High-throughput tongue electrophysiology detection system and method
By combining a high-density flexible sensor array with a high-performance processor, the problems of low resolution and limited processing power in the tongue electrophysiological signal acquisition system have been solved, enabling high-throughput and high-precision detection and analysis of tongue electrophysiological signals.
Patent Information
- Application Number
- CN202511720662.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-21
- Publication Date
- 2025-12-30
AI Technical Summary
Existing tongue electroacupuncture signal acquisition systems suffer from low resolution, low data transmission efficiency, and limited processing capabilities, making it difficult to meet the requirements for high-throughput and high-precision detection.
The system employs a high-density flexible sensor array and a working electrode made of highly biocompatible materials. Combined with a multi-channel preprocessing unit, an analog multiplexer, and an ADC, it achieves high-density signal acquisition and transmission. Signal processing and analysis are performed using a high-performance processor and algorithm library.
It achieves high-throughput, high-spatial-resolution detection of tongue electrophysiological signals, improves data acquisition and transmission efficiency, and supports tongue electrophysiological research, disease diagnosis, and rehabilitation training.
Smart Images
Figure CN121221118A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of biomedical engineering and electrophysiological detection technology, specifically to a high-throughput tongue electrophysiological detection system and method. Background Technology
[0002] In the fields of neuroscience and biomedicine, the detection and analysis of tongue electrophysiological signals is gradually becoming an important research direction. The tongue plays a crucial role not only in taste perception but is also closely related to various physiological functions such as language, chewing, and swallowing. Therefore, accurate detection and analysis of tongue electrophysiological signals can not only promote the understanding and diagnosis of oral cavity and related functional disorders (such as speech disorders and dysphagia) but also be applied to human-computer interaction technologies such as voice control and prosthetic control.
[0003] Although the study of tongue electrical signals has broad application potential, existing tongue electrical signal acquisition systems still have many limitations and cannot meet the growing demand for high-precision research.
[0004] (1) Low resolution: Most existing acquisition devices have low resolution, making it difficult to capture subtle changes in electrophysiological signals. This is a significant problem in the complex and subtle movements of the tongue, as tiny signals may contain crucial information.
[0005] (2) Low data transmission efficiency: Due to the slow data transmission speed in the existing system, the collected data may be delayed or lost during transmission, which affects the integrity and real-time performance of the signal.
[0006] (3) Limited processing capacity: Traditional processing systems are not efficient in dealing with large-scale data analysis and are difficult to process and interpret high-density electrophysiological data quickly, especially when performing multi-channel, large-scale parallel signal processing.
[0007] In summary, existing tongue electrophysiological signal acquisition systems cannot meet the requirements for high-throughput and high-precision detection. Therefore, it is particularly urgent to develop an integrated and efficient tongue electrophysiological detection system to solve the above problems. Summary of the Invention
[0008] To address the shortcomings of existing technologies, the present invention aims to provide a high-throughput tongue electrophysiological detection system and method that can acquire tongue electrophysiological signals in real time, achieve high-throughput and high spatial resolution signal detection through a high-density sensor array, and achieve accurate analysis of tongue electrophysiological signals through efficient data acquisition, transmission and processing, providing reliable support for tongue electrophysiological research, disease diagnosis and rehabilitation training.
[0009] This invention provides a high-throughput tongue electrophysiological detection system, comprising: The tongue electrosensing module includes a working electrode and a reference electrode. The working electrode is configured as a distributed flexible tongue electrode, which is configured to cover the tip, sides, middle and root of the tongue, and has a three-dimensional curved surface that matches the surface of the tongue. The tongue electrosensing module is configured to collect electrophysiological information from different regions of the tongue. A data acquisition and transmission module is used to acquire multiple channel information corresponding to microvolt-level electrophysiological signals from different regions of the tongue, and convert the multiple channel information into digital information. The data acquisition and transmission module further includes multiple channel preprocessing units, an analog multiplexer, an ADC, and a synchronization unit. Each of the multiple channel preprocessing units further includes an independent low-noise amplifier and a programmable upper and lower cutoff frequency filter for each channel, used to amplify and filter the analog signals corresponding to the microvolt-level electrophysiological signals from different regions of the tongue acquired by different channels. The multiple channels corresponding to the analog multiplexer MUXN are quickly switched via the MUXN, and the analog signals output from different channels are sequentially sent to the ADC for conversion to obtain the corresponding digital signals. The synchronization unit is configured to synchronize the sampling time of all channels. The data receiving and processing module is used to process and analyze the digital signals transmitted by the data acquisition and transmission module.
[0010] Furthermore, the data receiving and processing module includes: A digital signal receiving unit is used to receive the digital signal transmitted by the data acquisition and transmission module; A signal denoising unit is used to perform denoising processing on the digital signal to obtain a denoised digital signal; A signal filtering unit is used to filter the denoised digital signal to obtain a filtered digital signal. The feature extraction unit is used to perform feature extraction processing on the filtered digital signal to obtain the processed digital signal features. The classification and decoding unit is used to classify and decode the features of the processed digital signal to obtain the classification and decoding result. A data visualization unit is used to visualize the filtered digital signal.
[0011] Furthermore, the flexible tongue electrode is also etched with a third through-hole structure for discharging secretions, and the reference electrode is set as a patch electrode attached to both sides of the cheek; the distributed layout indicates that the electrode array is configured as a core functional area and an auxiliary monitoring area to perform targeted acquisition at different spatial resolutions, and the electrode array arrangement density in the core functional area is greater than the electrode array arrangement density in the auxiliary monitoring area.
