Evaluation system and evaluation method for electroencephalogram signal acquisition equipment

Abstract

The invention discloses an electroencephalogram signal acquisition equipment evaluation system and evaluation method. The system comprises a standard control signal generation module used for generating a standard control signal; the evaluation module is used for acquiring an electroencephalogram acquisition signal acquired by the electroencephalogram signal acquisition equipment, processing the electroencephalogram acquisition signal, displaying the electroencephalogram acquisition signal and a standard contrast signal, comparing waveforms and parameters of the electroencephalogram acquisition signal and the standard contrast signal, and determining whether the electroencephalogram acquisition signal and the standard contrast signal are in all sample points or not; the number of the samples with the difference relative amplitude exceeding the set threshold value accounts for the total number of the samples. The signal distortion degree is evaluated by calculating the normalized amplitude difference percentage, the difference between the collected signal and the standard contrast signal is comprehensively considered, the distortion condition is quantified, and a reliable index is provided for signal quality evaluation.

Description

Technical Field

[0001] This application relates to the technical field of electroencephalogram (EEG) signal processing, and particularly to an EEG signal acquisition device evaluation system and an evaluation method. Background Art

[0002] Electroencephalogram (EEG) is an important means of monitoring brain function by recording the electrical activities on the scalp. EEG is widely used in medical diagnosis, neuroscience research, brain-computer interface (BCI) and other fields. During the analysis of EEG signals, the time resolution of the signals is high. However, due to factors such as external interference, poor electrode contact, and signal processing algorithms, the acquired EEG signals often have problems such as noise, distortion, and phase deviation.

[0003] In the early production of EEG devices, in order to confirm the errors of each channel of the trial-produced EEG acquisition device, quality inspection personnel need to frequently operate each channel manually to evaluate the frequency domain error, phase error, and amplitude error of the device to be inspected, so as to facilitate production personnel to perform calibration detection and quality control in a timely manner. For example, production personnel often need to directly compare the differences between the standard reference signal and the acquired signal, but there is a lack of such an integrated detection system.

[0004] Traditionally, the detection of EEG acquisition devices is that production testers compare them through simple amplitude differences manually, and there is a lack of supporting quantitative precision detection means and system solutions to assist the work of production testers. Summary of the Invention

[0005] The technical objective of this application is to provide an EEG signal acquisition device evaluation system and an evaluation method for the current situation that mainly relies on manual testing of each channel of EEG signal acquisition devices and lacks an evaluation system that can directly batch and automatically compare the differences between standard reference signals and acquired signals.

[0006] To achieve the above technical objective, this application adopts the following technical solutions.

[0007] In the first aspect, an embodiment of this application provides an EEG signal acquisition device evaluation system, which is used to evaluate an EEG signal acquisition device. The evaluation system includes: A standard reference signal generation module, which is used to generate a standard reference signal; An evaluation module, which is used to obtain the EEG acquisition signal obtained by the EEG signal acquisition device, display the EEG acquisition signal and the standard reference signal, compare the waveforms and parameters of the EEG acquisition signal and the standard reference signal, and determine the percentage of the number of samples whose relative amplitude of the difference exceeds a set threshold in all sample points between the EEG acquisition signal and the labeled reference signal in the total number of samples.

[0008] Further, the system further includes a signal correction module; The signal correction module is configured to perform Fourier transform on the EEG acquisition signal and the standard reference signal, calculate the phase difference between the EEG acquisition signal and the standard reference signal after the transform, and use a sampling time scale factor sequence to phase-align the EEG acquisition signal and the standard reference signal according to the phase difference, so as to obtain a corrected EEG acquisition signal.

[0009] Further, the system further includes: an EEG characteristic waveform generation engine, which has an international standard EEG waveform signal template library built in, and is configured to select a corresponding template signal from the template library according to the type of the required signal and relevant physiological parameters, and the template signal can be used to generate a standard reference signal, so that the standard reference signal generation module generates the standard reference signal according to the standard reference signal.

[0010] Still further, the EEG characteristic waveform generation engine also has an intelligent parameter mapping rule library built in, which is configured to match the intelligent parameter mapping rule library according to the input physiological indicators to obtain key characteristic parameters; and use the key characteristic parameters to adjust the parameters of the template signal to generate the template signal that meets the target characteristics.

[0011] Further, the system further includes a standard reference signal generation module, which is configured to generate a standard reference signal, so that the standard reference signal generation module generates the standard reference signal according to the standard reference signal; The standard reference signal generation module includes an FPGA, a DSP, a DAC, a low-noise operational amplifier signal chain module, a power management module, and a physical isolation module; The FPGA is configured to generate waveform data; The DSP is communicatively connected to the FPGA, and the DSP is configured to preprocess and optimize the quality of the waveform data by controlling the PGA gain of the analog front end; The FPGA transmits the waveform data to the DAC, and the DAC is configured to convert the waveform data into an analog signal; The low-noise operational amplifier signal chain module is configured to amplify the analog signal; The power management module is configured to supply power to the FPGA, the DSP, the DAC, and the low-noise operational amplifier signal chain module; The physical isolation module adopts a three-layer PCB layout, with the top layer transmitting analog signals, the middle layer being a ground plane, and the bottom layer arranging digital circuits. The analog ground and the digital ground are connected by a 0Ω resistor at a single point, and the external uses an aluminum alloy shell and is sprayed with conductive paint.

[0012] Further, the system further includes an improved average filter, which is used to perform average filtering on the EEG acquisition signal.

[0013] In a second aspect, an evaluation method for an EEG signal acquisition device evaluation system provided by any possible implementation manner of the first aspect of the present application includes: For each sample point n , determine the ratio of the absolute value of the difference between the EEG acquisition signal of the sample point and the standard reference signal to the maximum of the absolute values of the two. ; Wherein, x gen n represents the EEG acquisition signal of the sample point n , x temp n represents the standard reference signal of the sample point n , max () is the maximum value function; Traverse all sample points. If the ratio is greater than or equal to a preset value, the sample point meets the distortion condition. Count the sample points that meet the distortion condition to obtain a count result Count Δp ; The ratio of the count result Count Δp to the number of all sample points is the percentage.

[0014] Further, when the system further includes a signal correction module; the method further includes: Use the signal correction module to convert the EEG acquisition signal and the standard reference signal into frequency domain signals; Compare the phase values of the EEG acquisition signal and the standard reference signal at the same frequency, and calculate the phase difference between the two; Use the sampling time scale factor sequence to phase-align the EEG acquisition signal and the standard reference signal according to the phase difference to obtain the corrected EEG acquisition signal.

[0015] Further, when the system further includes an improved average filter; The method further includes: According to the physical characteristics of different frequency bands of the EEG acquisition signal, adopt a Gaussian weighting strategy optimization and frequency band parameter dynamic adaptation method to assign different weights to the sample points in the neighborhood; Use the improved average filter to perform weighted average calculation on the sample points in the neighborhood according to the assigned weights to obtain the filtered EEG acquisition signal value.

[0016] ​​Further, the expression of the improved average filter is as follows: ; where y n represents the filtered signal value at the sample point n . x n + k represents the input signal value of the n -th sample point within the neighborhood centered on the sample point k . k ranges from - N / 2 to N / 2 , N represents the number of sample points extended to both sides centered on the sample point n , and N determines the size of the neighborhood; ω k represents the weight of the n -th sample point within the neighborhood at the sample point k .

