A method for transmitting electroencephalogram data combining lossless and lossy compression

Abstract

The application provides an electroencephalogram data transmission method combining lossless and lossy compression, belongs to the field of electroencephalogram data transmission and compression processing, and is used for solving the problems of low transmission efficiency of pure lossless compression and easy loss of key diagnostic information of pure lossy compression in the related art. The method extracts multi-dimensional features of electroencephalogram data, divides key areas and non-key areas, respectively adopts lossless compression and lossy compression processing, dynamically adjusts the area division proportion in combination with the transmission bandwidth, signal characteristics and application scene, and transmits corresponding data through double protocols. It can retain clinical diagnostic markers and core features while improving transmission efficiency, and has scene adaptability and transmission reliability.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electroencephalogram data processing and transmission, and particularly relates to an electroencephalogram data transmission method combining lossless and lossy compression. BACKGROUND

[0002] Electroencephalogram data has the characteristics of high sampling rate and multi-channel, and the data volume is huge, so the bandwidth requirement is high in the transmission process. In the prior art, pure lossless compression cannot meet the efficient transmission requirement, and pure lossy compression may lead to loss of key diagnostic information, and cannot balance the compression rate and data availability, so there is an urgent need for a method that can improve transmission efficiency while ensuring the effectiveness of key data. SUMMARY

[0003] The present application provides an electroencephalogram data transmission method combining lossless and lossy compression, which can balance the compression rate, transmission efficiency and data availability by distinguishing the key and non-key areas of electroencephalogram data and adapting different compression methods.

[0004] In a first aspect, the present application provides an electroencephalogram data transmission method combining lossless and lossy compression. Obtain original electroencephalogram data; divide the original electroencephalogram data into key area and non-key area; process the key area data after division by using a lossless compression algorithm, and process the non-key area data after division by using a lossy compression algorithm; transmit the key area data and non-key area data after compression processing.

[0005] By using the above technical solution, by obtaining the original electroencephalogram data and dividing the key and non-key areas, the key area is compressed losslessly to retain the core information, and the non-key area is compressed lossily to reduce the data volume, and finally the two types of compressed data are transmitted, so as to improve the overall compression rate and transmission efficiency on the basis of ensuring the availability of key data.

[0006] Further, the division of the original electroencephalogram data into key area and non-key area comprises: extracting time domain features, frequency domain features and spatial features of the original electroencephalogram data; determining a key feature threshold based on the time domain features, frequency domain features and spatial features; and dividing the original electroencephalogram data into key area and non-key area according to the key feature threshold.

[0007] By using the above technical solution, the key feature threshold is determined in combination with multi-dimensional features, so that the division of the key area and the non-key area is more accurate, the division deviation caused by a single feature is avoided, and it is ensured that the key area can accurately cover the core information.

[0008] Further, the key feature threshold is set based on a clinical diagnosis marker feature and a task-related core feature, the key region is an electroencephalogram data region containing the clinical diagnosis marker feature or the task-related core feature, and the non-key region is an electroencephalogram data region not containing the clinical diagnosis marker feature or the task-related core feature.

[0009] By adopting the technical solution, the key feature threshold is set based on clinical and task requirements, the key region is closely related to the actual application scenario, the region for lossless compression directly corresponds to core value information, and the compression strategy is improved in pertinence.

[0010] Further, the method further includes: monitoring a transmission bandwidth, an electroencephalogram signal feature, and an application scenario in real time; and dynamically adjusting a division ratio of the key region and the non-key region according to changes in the transmission bandwidth, the electroencephalogram signal feature, and the application scenario.

[0011] By adopting the technical solution, the division ratio is dynamically adjusted by real-time monitoring, and the compression strategy can adapt to changes in transmission conditions, signal features, and scenarios, thereby avoiding resource waste or loss of key information caused by a fixed ratio.

[0012] Further, dynamically adjusting the division ratio according to changes in the transmission bandwidth includes: when the transmission bandwidth is lower than a preset bandwidth threshold, reducing a proportion of the key region; and when the transmission bandwidth is higher than or equal to the preset bandwidth threshold, increasing the proportion of the key region.