[0012] Furthermore, the flexible tongue electrode includes: A flexible substrate layer configured to cover the tip, sides, middle and root of the tongue, and having a three-dimensional curved surface that matches the surface of the tongue. A conductive layer is configured as a patterned electrode array distributed on the flexible substrate. The conductive layer includes multiple electrode units and interconnecting wires. The interconnecting wires are used for electrical connections between and within the multiple electrode units. Each electrode unit is configured to cover the tip of the tongue, the side of the tongue, the middle of the tongue, and the root of the tongue, respectively. The electrode arrays of each electrode unit acquire the same or different spatial resolutions. A flexible encapsulation layer is in contact with the tongue and is configured to cover the tip, sides, middle and root of the tongue, and has a three-dimensional curved surface that matches the surface of the tongue. The flexible encapsulation layer is etched with at least a first through-hole structure for exposing electrode sites, a second through-hole structure for exposing back-end pads and a third through-hole structure for draining secretions. The signal processing module is electrically connected to each of the electrode units via the interconnecting wires and is used for signal processing.
[0013] Furthermore, the feature extraction unit includes: A time-domain feature extraction unit is used to extract time-domain features from the filtered digital signal; the time-domain features include the mean absolute value, mean, variance, skewness, and kurtosis of the filtered digital signal. A frequency domain feature extraction unit is used to extract frequency domain features from the filtered digital signal; the frequency domain features include the weighted average frequency, root mean square frequency, and relative frequency variability of the filtered digital signal. The time-frequency domain feature extraction unit is used to combine the time-domain features and the frequency-domain features, and extract the time-frequency domain features from the filtered digital signal using wavelet transform or short-time Fourier transform.
[0014] Furthermore, the calculation formula for extracting the time-frequency domain features in the filtered digital signal using the wavelet transform method in the time-frequency domain feature extraction unit is as follows: ; Wherein, CWT stands for Continuous Wavelet Transform. The original signal, Let a be the wavelet basis function, a be the scale, and b be the translation. Discrete Wavelet Transform (DWT) can decompose a signal into: ; ; in, The original signal, It is a low-pass filter. It is a high-pass filter; The calculation formula for extracting time-frequency domain features from the filtered digital signal using the short-time Fourier transform method in the time-frequency domain feature extraction unit is as follows: ; in, The original signal, For window functions, It is the Fourier kernel.
[0015] Furthermore, the classification and decoding unit is used to perform pattern classification and functional decoding on the time-domain features, frequency-domain features, and time-frequency-domain features in the filtered digital signal based on machine learning algorithms or deep learning algorithms, so as to obtain the classification and decoding result.
[0016] Furthermore, the classification and decoding unit is used to train the sample model based on logistic regression, decision tree, support vector machine or neural network using sample data and sample model, adjust the model parameters of the sample model until the sample model meets the training conditions to obtain the target model, and is used to perform pattern classification and functional decoding on the time domain features, frequency domain features and time-frequency domain features in the filtered digital signal through the target model to obtain the classification and decoding result; The sample data refers to data that has been labeled with taste classification information.
[0017] This invention also provides a high-throughput tongue electrophysiological detection method, applied to any of the high-throughput tongue electrophysiological detection systems described above, the method comprising: The analog signal is acquired using the tongue electrosensing module; the analog signal includes the signal acquired by the working electrode and the signal acquired by the reference electrode. The analog signal acquired by the tongue electrosensing module is processed by the data acquisition and transmission module and then converted from analog to digital to obtain a digital signal. The digital signal is transmitted to the data receiving and processing module using the data acquisition and transmission module. The digital signal is processed and analyzed using the data receiving and processing module.
[0018] Furthermore, the method includes: The analog signal is acquired using the tongue electrosensing module; the analog signal includes the signal acquired by the working electrode and the signal acquired by the reference electrode. The analog signal acquired by the tongue electrosensing module is amplified by the preamplifier unit in the data acquisition and transmission module to obtain the preamplified analog signal. The preamplified signal is filtered using the bandpass filter unit in the data acquisition and transmission module to obtain a filtered analog signal. The filtered analog signal is converted into the digital signal using the analog-to-digital conversion unit in the data acquisition and transmission module. The digital signal is transmitted to the data receiving and processing module using the transmission unit in the data acquisition and transmission module. The digital signal received by the data acquisition and transmission module is received by the digital signal receiving unit in the data receiving and processing module; The digital signal is denoised using the signal denoising unit in the data receiving and processing module to obtain a denoised digital signal. The signal filtering unit in the data receiving and processing module is used to filter the denoised digital signal to obtain a filtered digital signal. The feature extraction unit in the data receiving and processing module is used to perform feature extraction processing on the filtered digital signal to obtain the processed digital signal features. The classification and decoding unit in the data receiving and processing module is used to classify and decode the features of the processed digital signal to obtain the classification and decoding result. The filtered digital signal is visualized using the data visualization unit in the data receiving and processing module. The feature extraction process refers to extracting time-domain features, frequency-domain features, and time-frequency-domain features from the filtered digital signal. The time-domain features include the mean absolute value, mean, variance, skewness, and kurtosis of the filtered digital signal. The frequency-domain features include the weighted average frequency, root mean square frequency, and relative frequency variability of the filtered digital signal. The time-frequency-domain features are extracted using wavelet transform or short-time Fourier transform. The classification and decoding are implemented based on logistic regression, decision trees, support vector machines, or neural networks.
[0019] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention provides a high-throughput tongue electrophysiological detection system, comprising a tongue electrosensing module, a data acquisition and transmission module, and a data receiving and processing module. The system can acquire tongue electrophysiological signals in real time, achieve high-throughput and high spatial resolution signal detection through a high-density sensor array, and achieve accurate analysis of tongue electrophysiological signals through efficient data acquisition, transmission and processing, providing reliable support for tongue electrophysiological research, disease diagnosis and rehabilitation training.
[0020] (2) The present invention provides a high-throughput tongue electrophysiological detection system for real-time acquisition, analysis and application of tongue electrophysiological signals. It can solve the problems of low resolution, low data transmission efficiency and limited processing capacity of the existing tongue electrophysiological signal acquisition system, as well as the problem that it cannot meet the detection requirements of high throughput and high precision. It has the advantages of integration and high efficiency.
