Intelligent anesthesia depth detection system, method and device

Abstract

The invention provides an intelligent anesthesia depth detection system, method and device, and relates to the technical field of anesthesia depth detection, the system mainly comprises a data acquisition module, a data processing module and a result generation module; the data acquisition module is used for acquiring electroencephalogram data and cerebral blood oxygen data of the wounded; the data processing module comprises a preprocessing unit, a feature extraction unit, a multi-modal fusion unit and an evaluation unit; the feature extraction unit is used for extracting a first feature from the training set of the electroencephalogram data through a wavelet packet-empirical mode decomposition method; extracting a second feature from the training set of brain blood oxygen through a sample entropy method; the multi-modal fusion unit is used for fusing the first feature and the second feature to obtain a multi-modal anesthesia depth model; the evaluation unit is used for calculating an anesthesia depth index as an anesthesia depth result through a multi-mode anesthesia depth model on the basis of the electroencephalogram data and the cerebral blood oxygen data of the tested wounded; according to the scheme, anesthesia depth detection precision can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of anesthesia depth detection technology, and in particular to an anesthesia depth intelligent detection system, method and device. Background Art

[0002] Anesthesia equipment is essential medical equipment during surgery and emergency treatment. Its function is to provide oxygen and anesthetic agents to patients, and to manage their breathing, playing an invaluable role in ensuring their safety during surgery. In particular, due to the complexity of field disaster accident scenes and the individual differences among patients, even highly skilled and experienced anesthesiologists are prone to errors such as over- or under-dosing, which can cause harm to patients. Therefore, reliable anesthesia depth detection technology has important clinical implications.

[0003] Currently, there is no gold standard for assessing depth of anesthesia. While the BIS index has excellent validity, it still has some drawbacks: First, the BIS index is suitable for monitoring anesthesia scenarios in which intravenous and inhaled anesthetics are used in combination with small and medium-dose opioids. It is not sensitive to certain specific anesthetics or sedatives (such as nitrous oxide, ketamine, or high-dose opioids). Second, EEG is an electrical signal. During damage control surgery at disaster sites, electrosurgical scalpels can interfere with EEG signals, reducing the accuracy of the BIS index. Third, the BIS index is less effective for assessing depth of anesthesia in patients with acute traumatic brain injury and those with low EEG voltages. There is also controversy regarding its suitability for pediatric anesthesia. Fourth, the EEG primarily correlates with the patient's level of hypnosis but does not provide information on analgesia. While the BIS index can effectively predict anesthetic clearance and the level of sedation within anesthesia, it cannot effectively detect the transition from wakefulness to unconsciousness and is insensitive to analgesia levels.

[0004] In summary, the industry urgently needs to develop portable, safe, and easy-to-apply and popularize intelligent detection and assessment systems, methods, and devices for anesthesia depth that can be used at disaster accident sites. Summary of the Invention

[0005] The object of the present invention is to provide an intelligent anesthesia depth detection system, method and device to solve at least one of the above-mentioned technical problems existing in the prior art.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides an intelligent anesthesia depth detection system, including a data acquisition module, a data processing module and a result generation module.

[0007] The data acquisition module is used to collect the EEG data and cerebral blood oxygen data of the injured.

[0008] In a feasible implementation, the data acquisition module is further configured to acquire the ventilator data and monitor data of the wounded; the ventilator data includes respiratory data, end-tidal carbon dioxide data, tidal volume data, etc.; the monitor data includes blood pressure data, electrocardiogram data, body temperature data, etc.

[0009] The data processing module includes a preprocessing unit, a feature extraction unit, a multi-modal fusion unit, and an evaluation unit.

[0010] The preprocessing unit first performs One-Hot encoding and z-score standardization on the cerebral oxygenation data to obtain the minority class samples ω1; through K-means clustering, ω1 is divided into A clusters, and the sample quantity ratio of each cluster is statistically calculated; then the number of synthetic samples is assigned to the A clusters, and the lower the sample quantity ratio of the cluster, the more synthetic samples it gets. CTGAN sample synthesis is performed on the A clusters respectively to obtain the synthetic samples ω2; ω2 and ω1 are combined into the training set ω of cerebral oxygenation; thus, the sampling rates of the cerebral oxygenation data (sampling rate: 10Hz) and electroencephalogram data (sampling rate: 256Hz) are balanced to obtain a training set with comparable data volumes.

