Electroencephalogram epilepsy prediction system based on fuzzy neural network

Through the EEG epilepsy prediction system based on fuzzy neural network, the hydrogel electrode and cloud edge collaboration technology is used to solve the problem of information loss in signal preprocessing, improve the prediction accuracy and speed, and is suitable for portable EEG acquisition devices.

CN118902465APending Publication Date: 2024-11-08CHANGCHUN UNIV OF SCI & TECH
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
CN202411016193.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-27
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The existing EEG epilepsy prediction system will lead to the loss of useful information during the signal preprocessing step, resulting in a decrease in the accuracy of the prediction results.

Method used

Using an electroencephalogram prediction system based on fuzzy neural network, the signal acquisition is collected using hydrogel electrodes, the EEG signal is processed through the ADS1299 analog-to-digital conversion module and the ESP32-S3 microcontroller, and prediction and transmission are carried out through the cloud server and the CoAP protocol. , reduce information loss and improve prediction speed.

Benefits of technology

It improves the accuracy of epilepsy prediction, reduces the system reaction time, and accelerates the prediction speed through cloud-edge collaboration technology, and the convenience and low impedance of hydrogel electrodes make signal integrity better.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118902465A_ABST
    Figure CN118902465A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of cloud edge collaboration, in particular to an electroencephalogram epilepsy prediction system based on a fuzzy neural network, comprising: an electroencephalogram acquisition electrode, the electroencephalogram acquisition electrode adopts a hydrogel electrode, and the lower impedance of the electroencephalogram acquisition electrode is beneficial to the subsequent processing of a signal amplifier and an analog-to-digital converter; the wireless module adopts an ADS1299 and is used for receiving a signal acquired by the electroencephalogram electrode and transmitting data to the microcontroller through an SPI (Serial Peripheral Interface) communication protocol; the microcontroller adopts ESP32-S3 and is used for receiving the electroencephalogram signals processed by the analog-to-digital conversion module, performing data processing on the read electroencephalogram signals and sending the processed data to the cloud server through the Bluetooth module, and the cloud server is used for receiving the data sent by the microcontroller and sending the data to the microcontroller through the Bluetooth module. And the data are analyzed, processed and predicted. The electroencephalogram signal acquisition system has the advantages of high performance, high flexibility and low power consumption.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention is an electroencephalogram (EEG) epilepsy prediction system based on fuzzy neural network, belonging to the field of cloud-edge collaborative technology. Background Art

[0002] An epileptic seizure (ES) is caused by a sudden, abnormal, self-sustaining discharge in brain networks that usually lasts less than a few minutes. ES attacks are difficult to predict, and the severity and duration of the attack are also unpredictable. Therefore, the harm and safety issues caused by the incident are a major concern for patients and their families. Therefore, early prediction of epileptic seizures is crucial to avoid and cope with their adverse consequences. Brain activity in patients with epilepsy can be divided into different states: pre-ictal, intra-ictal, post-ictal, and interictal. ES prediction is a classification problem, that is, distinguishing between pre-ictal and interictal states. Due to the recurrent nature of epilepsy, ES occurs in groups, and patients with epilepsy groups can gain an advantage by predicting subsequent epileptic seizures.

[0003] Electroencephalography (EEG) is a particularly effective diagnostic tool for studying the functional anatomy of the brain during ES attacks. The prediction and treatment of epilepsy through EEG has been extensively studied. EEG signals are non-Gaussian and non-stationary, and are used to measure electrical activity in the brain, which in turn is used to diagnose the type of brain disease. Analysis of EEG measurements helps to distinguish between normal and abnormal brain function. In order to accurately predict epilepsy, it is necessary to examine EEG recordings of longer duration. Neurologists examine epilepsy by studying EEG signals recorded over several days, weeks, or even months, which requires a lot of manpower and time. Over the years, various studies have adopted prediction methods based on machine learning (ML) to solve this problem. Deep learning (DL) is an advanced ML technique that is able to process large amounts of data through a multi-layer hierarchy to learn patterns from it more accurately. The ability of DL to produce very accurate results has influenced researchers to use DL technology to handle many practical applications, and in the past five to six years, many researchers have proposed DL-based ES prediction methods.

[0004] At present, through the prediction of EEG epilepsy by different neural networks, good prediction results have been achieved for epileptic EEG signals, but most neural networks will cause some useful information to be lost when converting the initial signal into classification attributes in the signal preprocessing step. The loss of this information will lead to a decrease in the accuracy of the prediction results. Summary of the invention

[0005] In view of the above problems, the present invention provides an EEG epilepsy prediction system based on fuzzy neural network, the system comprising:

[0006] The EEG collection electrode adopts a hydrogel electrode, which can maintain an average impedance of no more than 100kΩ within 4 hours, and can reach 50kΩ at best, which is close to the impedance level of a wet electrode and meets basic BCI application scenarios. The lower impedance makes the signal integrity of the EEG signal better, which is beneficial to the subsequent processing of the signal amplifier and analog-to-digital converter.