[0017] Compared with the prior art, the following beneficial technical effects are achieved by the EEG signal acquisition device evaluation system and evaluation method provided by the embodiments of the present application: By calculating the percentage interval (i.e., the normalized amplitude difference percentage) of the number of samples whose relative amplitude of the difference exceeds a set threshold among all sample points between the EEG acquisition signal and the labeled reference signal, the signal distortion degree is evaluated, and the differences between the EEG acquisition signal and the standard reference signal are comprehensively considered to quantify the distortion situation, providing a reliable index for the signal quality evaluation of the EEG signal acquisition device. The Fourier transform and phase difference calculation correction module uses Fourier transform to obtain the signal phase information, calculates the phase difference and corrects the received signal (i.e., the EEG acquisition signal) accordingly, aligns the phase of the received signal with the standard reference signal, synchronizes in the time domain, and improves the signal accuracy. The improved average filter dynamically allocates neighborhood weights according to the characteristics of different frequency bands of EEG signals. For example, for low-frequency δ waves, the high-weight area is expanded to smooth the baseline drift, and for high-frequency γ waves, the edge weights are rapidly attenuated to suppress high-frequency noise, effectively suppressing noise spikes and improving the signal purity; it can process EEG signals in different frequency bands and different physiological states. Whether it is low-frequency slow waves, high-frequency fast waves, or normal and abnormal physiological state signals, it can be effectively evaluated, and the applicable range is wide. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] ​​​The accompanying drawings described herein are for illustrative purposes only and are not intended to limit the scope of the disclosure of the present application in any way. Additionally, the shapes and proportional dimensions of the various components in the figures are merely schematic and are used to assist in the understanding of the present application, rather than specifically defining the shapes and proportional dimensions of the various components of the present application. Those skilled in the art can, under the teachings of the present application, select various possible shapes and proportional dimensions according to specific circumstances to implement the present application. In the accompanying drawings: Figure 1 Schematic diagram of the structure of the electroencephalogram signal acquisition device evaluation system provided for the embodiment; Figure 2 Schematic diagram of the template signal generated by the electroencephalogram feature waveform generation engine in the electroencephalogram signal acquisition device evaluation system provided for the embodiment; Figure 3 Hardware flowchart of the standard reference signal generation module in the electroencephalogram signal acquisition device evaluation system provided for the embodiment of the application; Figure 4 External view schematic diagram of the standard reference signal generation module in the embodiment; Figure 5 Schematic diagram of the amplifier part of the electroencephalogram signal acquisition device in the embodiment; Figure 6 Electroencephalogram cap part of the electroencephalogram signal acquisition device in the embodiment; Figure 7 External view schematic diagram of the standard comparison signal generation module in the embodiment; Figure 8 Flowchart of the evaluation method of the electroencephalogram signal acquisition device evaluation system provided for the embodiment; Figure 9 Schematic diagram of the result with the phase and amplitude basically aligned output by the evaluation module in the embodiment; Figure 10 Schematic diagram of the result with the phase misaligned and the amplitude aligned output by the evaluation module in the embodiment; Figure 11 Schematic diagram of the result with the phase aligned and the amplitude misaligned output by the evaluation module in the embodiment; Figure 12 Schematic diagram of the result with the phase aligned and the amplitude slightly deviated output by the evaluation module in the embodiment; Figure 13 Schematic diagram of the output interface of the evaluation module in the embodiment. Detailed implementation manners

[0019] To enable those skilled in the art to better understand the technical solutions in this application, the following will clearly and completely describe the technical solutions in the embodiments of this application in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the scope of protection of this application.

[0020] As Figure 1 shown, the embodiment of this application provides an electroencephalogram (EEG) signal acquisition device evaluation system for evaluating an EEG signal acquisition device. The evaluation system includes a standard reference signal generation module and an evaluation module. Among them, the standard reference signal generation module is used to generate a standard reference signal.

[0021] The evaluation module is used to obtain the EEG acquisition signal obtained by the EEG signal acquisition device, display the EEG acquisition signal and the standard reference signal obtained by the EEG signal acquisition device, compare the waveforms and parameters of the EEG acquisition signal and the standard reference signal, and determine the percentage of the number of samples whose relative amplitude of the difference exceeds a set threshold among all sample points between the EEG acquisition signal and the labeled reference signal (i.e., the normalized amplitude difference percentage).

[0022] The EEG signal acquisition device is responsible for collecting EEG signals from the brain of the test subject. These signals reflect the electrical activities of brain neurons and are of great significance for studying brain functions and diagnosing nervous system diseases. The EEG signal acquisition device may include an EEG cap (such as Figure 6 ), the EEG cap includes electrodes, and the electrodes are made of high-conductivity materials and have good biocompatibility and conductivity, such as Figure 6 the gray electrode part. The cap body of the EEG cap is like Figure 6 the hat part.

[0023] The EEG signal acquisition device also includes an EEG cap amplifier (such as Figure 5 ), and the amplifier is used to connect to the EEG cap to amplify weak EEG signals and send them to the PC. At the same time, it suppresses noise, and the evaluation module set on the PC can be used to evaluate the accuracy of the EEG signals collected by the EEG signal acquisition device. Optionally, the EEG signal acquisition device also includes filters, such as low-pass, high-pass, and band-pass filters, for removing unwanted frequency components.

[0024] During use, the electrodes are placed on the scalp of the test subject and connected to the EEG cap, and the cap is connected to the amplifier through a wire. The amplifier amplifies the weak EEG signals and removes noise and interference through filters. Then, the amplifier converts the analog signals into digital signals, and these digital signals are transmitted to the data processing unit through the data interface.

[0025] Since the acquisition accuracy (frequency, amplitude, phase) of the electroencephalogram (EEG) signal acquisition device is unknown before leaving the factory, the device needs to acquire a standard reference signal (which can be based on the standard reference signal generated by the standard reference signal generation module), and compare the acquired EEG acquisition signal with the standard reference signal for subsequent processing (calibration or repair) of the acquisition signal.

[0026] Before receiving the EEG acquisition signal, the standard reference signal generation module generates a standard reference signal. The standard reference signal generation module (the appearance can be as shown in Figure 7 ) is a signal generation instrument running on a specific high-performance computing device. It can create standard reference signals for testing through arithmetic methods. These standard reference signals can be generated according to the standard reference signal. The standard reference signal is an arithmetic ideal signal used for comparison during evaluation, such as an absolutely pure square wave, sine wave, and triangular wave. These standard reference signals can help verify the effectiveness and accuracy of subsequent processing algorithms.

[0027] In an embodiment, the standard reference signal generation module may include a scientific computing module and a delay adjustment module. The scientific computing module is used for scientific computing to generate a dedicated standard reference signal for the standard signal. The delay adjustment module allows the user to adjust the delay of the signal to match the time characteristics of EEG signal acquisition.

[0028] The standard reference signal generated by the standard reference signal generation module can be copied to the standard reference signal generation module to generate a standard reference signal. The user can adjust the delay of the signal as needed to ensure that the standard reference signal is synchronized with the acquired EEG acquisition signal in time. The adjusted standard reference signal is transmitted to the evaluation module (such as the EEG signal acquisition accuracy evaluation software) for comparison and evaluation.