[0011] The feature extraction unit extracts the first feature from the training set of electroencephalogram data through the wavelet packet-empirical mode decomposition (EMD) method; and extracts the second feature from the training set of cerebral oxygenation through the sample entropy method.

[0012] In a feasible implementation, the specific method for extracting the first feature includes: Step a1: Filter the electroencephalogram data through a low-frequency filter. Step a2: Perform three-layer decomposition on the electroencephalogram data through wavelet packet transform to obtain a group of narrowband signals with respect to time ; Step a3: Read all the extreme value information in; Step a4: Calculate the upper envelope and lower envelope of the narrowband signal through a cubic spline function; Step a5: Calculate the mean value of the envelope line, and the specific formula is: ; Step a6: Calculate the difference between the original narrowband signal and the mean value, and the specific formula is: ; Step a7: Calculate the narrowband signal residue , and the specific formula is: ; Step a8, reconstruct to obtain the electroencephalogram signal , thereby removing the intrinsic mode functions (IMFs) with poor correlation. The specific formula is: ; where n represents the total number of electroencephalogram samples; i represents the i-th electroencephalogram sample point.

[0013] In a feasible implementation manner, the specific method for extracting the second feature includes: Step b1, based on the training set of cerebral blood oxygen, set the cerebral blood oxygen signal sequence as , where represents the total number of cerebral blood oxygen samples; after reconstruction, obtain -dimensional vector sequence: ; where represents the -th vector, and the specific expression is: ; where represents the selected spatial dimension vector (i.e., the window length, which is the consecutive values starting from the -th sample point); Step b2, calculate the Euclidean distance between any two vectors, and define the maximum Euclidean distance between each pair of vectors as the maximum contribution component distance. The specific formula is: ; where represents the maximum contribution component distance between the -th vector and the -th vector ; represents the coordinate value of a vector in the two-dimensional space, and ; Step b3, based on the preset threshold and (also known as the embedding dimension), calculate the measure representing the regularity probability magnitude, and take the average value of all . The specific formula is: ; Step b4, increment by 1, and iteratively execute steps b1 - b3 to obtain ; calculate the sample entropy of the cerebral blood oxygen signal sequence. The specific formula includes:​ ; Since is a finite value, there is: 。

[0014] Preferably, takes the value of 2, takes the value of 0.1 - 0.25 times the standard deviation of the electroencephalogram data.

[0015] In a feasible implementation manner, the feature extraction unit further includes: extracting a third feature from the training sets of the ventilator data and the monitor data through conventional time domain, frequency domain, and time-frequency domain analysis methods.

[0016] In this way, different and targeted feature extraction strategies are adopted for different types of data, which can fully and effectively mine the features of various data.

[0017] The multimodal fusion unit is used to fuse the first feature and the second feature to obtain the first multimodal anesthesia depth model.

[0018] In a feasible implementation manner, the multimodal fusion unit further includes fusing the third feature on the basis of the first multimodal anesthesia depth model to obtain the second multimodal anesthesia depth model; thus, a multimodal anesthesia depth model with more accurate prediction effect and better generalization performance than the first multimodal anesthesia depth model is obtained.

[0019] In a feasible implementation manner, the multimodal anesthesia depth model is a dynamic fuzzy neural network (DFNN).

[0020] In a feasible implementation manner, the dynamic fuzzy neural network includes an input layer, a membership function layer, a T-norm layer, a normalization layer, and an output layer arranged in sequence; The input layer is used to input the electroencephalogram data and cerebral blood oxygen data of the wounded; The membership function in the membership function layer is characterized as: ; Among them, represents the input data; represents the Gaussian function; represents the center of this Gaussian function; in this way, the deficiency of the single network width can be made up, different membership functions are set for each layer of input variables, so that the width of the signal is different, and thus the network has the ability to process wide-area data; The T-norm layer is used to perform weighted summation on the outputs of each fuzzy rule, so as to process the "and" relationship between the fuzzy rules; The normalization layer is used to map the data after weight summation into a standard normal distribution with a mean of 0 and a variance of 1, so as to reduce the influence of data distribution differences on the model training results; The output layer is used to output the anesthesia depth index (MADI).