[0007] The analog-to-digital conversion module uses ADS1299, which is used to receive EEG signals. The MCU module sends a start conversion command to the ADC through the SPI bus. Then the ADC works at the configured sampling frequency and informs the MCU through the pin signal that the data is ready. The MCU reads the data of all 8 channels from the ADC through the SPI bus again. The above process is repeated until a stop command is sent to inform the ADC to stop converting. And the processed signal is sent to the microcontroller.

[0008] The microcontroller adopts ESP32-S3, which is used to receive the EEG signal processed by the analog-to-digital conversion module, process the read EEG signal, and send the data to the cloud server.

[0009] The cloud server is used to receive the electroencephalogram (EEG) signals processed by the microcontroller and predict the processed signals.

[0010] The wireless transmission module uses the ESP8266 module of Espressif Technology to send the EEG signals processed by the microcontroller to the cloud server through the CoAP protocol.

[0011] A buzzer is used to alarm for predicted epilepsy.

[0012] In the fuzzy neural network-based EEG epilepsy prediction system of the present invention, the microcontroller communicates with the wireless transmission module via a serial communication protocol, and the baud rate is 115200.

[0013] Beneficial Effects

[0014] Compared with the existing EEG epilepsy prediction system technology, the present invention has the following beneficial effects:

[0015] 1) Compared with traditional wet electrodes, the hydrogel electrodes used in this EEG epilepsy prediction system are easy and quick to wear, and there is no need to clean the scalp and the residue on the electrodes after use. In addition, the production process of hydrogel is simple, with fewer steps, easy to operate, and easy to mass produce. In addition, hydrogel electrodes are more comfortable and convenient than wet electrodes.

[0016] 2) Compared with the previous EEG epilepsy prediction system, which loses some useful information when extracting features, this system loses less information during signal preprocessing, which can improve the accuracy of epilepsy prediction.

[0017] 3) By deploying the EEG epilepsy prediction algorithm on the cloud server using the cloud-edge collaboration method and transmitting information using the CoAP protocol, the response time of the system can be reduced and the prediction speed of the EEG epilepsy prediction system can be accelerated. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is the overall structure flow chart of the present invention;

[0019] Figure 2 This is the architecture diagram of the EEG acquisition system;

[0020] Figure 3 This is the pin diagram of the microprocessor ESP32-S3 chip;

[0021] Figure 4 This is the pin diagram of the analog-to-digital conversion chip ADS1299;

[0022] Figure 5 It is the structural diagram of fuzzy neural network; DETAILED DESCRIPTION

[0023] The embodiments of the present invention are further described below in conjunction with the accompanying drawings:

[0024] The overall process of the EEG epilepsy prediction system based on fuzzy neural network of the present invention is as follows: first, the EEG signal is collected through the EEG acquisition system mainboard, that is, the high-resolution analog-to-digital conversion chip ADS1299 is used to collect the EEG epilepsy signal in real time. Then the signal is transmitted to the main control module with ESP32-S3 as the core through the SPI interface. The main control module sends the data to the cloud server through the Bluetooth module ESP8266 to predict the EEG epilepsy signal. The EEG acquisition system mainboard includes all hardware modules except the EEG acquisition electrode, which is used to complete the actual EEG acquisition task. The EEG acquisition electrode is connected to the EEG acquisition system mainboard through the electrode interface, can be replaced independently, and is used to conduct scalp EEG signals. The EEG signal sampling rate of this system is set to 500Hz, and the sampling electrode position refers to the "International 10-20 System". The electrode placement positions of the 8 channels shown are F3, F4, P3, P4, T3, T4, T5, and T6 respectively.

[0025] Attached Figure 1It is the overall structural principle diagram of the present invention, including EEG acquisition electrodes, analog-to-digital conversion modules, microcontrollers, cloud server algorithms, and buzzers. Among them: the EEG acquisition electrodes use hydrogel electrodes, which can maintain an average impedance of no more than 100kΩ within 4 hours, and can reach 50kΩ at best, which is close to the impedance level of wet electrodes, meeting the basic BCI application scenarios. The lower impedance makes the signal integrity of the EEG signal better, which is conducive to the subsequent signal amplifier and analog-to-digital converter processing. The analog-to-digital conversion chip uses ADS1299, which is used to receive the EEG signal processed by the analog front end. The MCU module sends a start conversion command to the ADC through the SPI bus; then the ADC works at the configured sampling frequency and informs the MCU through the pin signal that the data is ready; the MCU reads the data of all 8 channels from the ADC again through the SPI bus; repeat the above process until the stop command is sent to notify the ADC to stop the conversion. And send the processed signal to the microcontroller. The microcontroller uses ESP32-S3, which is used to receive the EEG signal processed by the analog-to-digital conversion module and process the read EEG signal. And send the data to the cloud server. The cloud server is used to receive the EEG signals processed by the microcontroller and predict the processed signals. The wireless transmission module uses the ESP8266 module of Espressif Technology to send the EEG signals processed by the microcontroller to the cloud server through the CoAP protocol. The buzzer is used to alarm for predicted epilepsy.