[0029] The evaluation module is responsible for analyzing and evaluating the acquired EEG signal. This module calculates metrics such as errors by comparing the acquired EEG signal with the standard reference signal to evaluate the accuracy of the acquisition signal.

[0030] As an example, the software modules included in the evaluation module are: Signal comparison panel: This software panel is used to select the signals of the channels to be measured. It can compare the waveforms of the EEG acquisition signal and the standard reference signal in real-time plotting mode and visually display each core parameter.

[0031] Error calculation unit: Calculate the difference between the EEG acquisition signal and the standard reference signal, such as amplitude error and phase error, and correct the error of the waiting comparison signal for convenient calculation.

[0032] Correlation analysis unit: Evaluate the correlation between signals to determine the similarity of signals. For example, in some embodiments, the method of percentage of normalized amplitude difference interval is adopted.

[0033] The evaluation module inputs the collected EEG signals and standard reference signals into the signal comparison panel. The comparator analyzes the waveforms and parameters of the two signals, and the error calculation unit calculates the differences between the signals, such as amplitude error, phase error, etc. The signals after error correction will be subjected to correlation analysis. Finally, the evaluation result report is displayed in the form of a chart for the user's reference.

[0034] In the embodiment, the evaluation module may include several modules written using several software libraries to respectively implement a data loading module, a data cleaning module, and a data transformation module. First, the signal data is loaded from a CSV file, and columns 1 to n -1 are defined as the EEG acquisition signals received by n -1 channels, and column n is defined as the reference standard reference signal, and the standard reference signal is a reference signal calculated by a function based on the standard reference signal for evaluation.

[0035] In some embodiments, the evaluation system further includes a signal correction module; the signal correction module is used to perform Fourier transform on the EEG acquisition signal and the standard reference signal, calculate the phase difference between the EEG acquisition signal and the standard reference signal, use a sampling time scale factor sequence (the sampling time scale factor sequence is a time linear sequence related to the sampling rate), and align the phases of the EEG acquisition signal and the standard reference signal according to the phase difference to obtain the corrected EEG acquisition signal. The phase difference calculation is completed by comparing the main frequencies of the signals, and the core of this step lies in how to extract useful phase information from the frequency domain information. Based on the calculation result of the phase difference, a certain sampling time scale factor sequence is used to correct the received signal. This step converts the phase difference into an angular offset and applies it to the signal to generate the corrected EEG acquisition signal.

[0036] As an example, the generated corrected EEG acquisition signal is saved as a new CSV file, and the evaluation module is used to further analyze its differences from the standard reference signal. Especially within a certain range of ±, the numerical ratio of the signals is statistically analyzed. The corrected signal is compared with the original EEG acquisition signal (received signal) and the standard reference signal, and finally a chart is generated and saved as a PNG file.

[0037] In the fields of scientific research and medical treatment, generating accurate EEG waveforms that meet international standards has been a long-standing challenge. Traditional methods require manual adjustment of waveform parameters and it is difficult to simulate EEG signals that meet specific physiological conditions, which is not only time-consuming and laborious but also may lead to experimental errors. In addition, accurately simulating abnormal EEG signals (such as spikes and sharp waves in epilepsy patients) is particularly important for the testing of medical devices and the training of anti-interference algorithms, but existing simulation technologies often cannot achieve the required accuracy and stability. To address these difficulties, such asFigure 1 and Figure 2 As shown in Figure 2 , in some embodiments, the system further includes an electroencephalogram (EEG) feature waveform generation engine. The EEG feature waveform generation engine has an international standard EEG waveform signal template library built-in, which is used to select corresponding template signals from the template library according to the type of the required signal and relevant physiological parameters. The template signals can be used as standard reference signals, so that the standard comparison signal generation module can generate standard comparison signals based on the standard reference signals.

[0038] In some embodiments, the EEG feature waveform generation engine also has an intelligent parameter mapping rule library built-in, which is used to match the intelligent parameter mapping rule library according to the input physiological indicators to obtain key feature parameters; and use the key feature parameters to adjust the parameters of the template signals to generate template signals that meet the target features.

[0039] The EEG feature waveform generation engine is actually middleware software running on a PC. It has an international standard EEG waveform template library built-in, supports users to customize physiological parameters, and can automatically match and adjust the amplitude, frequency and noise floor of the EEG waveform to simulate EEG signals in a real environment. Through the collaborative optimization of FPGA and DSP, the EEG feature waveform generation engine can calculate and adjust the waveform in real time to achieve precise phase jitter and tiny amplitude adjustment. By inputting into the standard signal generating device, EEG signals with specified features can be generated. This intelligent generation method not only significantly shortens the experimental preparation time, but also improves the accuracy and stability of the signals. In addition, the EEG feature waveform generation engine can simulate abnormal EEG signals in a specific scenario, such as the discharge activity of epilepsy patients, which is crucial for training and verifying the anti-interference performance of medical devices.

[0040] In some embodiments, the system further includes an improved average filter, which is used to perform average filtering on the EEG acquisition signals.

[0041] In some embodiments, the improved average filter and the signal correction module can be built into the evaluation module.

[0042] In some embodiments, the system further includes a standard reference signal generation module for generating standard reference signals, so that the standard comparison signal generation module can generate standard comparison signal blocks based on the standard reference signals. As Figure 3 and Figure 4As shown, in the embodiment, the standard reference signal generation module is a standard reference signal generation system that combines FPGA and DSP. The standard reference signal generation device can generate electrical signals of several channels at a micro amplitude level with regular periods, such as standard reference signals like square waves, sine waves, and triangular waves. The generated signals can be input into the standard comparison signal generation module to generate standard comparison signals for evaluating and calibrating the internal parameters of the electroencephalogram (EEG) data acquisition device. At the same time, it can receive the template library and intelligent parameter mapping rules built into the EEG feature waveform generation engine on the PC and send specified waveforms.

[0043] The design requirements of the standard signal generation module are to provide high-stability and high-precision signal output to ensure the accuracy of subsequent signal acquisition and comparison. For example Figure 7 As shown, the standard signal generation module is a signal generation instrument responsible for generating a series of standard reference signals with known characteristics. These signals usually include sine waves, square waves, triangular waves, etc., and have precise frequency, amplitude, and phase characteristics. The standard signal generation module realizes frequency accuracy and multi-mode through the collaborative processing module of FPGA + DSP, realizes nano-volt level accuracy and low-noise output through the signal chain of 20-bit DAC + low-noise operational amplifier, and realizes anti-interference through multi-stage power supply filtering + electromagnetic shielding, achieving the three core requirements for EEG signal generation: 1. High precision: 20-bit DAC + segmented calibration, amplitude error < ±0.1% (nano-volt level).

[0044] 2. High purity: Noise floor < 500pVrms, distortion < 0.001% (EEG frequency band).

[0045] 3. High adaptability: High-impedance output / battery-powered / multi-mode waveform reception, matching EEG research and clinical scenarios.

[0046] Among them, the hardware modules included in the standard signal generation module are: I. Collaborative processing module of FPGA + DSP: DSP + FPGA ensure that the phase truncation error of low-frequency signals is small and the frequency stability is high. The integrated kernel of DSP supports real-time noise generation and modulation operations, meeting the requirements of complex waveforms.

