A method for selecting vestibular function evaluation indexes based on electroencephalogram data

Abstract

The present invention discloses a method for selecting vestibular function evaluation indexes based on electroencephalogram data, comprising the following steps: Step 1, collecting electroencephalogram data of each vestibular cold and hot stimulation cycle; Step 2, processing the electroencephalogram data obtained in Step 1; Step 3, calculating the relative power spectral density of different frequency bands in four periods; Step 4, comparing the differences between each period for the relative power spectral density of each lead in each frequency band, and taking the lead with significant changes in the relative power spectral density in the selected frequency band as the selected lead, and using the change trend of the relative power spectral density of the selected frequency band of the selected lead as the basis for vestibular function evaluation. The present invention provides more selectable analysis indexes for the evaluation of vestibular function.

Description

Technical Field

[0001] The present invention relates to the technical field of brain-computer interfaces, and particularly to a method for selecting vestibular function evaluation indexes based on electroencephalogram data. Background Art

[0002] The vestibular receptor is located in the inner ear and is an important sensory organ in the human body. It is mainly responsible for perceiving the acceleration and direction changes of the head, as well as maintaining balance and spatial orientation. The vestibular organ and its related nerves and brain structures together form the vestibular system. The vestibular system interacts with other sensory systems (such as vision and somatosensation) to jointly coordinate the balance and posture control of the body. When the vestibular system has a dysfunction, it may lead to balance disorders, dizziness and other balance problems. Understanding the functional status of the vestibular nervous system is of great significance for the diagnosis and treatment of related diseases and the screening of specific occupational personnel.

[0003] The caloric test of the vestibule is a clinical test method commonly used for evaluating the function of the vestibular system. In the caloric test of the vestibule, doctors will inject cold or hot water or air into the ear canal through the ear to stimulate the vestibular system. This kind of stimulation will cause nystagmus and dizziness. By observing and recording these reactions, the functional state of the vestibular system can be evaluated. If the patient's reaction is abnormal or asymmetric, it may indicate a problem with the vestibular system. However, this functional detection scheme of the vestibular system also has some disadvantages: some patients, especially elderly patients, are difficult to maintain the required eye-opening state during the temperature test, resulting in inaccurate test results. It can be seen that in addition to simply using the situation of nystagmus as an evaluation index for the function of the vestibular system, it is urgent to explore other physiological indexes to design a more perfect evaluation scheme for the function of the vestibular system.

[0004] When the vestibular receptor receives stimulation, it will transmit information to the central network related to the vestibule, which is also the key part of vertigo perception. Therefore, in addition to the nystagmus caused by vestibular stimulation, the nerve function indexes at the central network level also have important reference value for evaluating the vestibular function. By collecting and analyzing the nerve function indexes, the pathogenesis and pathophysiological process of vestibular diseases can be better understood, providing more accurate and effective reference for clinical treatment.

[0005] Electroencephalogram (EEG) signals are signals generated by the brain, which can reflect human brain activities and brain response characteristics, as well as electrophysiological, pathological and other information. At the same time, they have the advantages of being portable, low-cost in detection, and high time resolution. They have been widely used in fields such as cognitive neuroscience and psychiatry, and are very suitable for providing supplements from the perspective of neuroelectrophysiology for the evaluation of vestibular function. By recording and analyzing EEG signals, the neural mechanisms and functional characteristics of the vestibular system can be better understood, which helps in the diagnosis and treatment of vestibular system-related diseases. Currently, some studies have begun to combine EEG technology with problems related to the vestibular system. However, these research methods mostly collect and analyze the EEG of the vestibular cortex in the resting state, auditory evoked state, and visual evoked state. There is still a lack of a vestibular function assessment plan for directly collecting EEG data under vestibular evoked states. As a method of directly stimulating the vestibular system, vestibular evocation has many advantages in evaluating vestibular function. Compared with other evocation methods (such as visual evocation and auditory evocation), vestibular evocation can more accurately reflect the state of the vestibular system, and the analysis of EEG data under vestibular evoked states can also provide a more comprehensive and accurate assessment of diseases related to vestibular function. In summary, it is necessary to design a method for processing EEG data under vestibular evoked states. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for selecting vestibular function assessment indicators based on EEG data to address the technical deficiencies in the existing technology, so as to realize the evaluation of vestibular function using EEG signals under vestibular evoked states.

