters of node interference under a low-frequency sub-band; This is the normalized term for the total number of measurement channels, expressed as the natural logarithm. For the first The first analysis window The first low-frequency subband The coupling interference parameters of each measurement channel; For the first The first analysis window The first low-frequency subband The normalized interference distribution of each measurement channel is used to characterize the relative interference ratio of the measurement channels in the current low-frequency coupling structure; This is the sum of the coupling interference parameters of all measurement channels.

7. The brain oxygen signal processing method based on endogenous low-frequency oscillations according to claim 2, characterized in that: Step S34 is as follows: ； ； in, For the first Low-frequency coupling characteristic parameters of each analysis window; These are the coupling structure coefficients; For the first The first analysis window Low-frequency coupling structure strength under a low-frequency sub-band; The overall coupling strength; The dominant coupling strength parameter; For the ordered parameters of node interference; For the first Energy weights of each low-frequency subband within each analysis window; This represents the total number of low-frequency sub-bands. for total; For the first The measurement channel of the first A low-frequency sub-band signal.

8. The brain oxygen signal processing method based on endogenous low-frequency oscillations according to claim 1, characterized in that: Step S4 is as follows: S41: Low-frequency coupling characteristic parameters obtained in step S3 Used as the first The original brain functional state quantification parameters corresponding to each analysis window are used to realize the quantitative mapping of brain functional state. S42: Determine the reference prototype values ​​of awake characteristics based on brain functional state calibration samples. and unconscious feature reference prototype value , obtain the state quantization value of the current analysis window and the feature distance between the two state prototypes; S43: Based on the characteristic distance relationship between the current analysis window and the two types of state prototypes, determine whether the brain functional state corresponding to the current analysis window is the conscious reference prototype or the unconscious reference prototype.

9. A processing apparatus for the brain oxygen signal processing method based on endogenous low-frequency oscillations as described in any one of claims 1 to 8, characterized in that, It includes: a light source driving unit, a photoelectric detection unit, a signal acquisition unit, a main control processing unit, and a probe assembly; The light source driving unit is connected to the emitting probe in the probe assembly, and controls each near-infrared emitting probe to emit near-infrared light according to a preset driving sequence. The photoelectric detection unit is connected to the receiving probe in the probe assembly to receive near-infrared light signals and convert the received light signals into corresponding analog electrical signals for output. The signal acquisition unit is connected to the light source driving unit and the photoelectric detection unit respectively. It processes the analog electrical signals output by each receiving probe according to the light source driving timing to obtain the original light intensity signal corresponding to the measurement channel. The main control processing unit is connected to the signal acquisition unit to receive the raw light intensity signal of the measurement channel output by the signal acquisition unit, and to execute the brain oxygen signal processing method based on endogenous low-frequency oscillation to generate low-frequency coupling feature parameters and state recognition results corresponding to the brain functional state. The probe assembly includes a light-shielding part and a buffer contact part. The light-shielding part is used to reduce ambient light interference, and the buffer contact part is used to improve contact stability.

10. The processing device for the brain oxygen signal processing method based on endogenous low-frequency oscillations according to claim 9, characterized in that: The main control processing unit and signal acquisition unit include: a data preprocessing and concentration conversion module, a sliding window module, a low-frequency oscillation analysis module, a coupled structure feature calculation module, a state index generation module, and a state output module; The data preprocessing and concentration conversion module processes the original light intensity signal and converts the preprocessed light intensity signal into a hemoglobin concentration change signal; The sliding window module performs sliding window processing on the hemoglobin concentration change signal; The low-frequency oscillation analysis module performs low-frequency sub-band decomposition on the measurement channel window signal within each analysis window to obtain the sub-band signal on the low-frequency sub-band, performs transformation, and extracts instantaneous phase information and amplitude envelope information to obtain the analytical signal, instantaneous phase, and amplitude envelope. The coupled structure feature calculation module calculates the cooperative relationship between measurement channels based on the instantaneous phase information and amplitude envelope information between the measurement channels; The state index generation module generates low-frequency coupling characteristic parameters corresponding to the current analysis window based on the overall coupling strength, the dominant coupling strength parameter node interference ordered parameter, and the weights of each low-frequency sub-band. The state output module is used to perform state quantization mapping, state prototype distance discrimination, and discrimination confidence output on the low-frequency coupled feature parameters output by the state index generation module.

Images