[0026] The following is a detailed introduction to each module:

[0027] 1. EEG collection electrodes:

[0028] Compared with traditional wet electrodes, hydrogel electrodes have the advantages of being easy and quick to wear, and there is no need to clean the residue on the scalp and electrodes after use. At the same time, hydrogel electrodes have good elasticity, and the wearing experience is more comfortable than dry electrodes. The hydrogel electrode is mainly composed of two parts: the first part is the silver-silver chloride electrode and the second part is the hydrogel part. Among them, the sintered silver-silver chloride electrode is encapsulated in a plastic shell, and the shell can be repeatedly opened and closed to replace the hydrogel. The main function of the hydrogel is to replace the use of EEG paste in the wet electrode, eliminating the cumbersome and time-consuming process of injecting EEG paste. The main materials of the hydrogel are carrageenan, conductive salt, deionized water and water retaining agent. These raw materials are easy to obtain on the market and are low in cost. In addition, the production process of the hydrogel is simple, with few steps, easy to operate and easy to mass produce. Experiments show that the hydrogel electrode can maintain an average impedance of no more than 100kΩ within 4 hours, and can reach 50kΩ at best, which is close to the impedance level of the wet electrode, meeting the basic BCI application scenario; the lower impedance makes the signal integrity of the EEG signal better, which is conducive to the subsequent signal amplifier and analog-to-digital converter processing. The questionnaire survey showed that 80% of the subjects who participated in the experiment believed that hydrogel electrodes are better than wet electrodes considering comfort and convenience. The lower impedance makes the signal integrity of the EEG signal better, which is conducive to the subsequent signal amplifier and analog-to-digital converter processing. The better comfort and convenience make hydrogel electrodes more suitable for use in portable EEG acquisition equipment than traditional wet and dry electrodes.

[0029] 2. Analog-to-digital conversion module

[0030] The analog-to-digital conversion module converts the continuous analog signal generated by the acquisition electrode into a discrete digital signal that can be processed by the controller. The analog-to-digital conversion chip of this EEG acquisition system uses the ADS1299 integrated chip from Texas Instruments. This chip is a low-noise, 8-channel, 24-bit synchronous sampling analog-to-digital converter (ADC) designed specifically for biopotential measurement, with a built-in programmable gain amplifier. Each channel of the chip can be configured independently, and the sampling rate can reach 250Hz to 16kHz. The working principle of this module is that the MCU module sends a start conversion command to the ADC through the SPI bus; then the ADC works at the configured sampling frequency and informs the MCU through the pin signal that the data is ready; the MCU reads the data of all 8 channels from the ADC continuously through the SPI bus again; the above process is repeated until a stop command is sent to notify the ADC to stop converting.

[0031] 3. Microcontroller

[0032] 1) Introduction to ESP32-S3

[0033] ESP32-S3 is a low-power MCU system-on-chip (SoC) that supports 2.4GHz, Wi-Fi, and low-power Bluetooth ( LE) wireless communication. The chip integrates high-performance 32-bit LX7 dual-core processor, ultra-low power coprocessor, Wi-Fi baseband, Bluetooth baseband, RF module and peripherals.

[0034] 2) Introduction to ESP32-S3's SPI Function

[0035] SPI (Serial Peripheral Interface) is a synchronous serial communication interface, which is usually used to connect microcontrollers and external devices, such as memory chips, sensors, LCD drivers, etc. The SPI interface usually consists of four wires: master output (MOSI), master input (MISO), clock (SCLK) and slave select (CS). The SPI interface has the advantages of high data transmission speed, simple implementation and low hardware resource requirements, so it is widely used in many embedded systems.

[0036] The SPI interface of ESP32-S3 supports full-duplex communication, supports multiple clock polarity and phase settings to adapt to different peripheral requirements, supports hardware data flow control and DMA transfer mode, which can effectively reduce the burden on the CPU. Its SPI supports interrupt and polling data transmission methods.