[0007] The technical solution adopted to achieve the purpose of the present invention is as follows:

[0008] A method for selecting vestibular function assessment indicators based on EEG data, comprising the following steps:

[0009] Step 1, data acquisition:

[0010] The subject wears the infrared video eyepiece of the video electrooculogram instrument and the EEG cap, and undergoes vestibular caloric stimulation by perfusing gas into the external auditory canal of the subject. After each perfusion, when the nystagmus reaches its peak, the fixation light is turned on for a fixed time, and then the fixation light is turned off and maintained for a fixed time, and the EEG data of each vestibular caloric stimulation cycle is collected;

[0011] Step 2, data preprocessing:

[0012] First, the EEG data obtained in Step 1 is filtered using a high-pass filter and a low-pass filter, then downsampled, then the spherical curve method is used to interpolate the bad leads, and finally ICA is used to remove artifacts to obtain the preprocessed EEG data;

[0013] Step 3, feature selection and data analysis:

[0014] Step s31: Divide the preprocessed EEG data obtained in Step 2 into at least four periods, namely the pre-stimulus period, the nystagmus period, the fixation period, and the post-fixation period. The pre-stimulus period is the EEG data of a fixed duration before vestibular caloric stimulation. The nystagmus period is the EEG data of a fixed duration centered around the time point with the strongest nystagmus. The fixation period is the EEG data of a fixed duration when the fixation light is on. The post-fixation period is the EEG data of a fixed duration after the fixation light is turned off.

[0015] Step s32: Divide the EEG data of each period among the four periods into p time segments, then calculate the power spectral density of each time segment, and then average the power spectral densities of all time segments to obtain the average power spectral density of each period.

[0016] Step s33: In each period, use Fourier transform to transform the EEG data in Step s32 from the time domain to the frequency domain, and divide it into four frequency bands: delta wave (0.5 - 4 Hz), theta wave (4 - 8 Hz), alpha wave (8 - 14 Hz), and beta wave (14 - 30 Hz). Divide the average power spectral density within each frequency band by the sum of the average power spectral densities within all frequency bands to obtain the relative power spectral density of each frequency band.

[0017] Step s34: Display the relative power spectral density of each period and each frequency band obtained in Step s33 through the use of an electroencephalogram topographic map, which is generated by mapping the relative power spectral density of each lead to the scalp surface.

[0018] Step 4: Index judgment:

[0019] Compare the differences in the relative power spectral density of each frequency band between the nystagmus period and the pre-stimulus period, the fixation period and the nystagmus period, the post-fixation period and the fixation period, and the post-fixation period and the nystagmus period for each lead. For the relative power spectral density of each lead in each frequency band, compare the differences between each period. The leads with significant changes in the relative power spectral density in the selected frequency band (p < 0.05) are used as the selected leads, and the change trend of the relative power spectral density of the selected frequency band of the selected leads is used as the basis for judging vestibular function.

[0020] In the above technical solution, in Step 1, the vestibular caloric stimulation is successively cold air perfusion into the right ear, cold air perfusion into the left ear, hot air perfusion into the right ear, and hot air perfusion into the left ear. Each perfusion time is 60 s, and the time interval between every two perfusions is not less than 5 min. The temperature of the cold air is 24°C, and the temperature of the hot air is 50°C. Preferably, hot air is perfused into the left ear.

[0021] In the above technical solution, in step 2, the EEG data obtained in step 1 is processed using the open-source toolbox EEGLAB based on Matlab.