[0021] (3) The working electrode in the tongue electrosensing module is a high-density flexible sensor array made of highly biocompatible material. It can cover the entire surface of the tongue and achieve complete conformal attachment. The high-density sensor array is used to collect electrophysiological signals from different areas of the tongue and achieves high-throughput and high spatial resolution detection through optimized spatial layout. In addition, the working electrode is equipped with a third through-hole structure, which can effectively drain the liquid secreted on the tongue surface and reduce the accumulation of liquid between the electrode and the tongue.
[0022] (4) The data acquisition and transmission module is responsible for amplifying and filtering the analog signal acquired by the tongue electro-sensing module through multi-channel signal pre-amplification and filtering to improve signal quality. It also converts the analog signal into a digital signal through an analog-to-digital converter (ADC) to provide high-precision input for subsequent processing. Finally, it uses HDMI wired communication to transmit the quantized digital signal data to the digital signal receiving unit in real time without compression and with high quality.
[0023] (5) The data receiving and processing module receives high-throughput tongue electrophysiological digital signals acquired in real time and integrates a high-performance processor and algorithm library for processing and analyzing the transmitted tongue electrophysiological signals. The processing includes signal denoising, signal filtering, feature extraction, classification decoding and multi-dimensional data visualization, and supports machine learning, deep learning and other algorithms for pattern classification and functional decoding of complex signals, providing key data support for disease diagnosis and research. Attached Figure Description
[0024] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 A structural block diagram of a high-throughput tongue electrophysiological detection system provided in an embodiment of the present invention; Figure 2This is a circuit diagram of a data acquisition and transmission module provided in an embodiment of the present invention. Detailed Implementation
[0025] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.
[0026] The core of this invention: 1. Design a circuit to process electrophysiological signals at the microvolt level. The circuit consists of a low-noise amplifier, a programmable filter, a multiplexer (MUX), an analog-to-digital converter (ADC), and synchronization control logic.
[0027] First, a low-noise amplifier (LNA) is crucial because the signal is extremely weak, only in the microvolt range. Therefore, an ultra-low noise op-amp is required. High input impedance is necessary to prevent the signal source from being overloaded, necessitating an instrumentation amplifier architecture. Furthermore, extremely low bias current and noise levels are required, and 1 / f noise must also be considered, especially for low-frequency physiological signals.
[0028] Next is the filtering section. This invention uses a programmable upper and lower cutoff frequencies LFP.
[0029] Next, we discuss the coordination between the multiplexer (MUX) and the ADC. The MUX needs to quickly switch multiple channels, while the ADC's conversion rate must be high enough to process data from multiple channels without losing information. Specifically, the MUX and ADC use time-division multiplexing to sample and convert multiple channels. The high-speed analog multiplexer (MUX) allows the output signal after cascaded amplifiers and filters to be input to the ADC in turn, where the ADC samples and converts the voltage of each channel one by one. While meeting the required sampling rate for tongue electrophysiological signals (<1 kHz), the digital signals of each channel are output sequentially in a continuous SPI data stream, macroscopically approximating simultaneous synchronous acquisition of multiple channels. For synchronization control, a master clock is needed to drive the switching logic of the ADC and MUX, ensuring precise consistency in the sampling time of each channel. The ADC's internal clock and control logic need to be synchronized with the MUX switching to avoid inter-channel delays. Specifically, an FPGA or microcontroller is used to precisely control the timing of the MUX and ADC, ensuring consistent sampling intervals for each channel and compensating for switching time.
[0030] The technical challenge lies in maintaining multi-channel synchronization while simultaneously processing weak signals with low noise. Please refer to [link / reference]. Figure 2Taking a 256-channel tongue electrophysiological detection system as an example, this system adopts a modular parallel architecture, which has good channel independence and synchronization performance. The 256 channels are divided into modules of 32 channels each, consisting of 256 amplification and filtering units (LNA + LPF), 8 analog multiplexers (MUX1~MUX8), 8 analog-to-digital converters (ADC1~ADC8), and sampling timing and digital communication interfaces controlled by FPGA or microcontroller.
[0031] Each module contains 32 independent channels, with the following structure: a. Signal conditioning front end • Low-noise amplifier (LNA): Each channel is equipped with a high input impedance, low-noise instrumentation amplifier suitable for acquiring microvolt-level signals.
[0032] • Low-pass filter (LPF): Used for bandwidth limiting and anti-aliasing filtering; the cutoff frequency can be fixed or adjustable.
[0033] b. Multiplexer (MUX) • Use a 32:1 analog multiplexer to aggregate 32 analog signals into a single output channel.
[0034] • The MUX channel selection signal is generated uniformly by the FPGA or microcontroller to ensure accurate switching timing.
[0035] c. Analog-to-Digital Converter (ADC) Each MUX is connected to an independent ADC. The sampling results of all ADCs are centrally managed and packaged by the controller and transmitted to the host computer or storage system via the HDMI interface.
[0036] The system uses an FPGA or microcontroller to poll and sample the 32 channels of signals within each module. Simultaneously, it achieves parallel synchronous sampling of cross-module channel signals through a shared clock, enabling different modules to acquire corresponding channel signals at the same time, thus achieving synchronous control across modules. After understanding the core points and technical challenges of this invention, the technical solution is described below. Please refer to... Figure 1 The diagram shown is a structural block diagram of a high-throughput tongue electrophysiological detection system provided in an embodiment of the present invention. This high-throughput tongue electrophysiological detection system includes: The tongue electrosensing module includes a working electrode and a reference electrode. The working electrode is a flexible tongue electrode with a distributed layout. The flexible tongue electrode is configured to cover the tip, sides, middle and root of the tongue, and the surface has a three-dimensional curved surface that matches the surface of the tongue. The tongue electrosensing module is configured to collect electrophysiological information from different regions of the tongue. The data acquisition and transmission module is used to acquire multiple channel information corresponding to microvolt-level electrophysiological signals from different regions of the tongue, and convert the multiple channel information into digital information. The data acquisition and transmission module further includes multiple channel preprocessing units, an analog multiplexer, an ADC, and a synchronization unit. The multiple channel preprocessing units further include an independent low-noise amplifier and a programmable upper and lower cutoff frequency filter for each channel, which is used to amplify and filter the analog signals corresponding to the microvolt-level electrophysiological signals from different regions of the tongue acquired by different channels. The multiple channels corresponding to the analog multiplexer MUXN (N is a positive integer) are quickly switched through the analog multiplexer MUXN, and the analog signals output by different channels are sequentially sent to the ADC for conversion to obtain the corresponding digital signals. The synchronization unit is configured to synchronize the sampling time of all channels. The data receiving and processing module is used to process and analyze the digital signals transmitted from the data acquisition and transmission module.