[0047] Adopt a 20-bit high-precision DAC (digital-to-analog converter), combined with dynamic range extension technology, to achieve an ultra-wide amplitude output of 1nV - 10mV (6 orders of magnitude dynamic range). Through the segmented calibration algorithm (automatically calibrate offset / gain for every 100mV range), combined with the programmable gain amplifier (PGA) technology, achieve a 1nV resolution (full scale 10mV / 2 20). The non - linear distortion of the DAC is ≤0.001%. With a precise reference source (temperature coefficient < 1ppm / °C) and an error compensation algorithm, the amplitude error is controlled within ±0.1% (+1nV reference error). The signal chain design adopts a multi - stage amplification strategy: first, the PGA is used for pre - processing of nano - volt - level small signals, and then the low - noise operational amplifier is used to complete power driving, avoiding small signals being submerged by noise during transmission.

[0048] Among them, the FPGA part is responsible for: Real - time waveform generation: Based on DDS technology, with a built - in 32 - bit phase accumulator, it has high frequency resolution, supports μHz - level stepping, and can generate waveforms covering the entire electroencephalogram (EEG) frequency band.

[0049] High - speed data pre - processing: Caching waveform data, supporting dual - channel independent phase adjustment, and meeting the simulation of coherent / non - coherent EEG signals.

[0050] Multi - mode control: Communicating with the DSP through the AXI bus to achieve AM / FM modulation, noise superposition, and simple multi - waveform synthesis (such as sine, square, triangle waves, etc.).

[0051] Among them, the DSP part is responsible for: Algorithm processing: Running digital filtering and distortion compensation algorithms to ensure waveform quality.

[0052] Human - machine interaction: Connecting to the middleware, parsing PC - side instructions, updating waveform parameters in real - time, and controlling the PGA gain of the analog front - end.

[0053] II. DAC and low - noise operational amplifier signal chain module: Among them, the DAC module is part of the signal chain design and is used for nano - volt - level accuracy and low - noise core.

[0054] A 20 - bit high - precision DAC is adopted, supporting a wide dynamic range of 1nV - 10mV. An automatic range calibration algorithm (segmented every 100mV) is built - in, and the EEPROM stores offset / gain compensation coefficients. The amplitude error is < ±0.1% (+1nV reference error). The programmable gain amplification (PGA) technology (1 - 1000 times) is used to achieve pre - processing of nano - volt - level small signals, avoiding small signals being submerged by the noise of the operational amplifier. The combination of 20 - bit DAC + PGA has a resolution of 1nV (10mV / 2 20 ), meeting the precise generation of EEG signals (μV - nV level).

[0055] Among them, the analog front - end is part of the signal chain design and is responsible for low - noise amplification and filtering. It is equipped with a precise reference source (temperature coefficient < 1ppm / °C).

[0056] Buffered by a low - noise operational amplifier, with a driving ability of 20mA, ensuring the stability of the DAC reference.

[0057] The secondary is equipped with a low-noise operational amplifier circuit.

[0058] The differential-to-single-ended module constructs a differential amplifier circuit with a high input impedance (>10 GΩ) to match the high-impedance characteristics of EEG electrodes and suppress common-mode noise.

[0059] The last stage is equipped with a three-stage filtering system: ① Pre-filtering: LC low-pass (cutoff at 100 kHz) to filter out DAC switching noise; ② Active filtering: Second-order Sallen-Key low-pass (cutoff at 200 Hz), combined with digital filtering to achieve a roll-off of -80 dB / decade; ③ Noise cancellation: Active noise cancellation circuit (notch depth of 50 / 60 Hz > 60 dB) to suppress power supply ripple.

[0060] III. Power management module: The three-stage voltage regulation and battery redundant power supply architecture provides power through an AC adapter or a lithium battery. The isolated power supply uses digital isolation and buck technology, with an isolation noise greater than 150 dB. The precision voltage regulation part includes an analog power supply for the operational amplifier and a digital power supply for the FPGA / DSP, and these power supplies have extremely low noise characteristics. Ripple suppression is achieved by paralleling tantalum capacitors and ceramic capacitors at each chip power supply pin, with a ripple less than 1 μVrms (20 MHz bandwidth). The low-noise design ensures pure power supply for the reference source and the operational amplifier through multi-stage filtering in the battery-powered mode, reducing 1 / f noise.

[0061] IV. Physical isolation module: In terms of physical isolation, a three-layer PCB layout is adopted. The top layer is for analog signals, the middle layer is the ground plane, and the bottom layer is for digital circuits. The analog ground and the digital ground are connected at a single point through a 0 Ω resistor. The external uses an aluminum alloy shell (with a thickness of 2 mm), and the inside is sprayed with conductive paint, with a shielding effectiveness greater than 80 dB (10 kHz - 1 GHz). In terms of signal isolation, digital control signals are isolated through optocouplers to avoid digital noise coupling to the analog circuit. The Ethernet interface uses a pulse transformer for isolation to isolate common-ground noise. This anti-interference design comprehensively suppresses EMI / RFI interference through shielding, isolation, and hierarchical layout, ensuring that nanovolt-level signals are not polluted by environmental noise.

[0062] The embodiment of the present application also provides an evaluation method for the EEG signal acquisition device evaluation system as provided in the above embodiments, as Figure 8 shown.

[0063] In some embodiments, the evaluation module can perform data processing on the EEG acquisition signals obtained by the EEG signal acquisition device, including data loading, data cleaning, data conversion, etc. This module is used to process CSV files and provides powerful database framework operation functions, making it very convenient to read data from files, extract columns and rows, and perform simple data cleaning operations.

[0064] As an example, first, the evaluation module extracts EEG acquisition signal data from the file. Assuming that each signal is sampled as equally spaced time series data, the signal can be regarded as a discrete function in the time domain. After reading the data, it is first necessary to ensure that the EEG acquisition signal is clean, such as removing duplicate rows, missing values, noise, etc. Generally, common denoising methods include moving average filters, median filters, etc. If there are obvious outliers in the data, they can be identified and removed by calculating the deviation of each point relative to its neighboring points. Here, an improved average filter and median filter are proposed, which can better handle noise.

[0065] The read_csv() function directly reads the CSV file and extracts the required columns from the data through column positioning operations. received_data contains the received n columns of EEG acquisition signals collected from the EEG device, while true_data is the standard control signal as a reference value. The whole process is simple and efficient, demonstrating the convenience of this module in data processing.

[0066] In the embodiment, determine the percentage of the number of samples in all sample points where the relative amplitude of the difference between the EEG acquisition signal and the labeled control signal exceeds the set threshold (i.e., the percentage of the amplitude difference normalized by the interval), and calculate the signal correlation through this percentage of the amplitude difference normalized by the interval.

[0067] Calculating the amplitude difference is the simplest method to evaluate the amplitude gap between two sets of signals. Traditionally, the amplitude difference is usually calculated by simple difference comparison. In the evaluation module of the embodiment, the percentage index of the amplitude difference normalized by the interval is determined. In production practice, this index can more accurately and quickly measure the amplitude difference and phase difference between two signals in a period of time interval, which is very beneficial for the automatic detection of faulty or defective EEG acquisition devices.