[0022] In the above technical solution, in step 2, the high-pass filter is a high-pass filter with a half-amplitude cut-off frequency of 0.5 Hz and a slope of 24 dB / octave, and the low-pass filter is a low-pass filter with a half-amplitude cut-off frequency of 30 Hz and a slope of 24 dB / octave.

[0023] In the above technical solution, in step 2, downsample to 200 Hz.

[0024] In the above technical solution, in step s31, the pre-stimulus period is the EEG data of 50 s before the temperature stimulus, the nystagmus period is the EEG data of 20 s centered on the time point with the strongest nystagmus, the fixation period is the EEG data of 10 s when the fixation light is turned on, and the end of fixation period is the EEG data of 10 s after the fixation light is not seen to be turned off.

[0025] In the above technical solution, in step s32, the power spectral density J p (ω) is the modified periodogram of the p-th segment, x p (n) is the EEG signal of the p-th segment, containing M data, U is called the normalization factor, and w(n) is the window function;

[0026] The average power spectral density P is the number of time periods.

[0027] In the above technical solution, in step s34, the relative power spectral density is displayed in the brain topographic map through the eeglab toolbox in matlab.

[0028] In the above technical solution, in step 4, the SPSS 26.0 software is used to perform statistical analysis and test on the relative power spectral density, and the Wilcoxon signed-rank test is used for comparison.

[0029] In the above technical solution, in step 4, the change trend of the relative power spectral density of the beta frequency band of the leads in the central vertex and left occipital lobe regions during the nystagmus period and / or the change trend of the relative power spectral density of the alpha frequency band of the leads in the left occipital lobe region and right occipital lobe regions during the nystagmus period are used as the basis for judging the vestibular function.

[0030] In the above technical solution, in step 4, if the relative power spectral density suppression loss of the beta frequency band of the leads in the central vertex and left occipital lobe regions or the relative power spectral density enhancement loss of the alpha frequency band of the leads in the left occipital lobe region and right occipital lobe regions occurs, it indicates that the vestibular organ fails to participate in the motion perception process normally during the vestibular evoked state.

[0031] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0032] 1. In order to overcome the problems of the lack of biomarkers from the perspective of cortical electrophysiology and the lack of collection and analysis of electroencephalogram (EEG) data under vestibular evoked states in current vestibular function assessments, the present invention adopts a method combining EEG technology with temperature tests. Using the temperature test as a vestibular evoked means, EEG data collection is synchronously carried out under vestibular evoked states. After implementing the solution of the present invention, the EEG rhythm characteristics in different periods are compared, and the correlation between the change values of EEG rhythm characteristics and the nystagmus value, a commonly used clinical index, is further studied. The results show that the suppression of the relative power of beta waves in the central parietal and left occipital regions and the enhancement of the relative power of alpha waves in the occipital region can be used as cortical electrophysiological characteristics in response to temperature vestibular stimulation. And in specific leads, it is found that under the condition of left thermal stimulation, the change values of the relative power of alpha and beta rhythms are significantly correlated with the intensity of nystagmus caused by vestibular temperature stimulation, verifying the effectiveness of the above-mentioned EEG rhythm characteristics for evaluating vestibular function.

[0033] 2. The present invention collects nystagmus and EEG data under vestibular evoked states, analyzes the collected EEG data, conducts a study on the influence of EEG rhythm characteristics under vestibular evoked states, excavates specific EEG characteristics related to the response to vestibular temperature stimulation, and conducts a correlation analysis with nystagmus data, proving that the experimental scheme of the present invention can successfully collect nystagmus and EEG data related to the vestibular function status, and this data can be used for vestibular function assessment. This provides more optional analysis indicators for the assessment of vestibular function and provides valuable references for subsequent research. At the same time, it provides new ideas and directions for conducting more in-depth research on vestibular function. Description of the Drawings

[0034] Figure 1 It is a paradigm illustration of the experimental scheme of the present invention, where (a) is a diagram of the experimental scenario and (b) is a schematic diagram of the experimental process. Right cold, left cold, right hot, and left hot respectively represent the perfusion of cold air into the right ear, cold air into the left ear, hot air into the right ear, and hot air into the left ear;