[0013] By adopting the technical solution, the proportion of the key region is dynamically adjusted based on the transmission bandwidth, transmission efficiency is preferentially ensured when the bandwidth is insufficient, key data integrity is preferentially ensured when the bandwidth is sufficient, and optimal configuration is achieved under different bandwidth conditions.

[0014] Further, dynamically adjusting the division ratio according to changes in the electroencephalogram signal feature includes: when a preset transient event occurs in the electroencephalogram signal feature, expanding a range of the key region; and when the electroencephalogram signal feature is in a steady state, reducing the range of the key region.

[0015] By adopting the technical solution, the range of the key region is dynamically adjusted according to the electroencephalogram signal feature, the lossless compression range is expanded when a transient event (such as a pathological wave) occurs, the compression rate is increased when in a steady state, and dynamic changes in the signal feature are adapted.

[0016] Further, dynamically adjusting the division ratio according to changes in the application scenario includes: when the application scenario is a clinical diagnosis scenario, controlling the proportion of the key region to be not lower than a preset high threshold; and when the application scenario is a real-time monitoring scenario, controlling the proportion of the key region to be not lower than a preset low threshold, the preset low threshold being lower than the preset high threshold.

[0017] By adopting the technical scheme, the key area proportion threshold is set according to different application scenarios, the demand of high fidelity for clinical diagnosis and the demand of efficiency for real-time monitoring are met, and the scene adaptability of the method is improved.

[0018] Further, the divided key area data is processed by using a lossless compression algorithm, including: performing linear prediction coding on the key area data to obtain a prediction error sequence; and processing the prediction error sequence by using an entropy coding algorithm to obtain lossless compression data.

[0019] By adopting the technical scheme, the time sequence redundancy of the key area data is reduced by linear prediction coding, and the statistical redundancy is eliminated by entropy coding, so that the compression efficiency is improved under the premise of losslessness, and the transmission amount of the key data is reduced.

[0020] Further, the divided non-key area data is processed by using a lossy compression algorithm, including: performing wavelet transform on the non-key area data to obtain low-frequency coefficients and high-frequency coefficients; determining a quantization step based on an energy value of the non-key area data, the quantization step being positively correlated with the energy value; and performing quantization processing on the high-frequency coefficients by using the quantization step to obtain lossy compression data.

[0021] By adopting the technical scheme, the quantization step is dynamically adjusted based on the energy value of the non-key area data, the quantization distortion is reduced when the energy is high, and the compression rate is improved when the energy is low, so that the compression effect of the non-key data is maximized within an acceptable range.

[0022] Further, the compressed key area data and the compressed non-key area data are transmitted, including: configuring a first transmission protocol for the compressed key area data and a second transmission protocol for the compressed non-key area data; and respectively transmitting the corresponding compression data based on the first transmission protocol and the second transmission protocol, wherein the reliability of the first transmission protocol is higher than that of the second transmission protocol.

[0023] By adopting the technical scheme, the key data is transmitted by using a high-reliability protocol, and the non-key data is transmitted by using an adaptive protocol, so that the overall transmission efficiency is improved while the integrity of the key data is ensured, and the reliability and the transmission speed are balanced.

[0024] In summary, the present application at least has the following beneficial effects:

[0025] 1. A brain electrical data transmission method is provided, which balances the availability and transmission efficiency of key data and balances the compression rate and core information retention.

[0026] 2. By dynamically adjusting the mechanism and multi-dimensional division, the adaptability of the method to different scenes and signal characteristics is improved;

[0027] 3. By differentiating compression and transmission protocol configuration, the overall performance of data processing and transmission is further optimized.

[0028] It should be understood that the content described in the summary section is not intended to limit the key or important features of the embodiments of the present application, nor to limit the scope of the present application. Other features of the present application will become apparent through the following description. BRIEF DESCRIPTION OF DRAWINGS

[0029] The above and other features, advantages and aspects of the embodiments of the present application will become more apparent by describing in detail preferred embodiments thereof with reference to the attached drawings in which:

[0030] Figure 1 An exemplary operating environment in which embodiments of the present application can be implemented is shown.