[0037] 4. CoAP protocol

[0038] 1) Introduction to CoAP Protocol

[0039] CoAP (Constrained Application Protocol) is a communication protocol designed for the Internet of Things. It is based on the RESTful architecture and uses the UDP transmission protocol to achieve fast and reliable data transmission. The CoAP protocol has the advantages of being lightweight, efficient, and easy to expand, and is suitable for communication between resource-constrained embedded devices.

[0040] 2) Advantages of CoAP

[0041] Lightweight: The CoAP protocol data packet format is concise, which effectively reduces the amount of data transmission and processing complexity, and reduces the consumption of device resources. Efficient: The CoAP protocol uses the UDP transmission protocol to improve the efficiency of data transmission and is suitable for IoT scenarios with high real-time requirements. Easy to expand: The CoAP protocol supports the RESTful architecture, and can easily expand or delete instructions by designing appropriate URIs to meet the needs of different devices. Reliable: The CoAP protocol provides two message types: confirmation message (CON) and non-confirmation message (NON), which ensures the reliable transmission of messages. Asynchronous communication: The CoAP protocol supports asynchronous communication mode, and devices can transmit and interact data without maintaining a long connection. Security: The CoAP protocol supports encryption and authentication functions to ensure the security of data transmission.

[0042] 5. Fuzzy Neural Network

[0043] The network structure of the fuzzy neural network is as follows Figure 5 As described above, data input into the fuzzy neural network needs to go through a total of five layers of network structure, the first of which is the input layer. The number of nodes in this layer is the characteristic dimension of the input data, that is, when the characteristic dimension of the data is n, the number of nodes in the input layer is n.

[0044] Then the data is passed from the input layer to the membership function calculation layer. In this layer, the membership function is needed to calculate the membership of each node. Each node represents a membership function. The number of nodes in this layer is the number of fuzzy conditions that the input variables may constitute. The membership function generally uses the Gaussian function, and the formula is as follows:

[0045]

[0046] where u ij is the output of the node, x i is the input data, c ij is the center point of the membership function, b ij is the width vector of the membership function, and m is the number of fuzzy classifications of the input. The data is then passed to the rule generation layer. The number of nodes in this layer is the number of fuzzy rules. The output of this layer is the applicability of each rule, which is generally calculated using the following formula:

[0047] α j =Πu ij

[0048] Next, the node outputs of the rule generation layer are normalized in the normalization layer. The nodes in this layer are the same as those in the rule generation layer, and are normalized using the following formula:

[0049]

[0050] Finally, the data is weighted in the output layer to obtain the final output. The calculation formula is:

[0051]

[0052] where ω j is the connection weight of the output layer. The above formula is the weight set when the dimension of the output variable of the network is 1. When the dimension of the output variable increases, the weight will be adjusted accordingly.

Claims

1. A fuzzy neural network-based EEG epilepsy prediction system, which is characterized by include: The electroencephalogram (EEG) collecting electrode adopts a hydrogel electrode, and its lower impedance makes the signal integrity of the EEG signal better, which is beneficial to the subsequent processing of the signal amplifier and analog-to-digital converter. The analog-to-digital conversion module includes an ADS1299 chip, which is used to collect the EEG signals processed by the analog front end and send the collected EEG signals to the microcontroller. The microcontroller, whose model is ESP32-S3, is used to receive the EEG signal data from the analog-to-digital conversion module, and process and package the data before sending it to the host computer. The cloud server is used to receive the electroencephalogram (EEG) signals processed by the microcontroller and predict the processed signals. The wireless transmission module uses the ESP8266 module of Espressif Technology to send the EEG signals processed by the microcontroller to the cloud server through the CoAP protocol. A buzzer is used to alarm for predicted epilepsy.

2. The EEG epilepsy prediction system based on fuzzy neural network as claimed in claim 1, Features: By deploying the EEG epilepsy prediction algorithm on the cloud server using the cloud-edge collaboration method and transmitting information using the CoAP protocol, the system's response time can be reduced and the prediction speed of the EEG epilepsy prediction system can be accelerated.

3. The EEG epilepsy prediction system based on fuzzy neural network as claimed in claim 1, Features :The microcontroller communicates with the wireless transmission module through the serial communication protocol. The baud rate is 115200.

Citation Information

Patent Citations

  • Pre-epileptic EEG signal recognition method based on GBDT model and cloud system

    CN109924973A

  • Epileptic seizure early warning system based on portable electroencephalogram collection equipment and application thereof

    CN111616682A

  • Electroencephalogram acquisition system based on ADS1299, STM32 and ESP8266

    CN113274036A

  • Prediction system, method and device for predicting obsessive-compulsive disorder treatment effect of tDCS based on EEG, processor and storage medium thereof

    CN113694374A

  • Wearable minimally invasive closed-loop regulation and control device for infant epilepsy

    CN115177209A