[0035] Figure 2 It is a schematic diagram of the Net Amps 400 device of EGI Company;

[0036] Figure 3 It is a schematic diagram of the electrode positions of the saline EEG cap of the EGI device with 64 leads (blue are the effective leads);

[0037] Figure 4 It is a schematic diagram of the Verti - Goggles device of Shanghai Zhiting Company;

[0038] Figure 5Schematic diagram of the ICS AirCal device of Ertina Medical Devices Co., Ltd.;

[0039] Figure 6 It is the correlation fitting between the change value of the relative power spectral density in the alpha band and the nystagmus value under the condition of left thermal stimulation;

[0040] Figure 7 It is the correlation fitting between the change value of the relative power spectral density of the relative power spectral density in the beta band and the nystagmus value under the condition of left thermal stimulation. Specific implementation manner

[0041] The present invention will be further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0042] The present invention aims to use temperature experiments for vestibular induction, and then synchronously collect nystagmus data and electroencephalogram data in the vestibular induced state and analyze them. The analysis results can be used for further vestibular function evaluation.

[0043] A method for processing electroencephalogram data of nystagmus in the vestibular induced state, comprising the following steps:

[0044] Step 1, data acquisition:

[0045] (1) Preparation: Before collecting data, the experimenter must communicate with the subject to confirm basic information and record it, including age, gender, previous medical history, etc., to ensure that the subject meets the inclusion criteria. After communicating with the subject about the experiment time, the saline electrode cap needs to be soaked in the prepared potassium chloride solution for no less than 15 minutes before the subject arrives. Before the experiment starts, the experimenter needs to explain the experiment process and the possible sense of dizziness during the experiment to the subject and obtain the consent of the subject, and remind the subject to avoid unnecessary body and head movements during data collection to reduce the artifact contamination of electroencephalogram data.

[0046] (2) Calibration: First, before collecting the nystagmus values in the experiment, the experimenter wears the infrared video eyepiece of the video electro-oculogram instrument for the subject and performs calibration. The purpose of calibration is to measure the relationship between eye movements in a specific direction and the corresponding recorded signals (i.e., the displacement of the eye movement curve). Calibration is the basis for calculating parameters such as the amplitude and speed of eye movements. When calibrating, it is required that the subject sit upright on the chair with the head straight and immobile and alternately fixate on the front view target and the visual targets at 20° left and right viewing angles according to the experimenter's verbal reminder, and return to the front view target after no less than 8 alternations.

[0047] (3) EEG cap wearing: After the calibration is completed, the experimenter needs to guide the subject to the examination bed, put on the EEG cap for the subject, drop potassium chloride solution into the sponge tips of the leads with too high impedance, readjust the position to make them close to the scalp, and adjust until the impedance of each lead meets the acquisition standard. Guide the subject to lie flat on the examination bed, place the head on a 30° examination pillow, and align the shoulders with the edge of the examination pillow. Remind the subject to keep their eyes open throughout the data acquisition for the collection of nystagmus data.

[0048] After the nystagmus data is recorded by the video electro-oculogram instrument, the data results can be directly obtained by the supporting software without additional processing.

[0049] (4) Data recording before the experiment: After the experiment starts, it is first necessary to record the EEG data and spontaneous nystagmus data before the stimulation.

[0050] (5) Perform vestibular caloric stimulation: This experimental protocol includes a total of four gas perfusion situations: namely, cold air (24°C) is perfused into the right ear, cold air (24°C) is perfused into the left ear, hot air (50°C) is perfused into the right ear, and hot air (50°C) is perfused into the left ear.