[0021] In a feasible implementation, the parameters of the dynamic fuzzy neural network are optimized by the particle swarm optimization (PSO).

[0022] The evaluation unit calculates the anesthesia depth index (MADI) based on the EEG data and cerebral blood oxygen data of the injured person to be measured through a multi-modal anesthesia depth model as the anesthesia depth result.

[0023] The result generation module is used to send out the anesthesia depth result.

[0024] In a second aspect, based on the same inventive concept, the present application also provides an intelligent anesthesia depth detection device, including a processor, a memory and a bus. The memory stores instructions and data read by the processor. The processor is used to call the instructions and data in the memory to implement the intelligent anesthesia depth detection system as described above. The bus is connected between each functional component for transmitting information.

[0025] In a feasible implementation, the device further includes a sensor for simultaneously collecting the EEG data and cerebral blood oxygen data of the injured person. The main body of the sensor is a flexible substrate. A circuit board is arranged inside the flexible substrate, and a first measurement electrode and a fifth measurement electrode are symmetrically arranged on the outermost sides at both ends of the flexible substrate. The first measurement electrode is used to detect the EEG signal above the outer side of the left eyebrow bone of the injured person's forehead. The fifth measurement electrode is used to detect the EEG signal above the outer side of the right eyebrow bone of the injured person's forehead. Between the first measurement electrode and the fifth measurement electrode, a second measurement electrode and a fourth measurement electrode are symmetrically arranged. The second measurement electrode is used to detect the EEG signal above the inner side of the left eyebrow bone of the injured person's forehead. The fourth measurement electrode is used to detect the EEG signal above the inner side of the right eyebrow bone of the injured person's forehead. Between the second measurement electrode and the fourth measurement electrode, a third measurement electrode is arranged as a ground electrode; Between the first measurement electrode and the second measurement electrode, a first phototube and a second phototube are arranged to collect the cerebral blood oxygen saturation of the left forehead of the injured person. Between the second measurement electrode and the third measurement electrode, a first light source is arranged to provide the light source required for collecting the cerebral blood oxygen saturation for the skin of the left forehead of the injured person; Between the fourth measurement electrode and the fifth measurement electrode, a third phototube and a fourth phototube are provided for collecting the cerebral blood oxygen saturation on the right side of the forehead of the wounded; between the third measurement electrode and the fourth measurement electrode, a second light source is provided for providing the light source required for collecting the cerebral blood oxygen saturation for the skin on the right side of the forehead of the wounded.

[0026] In a feasible implementation manner, the light source is an LED light source.

[0027] In a feasible implementation manner, the LED light source uses near-infrared light with wavelengths of 735 nm and 850 nm respectively, so as to collect the concentration change signals of Hb (deoxyhemoglobin) and (oxyhemoglobin).

[0028] In a feasible implementation manner, based on the actual spectral distribution function of the LED, the absorption coefficients of Hb and are corrected to reduce the error caused by directly using the absorption coefficient obtained from the LED peak.

[0029] In a feasible implementation manner, the LED light source adopts LED optical power adjustable technology to automatically adapt to the skin detection depth of different skin color populations.

[0030] In a feasible implementation manner, the preamplifier circuit connected to the phototube includes two-stage amplifiers: the first stage is a transimpedance amplifier for converting the input current signal of the phototube into a voltage signal; the second stage is a voltage follower circuit amplifier for reducing the output impedance of the phototube.

[0031] In a feasible implementation manner, the preamplifier circuit includes an equipotential protection ring for reducing the noise of the input current signal.

[0032] In a feasible implementation manner, a voltage follower is pre-placed at the measurement electrode for increasing the input impedance and common-mode rejection ratio of the electroencephalogram signal and reducing the output impedance.

[0033] In a feasible implementation manner, the measurement electrode includes an electrode column with low AC impedance, low DC polarization potential and stable high polarization potential.

[0034] In a feasible implementation manner, the flexible substrate is made of silica gel, and the circuit board is potted with epoxy resin to protect the stable electrical performance of the circuit board.