[0037] Specifically, the flexible tongue electrode is also etched with a third through-hole structure for expelling secretions, and the reference electrode is set as a patch electrode attached to both sides of the cheek; the distributed layout indicates that the electrode array is configured as a core functional area (there can be one or more core functional areas. When there are multiple core functional areas, the spatial resolution of each core functional area may be the same or different, that is, the electrode array arrangement of multiple core functional areas can be partially the same, partially different, or completely different) and an auxiliary monitoring area (there are generally multiple auxiliary monitoring areas, and the spatial resolution of each auxiliary monitor may be the same or different, that is, the electrode array arrangement of multiple auxiliary monitoring areas can be partially the same). (Same, partially different, or completely different) to selectively acquire data at different spatial resolutions. The electrode array density in the core functional area is greater than that in the auxiliary monitoring area. The working electrode in the tongue electrosensing module is a high-density flexible sensor array made of highly biocompatible materials. It can cover the entire surface of the tongue and achieve complete conformal attachment. The high-density sensor array is used to acquire electrophysiological signals from different areas of the tongue, and high-throughput, high-spatial-resolution detection is achieved through optimized spatial layout. In addition, the working electrode is equipped with a third through-hole structure, which can effectively drain the liquid secreted by the tongue surface and reduce the accumulation of liquid between the electrode and the tongue.
[0038] The data acquisition and transmission module is responsible for pre-amplifying and filtering the analog signals acquired by the tongue electro-sensing module to improve signal quality. It also converts the analog signals into digital signals through an analog-to-digital converter (ADC) to provide high-precision input for subsequent processing. Finally, it uses HDMI wired communication to transmit the quantized digital signal data to the digital signal receiving unit in real time with high quality and without compression.
[0039] The data receiving and processing module receives high-throughput digital electrophysiological signals of the tongue in real time and integrates a high-performance processor and algorithm library for processing and analyzing the transmitted electrophysiological signals. The processing includes signal denoising, signal filtering, feature extraction, classification decoding, and multi-dimensional data visualization. It also supports machine learning, deep learning, and other algorithms for pattern classification and functional decoding of complex signals, providing key data support for disease diagnosis and research.
[0040] In an optional embodiment, the flexible tongue electrode described above includes: A flexible substrate layer is configured to cover the tip, sides, middle and root of the tongue, and has a three-dimensional curved surface that matches the surface of the tongue. The conductive layer is configured as a patterned electrode array distributed on a flexible substrate. The conductive layer includes multiple electrode units and interconnecting wires. The interconnecting wires are used for electrical connections between and within the multiple electrode units. Each electrode unit is configured to cover the tip, side, middle, and root of the tongue. The electrode arrays of each electrode unit may be arranged the same or different to acquire the same or different spatial resolutions. The flexible encapsulation layer is in contact with the tongue and is configured to cover the tip, sides, middle and root of the tongue. The surface has a three-dimensional curved surface that matches the surface of the tongue. The flexible encapsulation layer is etched with at least a first through-hole structure for exposing electrode sites, a second through-hole structure for exposing back-end pads and a third through-hole structure for draining secretions. The signal processing module is electrically connected to each electrode unit via interconnecting wires and is used for signal processing.
[0041] The third through-hole structure connects to the drainage channel for expelling secretions on the conductive layer, or the third through-hole structure penetrates the conductive layer and the flexible substrate layer to form a drainage channel for expelling secretions. The thickness of the flexible tongue electrode matches the surface curvature of the tongue. The flexible substrate layer and flexible encapsulation layer are made of flexible polymer materials, such as polyimide. The electrode units of the conductive layer adopt a distributed layout, and the electrode arrays of the tongue tip, sides, middle, and root can be divided into core functional areas and auxiliary monitoring areas according to specific needs. The interconnecting wires are designed with multi-layer wiring. Specifically, the core functional area is a densely arranged electrode array, and the electrode array arrangement can be optimized according to the shape of the lesion (such as square array, ring array, honeycomb array, etc.). The auxiliary monitoring area is a sparsely arranged electrode array. The definition of dense and sparse can be obtained according to preset conditions, such as dense within one threshold range and sparse within another threshold range. The electrode array arrangement includes the arrangement of the number of electrodes per unit area, electrode size, and electrode spacing.
[0042] Specifically, the flexible tongue electrode can be prepared using the following method: A silicon substrate is provided, on which a flexible substrate layer of polyimide is fabricated. A conductive gold layer is formed on the flexible substrate layer, the conductive layer including tongue electrophysiological signal acquisition electrode sites, connecting lines, and rear pads. A top encapsulation layer of polyimide is formed on the conductive layer. The top encapsulation layer is patterned and etched to form a through-hole structure. The through-hole structure: 1) exposes the tongue electrophysiological signal acquisition electrode sites, allowing them to directly contact the tongue surface and acquire tongue electrophysiological signals; 2) exposes the rear pads, enabling them to connect to a data acquisition and transmission module via a flexible printed circuit board; 3) effectively drains secreted fluid from the tongue surface, reducing fluid accumulation between the electrodes and the tongue. Finally, the flexible electrodes are released to obtain a high-density flexible sensing array.