[0068] Traditional amplitude difference calculation only relies on the simple subtraction of signal values and does not consider the change amplitude of signal phase at different moments. For example, when the amplitudes of two signals are at different orders of magnitude, the simple difference calculation alone cannot effectively reflect the similarity degree of the two signals. Affected by the period and amplitude, the random simple difference may be large or small. Therefore, the traditional method is prone to mislead in the comparison of signals with different orders of magnitude. To overcome this limitation, an improved method of calculating the percentage of amplitude difference with interval normalization is adopted in the embodiment, including: for each sample point n , determine the absolute value of the difference between the EEG acquisition signal and the standard reference signal at the sample point, relative to the ratio of the maximum of the absolute values of the two. Traverse all sample points. If the ratio is greater than or equal to the preset value P, the sample point meets the distortion condition. Count the sample points that meet the distortion condition to obtain the counting result Count Δp ; Count Δp The ratio of the counting result ; to the number of all sample points is the percentage (percentage of amplitude difference with interval normalization). Its expression is as follows: x gen n represents the EEG acquisition signal of the sample point n , x temp n represents the standard reference signal of the sample point n , max () is the maximum value function; | x gen n - x temp n | respectively represent the amplitudes of the received signal (i.e., EEG sampling signal) and the standard reference signal at the sampling point n (at time n ), Count Δp counts the number of signals with amplitude limited by the preset value P amplitude range. The preset value P amplitude is a limited amplitude, indicating the signal accuracy range that should be achieved after acquisition, Count all counts the total number of received signals. max(| x gen n |,| x temp n |) represents the maximum amplitude of the two signals at this moment and is used for calculating the within-group difference normalization.

[0069] ​​​​​​One of the advantages of this method is that it clarifies the concept of time interval. It is necessary to collect and record the standard reference signal and the EEG acquisition signal over a long period. Total large interval analysis can be carried out according to different needs, or key interval analysis can be carried out, which is especially suitable for long-range EEG datasets.

[0070] Secondly, the normalization of the within-group differences solves the error caused by the inconsistent amplitude when multiple groups are compared, especially suitable for processing signals with a large amplitude change range such as electroencephalogram (EEG) signals. (It may be necessary to compare signals of 1 mV or 40 μV, avoiding interference from the range configuration.) Finally, by calculating the relative percentage, the compliance probability level of the signal can be better judged. In the underlying signal verification scenario of EEG signal acquisition equipment, it can intuitively and accurately show a lot of physical meanings, that is, whether there is a loss in the number of discrete digital signals obtained by the standard signal after passing through the acquisition system, how many qualified signals are there in total within the specified time interval and the allowed fluctuation range, and what is the proportion of the total number of signals, which reflects the compliance level of this EEG acquisition system.

[0071] In the EEG signal analysis scenario, the calculation of the percentage of the amplitude difference of interval normalization also has certain significance, which can reveal the subtle changes in brain activities in different states. For example, when comparing pathological EEG signals with healthy EEG signals, this percentage difference can provide quantitative support for clinical diagnosis, and combined with the spatial position of the channels, it can also help doctors better identify the lesion area or abnormal activities in specific brain regions.

[0072] In some embodiments, the evaluation system includes a signal correction module, and the evaluation method of the evaluation system further includes the following method: using the signal correction module to convert the EEG acquisition signal and the standard reference signal into frequency domain signals; comparing the phase values of the EEG acquisition signal and the standard reference signal at the same frequency, and calculating the phase difference between the two; using the sampling time scale factor sequence, aligning the phases of the EEG acquisition signal and the standard reference signal according to the phase difference, so as to achieve phase compensation and obtain the corrected EEG acquisition signal.

[0073] The calculation of the phase difference depends on the Fourier transform. The Fourier transform can convert the time-domain signal into a frequency-domain signal, which is convenient for analyzing the frequency components and phase differences of the signal. The fft function can directly calculate the fast Fourier transform of the signal and return an array of complex numbers in the frequency domain. The phase information can be obtained through the angle (phase angle) of the complex number, that is, using the angle function.

[0074] ; Here, the phase information is extracted from the first frequency component (ignoring the DC component) of the Fourier transform result, and the phase difference between the two is calculated. This phase difference is used to measure the frequency asynchrony of the signal and provide a basis for subsequent phase correction.

[0075] Specifically, the Fourier transform is a common tool for converting a signal from the time domain to the frequency domain. It can decompose a time-varying signal into a series of sine waves of different frequencies. The formula for the Fourier transform is as follows: ; where: x n is the original time series signal, N is the total length of the signal. F ( f ) is the frequency domain representation of the signal, representing the amplitude and phase information at different frequencies f . The in the exponential term is a complex exponential, representing the weighting term for each frequency component. The result of the Fourier transform is a complex number, where the real part and the imaginary part represent the amplitude and phase of the signal at a specific frequency respectively. By converting the complex number to polar coordinate form, the amplitude and phase information can be obtained.

[0076] The amplitude can be calculated by the following formula: ; The phase can be calculated by the following formula: ; The atan2 function here is a two-parameter arctangent function, which is used to ensure that the calculated phase angle is in the correct quadrant.

[0077] After obtaining the Fourier transform results of the received signal (i.e., the EEG acquisition signal) and the standard reference signal, the next step is to calculate the phase difference between them. The calculation of the phase difference is based on the difference in the phase values of the two signals at the same frequency: ; This phase difference is usually calculated at the main frequency of the signal, that is, the frequency component with the largest amplitude is selected. The selection of the main frequency can be determined by finding the maximum value in the Fourier transform result.

[0078] The goal of phase correction is to align the phase of the received signal with the phase of the standard signal. The idea of phase correction is to adjust the received signal according to the calculated phase difference so that it is synchronized with the standard signal in the time domain. The way of correction is to apply a corresponding phase shift to the signal. Assuming the phase difference is Δθ , then the signal can be corrected in the following way: ​ ; The intuitive meaning of this formula is that by adjusting the phase of the received signal, it is aligned with the reference signal. The corrected signal has the same phase as the standard signal in time, thus reducing the phase error between the two.

[0079] For the magnitude of the phase difference, which is usually an angular value, it can be converted into an offset in time. If the sampling frequency is f s , and the time length of the signal is T , then the phase difference Δθ can be converted into a time difference Δt : ; where f is the main frequency. This time difference Δ t represents the displacement that the signal needs to make in time to align with the reference signal.

[0080] As an example, in this embodiment, the sampling time scale factor sequence can be expressed as: t = np.arange(N) / fs; where np.arange(N) is used to create an array of length N , whose elements are integers from 0 to N - 1, and these integers represent the indices of each sampling point. After dividing this array by the sampling frequency fs, the signal time vector t in seconds is obtained. This time vector t is used to represent the moment corresponding to each sampling point. If each element in the sampling time scale factor sequence t is regarded as a sampling time scale factor, the sampling time scale factor sequence t represents the time from the start moment of the signal to each sampling point. In other words, t is a linearly increasing time series related to the sampling rate, accurately reflecting the sampling moments of the signal.

[0081] During the phase compensation process, in this embodiment, complex multiplication is used to rotate the phase of the signal. Since the real and imaginary parts of the signal are composed of cosine and sine parts respectively, using complex multiplication for phase adjustment can not only effectively achieve signal alignment but also avoid directly operating on the signal phase, ensuring the simplicity and efficiency of the signal alignment operation.