[0031] Figure 2 A flowchart of an electroencephalogram data transmission method combining lossless and lossy compression in embodiments of the present application is shown. DETAILED DESCRIPTION

[0032] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0033] In addition, the term "and / or" herein is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent the three cases of existence of A alone, existence of A and B together, and existence of B alone. In addition, the character " / " herein generally represents an "or" relationship between the front and rear associated objects.

[0034] The present application provides an electroencephalogram data transmission method combining lossless and lossy compression, which can guarantee the integrity of key electroencephalogram data to ensure availability, improve compression ratio and transmission efficiency, and dynamically adapt to scene changes to balance data quality and transmission performance.

[0035] Figure 1 An exemplary operating environment in which embodiments of the present application can be implemented is shown.

[0036] Referring toFigure 1 The running environment includes a brain electrical signal acquisition device for acquiring raw brain electrical data, which comprises a multi-lead electrode array and a signal preprocessing module. The multi-lead electrode array can synchronously acquire brain electrical signals of multiple channels, and the signal preprocessing module is used for filtering and denoising the raw signals to remove power frequency interference and environmental noise.

[0037] The running environment also includes a data processing unit connected with the brain electrical signal acquisition device, which is used for receiving the preprocessed brain electrical data and performing division of key areas and non-key areas, as well as corresponding lossless compression and lossy compression processing. The unit is built-in with an algorithm module for feature extraction and a compression algorithm module. The feature extraction module can extract time domain, frequency domain and spatial features of the brain electrical data, and the compression algorithm module includes lossless compression algorithms and lossy compression algorithms.

[0038] The running environment includes a communication transmission module connected with the data processing unit, which is used for receiving compressed key area data and non-key area data and transmitting them. The module supports multiple transmission protocols and can select corresponding protocols according to data types. High-reliability protocols are used for key area data, and high-efficiency transmission protocols are used for non-key area data.

[0039] The running environment also includes a monitoring and adjusting unit connected with the data processing unit and the communication transmission module, which is used for real-time monitoring of transmission bandwidth, brain electrical signal features and application scenarios, and sending adjustment instructions to the data processing unit according to the monitoring results to dynamically adjust the division ratio of key areas and non-key areas.

[0040] In addition, the running environment includes a terminal device connected with the communication transmission module, which is used for receiving transmitted compressed data and decoding and displaying them. The terminal device can further analyze or display the decoded brain electrical data according to the requirements of the application scenario.

[0041] Figure 2 A flowchart of a brain electrical data transmission method combining lossless and lossy compression in an embodiment of the application is shown. The method can be specifically executed by the running environment in Figure 1 .

[0042] Referring to Figure 2 , the method specifically includes the following steps:

[0043] S1: Obtain raw brain electrical data.

[0044] The method of this step is realized by a multi-channel brain electrical acquisition device. The device needs to meet the following technical parameters: sampling rate is (According to the Nyquist sampling theorem, to ensure that the effective frequency band of the brain electrical signal is covered ), the number of channels (Clinical scenarios commonly use Channel), analog-to-digital conversion resolution bit (to ensure microvolt-level signal acquisition accuracy, such as the amplitude of an epileptic spike usually During the acquisition process, the device suppresses common-mode interference through active electrode technology, and the common-mode rejection ratio (CMRR) to reduce the influence of power frequency and electromyographic noise .

[0045] The raw data is stored in the form of a time series matrix, denoted as , where is the number of time points is the acquisition duration, in seconds), and the matrix element represents the amplitude of the electroencephalogram signal of the channel at the time point (unit: ). To ensure data quality, the signal-to-noise ratio of each channel needs to be calculated in real time , where is the signal power of the channel (calculated by integrating the signal frequency band), and is the noise power (calculated by integrating the frequency band). When , the device gain adjustment or reacquisition is triggered to ensure the reliability of subsequent processing.