[0035] In a feasible implementation manner, the device further includes a Bluetooth module for connecting external devices such as an anesthesia machine, a ventilator and a monitor, so as to collect external data such as ventilator data and monitor data.

[0036] Adopting the above technical solution, the present invention has the following beneficial effects: An intelligent anesthesia depth detection system, method and device provided by the present invention take into account the characteristics of high temporal resolution of electroencephalogram data and high spatial resolution of cerebral blood oxygen data, etc., realize the signal fusion between electroencephalogram data and cerebral blood oxygen data, and effectively avoid problems such as electrocautery interference and insensitive anesthesia depth detection caused by anesthetics when using the BIS index; through the K-means clustering and CTGAN sample synthesis methods, the adoption rate difference between cerebral blood oxygen data and electroencephalogram data is balanced; through a targeted feature extraction strategy, the features of different data are extracted; through the dynamic fuzzy neural network combined with the particle swarm optimization method, the prediction accuracy of the multi-modal anesthesia depth model is improved, and the transitional changes from wakefulness to loss of consciousness can be well detected; by collecting external data such as ventilator data and monitor data, the multi-modal anesthesia depth model is optimized to further improve the prediction accuracy. Description of the Drawings

[0037] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0038] Figure 1 A diagram of an intelligent anesthesia depth detection system provided by an embodiment of the present invention; Figure 2 A flowchart of the intelligent anesthesia depth detection algorithm; Figure 3 A diagram of the composition of the training set of cerebral blood oxygen provided by an embodiment of the present invention; Figure 4 A schematic structural diagram of the dynamic fuzzy neural network provided by an embodiment of the present invention; Figure 5 A diagram of the construction process of the particle swarm optimization-based dynamic fuzzy neural network provided by an embodiment of the present invention; Figure 6 A schematic layout diagram of the sensors provided by an embodiment of the present invention; Reference Signs: 1 - First measurement electrode; 2 - Second measurement electrode; 3 - Third measurement electrode; 4 - Fourth measurement electrode; 5 - Fifth measurement electrode; 6 - First phototube; 7 - Second phototube; 8 - First light source; 9 - Third phototube; 10 - Fourth phototube; 11 - Second light source. Detailed Embodiments

[0039] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work belong to the scope of protection of the present invention.

[0040] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, the terms "first", "second", "third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0041] In the description of the present invention, it should be noted that unless otherwise clearly specified and defined, the terms "installation", "connection", "connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0042] The following further explains the present invention in conjunction with specific implementation manners.

[0043] It should also be noted that the following specific embodiments or specific implementation manners are a series of optimized setting methods listed by the present invention to further explain the specific content of the invention, and these setting methods can be combined with each other or used in association with each other.

[0044] Embodiment 1: As Figure 1-2 shown, an intelligent anesthesia depth detection system provided in this embodiment includes a data acquisition module, a data processing module, and a result generation module.

[0045] The data acquisition module is used to acquire the electroencephalogram data and cerebral blood oxygen data of the wounded.

[0046] Furthermore, the data acquisition module is also used to acquire the ventilator data and monitor data of the wounded; the ventilator data includes respiratory data, end-tidal carbon dioxide data, tidal volume data, etc.; the monitor data includes blood pressure data, electrocardiogram data, body temperature data, etc.

[0047] The data processing module includes a preprocessing unit, a feature extraction unit, a multi-modal fusion unit, and an evaluation unit.

[0048] As shown in Figure 3 , the preprocessing unit first performs One-Hot encoding and z-score standardization on the cerebral blood oxygen data to obtain the minority class samples ω1; through K-means clustering, ω1 is divided into A clusters, and the sample quantity ratio of each cluster is statistically calculated; then the number of synthetic samples is allocated to the A clusters, and the lower the sample quantity ratio of a cluster, the more synthetic samples it gets. CTGAN sample synthesis is performed on the A clusters respectively to obtain the synthetic samples ω2; ω2 and ω1 are combined into the training set ω of cerebral blood oxygen; thus, the sampling rates of the cerebral blood oxygen data (sampling rate: 10Hz) and the electroencephalogram data (sampling rate: 256Hz) are balanced to obtain a training set with comparable data volume.