[0043] In an optional embodiment, the development process for the data acquisition and transmission module is as follows: For weak tongue electrophysiological signals at the microvolt level across multiple channels, a dual-layer stacked printed circuit board (PCB) design integrates a low-noise operational amplifier array, a 16-bit high-precision analog-to-digital converter, and programmable high-pass and low-pass filters. It supports simultaneous acquisition and transmission of up to 256 channels of electrophysiological signals, with a sampling rate of up to 30 kHz per channel. A pre-amplified analog filtering module removes artifacts such as power frequency noise, optimizing signal quality. Employing HDMI wired transmission, compared to Bluetooth or Wi-Fi wireless transmission, HDMI offers higher bandwidth, ensuring uncompressed real-time transmission of all 256 channels without data loss or delay. Compared to wireless signals, which are limited by environmental noise, HDMI transmission is more stable.
[0044] Please see Figure 2 The circuit diagram of the data acquisition and transmission module in a practical application is as follows: Because electrophysiological signals are relatively weak, at the microvolt level, each channel is first configured with an independent low-noise amplifier (LNA) to avoid inter-channel crosstalk caused by shared amplification circuitry. The LNA features high input impedance and low noise characteristics to ensure the signal is not distorted during amplification. The amplified signal passes through a programmable upper and lower cutoff frequency filter (LFP) to remove noise and interference components. The amplified and filtered analog signal then enters the analog-to-digital converter (ADC) via an analog multiplexer (MUX) to be converted into a digital signal. The MUX can quickly switch multiple input channels, sequentially sending signals from different channels to the ADC for conversion. Simultaneously, the ADC's internal clock and control logic ensure strict synchronization of sampling times across all channels, thus avoiding signal distortion and time deviation. The ADC's performance determines the accuracy and speed of data acquisition. These modules collectively determine the high-precision, high-sampling-rate synchronous acquisition of multi-channel data.
[0045] In an optional embodiment, the data receiving and processing module includes: A digital signal receiving unit is used to receive digital signals transmitted by a data acquisition and transmission module. The signal denoising unit is used to denoise the digital signal to obtain the denoised digital signal. The signal filtering unit is used to filter the denoised digital signal to obtain the filtered digital signal; wherein, when extracting the tongue electrophysiological signal related to taste, the 10-500 Hz frequency band is used. The feature extraction unit is used to perform feature extraction processing on the filtered digital signal to obtain the processed digital signal features. The classification and decoding unit is used to classify and decode the features of the processed digital signal to obtain the classification and decoding results. The data visualization unit is used to visualize the filtered digital signal.
[0046] Specifically, the feature extraction unit includes: The time-domain feature extraction unit is used to extract time-domain features from the filtered digital signal; the time-domain features include the mean absolute value, mean, variance, skewness, and kurtosis of the filtered digital signal. The frequency domain feature extraction unit is used to extract frequency domain features from the filtered digital signal; the frequency domain features include the weighted average frequency, root mean square frequency, and relative frequency variability of the filtered digital signal. The time-frequency domain feature extraction unit is used to combine time-domain features and frequency-domain features, and extract time-frequency domain features from the filtered digital signal using wavelet transform or short-time Fourier transform.
[0047] Preferably, in the time domain, the following features can be directly extracted from the waveform of the tongue electrophysiological signal: a) Average absolute value: Reflects the overall activity level of taste bud cells and lingual nerve fibers under baseline and different taste stimuli. The calculation formula is as follows: Average absolute value ; in, denoted as , where is the value of the i-th sampling point, and N is the total number of samples. b) Mean: Represents the average level of the tongue's electrophysiological signal, revealing the electrophysiological activity of taste bud cells and tongue nerve fibers at baseline and under different taste stimuli. The calculation formula is as follows: mean ; in, Here, is the value of the i-th sampling point, and N is the total number of samples. c) Variance: Used to measure the fluctuation range of electrophysiological activity of taste bud cells and lingual nerve fibers under baseline and different taste stimuli. The calculation formula is as follows: variance ; in, Let N be the value of the i-th sampling point, and N be the total number of samples. This is the sample mean.
[0048] d) Skewness: In the process of taste perception, the responses of taste bud cells and lingual nerve fibers may be asymmetrical. For example, certain taste stimuli may lead to rapid changes in electrical activity, resulting in negative skewness; while in other cases, the signal may exhibit positive skewness, indicating positive changes in electrophysiological activity. The calculation formula is as follows: Skewness ; in, Let N be the value of the i-th sampling point, and N be the total number of samples. The sample mean. This represents the sample standard deviation.
[0049] e) Kuroki: During taste stimulation and perception, the responses of taste bud cells and lingual nerve fibers may be accompanied by some dramatic fluctuations, resulting in a high kurtosis of the corresponding signal. Therefore, kurtosis can reveal transient changes in electrophysiological signals. The calculation formula is as follows: Kudo ; in, Let N be the value of the i-th sampling point, and N be the total number of samples. The sample mean. This represents the sample standard deviation.
[0050] Frequency domain features, by converting tongue electrophysiological signals to frequency space for analysis, can extract the following information: a) Weighted average frequency: Reflects the overall frequency distribution of the tongue's electrophysiological signals induced by taste stimulation. The calculation formula is as follows: Weighted average frequency ; in, For the i-th frequency component, This represents the power spectral density of the corresponding frequency component. This represents the total number of frequency components.