[0046] During the acquisition process, external event markers (such as manual markers of epileptic seizures and sleep staging state switching) are recorded synchronously, and the marker signal is stored in binary sequence , where indicates that there is an event trigger at the time point, which is used for time series anchoring during subsequent key area division. The marker signal can also be stored in a digital sequence, indicates that there is no time trigger at the time point, and non-zero indicates that there is an event trigger at the time point, and the specific event type of the event trigger can be determined by the specific number . After the data acquisition is completed, the raw data is transmitted to the edge computing unit through USB3.0 or wireless transmission (such as Wi-Fi6), and the raw data is verified before transmission, with the verification formula

[0047] to ensure data transmission integrity.

[0048] The specific methods of this step include: extracting the temporal, frequency, and spatial features of the original EEG data; determining key feature thresholds based on the temporal, frequency, and spatial features; and dividing the original EEG data into key and non-key regions according to the key feature thresholds.

[0049] The key feature threshold is set based on clinical diagnostic biomarker features and task-related core features. The key region is an EEG data region that contains the clinical diagnostic biomarker features or task-related core features, and the non-key region is an EEG data region that does not contain the clinical diagnostic biomarker features or task-related core features.

[0050] When extracting time-domain features, for the original EEG data matrix Calculate the local peak value for each channel. : ( For window size, take Each sampling point corresponds to (Adapting to the time scale of transient EEG events) and defining peak thresholds. (For example, the peak threshold for epileptic spikes in clinical diagnosis is set to...) Simultaneously calculate the absolute value of the first-order difference. Used to characterize the degree of signal abrupt change, abrupt change threshold. Set as (Based on dynamic adjustment of the mean aberration rate over all time periods) To find the mean function.

[0051] Frequency domain feature extraction employs Short Time Fourier Transform (STFT), dividing each channel's data into frames (frame length 256 sampling points, overlap rate...). ), calculate frequency resolution ( (where the frame length is 1), the frequency domain matrix is ​​obtained. ( For frequency points, (Number of frames). Extracted features include the energy percentage of each frequency band: ( The energy proportion of each band corresponds to key characteristics of sleep stages. ( The energy percentage of each frequency band corresponds to the core characteristics of the motion visualization task, and a frequency band energy threshold is set. (like The proportion of energy in the band exceeds Time markers are key characteristics of sleep.

[0052] Spatial features are defined by cross-channel correlation calculation. and Pearson correlation coefficient where is the covariance, is the standard deviation. The synchrony threshold is set to 0.7 (based on clinical data statistics, the synchrony between adjacent channels during seizure is usually ), when it is determined that there is strong spatial correlation between the two channels.

[0053] The determination of the key feature threshold needs to integrate clinical priori and data statistics: the peak threshold takes the quantile of the peak distribution of the healthy population (e.g. ), the frequency band energy threshold is set based on the task requirement (e.g. wave band in sleep staging), the synchrony threshold . When dividing, a feature vector is generated for each piece of data (time window length 1s) (the is the average correlation coefficient with the adjacent channel, the is the vector combination, which is used to combine multiple features in the parentheses into a vector as needed), when at least one feature in exceeds the corresponding threshold, mark this period as a key area, otherwise as a non-key area, finally generate a binary mask matrix ( indicates that the channel is a key area at the moment).

[0054] S3: the key area data after division is processed by a lossless compression algorithm, and the non-key area data after division is processed by a lossy compression algorithm.

[0055] In the method of this step, the key area data after division is processed by a lossless compression algorithm, including: performing linear predictive coding on the key area data to obtain a prediction error sequence; processing the prediction error sequence by an entropy coding algorithm to obtain lossless compression data.

[0056] The non-key area data after division is processed by a lossy compression algorithm, further including: performing wavelet transform on the non-key area data to obtain low-frequency coefficients and high-frequency coefficients; determining a quantization step based on the energy value of the non-key area data, the quantization step being positively correlated with the energy value; and performing quantization processing on the high-frequency coefficients by using the quantization step to obtain lossy compression data.