[0049] The feature extraction unit extracts the first feature from the training set of electroencephalogram data through the wavelet packet-empirical mode decomposition (EMD) method; and extracts the second feature from the training set of cerebral blood oxygen through the sample entropy method.

[0050] Furthermore, the specific method for extracting the first feature includes: Step a1: Filter the electroencephalogram data through a low-frequency filter; Step a2: Perform three-layer decomposition on the electroencephalogram data through wavelet packet transform to obtain a group of narrowband signals with respect to time ; Step a3: Read all the extreme value information in ; Step a4: Calculate the upper envelope and the lower envelope of the narrowband signal through a cubic spline function; Step a5: Calculate the mean of the envelope line, and the specific formula is: ; Step a6: Calculate the difference between the original narrowband signal and the mean, and the specific formula is: ; Step a7: Calculate the narrowband signal residue , and the specific formula is: ; Step a8: Reconstruct to obtain the electroencephalogram signal , thereby removing the intrinsic mode functions (IMFs) with poor correlation, and the specific formula is: ; Among them, n represents the total number of EEG samples; i represents the i-th EEG sample point.

[0051] Furthermore, the specific method for extracting the second feature includes: Step b1: Based on the training set of cerebral blood oxygen, set the cerebral blood oxygen signal sequence as , where represents the total number of cerebral blood oxygen samples; after reconstruction, obtain the -dimensional vector sequence: ; Among them, represents the -th vector, and the specific expression is: ; Among them, represents the selected spatial dimension vector (i.e., the window length, which is the continuous values starting from the -th sample point); Step b2: Calculate the Euclidean distance between any two vectors, and define the maximum Euclidean distance between each pair of vectors as the maximum contribution component distance. The specific formula is: ; Among them, represents the maximum contribution component distance between the -th vector and the -th vector ; represents the coordinate value of a vector in a two-dimensional space, and ; Step b3: Based on the preset threshold and (also known as the embedding dimension), calculate the measure of the probability size representing regularity , and take the average value for all . The specific formula is: ; Step b4: Increment by 1, and iteratively execute steps b1 - b3 to obtain ; calculate the sample entropy of the cerebral blood oxygen signal sequence. The specific formula includes: ; Since is a finite value, there is: .

[0052] Preferably, The value is 2, and is 0.1 - 0.25 times the standard deviation of the electroencephalogram data.

[0053] Furthermore, the feature extraction unit further includes: extracting third features from the training sets of ventilator data and monitor data through conventional time domain, frequency domain, and time - frequency domain analysis methods.

[0054] The multi - modal fusion unit is used to fuse the first feature and the second feature to obtain a first multi - modal anesthesia depth model.

[0055] Furthermore, the multi - modal fusion unit further includes fusing the third feature on the basis of the first multi - modal anesthesia depth model to obtain a second multi - modal anesthesia depth model.

[0056] Furthermore, the multi - modal anesthesia depth model is a dynamic fuzzy neural network (DFNN).

[0057] Furthermore, as Figure 4-5 shown, the dynamic fuzzy neural network includes an input layer, a membership function layer, a T - norm layer, a normalization layer, and an output layer arranged in sequence; The input layer is used to input the electroencephalogram data and cerebral blood oxygen data of the wounded. The membership function in the membership function layer is characterized as: ; where represents the input data; represents the Gaussian function; represents the center of this Gaussian function; in this way, the deficiency of a single network width can be made up for, different membership functions are set for each layer of input variables, so that the width of the signal is different, and thus the network has the ability to process wide - area data; The T - norm layer is used to perform weighted summation on the outputs of each fuzzy rule, so as to process the "and" relationship between fuzzy rules; represents the width of different Gaussian functions; The normalization layer is used to map the data after weighted summation into a standard normal distribution with a mean of 0 and a variance of 1, so as to reduce the influence of data distribution differences on the model training results; represents the normalization value; The output layer is used to output the anesthesia depth index.

[0058] Furthermore, the parameters of the dynamic fuzzy neural network are optimized by the particle swarm optimization algorithm (PSO).