[0051] After recording the relevant data before the stimulation, the experimenter needs to turn on the vestibular caloric stimulator. After the gas is heated or cooled to the specified temperature, perfuse the gas into the subject's external auditory canal for the temperature test. The gas perfusion is carried out in the above order. To ensure that the subject's nystagmus completely disappears before the next gas perfusion, the interval between every two gas perfusions needs to be no less than 5 minutes, and the time for each perfusion is 60 seconds. The nystagmus reaches its peak within 50 - 60 seconds. When observing the attenuation of the nystagmus intensity, turn on the fixation light to inhibit the activation of the vestibule. The experimenter needs to check the impedance of the EEG cap leads during the rest time and adjust it in time to ensure the quality of the collected EEG data. After completing the perfusion in the four situations, end the experiment and remove the equipment worn by the subject.

[0052] (6) Discomfort recording and subsequent observation:

[0053] After the experiment is over, the experimenter needs to ask the subject about the discomfort reactions during the experiment and record them, and let the subject rest until the discomfort subsides. Before the subject leaves, it is necessary to remind the subject to continuously pay attention to whether there is dizziness or ear discomfort in the following days. If so, it needs to be reported in time.

[0054] Step 2, data preprocessing:

[0055] Although unnecessary interference is reduced as much as possible during the EEG signal collection process, EEG signals are very weak, with an extremely low signal-to-noise ratio, and are easily affected by various physiological and non-physiological factors, such as electromyography, electrooculography, power frequency interference, poor electrode contact, etc. Therefore, there are a large number of non-EEG signals in the collected data, which are called artifacts. In order to mine specific features in EEG signals, the recorded EEG signals must be strictly processed for noise reduction and artifact removal to eliminate or weaken the influence of these artifacts as much as possible and retain the most authentic and original EEG signals.

[0056] The present invention uses the open source toolbox EEGLAB based on Matlab (R2021a) to process the recorded raw EEG data. First of all, since the EEG data in the experiment is collected by the EGI device, its original data format already contains the position information of each electrode, so there is no need to locate the electrode lead during data processing. In order to reduce the slow voltage drift caused by skin potential, as well as attenuate muscle artifacts and mains power frequency interference, referring to the commonly used EEG processing process and existing research results in related fields, the EEG data is filtered using a high-pass filter with a half-amplitude cutoff frequency of 0.5Hz and a slope of 24dB / octave and a low-pass filter with a half-amplitude cutoff frequency of 30Hz and a slope of 24dB / octave.

[0057] When collecting EEG data in the experiment, the sampling rate was 1000Hz. In the preprocessing, this solution downsamples the EEG data to 200Hz. Taking into account that the EEG signal has been filtered to the pre-analyzed 0.5-30Hz frequency band, according to the Nyquist sampling theorem, it is concluded that the minimum sampling rate for analyzing the data in this frequency band is 60Hz. Therefore, downsampling the data to 200Hz can not only meet the accuracy requirements of data analysis, but also increase the running speed and reduce unnecessary calculations. This processing method can better retain the information of EEG signals, while eliminating or reducing the influence of interference noise, so as to more accurately analyze EEG features. In the present invention, the original EEG data is downsampled and can be used for subsequent data analysis and mining.

[0058] If an electrode is affected by the subject's movements during the experiment and has poor contact with the scalp, resulting in increased impedance or obvious signal drift, this electrode is classified as a bad lead. Although some EEG preprocessing tutorials will directly delete the bad leads, the number and position of bad leads may be different between different subjects. Directly deleting the bad leads will result in inconsistent numbers of electrodes between different subjects, which may increase variables when comparing EEG responses and is not rigorous enough. Therefore, the present invention manually removes the bad leads, and then uses the spherical curve method to interpolate the bad leads, and uses other electrode values ​​with better signal quality to replace them after calculation. In actual processing, if the number of bad derivatives is greater than five, these data need to be excluded.

[0059] By manual inspection and marking, data segments with severe motion artifacts caused by accidental movements of the subjects can be excluded. These data segments can be manually selected and deleted to reduce the impact of artifacts, thereby improving the quality and reliability of the data.