[0051] b) Root mean square frequency: This reflects the concentration of tongue electrophysiological activity induced by different taste stimuli, thus revealing the frequency response pattern of the tongue to different taste stimuli. The calculation formula is as follows: Root mean square frequency ; in, For the i-th frequency component, This represents the power spectral density of the corresponding frequency component. c) Relative frequency variability: This reveals the frequency fluctuations of the electrophysiological signal elicited by taste stimulation. For example, when taste stimulation produces sudden or irregular neural responses, the relative frequency variability may be high, indicating instability in the frequency components of the signal; while a lower relative frequency variability may mean a more stable or regular response. The calculation formula is as follows: Relative frequency variability ; in, For the i-th frequency component, This represents the power spectral density of the corresponding frequency component. The total number of frequency components. This represents the average frequency.
[0052] Time-frequency domain features combine information from both the time and frequency domains, enabling the capture of dynamic changes in tongue electrophysiological signals over time and frequency. These time-frequency domain features can be effectively extracted using wavelet transform or short-time Fourier transform (STFT).
[0053] a) Wavelet Transform: By selecting appropriate wavelet basis functions (such as Morlet wavelet or Haar wavelet), time-frequency analysis of tongue electrophysiological signals can be performed. Continuous Wavelet Transform (CWT) is suitable for analyzing short-term electrical activity induced by taste stimuli, while Discrete Wavelet Transform (DWT) is suitable for multi-scale decomposition, extracting the long-term responses of the tongue and brain to different taste stimuli. Furthermore, time-frequency domain features such as wavelet energy distribution (energy changes of electrophysiological signals caused by different taste stimuli at different wavelet scales) and wavelet entropy (changes in entropy values caused by different taste stimuli) can be extracted. The formula for calculating CWT is as follows: ; in, The original signal, Let a be the wavelet basis function, a be the scale, and b be the translation.
[0054] DWT can decompose a signal into: ; ; in, The original signal, It is a low-pass filter. It is a high-pass filter.
[0055] b) Short-Time Fourier Transform (STFT): By performing a Fourier transform on the tongue's electrophysiological signal within a fixed-length time window, the frequency components of that time period are extracted. The time-frequency distribution of the entire signal is obtained by moving the window, thus generating a time-frequency energy map (time spectrum) induced by different taste stimuli such as salty, sweet, sour, umami, and bitter. Further, time-frequency features such as power spectral density distribution (active frequency bands of electrophysiological activity after taste stimulation), peak frequency changes (peak frequency shift and response pattern of the electrophysiological signal before and after taste stimulation), and time-frequency entropy (changes in entropy values of electrophysiological activity caused by different taste stimuli, i.e., complexity) can be extracted. The calculation formula is as follows: ; in, The original signal, For window functions, It is the Fourier kernel.
[0056] Specifically, the classification and decoding unit is used to perform pattern classification and functional decoding on the time-domain features, frequency-domain features, and time-frequency-domain features in the filtered digital signal based on machine learning algorithms or deep learning algorithms, so as to obtain the classification and decoding results.
[0057] Furthermore, the classification and decoding unit is used to train the sample model based on logistic regression, decision tree, support vector machine or neural network using sample data and sample model, adjust the model parameters of the sample model until the sample model meets the training conditions to obtain the target model, and perform pattern classification and functional decoding on the time domain features, frequency domain features and time-frequency domain features in the filtered digital signal through the target model to obtain the classification and decoding results. The sample data refers to data that has been labeled with taste classification information.
[0058] In a practical application scenario, classification decoding includes a training phase and a classification decision phase. The first step, the training phase, uses data with labeled taste information, i.e., data where input features correspond to a specific taste. During training, model parameters are adjusted using a specific model (such as logistic regression, decision trees, support vector machines, neural networks, etc.) until the model can accurately map input features to their corresponding taste types. Cross-validation is used to evaluate the model's performance and prevent overfitting. The second step, the classification decision phase, takes in new samples, i.e., new features, and uses the trained classification model to predict their taste, outputting the classifier's predicted taste type.
[0059] This invention also provides a high-throughput tongue electrophysiological detection method, applied to the aforementioned high-throughput tongue electrophysiological detection system, the method comprising: Analog signals are acquired using a tongue electrosensor module; the analog signals include signals acquired by the working electrode and signals acquired by the reference electrode. The analog signal acquired by the tongue electro-sensing module is processed by the data acquisition and transmission module and then converted from analog to digital to obtain a digital signal. The data acquisition and transmission module transmits digital signals to the data receiving and processing module. The data receiving and processing module is used to process and analyze digital signals.
[0060] In one specific embodiment, the method includes: Analog signals are acquired using a tongue electrosensor module; the analog signals include signals acquired by the working electrode and signals acquired by the reference electrode. The analog signal acquired by the tongue electrosensing module is amplified by the preamplifier unit in the data acquisition and transmission module to obtain the preamplified analog signal. The preamplified signal is filtered by the bandpass filter unit in the data acquisition and transmission module to obtain the filtered analog signal. The filtered analog signal is converted into a digital signal using the analog-to-digital conversion unit in the data acquisition and transmission module. The digital signal is transmitted to the data receiving and processing module using the transmission unit in the data acquisition and transmission module. The digital signal receiving unit in the data receiving and processing module receives the digital signal transmitted from the data acquisition and transmission module; The signal denoising unit in the data receiving and processing module is used to denoise the digital signal to obtain the denoised digital signal. The signal filtering unit in the data receiving and processing module is used to filter the denoised digital signal to obtain the filtered digital signal. The feature extraction unit in the data receiving and processing module is used to perform feature extraction processing on the filtered digital signal to obtain the processed digital signal features. The classification and decoding unit in the data receiving and processing module is used to classify and decode the features of the processed digital signal to obtain the classification and decoding results. The filtered digital signal is visualized using the data visualization unit in the data receiving and processing module; The feature extraction process involves extracting time-domain, frequency-domain, and time-frequency-domain features from the filtered digital signal. The time-domain features include the mean absolute value, mean, variance, skewness, and kurtosis of the filtered digital signal. The frequency-domain features include the weighted average frequency, root mean square frequency, and relative frequency variability of the filtered digital signal. The time-frequency-domain features are extracted using wavelet transform or short-time Fourier transform. Classification decoding is implemented based on logistic regression, decision trees, support vector machines, or neural networks.