[0057] The key area data is extracted by the mask matrix , that is ( (elemental product), where ( (Number of key time points in the critical region). In linear predictive coding, key data for each channel. use Linear prediction model: ,in For prediction coefficients, Characterize the key data for each channel by minimizing the mean square error. Please solve. This is the mathematical expectation symbol, used to average the random variables within the brackets (such as the squared error term) over a probability space or sample set. It provides a holistic and stable measure of the average level of random quantities, offering an indicator of long-term or global performance for algorithm optimization (such as minimizing the mean squared error). It is estimated using the Yule-Walker equation.

[0058] , , For the first The autocorrelation function of key channel data, The prediction error sequence is as follows: When performing Huffman entropy encoding on it, first statistically analyze the frequency distribution of the error values. , This is a counting function used to count the number of times the error value e occurs. To obtain the number of key time points in the aforementioned critical region, a prefix code tree is constructed based on the distribution, so that high-frequency error values ​​correspond to short codewords, and the length of the encoded data meets the requirements. (Approaching the lower limit of information entropy).

[0059] Non-critical area data , For non-critical region data, X is the previously mentioned raw EEG data matrix, and M is the previously mentioned mask matrix for extracting critical region data. First, a two-dimensional wavelet transform (time-weighted transformation) is performed. (channel), using DB4 wavelet basis decomposition in 3 layers, to obtain low-frequency approximation coefficients. (Low-frequency approximation coefficients are used to preserve the overall trend of the signal) and high-frequency detail coefficients ( The number of decomposition layers, (for direction) This is the number of non-critical region time dimension samples naturally determined after masking and filtering based on the original EEG time sampling points. It's an inherent attribute of the EEG data after dimensional decomposition (similar to the logic of critical region time point counts, implicit in the data processing flow). The energy value of the non-critical region data is then calculated. C represents the number of channels mentioned earlier, and the quantization step size. according to Sure These are empirical coefficients; high-frequency detail coefficients are used to ensure finer quantization at high energies. (For high-frequency detail coefficients...) Quantification: (Round to the nearest whole number) (integer multiples) For high-frequency detail coefficients The quantized result is the low-frequency approximation coefficient. Unquantized data is used to preserve basic features. Quantized data is then concatenated with low-frequency coefficients to form intermediate data with lossy compression, where the quantization error of the high-frequency coefficients satisfies... (Based on the uniform distribution assumption), that is, assuming that the original data (such as high-frequency detail coefficients) takes values ​​that are uniformly distributed within the quantization interval. Based on the assumption of uniform distribution, this is the mathematical expectation of the squared error generated when quantizing high-frequency detail coefficients. Theoretically, it is approximately 1 / 12 of the square of the quantization step size and is used to measure the degree of quantization distortion.

[0060] S4: Transmit the compressed critical area data and non-critical area data.

[0061] The method in this step specifically includes: configuring a first transmission protocol for the compressed critical area data, configuring a second transmission protocol for the compressed non-critical area data; and transmitting the corresponding compressed data based on the first transmission protocol and the second transmission protocol, wherein the reliability of the first transmission protocol is higher than that of the second transmission protocol.

[0062] The compressed key area data is denoted as (Lossless compressed data blocks), non-critical area data is denoted as (Lossively compressed data blocks), both need to be encapsulated with additional metadata: the metadata includes compression parameters (prediction order of key regions). Entropy-encoded dictionary version; wavelet decomposition levels for non-critical regions. Quantization step size ), timestamp (Data start time), verification information (critical areas use CRC32 checksum) Non-critical areas use simple checksums

[0063] The first transmission protocol uses TCP (Transmission Control Protocol), configured as follows: Sliding window size (Unit: bytes), where Real-time bandwidth (unit: bps). Round-trip time (measured via ICMP protocol, averaged over the last 10 times). Indicates rounding down; retransmission timeout. When key data packets are lost (more than Immediate retransmission is triggered when key data packets are lost (more than The second transmission protocol uses UDP (User Datagram Protocol), closes the checksum field (reduces protocol overhead), and sets the maximum length of the packet bytes (adapted to the Ethernet MTU limit), and does not trigger retransmission when non-key data packets are lost.