[0060] Step 3, Feature selection and data analysis:

[0061] Step s31, To better evaluate the vestibular function of the subjects, the collected EEG data is divided into four periods for analysis, corresponding to the pre-stimulus period, nystagmus period, fixation period, and end of fixation period (the EEG data of 50 s before the temperature stimulus is defined as the pre-stimulus period, the EEG data of 20 s centered on the time point with the strongest nystagmus is defined as the nystagmus period, the EEG data of 10 s when the fixation light is on is defined as the fixation period, and the EEG data of 10 s after the fixation light is turned off is defined as the end of fixation period). Thanks to the high temporal resolution characteristics of scalp EEG, EEG rhythm response analysis allows us to deeply understand the brain response characteristics of the subjects in each period during the experiment. Therefore, this protocol conducts EEG rhythm response analysis for each period of the vestibular temperature test, and the EEG rhythms include four frequency bands: delta wave (0.5 - 4 Hz), theta wave (4 - 8 Hz), alpha wave (8 - 14 Hz), and beta wave (14 - 30 Hz).

[0062] Step s32, Divide the EEG data of each of the four periods into p time segments, then calculate the power spectral density of each time segment, and then average the power spectral densities of all time segments to obtain the average power spectral density of each period. Specifically:

[0063] Calculate the power spectral density of each segment of EEG data using the Welch method. This method divides the EEG signal x(n) of length N into p segments, performs Fourier transform on M data in each time segment to obtain the spectral density of each time segment, and squares it to obtain its power spectral density J p (ω). Then average the power spectral densities of all time segments to obtain the average power spectral density within the frequency resolution The specific calculation formula is:

[0064]

[0065]

[0066] In formula 1, J p (ω) is the modified periodogram of the p-th segment, x p (n) is the EEG signal of the p-th segment, containing M data, U is called the normalization factor, w(n) is the window function, and in formula 2 The result calculated by Welch's method, where P is the number of data segments, and the signals of each lead are calculated separately.

[0067] Step s33, in each period, use Fourier transform to transform the EEG data in step s32 from the time domain to the frequency domain, and divide it into four frequency bands: delta wave (0.5 - 4 Hz), theta wave (4 - 8 Hz), alpha wave (8 - 14 Hz), and beta wave (14 - 30 Hz). Divide the average power spectral density within each frequency band by the sum of the average power spectral densities within all frequency bands to obtain the relative power spectral density of each frequency band.

[0068] Step s34, in order to investigate the EEG distribution of specific rhythms, use EEG topographic maps in the EEGLAB toolbox of Matlab to present the relative power spectral density of each frequency band and each period. The EEG topographic map is generated by mapping the relative power spectral density calculated above to the scalp surface. These EEG topographic maps can show the distribution of EEG rhythms in specific frequency bands on the scalp, helping us visually view the activation of each brain region. This scheme uses the relative power spectral density of all leads in each period to draw EEG topographic maps of different periods to investigate the distribution of specific EEG rhythms. By comparing the EEG topographic maps of each period, the whole-brain distribution differences of EEG rhythms in each period can be observed and analyzed, so as to understand whether the activation of brain regions changes with the period.

[0069] Step 4, index judgment:

[0070] To study whether the above changes are significant, statistical analysis and testing of the relative power spectral density are required. In the part of statistical analysis, SPSS 26.0 software is used for calculation. For the relative power spectral density of each lead of each EEG rhythm, the differences between each period are compared. Since it is found that some groups of data do not follow the normal distribution during the normality test of the data, the Wilcoxon signed-rank test is used for comparison. In order to find the differences brought about by temperature vestibular evoked, fixation inhibition, removal of fixation inhibition, and after the fixation inhibition process, the nystagmus period and the pre-stimulus period, the fixation period and the nystagmus period, the end of fixation period and the fixation period, and the end of fixation period and the nystagmus period are compared respectively. The leads with significant changes in each period (p < 0.05) are used as the selected leads, and the average relative power spectral change trend of the selected leads is calculated.

[0071] Through the above comparison, the change trend of the relative power spectral density of the beta frequency band of the leads in the central parietal and left occipital regions during the nystagmus period shows significant changes, and the change trend of the relative power spectral density of the alpha frequency band of the leads in the left occipital region and the right occipital region during the nystagmus period shows significant changes.