[0059] In this way, the dynamic fuzzy neural network is combined with the particle swarm optimization algorithm to enhance the model training ability through the dynamic structure and improve the model's ability to process wide-area data by optimizing the membership function layer.

[0060] The evaluation unit calculates the anesthesia depth index (MADI) based on the electroencephalogram data and cerebral blood oxygen data of the injured patient to be measured through a multi-modal anesthesia depth model as the anesthesia depth result.

[0061] The result generation module is used to send out the anesthesia depth result.

[0062] Embodiment 2: This embodiment provides an intelligent anesthesia depth detection device, including a processor, a memory, and a bus. The memory stores instructions and data read by the processor, and the processor is used to call the instructions and data in the memory to implement the intelligent anesthesia depth detection system as described above. The bus is connected between the functional components for transmitting information.

[0063] Further, as Figure 6 shown, the device further includes a sensor for simultaneously collecting the electroencephalogram data and cerebral blood oxygen data of the injured patient. The main body of the sensor is a flexible substrate, and a circuit board is arranged inside the flexible substrate. The first measurement electrode 1 and the fifth measurement electrode 5 are symmetrically arranged at the outermost sides of both ends of the flexible substrate. The first measurement electrode 1 is used to detect the electroencephalogram signal above the outer side of the left eyebrow bone on the forehead of the injured patient. The fifth measurement electrode 5 is used to detect the electroencephalogram signal above the outer side of the right eyebrow bone on the forehead of the injured patient. Between the first measurement electrode 1 and the fifth measurement electrode 5, the second measurement electrode 2 and the fourth measurement electrode 4 are symmetrically arranged. The second measurement electrode 2 is used to detect the electroencephalogram signal above the inner side of the left eyebrow bone on the forehead of the injured patient. The fourth measurement electrode 4 is used to detect the electroencephalogram signal above the inner side of the right eyebrow bone on the forehead of the injured patient. Between the second measurement electrode 2 and the fourth measurement electrode 4, a third measurement electrode 3 is arranged to be used as a grounding electrode. Between the first measurement electrode 1 and the second measurement electrode 2, a first phototube 6 and a second phototube 7 are arranged to collect the cerebral blood oxygen saturation of the left side of the forehead of the injured patient. Between the second measurement electrode 2 and the third measurement electrode 3, a first light source 8 is arranged to provide the illumination required for collecting the cerebral blood oxygen saturation for the skin on the left side of the forehead of the injured patient. Between the fourth measurement electrode 4 and the fifth measurement electrode 5, a third phototube 9 and a fourth phototube 10 are arranged to collect the cerebral blood oxygen saturation of the right side of the forehead of the injured patient. Between the third measurement electrode 3 and the fourth measurement electrode 4, a second light source 11 is arranged to provide the illumination required for collecting the cerebral blood oxygen saturation for the skin on the right side of the forehead of the injured patient.

[0064] Further, the light source is an LED light source.

[0065] Further, the LED light source uses near-infrared light with wavelengths of 735 nm and 850 nm respectively, in order to collect the concentration change signals of Hb (deoxyhemoglobin) and (oxyhemoglobin).

[0066] Further, based on the actual spectral distribution function of the LED, the absorption coefficients of Hb and are corrected, so as to reduce the error caused by directly using the absorption coefficients obtained from the LED wave peaks.

[0067] Further, the LED light source adopts LED optical power adjustable technology, so as to automatically adapt to the skin detection depths of people with different skin colors.

[0068] Further, the preamplifier circuit connected to the phototube includes two-stage amplifiers: the first stage is a transimpedance amplifier, which is used to convert the input current signal of the phototube into a voltage signal; the second stage is a voltage follower circuit amplifier, which is used to reduce the output impedance of the phototube.

[0069] Further, the preamplifier circuit includes a conventional equipotential protection ring, which is used to reduce the noise of the input current signal.

[0070] Further, a conventional voltage follower is pre-positioned at the measurement electrode, which is used to increase the input impedance and common-mode rejection ratio of the electroencephalogram signal and reduce the output impedance.

[0071] Further, the measurement electrode includes an electrode column with low AC impedance, low DC polarization potential and stable high polarization potential.

[0072] Further, the flexible substrate is made of silica gel, and the circuit board is potted with epoxy resin to protect the stable electrical performance of the circuit board.