[0064] Bandwidth allocation is required before transmission, and the priority weight of key data is defined , and the weight of non-key data is , then the actual allocated bandwidth satisfies ( is the current available bandwidth, is the bandwidth allocated for key data transmission, is the bandwidth allocated for non-key data transmission). When ( is the minimum required bandwidth for key data, , is the maximum allowed transmission delay), key data transmission is prioritized, and non-key data transmission is suspended until the bandwidth is restored, is the operation of calculating the size of the data, represents the size of the data of the key data .

[0065] During data transmission, control frames are sent synchronously, including current compression parameters (such as entropy encoding dictionary index for lossless compression of key areas, quantization step ) and transmission state flags (such as indicates key data transmission, indicates non-key data transmission). The receiving end calls the corresponding decoding algorithm by parsing the control frame, wherein key data decoding needs to verify the CRC32 check code , and requests retransmission when , non-key data is directly reconstructed based on wavelet inverse transform without the need for verification matching.

[0066] Further, the method further comprises: monitoring the transmission bandwidth, the electroencephalogram signal characteristics, and the application scenario in real time; and dynamically adjusting the division proportion of the key area and the non-key area according to changes in the transmission bandwidth, the electroencephalogram signal characteristics, and the application scenario.

[0067] Among them, dynamically adjusting the division proportion according to the change of the transmission bandwidth comprises: when the transmission bandwidth is lower than a preset bandwidth threshold, reducing the proportion of the key area; and when the transmission bandwidth is higher than or equal to the preset bandwidth threshold, increasing the proportion of the key area.

[0068] The division ratio is dynamically adjusted based on changes in the EEG signal characteristics, including: expanding the range of the key region when a preset transient event occurs in the EEG signal characteristics; and reducing the range of the key region when the EEG signal characteristics are in a stable state.

[0069] The division ratio is dynamically adjusted according to changes in the application scenario, including: when the application scenario is a clinical diagnosis scenario, the proportion of the key area is controlled to be no less than a preset high threshold; when the application scenario is a real-time monitoring scenario, the proportion of the key area is controlled to be no less than a preset low threshold, wherein the preset low threshold is less than the preset high threshold.

[0070] When monitoring transmission bandwidth in real time, the effective bandwidth is calculated using the sliding window averaging method. ( To determine the window size, take 10 sampling points. For the first Instantaneous bandwidth of this measurement (unit: Mbps). Preset bandwidth threshold. Based on the minimum transmission requirements for critical data: ,in Minimum rate for compressed critical data (The time window is 1 second). When At the same time, by increasing the threshold of key features (such as the peak threshold) from Adjusted to mutation threshold improve Reduce the proportion of key areas, and adjust the formula as follows: (for adjustment coefficients); when When, lower the threshold (e.g.) To increase the proportion of key areas.

[0071] EEG signal feature monitoring is achieved by real-time calculation of transient event detection indicators: defining transient event scores. , As weight, when The event is then identified as a preset transient event (such as epileptic spikes or sleep spindles). At this point, the critical region is expanded to include the events before and after the event. (51 sampling points corresponding to a 512Hz sampling rate) are included in the key region, i.e., the mask matrix. Expand within the event time window (Before and after the original event window) (within the range). When If the duration exceeds 5 seconds, it is determined to be a stable state, and the critical area is narrowed: the mask value of the non-event area is adjusted. (Original key area) (part of the text).

[0072] Application scenario monitoring through scenario identification tags (0 for real-time monitoring, 1 for clinical diagnosis) is realized, and the tags are generated by external input (such as manual switching by the user) or a scenario recognition model (based on device location, data sampling rate, and other features). A preset high threshold (clinical diagnosis), a preset low threshold (real-time monitoring), key area proportion ( total time points) must be met: when , , all feature thresholds are lowered (such as ) to ensure compliance; when , , the threshold is allowed to be raised to reduce the proportion while ensuring that core features are not lost (such as the wave band energy area in the motor imagery task is still retained), is the scenario identification tag mentioned above, taking the value of 0 (real-time monitoring) or 1 (clinical diagnosis), used to distinguish the application scenario of electroencephalogram data, generated by external input or a scenario recognition model, is the value of the original time threshold adjusted to meet the key area proportion requirement in the clinical diagnosis scenario, , used for dynamic adaptation of key area determination.