[0082] The specific phase alignment method in the embodiment is implemented through the following formula: aligned_signal = signal * np.cos(-phase_diff) - np.imag(signal *np.sin(-phase_diff)); This formula uses a phase compensation method in complex form to achieve alignment by rotating the signal (i.e., phase adjustment). In the NumPy library of Python, np.sin() and np.cos() are functions used to calculate sine and cosine values respectively. Among them, np.sin(-phase_diff) calculates the sine value of the negative phase difference -phase_diff, and np.cos(-phase_diff) calculates the cosine value of the negative phase difference -phase_diff.

[0083] sin(-phase_diff) and cos(-phase_diff) use the negative phase difference -phase_diff to adjust the signal phase. cos(-phase_diff) and sin(-phase_diff) correspond to the real and imaginary parts of the phase difference respectively. Based on the calculation results of these functions, the signal phase can be rotated in the negative direction by phase_diff, thereby achieving signal alignment.

[0084] The operation of signal * np.cos(-phase_diff) adjusts the phase of the real part of the signal, that is, the amplitude of the signal is multiplied by np.cos(-phase_diff).

[0085] np.imag(signal * np.sin(-phase_diff)) uses the np.imag() function to obtain the imaginary part of the signal. In fact, this is to adjust the phase of the imaginary part of the signal. The principle is similar to that of the real part adjustment, but it is necessary to extract the imaginary part first and then multiply it by np.sin(-phase_diff).

[0086] In electroencephalogram (EEG) signal processing, various noise interferences inevitably exist in the signal. The traditional average filter smooths the signal by calculating the simple mean of data points, but this method is prone to losing valuable detailed information when processing high-noise or complex signals. Therefore, in the evaluation system of the EEG signal acquisition device provided in the embodiments of the present application, an improved average filter is included to improve the filtering accuracy.

[0087] In some embodiments, the evaluation method further includes: the improved average filter adopts a Gaussian weighting strategy and a dynamic adaptation method for frequency band parameters according to the physical characteristics of different frequency bands of the EEG acquisition signal, and assigns different weights to the sample points in the neighborhood; uses the improved average filter to perform a weighted average calculation on the sample points in the neighborhood according to the assigned weights to obtain the filtered EEG acquisition signal value.

[0088] In EEG signal processing, due to the "simple mean calculation" characteristic of traditional average filtering, it is easy to blur the key details of the signal. The improved weighted average filtering effectively alleviates this problem by allocating weights to each EEG frequency band and increasing the weights for the concerned frequency bands. The following is an explanation in combination with formulas and EEG scenarios.

[0089] The formula for the improved mean filter is as follows: ; Among them, x n + k is the original EEG sampling signal, such as the voltage values (in μV level) of sleep EEG delta waves (0.1–4 Hz) and wakefulness alpha waves (8–13 Hz). w k is the weighting factor, generated based on the Gaussian distribution. For example, when filtering a certain alpha wave signal, the sampling points closer to the window center (representing the continuous fluctuations of the same rhythm) have higher weights, highlighting the continuity of normal EEG rhythms. N is the window size, which needs to match the characteristics of the EEG frequency band. When processing delta waves (low frequency), N can be set to 200 (corresponding to 2 - second sampling, the signal period of 0.1 Hz is 10 seconds, covering the complete fluctuations); when processing gamma waves (30–100 Hz) N set to 50 (adapting to the rapid changes at high frequencies). y n is the EEG signal after filtering, retaining the key rhythms (such as sleep slow waves) and suppressing noise spikes. In the improved mean filter, the Gaussian weighting is achieved through the formula (σ controls the weight attenuation speed): - For low - frequency signals (such as delta waves): σ is set to a larger value (such as σ = 10), expanding the region with high weights, smoothing the long - time baseline drift, and retaining the slow - wave trend.

[0090] - For high - frequency signals (such as gamma waves): σ is set to a smaller value (such as σ = 3), rapidly attenuating the edge weights and suppressing high - frequency noise (such as EMG interference).

[0091] For example, when processing an alpha wave signal containing EMG noise, the alpha wave sampling points near the window center obtain high weights, and the weights of the edge noise points are low. After filtering, the alpha wave rhythm is clearer.

[0092] Dynamic adaptation of frequency - band parameters: By dynamically adjusting the filtering window size ( N ) and the Gaussian weight parameter ( σ according to the physical characteristics (such as frequency, period) of different frequency bands (delta waves, alpha waves, gamma waves) of the EEG signal​​), it solves the problem of signal blurring caused by the "one-size-fits-all" approach of traditional average filtering in EEG processing. This frequency-band adaptive parameter design significantly improves the pertinence of filtering and belongs to a substantial improvement over the prior art.

[0093] Gaussian weighting strategy optimization: For the different noise characteristics of low-frequency and high-frequency signals, the σ value is set differently (such as using σ = 15 to expand the smoothing range for low-frequency δ waves and using σ = 3 for high-frequency γ waves to quickly attenuate the noise weight), optimizing the weight distribution and achieving efficient noise reduction while retaining signal characteristics. The application of this strategy in the EEG field is innovative.

[0094] Application examples show that the improved method can increase the noise suppression rate in δ wave extraction while retaining the corresponding amplitude, showing certain advantages compared with traditional methods. Weighted average filtering itself is a conventional method in signal processing, but its combination with the characteristics of EEG frequency bands and the application of dynamically adjusting parameters have not been widely disclosed. Among existing EEG filtering methods, fixed window or fixed weight design is more common, while the strategy of dynamically adjusting N and σ is novel, especially the design of a long window ( N= 200) for low-frequency δ waves and a short window ( N= 50) for high-frequency γ waves, which may be proposed for the first time. This method is specifically optimized for the characteristics of EEG signals (low amplitude, multi-noise, multi-frequency bands), and compared with general signal processing solutions, its adaptability in the EEG field enhances the novelty of the technology.

[0095] The embodiments of this application can implement a weighted average filtering method that dynamically allocates neighborhood weights based on the main effective frequency bands of EEG signals, as well as its supporting EEG signal phase extraction and correction methods based on Fourier transform. Combining steps such as phase information extraction, phase difference calculation, and correction can effectively reduce the phase error in EEG signal acquisition to improve the stability and accuracy of EEG signals.

[0096] In the embodiment, the evaluation module can compare the corrected EEG acquisition signal with a standard reference signal and usually cares about the deviation of the corrected signal within a certain range. For example, assume that it is desired that the corrected signal and the standard signal are within a certain amplitude range of ±. The proportion can be calculated by counting the number of points in the corrected signal that meet this condition. The calculation steps are as follows: (1) Calculate the deviation: Calculate the difference between the corrected signal and the standard signal at each time point: ; (2) Range statistics: According to the given deviation range, count the number of points where the signal falls within this range. For example, for a given range of ±10, calculate whether each time point satisfies: |δ ( t )|≤10; (3) Proportion calculation: Finally, divide the number of points that meet the conditions by the total number of points of the signal to obtain the compliance percentage of the corrected signal: ; The calculation result of this step can be used to measure the effect of signal correction. If the deviation of most points is within the given range, it indicates that the phase correction is effective and the signal is successfully corrected. After signal correction, the code statistically analyzes the deviation of the corrected signal. The goal of the statistics is to determine the proportion of the corrected signal within the given range.