[0073] Further, the device further includes a Bluetooth module, which is used to connect external devices such as anesthetic machines, ventilators and monitors, so as to collect external data such as ventilator data and monitor data.

[0074] In another implementation manner of this solution, it can also be implemented in the form of an integrated device, and the device can include corresponding modules that execute each or several steps in the above-mentioned various implementation manners. The module can be one or more hardware modules specifically configured to execute the corresponding steps, or be implemented by a processor configured to execute the corresponding steps, or be stored in a computer-readable medium for implementation by the processor, or be implemented through a certain combination.

[0075] The processor executes the various methods and processes described above. For example, the method embodiments in this solution can be implemented as a software program, which is tangibly included in a machine-readable medium, such as a memory. In some embodiments, part or all of the software program can be loaded and / or installed via the memory and / or communication interface. When the software program is loaded into the memory and executed by the processor, one or more steps of the methods described above can be executed. Alternatively, in other embodiments, the processor can be configured to execute one of the above methods by any other suitable means (e.g., by means of firmware).

[0076] The device can be implemented using a bus architecture. The bus architecture can include any number of interconnected buses and bridges, depending on the specific application of the hardware and overall design constraints. The bus connects various circuits including one or more processors, memories, and / or hardware modules together. The bus can also connect various other circuits such as peripheral devices, voltage regulators, power management circuits, external antennas, etc.

[0077] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Component (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0078] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent anesthesia depth detection system, characterized in that It includes a data acquisition module, a data processing module and a result generation module; The data acquisition module is used to acquire the electroencephalogram data and cerebral blood oxygen data of the wounded; The data processing module includes a preprocessing unit, a feature extraction unit, a multi-modal fusion unit and an evaluation unit; The preprocessing unit first performs One-Hot encoding and z-score normalization on the cerebral blood oxygen data to obtain the minority class samples ω1; through K-means clustering, ω1 is divided into A clusters, and the sample quantity ratio of each cluster is statistically calculated; then the number of synthetic samples is allocated to the A clusters, and the clusters with lower sample quantity ratios get more synthetic samples. CTGAN sample synthesis is performed on the A clusters respectively to obtain the synthetic samples ω2; ω2 and ω1 are combined into the training set ω of cerebral blood oxygen; The feature extraction unit extracts the first feature from the training set of electroencephalogram data through the wavelet packet-empirical mode decomposition method; and extracts the second feature from the training set of cerebral blood oxygen through the sample entropy method; The multi-modal fusion unit is used to fuse the first feature and the second feature to obtain the first multi-modal anesthesia depth model; The evaluation unit calculates the anesthesia depth index based on the electroencephalogram data and cerebral blood oxygen data of the measured wounded through the multi-modal anesthesia depth model as the anesthesia depth result; The result generation module is used to transmit the anesthesia depth result externally.

2. The system according to claim 1, wherein The data acquisition module is also used to acquire the ventilator data and monitor data of the wounded; the ventilator data includes respiratory data, end-tidal carbon dioxide data and tidal volume data; the monitor data includes blood pressure data, electrocardiogram data and body temperature data.

3. The system according to claim 1, wherein The specific method for extracting the first feature includes: Step a1: Filter the electroencephalogram data through a low-frequency filter; Step a2: Through wavelet packet transform, perform three-layer decomposition processing on the EEG data to obtain a set of narrowband signals with respect to time ; ; Step a3, read all the extreme value information in; Step a4: Calculate the upper envelope and the lower envelope of the narrowband signal through a cubic spline function and the lower envelope ; Step a5, calculate the mean value of the envelope line , and the specific formula is as follows: ； Step a6, calculate the difference between the original narrowband signal and the mean value , and the specific formula is: ； Step a7, calculate the remaining amount of the narrowband signal , and the specific formula is: ； Step a8, reconstruct to obtain the electroencephalogram signal , and the specific formula is: ； where n represents the total number of electroencephalogram samples; i represents the i-th electroencephalogram sample point.