[0073] Dynamic adjustment smooth transition mechanism: all threshold and mask matrix changes are realized by linear interpolation, that is, ( is the adjustment trigger time, is the completion time), avoiding data feature distortion caused by sudden changes.

[0074] Based on the foregoing, the method of this step acquires original electroencephalogram data through multi-channel high-precision sampling to provide a high-fidelity data source for subsequent processing, combines time-space feature extraction and quantitative threshold division to identify core information containing clinical diagnosis markers; through linear prediction and entropy coding, the key area is losslessly compressed, eliminating temporal redundancy while retaining feature details; wavelet transform and dynamic quantization step are used for lossy compression of non-key areas, utilizing spatial and frequency domain redundancy to achieve efficient data reduction; a dual-protocol transmission mechanism uses a reliable retransmission strategy for key data to ensure that core information is not lost, and a lightweight protocol is used for non-key data to improve transmission efficiency; a dynamic adjustment mechanism optimizes the area division proportion flexibly by monitoring bandwidth, signal features, and scenario changes in real time, so that the compression strategy adapts to different transmission environments and data characteristics.

[0075] The above technical means synergistically act to realize, in the transmission process of electroencephalogram data, significant reduction of data volume by hierarchical compression to improve transmission efficiency, and ensure that core features related to clinical diagnosis and tasks are not lost by key area lossless processing and dynamic adaptation mechanism, while balancing the transmission reliability and real-time performance of different data by means of dual protocol transmission and bandwidth allocation strategy, ultimately forming a set of electroencephalogram data transmission scheme with high efficiency, reliability and scene adaptability, to meet the diversified application requirements from clinical diagnosis to real-time monitoring.

[0076] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited by the order of the described actions, because according to the embodiments of the present application, certain steps can be performed in other order or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present application.

[0077] In summary, the present application at least contains the following beneficial effects:

[0078] 1. By using the hierarchical strategy of lossless compression for key areas and lossy compression for non-key areas, the data transmission volume is significantly reduced while ensuring the integrity of clinical diagnosis markers and task core features, balancing data availability and transmission efficiency;

[0079] 2. The dynamic adjustment mechanism based on transmission bandwidth, signal characteristics and application scenarios can flexibly optimize the division ratio of key areas and non-key areas, so that the method can adapt to different transmission environments and data characteristics, improving the scene adaptability;

[0080] 3. High-reliability protocol and lightweight protocol are used for transmission of key data and non-key data respectively, which ensures the lossless of core information and improves the overall transmission speed, balancing the reliability and real-time performance of transmission, and meeting the diversified application requirements.

[0081] The above description is only the preferred embodiments of the present application and the explanation of the technical principles used. Those skilled in the art should understand that the disclosed range in the present application is not limited to the technical solutions formed by the specific combination of the above technical features, and also covers other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the disclosed concept. For example, the above features are replaced with the technical features disclosed in the present application (but not limited to) with similar functions to form technical solutions.