[0061] Preferably, in the time domain, the following features can be directly extracted from the waveform of the tongue electrophysiological signal: a) Average absolute value: reflects the overall activity level of taste bud cells and lingual nerve fibers at baseline and under different taste stimulation conditions.
[0062] b) Mean: Represents the average level of tongue electrophysiological signals, which can reveal the electrophysiological activity of taste bud cells and tongue nerve fibers at baseline and under different taste stimulation conditions.
[0063] c) Variance: Used to measure the fluctuation range of electrophysiological activity of taste bud cells and lingual nerve fibers at baseline and under different taste stimuli.
[0064] d) Skewness: In the process of taste perception, the responses of taste bud cells and lingual nerve fibers may be asymmetrical. For example, certain taste stimuli may lead to rapid changes in electrical activity, resulting in negative skewness; while in other cases, the signal may exhibit positive skewness, indicating positive changes in electrophysiological activity.
[0065] e) Kurometry: During taste stimulation and perception, the response of taste bud cells and lingual nerve fibers may be accompanied by some dramatic fluctuations, and the kurometry of the corresponding signal will be high. Therefore, kurometry can reveal the transient changes that occur in electrophysiological signals.
[0066] Frequency domain features, by converting tongue electrophysiological signals to frequency space for analysis, can extract the following information: a) Weighted average frequency: Reflects the overall frequency distribution of the tongue's electrophysiological signals caused by taste stimulation.
[0067] b) Root mean square frequency: reflects the degree of concentration of tongue electrophysiological activity induced by different taste stimuli, thereby revealing the frequency response pattern of the tongue to different taste stimuli.
[0068] c) Relative frequency variability: This reveals the frequency fluctuations of the electrophysiological signals elicited by taste stimuli. For example, when taste stimuli produce sudden or irregular neural responses, the relative frequency variability may be high, indicating that the frequency components of the signal are unstable; while a lower relative frequency variability may mean that the response is more stable or regular.
[0069] Time-frequency domain features combine information from both the time and frequency domains, enabling the capture of dynamic changes in tongue electrophysiological signals over time and frequency. These time-frequency domain features can be effectively extracted using wavelet transform or short-time Fourier transform (STFT).
[0070] a) Wavelet Transform: By selecting appropriate wavelet basis functions (such as Morlet wavelet or Haar wavelet), time-frequency analysis of tongue electrophysiological signals can be performed. Continuous wavelet transform (CWT) is suitable for analyzing short-term electrical activity induced by taste stimulation, while discrete wavelet transform (DWT) is suitable for multi-scale decomposition to extract the long-term response of the tongue and brain to different taste stimuli. Furthermore, time-frequency domain features such as wavelet energy distribution (energy changes of electrophysiological signals caused by different taste stimuli at different wavelet scales) and wavelet entropy (different entropy changes caused by different taste stimuli) can be extracted.
[0071] b) Short-Time Fourier Transform (STFT): By performing a Fourier transform on the tongue electrophysiological signal within a fixed-length time window, the frequency components of that time period are extracted, and the time-frequency distribution of the entire signal is obtained by moving the window, thereby generating a time-frequency energy map (time spectrum) induced by different taste stimuli such as salty, sweet, sour, umami, and bitter. Further, time-frequency features such as power spectral density distribution (active frequency band of electrophysiological activity after taste stimulation), peak frequency change (peak frequency shift and response pattern of electrophysiological signal before and after taste stimulation), and time-frequency entropy (entropy change of electrophysiological activity caused by different taste stimuli, i.e., complexity) can be extracted.
[0072] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features of the present invention can be arbitrarily combined with each other.
Claims
1. A high-throughput electrogustomography system, comprising: The tongue electrode sensing module comprises a working electrode and a reference electrode, the working electrode is arranged as a flexible tongue electrode in a distributed layout, the flexible tongue electrode is configured to cover the tongue tip, the tongue side, the tongue middle and the tongue root, and the surface has a three-dimensional curved surface matching the surface of the tongue part; the tongue electrode sensing module is configured to collect the electrophysiological information of different areas of the tongue part; The data acquisition and transmission module is used for collecting multiple channel information corresponding to the microvolt-level electrophysiological signals of different areas of the tongue part, and converting the multiple channel information into digital information; the data acquisition and transmission module further comprises multiple channel pre-processing units, an analog multiplexer, an ADC and a synchronization unit; wherein the multiple channel pre-processing units further comprise a low-noise amplifier and a programmable upper and lower cutoff frequency filter configured for each channel, which are used to amplify and filter the analog signals corresponding to the microvolt-level electrophysiological signals collected by different channels in different areas of the tongue part; the analog multiplexer MUXN is used for quickly switching the multiple channels corresponding to the analog multiplexer MUXN, and the analog signals output by different channels are sequentially sent into the ADC for conversion to obtain corresponding digital signals; the synchronization unit is configured to realize the sampling time synchronization of all channels; The data receiving and processing module is used for processing and analyzing the digital signals transmitted by the data acquisition and transmission module. The data receiving and processing module comprises:
2. The high-throughput electrogastrophysiological detection system according to claim 1, wherein, A digital signal receiving unit is used for receiving the digital signals transmitted by the data acquisition and transmission module; A signal denoising unit is used for denoising the digital signals to obtain denoised digital signals; A signal filtering unit is used for filtering the denoised digital signals to obtain filtered digital signals; A feature extraction unit is used for extracting features from the filtered digital signals to obtain processed digital signal features; A classification decoding unit is used for classifying and decoding the processed digital signal features to obtain classification decoding results; A data visualization unit is used for visualizing the filtered digital signals. The flexible tongue electrode further has a third through-hole structure for discharging secretions, and the reference electrode is arranged as a patch electrode attached to the cheeks; the distributed layout indicates that the electrode array is arranged as a core functional area and an auxiliary monitoring area for targeted collection of different spatial resolutions, and the electrode array arrangement density of the core functional area is greater than that of the auxiliary monitoring area.