[0097] First, calculate whether the absolute value of each value in the corrected data array is within the specified range, and store the result in the withinrange variable. Then, use the sum function to count the number of signals that meet the range condition, and save the result in the counwithirange variable. Finally, by dividing the number of signals that meet the conditions by the total number of signals totalpoints , calculate the proportion of these signals in the total signals, and multiply it by 100 to get the percentage, and store the result in the percentagwithirange variable. The purpose of this process is to evaluate the accuracy and reliability of the corrected signal.

[0098] withinrange First, calculate whether each corrected signal is within the specified range, then use the sum function to count the number of signals that meet the conditions, and finally calculate the proportion of these signals. This step can provide an intuitive value to evaluate the matching degree of the corrected signal and the standard signal.

[0099] In some embodiments, the evaluation module can also implement data visualization. The data visualization part is completed through the matplotlib library, which provides flexible plotting functions. Multiple graphs are created in the code, representing the corrected EEG acquisition signal, the original EEG acquisition signal, the phase difference, and the standard reference signal respectively, and they are plotted in the same chart.

[0100] The last step of signal processing is visual analysis. By plotting the comparison graphs of the corrected signal (corrected EEG acquisition signal), the original signal (original EEG acquisition signal), and the standard signal (standard reference signal), the effect of correction can be intuitively seen. Generally, the following information needs to be shown in the chart: Comparison of the original signal and the corrected signal: Plot the originally received signal and the signal after phase correction in the time domain to observe the phase alignment of the signal.

[0101] Difference between the signal and the reference signal: The difference between the corrected signal and the standard signal can be plotted to show the deviation distribution.

[0102] Visualization of the statistical analysis results: If there are statistical results of the number of points within the deviation range, the compliance of the signal within the specified range can be shown through bar charts or pie charts. This part shows the plotting results of multiple curves, and the phase alignment of the signal can be intuitively seen. Different colored lines in the chart represent different data columns, and the dashed line represents the allowed signal range. The finally generated chart is saved as a PNG file, and this step is very important for analyzing the comparison effect of the signal.

[0103] In some embodiments, the evaluation system of the electroencephalogram (EEG) signal acquisition device implements a waveform template library and its intelligent parameter mapping rules in the middleware, and this component is the EEG feature waveform generation engine.

[0104] As an example, the international standard EEG waveform signal template library includes international standard EEG waveforms, such as the 10 - 20 system α theta wave templates, clinical abnormal wave (spike wave, sharp wave) models, etc. The template library uses a structured storage method, with waveform type, physiological parameter range, etc. as indexes. For user-defined physiological parameters, such as age-related α theta wave frequency drift, it is achieved by establishing a mathematical model. Assume α the relationship between the theta wave frequency and age is a linear model f = f 0 + k · ( age - age 0), where f is the theta wave frequency corresponding to the current age, α theta, f 0 is the reference age age at α the theta wave frequency, k is the frequency drift coefficient. For example, the reference age age 0 = is 30 years old α the theta wave frequency f 0 = 10 Hz, and the frequency drift coefficient k = -0.05 Hz / year. When the user inputs an age of 40 years old, the calculated α theta wave frequency f = 10 - 0.05 · (40 - 30) = 9.5 Hz.

[0105] As an example, the intelligent parameter mapping rule library can be like: input physiological indicators, such as "mild anxiety β theta wave enhancement", and the system automatically matches the amplitude, frequency, and noise floor through a predefined rule library. The rule library is established based on a large amount of clinical data and research results.β Taking the enhancement of the wave amplitude as an example, assume normal β The wave amplitude range is A normal =[5,10] μV , and the enhancement ratio of the amplitude during mild anxiety is r =1.5, then the matching amplitude range is . The frequency range is also determined according to clinical data. For example, normal β The wave frequency range is f normal =[12,20] Hz, and it may become f anxiety =[13,22] Hz during mild anxiety. The noise floor simulates the real EEG noise distribution and adopts a 1 / f noise model. Its power spectral density formula is , where S 0 is the power spectral density at low frequencies, γ usually between 0.5 - 2. For EEG noise, γ ≈ 1. By adjusting S 0 to adapt to the noise levels in different physiological states.

[0106] The EEG feature waveform generation engine can also adopt hardware co - optimization, such as pre - loading the waveform feature code table, such as α The phase jitter characteristic of the wave. The phase jitter can be described by the standard deviation δ φ to describe. Assume α The ideal phase of the wave is φ 0, and the actual phase φ 0= φ 0 + δφ, where . The FPGA adjusts the phase in real - time according to the predefined δ φ . The DSP calculates the waveform correction amount in real - time. The amplitude micro - adjustment adopts a proportional - integral (PI) control algorithm. Let the desired amplitude be A d , and the current amplitude be A c , then the adjustment amount: , where K p and K i are the proportional and integral coefficients. The frequency micro - stepping is achieved by adjusting the phase increment of the DDS (Direct Digital Synthesis). Let the initial phase increment be Δφ 0, and the frequency micro - stepping amount be Δf , then the adjusted phase increment is: , where f clk is the system clock frequency, N is the number of bits of the phase accumulator.

[0107] By setting up an electroencephalogram (EEG) feature waveform generation engine in the embodiment, the preparation time of the calibration experiment can be shortened, and the calibration efficiency can be improved; for accurately simulating abnormal EEG signals during epileptic seizures, it is used for anti-interference algorithm training and pathological exploration of EEG devices. The scenario value of this embodiment may include: in the scientific research field, through the built-in template library and intelligent parameter mapping, it can be paired with a standard reference signal generation device, and researchers can generate standard waveforms that conform to the set templates with one click. For example, when conducting a calibration experiment on an EEG signal acquisition system, a large amount of time was previously required to manually set waveform parameters. Now, through this engine, the experiment preparation time can be correspondingly shortened, greatly improving the experimental efficiency. In the medical field, it can simulate patient-specific EEG, such as abnormal discharges from epileptic foci. By adjusting parameters such as the amplitude, frequency, and phase of the waveform, abnormal EEG signals during epileptic seizures can be accurately simulated, which is used for anti-interference algorithm training of EEG devices to improve the device's ability to recognize and process clinical EEG signals.

[0108] Figure 9 is a schematic diagram of the result of the phase-amplitude basic alignment output by the evaluation module in the embodiment; first, the phase-amplitudes of the three-channel EEG acquisition signals (received signal 1, received signal 2, received signal 3) are aligned. The yellow line represents the phase difference. At this time, the correction strength of the phase difference is relatively large. The comparison between the corrected signal and the standard reference signal is made. By calculating the amplitude differences between the three corrected signals and the standard reference signal, the three blue amplitude difference curves (amplitude difference curve 1, amplitude difference curve 2, amplitude difference curve 3) in the above figure are further obtained, which show the corrected amplitude differences at the corresponding moments. Finally, the percentage of the three amplitude differences within the range of 10 is calculated to further evaluate the correction effect. From Figure 9 it can be seen that the percentage of the differences of the three signals corrected by ADP has all reached more than 97%, which means that 97% of the signal error range is within the specified accuracy, indicating the effectiveness of the research method in this study.