Claims

1. A method for transmission of electroencephalography data combining lossless and lossy compression, characterized in that, The method comprises the following steps: acquiring original electroencephalogram data; dividing the original electroencephalogram data into key region and non-key region; processing the key region data after division by using a lossless compression algorithm, and processing the non-key region data after division by using a lossy compression algorithm; transmitting the key region data and the non-key region data after compression processing; dividing the original electroencephalogram data into key region and non-key region, comprising the following steps: extracting time domain features, frequency domain features and spatial features of the original electroencephalogram data; determining a key feature threshold based on the time domain features, the frequency domain features and the spatial features; dividing the original electroencephalogram data into key region and non-key region according to the key feature threshold; When extracting time domain features, for the original EEG data matrix , the local peak value of each channel is calculated : , is the window size, and the peak threshold value is defined; At the same time, the first order difference absolute value is calculated, which is used to represent the mutation degree of the signal, and the mutation threshold value is set to , is the mean function; Frequency domain feature extraction adopts short-time Fourier transform to frame each channel data, and calculates frequency resolution , is the frame length, and a frequency domain matrix is obtained , is the frequency point number, is the frame number; the extracted features include the energy proportion of each frequency band: Energy , of the wave band, which corresponds to the key feature of sleep staging, Energy , of the wave band, which corresponds to the core feature of motor imagery tasks, and a frequency band energy threshold is set ; Spatial features are defined by cross-channel correlation computation With Pearson correlation coefficient Where Covariance, Standard deviation; synchrony threshold Set to 0.7, when Strong spatial association between two channels is determined. When dividing, a feature vector is generated for each segment of data , is the average correlation coefficient with adjacent channels, is the vector combination compliance, which is used to combine multiple features in parentheses into a vector as needed, when When at least one feature in the feature vector exceeds the corresponding threshold value, mark this time period as a key area, otherwise as a non-key area, and finally generate a binary mask matrix , represents the first channel is a key area at the first moment.

2. The electroencephalography data transmission method incorporating lossless and lossy compression according to claim 1, wherein, the key feature threshold is set based on clinical diagnosis marker features and task-related core features, the key region is an electroencephalogram data region containing the clinical diagnosis marker features or the task-related core features, and the non-key region is an electroencephalogram data region not containing the clinical diagnosis marker features and the task-related core features.

3. The electroencephalography data transmission method incorporating lossless and lossy compression according to claim 1, wherein, Further comprising: real-time monitoring of transmission bandwidth, electroencephalogram signal features and application scenarios; dynamically adjusting the division proportion of the key region and the non-key region according to changes in the transmission bandwidth, the electroencephalogram signal features and the application scenarios.

4. The electroencephalography data transmission method incorporating lossless and lossy compression according to claim 3, wherein, dynamically adjusting the division proportion according to changes in the transmission bandwidth, comprising: when the transmission bandwidth is lower than a preset bandwidth threshold, reducing the proportion of the key region; when the transmission bandwidth is higher than or equal to the preset bandwidth threshold, increasing the proportion of the key region.

5. The electroencephalography data transmission method incorporating lossless and lossy compression according to claim 3, wherein, dynamically adjusting the division proportion according to changes in the electroencephalogram signal features, comprising: when the electroencephalogram signal features appear a preset transient event, expanding the range of the key region; when the electroencephalogram signal features are in a steady state, reducing the range of the key region.

6. The electroencephalography data transmission method incorporating lossless and lossy compression according to claim 3, wherein, dynamically adjusting the division proportion according to changes in the application scenarios, comprising: when the application scenario is a clinical diagnosis scenario, controlling the proportion of the key region to be not lower than a preset high threshold; when the application scenario is a real-time monitoring scenario, controlling the proportion of the key region to be not lower than a preset low threshold, the preset low threshold being smaller than the preset high threshold.

7. The electroencephalography data transmission method incorporating lossless and lossy compression according to claim 1, wherein, processing the key region data after division by using a lossless compression algorithm, comprising: performing linear predictive coding on the key region data to obtain a prediction error sequence; processing the prediction error sequence by using an entropy coding algorithm to obtain lossless compression data.

8. The electroencephalography data transmission method incorporating lossless and lossy compression according to claim 1, wherein, processing the non-key region data after division by using a lossy compression algorithm, comprising: performing wavelet transform on the non-key region data to obtain low-frequency coefficients and high-frequency coefficients; determining a quantization step based on the energy value of the non-key region data, the quantization step being positively correlated with the energy value; quantizing the high-frequency coefficients by using the quantization step to obtain lossy compression data.

9. The electroencephalography data transmission method incorporating lossless and lossy compression according to claim 1, wherein, transmitting the key region data and the non-key region data after compression processing, comprising: configuring a first transmission protocol for the key region data after compression processing, and configuring a second transmission protocol for the non-key region data after compression processing; transmit corresponding compressed data based on the first transmission protocol and the second transmission protocol, respectively, wherein reliability of the first transmission protocol is higher than reliability of the second transmission protocol.

Images