3. The high-throughput electrogastrophysiological detection system according to claim 1 or 2, characterized in that, The flexible tongue electrode comprises:
4. The high-throughput electrogastrography system of claim 3, wherein, A flexible substrate layer is configured to cover the tongue tip, the tongue side, the tongue middle and the tongue root, and the surface has a three-dimensional curved surface matching the surface of the tongue part; A conductive layer configured to be disposed on the flexible substrate layer in a patterned electrode array in a distributed layout, the conductive layer comprising a plurality of electrode units and interconnecting wires for electrical connection among the plurality of electrode units and within each of the electrode units, each of the electrode units being configured to cover the tongue tip, the tongue side, the tongue middle and the tongue root respectively, and the electrode array of each of the electrode units collecting the same or different spatial resolution; A flexible encapsulation layer in contact with the tongue portion, configured to cover the tongue tip, the tongue side, the tongue middle and the tongue root, and having a surface with a three-dimensional curved surface matching the surface of the tongue portion, the flexible encapsulation layer being etched with at least a first via structure for exposing an electrode site, a second via structure for exposing a back-end pad and the third via structure for discharging secretions; A signal processing module electrically connected to each of the electrode units through the interconnecting wires for signal processing.
5. The high-throughput electrogastrography system of claim 2, wherein, The feature extraction unit comprises: A time domain feature extraction unit for extracting time domain features in the filtered digital signal; the time domain features include the average absolute value, mean, variance, skewness and kurtosis of the filtered digital signal; A frequency domain feature extraction unit for extracting frequency domain features in the filtered digital signal; the frequency domain features include the weighted average frequency, root mean square frequency and relative frequency variability of the filtered digital signal; A time-frequency domain feature extraction unit for extracting time-frequency domain features in the filtered digital signal by combining the time domain features and the frequency domain features using wavelet transform or short-time Fourier transform.
6. The high-throughput electrogastrophysiological testing system according to claim 5, wherein, The calculation formula for extracting time-frequency domain features in the filtered digital signal using the wavelet transform method of the time-frequency domain feature extraction unit is as follows: ； where CWT is a continuous wavelet transform, is the original signal, is a wavelet basis function, a is a scale, and b is a translation. The discrete wavelet transform (DWT) can decompose the signal into: ； ； wherein is the original signal, is a low-pass filter, is a high-pass filter; The calculation formula for extracting time-frequency domain features in the filtered digital signal using the short-time Fourier transform method of the time-frequency domain feature extraction unit is as follows: ； wherein is the original signal, is a window function, is a Fourier kernel.
7. The high-throughput electrogastrophysiological testing system according to claim 6, wherein, The classification decoding unit is used to perform pattern classification and functional decoding of the time domain features, the frequency domain features and the time-frequency domain features in the filtered digital signal based on machine learning algorithms or deep learning algorithms to obtain the classification decoding results.
8. The high-throughput electrogastrophysiological testing system according to claim 7, wherein, The classification decoding unit is used to train the sample model based on logistic regression, decision tree, support vector machine or neural network using sample data and sample model, adjust the model parameters of the sample model until the sample model meets the training conditions to obtain a target model, and perform pattern classification and functional decoding of the time domain features, the frequency domain features and the time-frequency domain features in the filtered digital signal through the target model to obtain the classification decoding results. The sample data indicates data with labeled taste classification information.
9. A high-throughput electrogustomography method applied to a high-throughput electrogustomography system according to any one of claims 1 to 8, characterized in that, The method comprises: Collecting the analog signal using the tongue electrical sensing module; the analog signal includes the signal collected by the working electrode and the signal collected by the reference electrode; The data acquisition and transmission module is used for processing the analog signal collected by the tongue electrical sensing module and performing analog-digital conversion to obtain a digital signal; The data acquisition and transmission module is used for transmitting the digital signal to the data receiving and processing module; The data receiving and processing module is used for processing and analyzing the digital signal.
10. The high-throughput electrogastrophysiological detection method according to claim 9, wherein, The method comprises: The tongue electrical sensing module is used for collecting the analog signal, which includes the signal collected by the working electrode and the signal collected by the reference electrode; The preamplification unit in the data acquisition and transmission module is used for preamplifying the analog signal collected by the tongue electrical sensing module to obtain a preamplified analog signal; The band-pass filter unit in the data acquisition and transmission module is used for filtering the preamplified signal to obtain a filtered analog signal; The analog-digital conversion unit in the data acquisition and transmission module is used for converting the filtered analog signal into the digital signal; The transmission unit in the data acquisition and transmission module is used for transmitting the digital signal to the data receiving and processing module; The digital signal receiving unit in the data receiving and processing module is used for receiving the digital signal transmitted by the data acquisition and transmission module; The signal denoising unit in the data receiving and processing module is used for denoising the digital signal to obtain a denoised digital signal; The signal filtering unit in the data receiving and processing module is used for filtering the denoised digital signal to obtain a filtered digital signal; The feature extraction unit in the data receiving and processing module is used for extracting features from the filtered digital signal to obtain a processed digital signal feature; The classification decoding unit in the data receiving and processing module is used for classifying and decoding the processed digital signal feature to obtain a classification decoding result; The data visualization unit in the data receiving and processing module is used for visualizing the filtered digital signal. The feature extraction processing indicates extracting time domain features, frequency domain features and time-frequency domain features from the filtered digital signal; the time domain features include the average absolute value, mean, variance, skewness and kurtosis of the filtered digital signal, the frequency domain features include the weighted average frequency, root mean square frequency and relative frequency variability of the filtered digital signal, and the time-frequency domain features are extracted using the wavelet transform or short-time Fourier transform method; The classification decoding is based on logic regression, decision tree, support vector machine or neural network.