[0109] Figure 10 shows the result of only performing amplitude normalization on the received signal without phase alignment. The observed comparison between the corrected signal and the standard reference signal cannot simply obtain the effect of the corrected signal. However, by further calculating the amplitude difference, the differences between the three corrected curves and the standard signal are obtained, and it is found that the percentage of the amplitude differences within 10 is only 36%. Therefore, the effect of only performing amplitude correction on the received signal cannot reach the ideal effect.

[0110] There will beFigure 11 In the shown situation, there are significant differences in both the phase and amplitude between the acquired signal and the standard reference signal. At this time, without correcting these signals, by calculating the percentage within an amplitude difference of 10, it can be obtained that the percentages of the three signals are only 12%, indicating that the difference between the received signal and the standard reference signal is large at this time, and the signals need to be corrected and aligned.

[0111] Figure 12 It shows a situation where the phase difference between the directly received signal and the standard reference signal is small. In this case, good results can be achieved without performing amplitude difference normalization correction. However, as shown above Figure 9 The result reaches 98%. It can be concluded that even when the phase and amplitude differences between the received signal and the standard signal are very small, the ADP method used in this study can further reduce the difference between the received signal and the standard reference signal.

[0112] Figure 13 It shows the interface of the evaluation module (evaluation software) for the acquisition accuracy of partial EEG signals.

[0113] In view of the current problem of system acquisition error detection in the production of EEG devices, this application provides a set of EEG signal acquisition device evaluation system and evaluation method. The evaluation module can be an upper computer evaluation software capable of issuing an evaluation report. This set of evaluation system is of great significance for guiding the automation of EEG device production testing.

[0114] The above has introduced in detail the EEG signal acquisition device evaluation system and evaluation method provided by this application. Specific examples are used in this article to elaborate on the principle and implementation manner of this application. The description of the above embodiments is only used to help understand the concept of this application and should not be construed as a limitation on the protection scope of this application.

Claims

1. An evaluation system for electroencephalogram signal acquisition devices, characterized in that, The evaluation system is used to evaluate an electroencephalogram (EEG) signal acquisition device, and the evaluation system includes: A standard reference signal generation module for generating a standard reference signal; An evaluation module for acquiring the EEG acquisition signal obtained by the EEG signal acquisition device, displaying the EEG acquisition signal and the standard reference signal, comparing the waveforms and parameters of the EEG acquisition signal and the standard reference signal, and determining the percentage of the number of samples in all sample points where the relative amplitude of the difference between the EEG acquisition signal and the standard reference signal exceeds a set threshold in the total number of samples.

2. The electroencephalogram signal acquisition device evaluation system according to claim 1, wherein The system further includes a signal correction module; The signal correction module is configured to perform Fourier transform on the EEG acquisition signal and the standard reference signal, calculate the phase difference between the transformed EEG acquisition signal and the standard reference signal, and use a sampling time scale factor sequence to phase-align the EEG acquisition signal and the standard reference signal according to the phase difference to obtain a corrected EEG acquisition signal.

3. The electroencephalogram signal acquisition device evaluation system according to claim 1, wherein, The system further includes: An EEG characteristic waveform generation engine, which has an international standard EEG waveform signal template library built-in, and is used to select a corresponding template signal from the template library according to the type of the required signal and relevant physiological parameters. The template signal can be used to generate a standard reference signal so that the standard reference signal generation module generates the standard reference signal according to the standard reference signal.

4. The electroencephalogram signal acquisition device evaluation system according to claim 3, wherein The EEG characteristic waveform generation engine also has an intelligent parameter mapping rule library built-in, which is used to match the intelligent parameter mapping rule library according to the input physiological index to obtain key characteristic parameters; and use the key characteristic parameters to adjust the parameters of the template signal to generate the template signal that meets the target characteristics.

5. The electroencephalogram signal acquisition device evaluation system according to claim 1, characterized in that, The system further includes a standard reference signal generation module for generating a standard reference signal so that the standard reference signal generation module generates the standard reference signal according to the standard reference signal; The standard reference signal generation module includes an FPGA, a DSP, a DAC, a low-noise operational amplifier signal chain module, a power management module, and a physical isolation module; The FPGA is used to generate waveform data; The DSP is communicatively connected to the FPGA, and the DSP is used to preprocess and optimize the quality of the waveform data by controlling the FPGA gain of the analog front end; The FPGA transmits the waveform data to the DAC, and the DAC is used to convert the waveform data into an analog signal; The low-noise operational amplifier signal chain module is used to amplify the analog signal; The power management module is used to supply power to the FPGA, DSP, DAC, and low-noise operational amplifier signal chain module; The physical isolation module adopts a three-layer PCB layout, with the top layer transmitting analog signals, the middle layer being a ground plane, and the bottom layer arranging digital circuits. The analog ground and the digital ground are connected by a 0Ω resistor at a single point, and the external uses an aluminum alloy shell and is sprayed with conductive paint.

6. The electroencephalogram signal acquisition device evaluation system according to claim 1, characterized in that, The system further includes an improved average filter, and the improved average filter is used to perform average filtering on the EEG acquisition signal.

7. The evaluation method of the electroencephalogram signal acquisition device evaluation system according to any one of claims 1 to 6; characterized in that, The method for determining the percentage includes: For each sample point n, determine the ratio of the absolute value of the difference between the electroencephalogram acquisition signal and the standard reference signal of the sample point to the maximum of the absolute values of the two ; Among them, represents the electroencephalogram acquisition signal of sample point n, represents the standard control signal of sample point n, and max() is the maximum value function; Traverse all sample points. If the ratio is greater than or equal to the preset value, then the sample point meets the distortion condition, count the sample points that meet the distortion condition, and obtain the counting result ; The ratio of the counting result to the number of all sample points is the said percentage.

8. The evaluation method of the electroencephalogram signal acquisition device evaluation system according to claim 7, characterized in that, When the system further includes a signal correction module; the method further includes: Using the signal correction module to convert the EEG acquisition signal and the standard reference signal into frequency-domain signals; Comparing the phase values of the EEG acquisition signal and the standard reference signal at the same frequency, and calculating the phase difference between the two; Using the sampling time scale factor sequence, aligning the phases of the EEG acquisition signal and the standard reference signal according to the phase difference to obtain the corrected EEG acquisition signal.

9. The evaluation method of the electroencephalogram signal acquisition device evaluation system according to claim 7, characterized in that, When the system further includes an improved mean filter; The method further includes: according to the physical characteristics of different frequency bands of the EEG acquisition signal, adopting a Gaussian weighting strategy optimization and frequency band parameter dynamic adaptation method to assign different weights to the sample points in the neighborhood; Using the improved mean filter to perform weighted average calculation on the sample points in the neighborhood according to the assigned weights to obtain the filtered EEG acquisition signal value.

10. The evaluation method of the electroencephalogram signal acquisition device evaluation system according to claim 9, characterized in that, The expression of the improved mean filter is as follows: ； Where y[n] represents the filtered signal value at the sample point n, x[n + k] represents the input signal value of the k-th sample point in the neighborhood centered on the sample point n, the value range of k is from -N / 2 to N / 2, N represents the number of sample points extended to both sides centered on the sample point n, and N determines the size of the neighborhood; ω[k] represents the weight of the k-th sample point in the neighborhood at the sample point n.

Images