[0072] Example 2

[0073] This embodiment conducts a correlation test on the results obtained in Embodiment 1:

[0074] The nystagmus value is an indicator for evaluating the vestibular evoked response clinically. The correlation analysis between electroencephalogram (EEG) data and nystagmus data can prove the effectiveness of EEG rhythm characteristics in evaluating vestibular function. According to the analysis results of the EEG rhythms in each stage, the change trend of the relative power spectral density in the beta frequency band of the leads in the central vertex and left occipital lobe regions during the nystagmus period shows significant changes, and the change trend of the relative power spectral density in the alpha frequency band of the leads in the left occipital lobe region and right occipital lobe region during the nystagmus period shows significant changes.

[0075] To verify whether the EEG rhythm characteristics (i.e., relative power spectral density) have the potential to evaluate vestibular function, under four vestibular temperature stimulation conditions (right cold, left cold, right hot, left hot), the correlation analysis was performed between the differences in the relative power spectral density of the alpha rhythm in the left and right occipital lobe regions and the beta rhythm in the central vertex and left occipital lobe regions and the nystagmus value respectively.

[0076] The normality test was performed on the above data, and it was found that the data conformed to the normal distribution. Therefore, Pearson correlation test was selected to explore the correlation between the relative power change value of a specific EEG rhythm and the nystagmus value.

[0077] The test results showed that under the left hot condition, there was a significant correlation between the enhancement of the relative power spectral density in the alpha frequency band in the region of interest (the regions where the O1, Oz, O2, P10, and P6 leads are located) (the average value of the relative power spectral density of the five leads increased) and the nystagmus value. The fitting line of the correlation coefficient is shown in Figure 6 . That is, the difference in the relative power spectral density in the alpha frequency band in the region of interest was significantly correlated with the nystagmus value (r = -0.527, p = 0.025). No statistically significant correlation was found in the remaining stimulation cases. Therefore, it is suggested that the degree of enhancement of the relative power spectral density in the alpha frequency band under the left hot condition can be used to evaluate the activation degree of the vestibule in clinical practice.

[0078] It can be found from the results of the correlation analysis that under the left hot condition, there was an extremely significant correlation between the decrease in the relative power in the beta frequency band and the nystagmus value. The fitting line of the correlation coefficient is as shown in Figure 7That is, the suppression of the relative power spectral density in the beta band in the region where the interested leads (leads 63, 62, 64, F9, FP1, AFz, F10, 61, F7, F5, T9, FT7, F1, Fz, F2, F4, FC6, FT8, TP9, P9, O1, Oz, P10, TP10, P7, P5, P3, P1, PO3, POz, Pz, P2, P4, TP7, C3, C5, CP1, C1, FC1, FCz, FC2, C2, CP2, C4, FC4 are located), i.e., the suppression of the average value of the relative power spectral density of forty-five leads, is significantly correlated with the nystagmus value (r = 0.708, p = 0.001). No statistically significant correlation was found in other stimulation conditions. The change regulation of the relative power spectral density in the beta band found in this study coincides with the activation and inhibition phases in the caloric test. It can be speculated that the beta band is closely related to the vestibular activation state, and the degree of suppression of the relative power spectral density in the beta band, especially under left thermal stimulation, is related to the intensity of nystagmus. Therefore, it is suggested that the relative power spectral density in the beta band induced by the vestibular stimulation reflects the functional status of the vestibule.