4. The system according to claim 3, wherein The specific method for extracting the second feature includes: Step b1: Based on the training set of cerebral blood oxygen, set the cerebral blood oxygen signal sequence as , where represents the total number of cerebral blood oxygen samples; after reconstruction, obtain -dimensional vector sequence: ； Among them, represents the th vector, and the specific expression is: ； Among them, represents the selected spatial dimension vector; Step b2: Calculate the Euclidean distance between any vectors, and define the maximum Euclidean distance between each vector as the maximum contribution component distance. The specific formula is: ； Among them, represents the th vector and the th vector the maximum contribution component distance therebetween; represents the coordinate value of a vector in a two-dimensional space, and ; Step b3: Based on a preset threshold and , calculate the measure of the probability magnitude representing regularity , and take the average value of all . The specific formula is as follows: ​ ； Step b4: Add 1 to , and iteratively execute steps b1 - b3 to obtain ; calculate the sample entropy of the cerebral blood oxygen signal sequence , and the specific formula is as follows: 。 5. The system according to claim 4, characterized in that, The value is 2, which is 0.1 - 0.25 times the standard deviation of the electroencephalogram data.

6. The system according to claim 2, wherein The feature extraction unit also includes: extracting the third feature from the training sets of ventilator data and monitor data through time domain, frequency domain and time-frequency domain analysis methods.

7. The system according to claim 6, wherein The multi-modal fusion unit also includes fusing the third feature on the basis of the first multi-modal anesthesia depth model to obtain the second multi-modal anesthesia depth model.

8. The system according to claim 1, characterized in that The multi-modal anesthesia depth model is a dynamic fuzzy neural network.

9. The system according to claim 8, wherein The dynamic fuzzy neural network includes an input layer, a membership function layer, a T-norm layer, a normalization layer and an output layer arranged in sequence; The input layer is used to input the electroencephalogram data and cerebral blood oxygen data of the wounded; The membership function in the membership function layer is characterized as: ； Among them, represents the input data; represents the Gaussian function; represents the center of this Gaussian function; The T-norm layer is used to perform weighted summation on the outputs of each fuzzy rule; The normalization layer is used to map the data after weighted summation to a standard normal distribution with a mean of 0 and a variance of 1; The output layer is used to output the anesthesia depth index.

10. An intelligent anesthesia depth detection device, characterized in that, It includes a processor, a memory and a bus. The memory stores instructions and data read by the processor. The processor is used to call the instructions and data in the memory to implement the system as described in any one of claims 1-9. The bus is connected between each functional component for transmitting information; The device further includes sensors for simultaneously collecting electroencephalogram data and cerebral blood oxygen data of the wounded. The main body of the sensor is a flexible substrate. A circuit board is arranged inside the flexible substrate. A first measurement electrode and a fifth measurement electrode are symmetrically arranged on the outermost sides at both ends of the flexible substrate. The first measurement electrode is used to detect the electroencephalogram signal on the outer side above the left eyebrow bone of the wounded's forehead. The fifth measurement electrode is used to detect the electroencephalogram signal on the outer side above the right eyebrow bone of the wounded's forehead. Between the first measurement electrode and the fifth measurement electrode, a second measurement electrode and a fourth measurement electrode are symmetrically arranged. The second measurement electrode is used to detect the electroencephalogram signal on the inner side above the left eyebrow bone of the wounded's forehead. The fourth measurement electrode is used to detect the electroencephalogram signal on the inner side above the right eyebrow bone of the wounded's forehead. Between the second measurement electrode and the fourth measurement electrode, a third measurement electrode is arranged and used as a grounding electrode; Between the first measurement electrode and the second measurement electrode, a first phototube and a second phototube are arranged for collecting the cerebral blood oxygen saturation of the left side of the wounded's forehead. Between the second measurement electrode and the third measurement electrode, a first light source is arranged for providing the light source required for collecting the cerebral blood oxygen saturation for the skin on the left side of the wounded's forehead; Between the fourth measurement electrode and the fifth measurement electrode, a third phototube and a fourth phototube are arranged for collecting the cerebral blood oxygen saturation of the right side of the wounded's forehead; Between the third measurement electrode and the fourth measurement electrode, a second light source is arranged for providing the light source required for collecting the cerebral blood oxygen saturation for the skin on the right side of the wounded's forehead.

Images