[0079] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for selecting vestibular function evaluation indicators based on EEG data, characterized in that: The following steps are involved: Step 1, data collection: The subjects wore the infrared video eyepieces of the video electronystagmograph and the EEG cap, and vestibular cold and heat stimulation was performed by perfusing gas into the external auditory canal of the subjects. The vestibular cold and heat stimulation was cold air perfused into the right ear, cold air perfused into the left ear, hot air perfused into the right ear, and hot air perfused into the left ear in sequence. Each perfusion time was 60 seconds, and the time interval between each two perfusions was not less than 5 minutes. The temperature of the cold air was 24°C, and the temperature of the hot air was 50°C. After each perfusion, when the nystagmus reached its peak, the fixation lamp was turned on for a fixed time, and then the fixation lamp was turned off for a fixed time, and the EEG data of each vestibular cold and heat stimulation cycle was collected; Step 2, data preprocessing: First, the EEG data obtained in step 1 is filtered using a high-pass filter and a low-pass filter, then downsampled, and then the bad conduction is interpolated using the spherical curve method, and finally the artifacts are removed using ICA to obtain the preprocessed EEG data; Step 3, feature selection and data analysis: Step s31, dividing the pre-processed EEG data obtained in step 2 into at least four periods, namely, pre-stimulation period, nystagmus period, fixation period and end of fixation period, wherein the pre-stimulation period is the EEG data of 50 seconds before the temperature stimulation, the nystagmus period is the EEG data of 20 seconds centered on the time point of the strongest nystagmus, the fixation period is the EEG data of 10 seconds when the fixation light is turned on, and the end of fixation period is the EEG data of 10 seconds after the fixation light is turned off; Step s32, dividing the EEG data of each of the four periods into p time periods, then calculating the power spectrum density of each time period, and then averaging the power spectrum densities of all time periods to obtain the average power spectrum density of each period; Step s33, in each period, using Fourier transform to transform the EEG data of step s32 from the time domain to the frequency domain, and divide it into four frequency bands: delta wave, theta wave, alpha wave and beta wave, and divide the average power spectral density in each frequency band by the sum of the average power spectral densities in all frequency bands to obtain the relative power spectral density of each frequency band; Step s34, displaying the relative power spectrum density of each frequency band in each period obtained in step s33 by using a brain topography map, wherein the brain topography map is generated by mapping the relative power spectrum density of each lead onto the scalp surface; Step 4, indicator judgment: The differences in relative power spectral density of each frequency band between the nystagmus period and the pre-stimulation period, the fixation period and the nystagmus period, the end of fixation period and the fixation period, and the end of fixation period and the nystagmus period of each lead were compared. For the relative power spectral density of each lead in each frequency band, the differences between each period were compared, and the leads with significant changes in the relative power spectral density of the selected frequency band were selected as the selected leads, and the change trend of the relative power spectral density of the selected frequency band of the selected leads was used as the basis for judging vestibular function.

2. The method for selecting vestibular function evaluation indicators based on EEG data according to claim 1, characterized in that: In the step 2, the EEG data obtained in the step 1 is processed based on the Matlab open source toolbox EEGLAB.

3. The method for selecting vestibular function evaluation indicators based on EEG data according to claim 1, characterized in that: In step 2, the high-pass filter is a high-pass filter with a half-amplitude cutoff frequency of 0.5 Hz and a slope of 24 dB / octave, and the low-pass filter is a low-pass filter with a half-amplitude cutoff frequency of 30 Hz and a slope of 24 dB / octave.

4. The method for selecting vestibular function evaluation indicators based on EEG data according to claim 1, characterized in that: In step 2, the sampling is downsampled to 200 Hz.

5. The method for selecting vestibular function evaluation indicators based on EEG data according to claim 1, characterized in that: In step s32, the power spectral density J p (ω) is the modified periodogram of the pth segment, x p (n) is the EEG signal of the pth segment, containing M data, U is called the normalization factor, and w(n) is the window function; Average power spectral density P is the number of time periods.

6. The method for selecting vestibular function evaluation indicators based on EEG data according to claim 1, characterized in that: In the step s34, the relative power spectrum density is displayed in the brain topography map by using the eeglab toolbox in matlab.

7. The method for selecting vestibular function evaluation indicators based on EEG data according to claim 1, characterized in that: In step 4, SPSS 26.0 software was used to perform statistical analysis on the relative power spectral density, and the Wilcoxon signed rank test was used for comparison.

Images