Intelligent anesthesia depth regulation and control system based on electroencephalogram signal closed-loop feedback

Abstract

The invention relates to the field of anesthesia, in particular to an intelligent anesthesia depth regulation and control system based on electroencephalogram signal closed-loop feedback, which comprises a data acquisition module, a BIS monitoring module, a video acquisition module, a motion state recognition module, a comparison processing module, a control module and a driving module, when the anesthesia depth value is higher than the preset anesthesia depth value, the control module controls the driving module to increase the anesthetic dosage, and when the anesthesia depth value is lower than the preset anesthesia depth value or the descending speed of the anesthesia depth value exceeds the preset anesthesia depth threshold value, the control module controls the driving module to reduce the anesthetic dosage, so that regulation and control are implemented; when the interval signal falls into the interval threshold value, the control module controls the driving module to adjust the anesthesia depth value to the anesthesia depth value corresponding to the special time point, and the patient can enter the appropriate anesthesia depth in advance when the operation at the special time point is carried out.

Description

Technical Field

[0001] The present invention relates to the field of anesthesiology, and particularly to an intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals. Background Art

[0002] Anesthesiology is a science that studies clinical anesthesia, regulation of vital functions, intensive care monitoring and treatment, and pain diagnosis and treatment. It is usually used during surgery or first aid. During traditional anesthesia, anesthesiologists need to frequently observe the patient's vital signs, judge the anesthesia depth, and manually adjust the infusion of anesthetic drugs. The work intensity is relatively high. Moreover, due to individual differences in the response to anesthetic drugs, it is difficult for traditional anesthesia methods to achieve complete personalization.

[0003] Currently, electroencephalogram signals have been applied to the field of anesthesia. That is, through Bispectral Index (BIS) monitoring, based on the analysis of the electroencephalogram power spectrum, the linear components of the electroencephalogram, such as frequency and power, are measured, and at the same time, the non-linear relationships between component waves, such as phase and harmonic, are analyzed. Various electroencephalogram signals representing different sedation levels are selected, standardized and digitized, and finally converted into a simple quantitative index. By analyzing the clinical responses of a large number of anesthetized patients and volunteers, such as body movement, hemodynamic changes, drug concentration, and electroencephalogram, the anesthesia depth number is obtained, that is, 0 - 100, where 0 represents complete absence of electroencephalogram activity and 100 represents a fully conscious state. Currently, this technology is the most widely used. Currently, there are also auditory evoked potential index and entropy index used to reflect the anesthesia depth.

[0004] Currently, based on the above technical principles, a number of patents such as application numbers 2021110339623, 2021112646024, 2022100101327, etc. have also disclosed how to use electroencephalogram signals to monitor anesthesia depth. However, due to different requirements for anesthesia depth at each stage of the operation, such as before strong stimulation operations like "surgical incision" and "exploration", it is necessary to deepen anesthesia in advance, and during stages with less stimulation like suturing, it is necessary to appropriately shallow anesthesia to accelerate the patient's recovery. Currently, the existing anesthesia depth regulation systems cannot perform intelligent adjustment for this, and anesthesiologists still need to adjust parameters in a timely manner according to experience. Therefore, the inventor proposes an intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals to solve the problems raised in the background art. The specific technical solutions are as follows:

[0006] To achieve the above and other related objectives, the present invention provides an intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals, including a data acquisition module, a BIS monitoring module, a video acquisition module, a motion state recognition module, a comparison processing module, a control module, and a driving module; wherein,

[0007] The data acquisition module is used to collect the physical information of the patient;

[0008] The BIS monitoring module is used to receive the physical information sent by the data acquisition module, monitor the anesthesia depth value of the patient in real time according to the physical information, and is electrically connected to the control module to send the anesthesia depth value to the control module;

[0009] The video acquisition module is used to record the surgical process in real time and send it to the motion state recognition module for processing;

[0010] The motion state recognition module is used to capture the motion state of the person in the video captured by the video acquisition module and send the motion state to the comparison processing module;

[0011] The comparison processing module is used to identify the time point of the current motion state during the surgical process according to the motion state, and judge the interval between this time point and the special time point. The special time point is the time point during the surgical process when the anesthesia depth needs to be changed. The comparison processing module sends the interval signal between this time point and the special time point to the control module;

[0012] The control module is used to set the anesthesia depth value and the interval threshold. The control module receives the anesthesia depth value and the interval signal sent by the comparison processing module. When the anesthesia depth value is higher than the preset anesthesia depth value, the control module controls the driving module to increase the dosage of the anesthetic. When the anesthesia depth value is lower than the preset anesthesia depth value or the decreasing speed of the anesthesia depth value exceeds the preset anesthesia depth threshold, the control module controls the driving module to reduce the dosage of the anesthetic, so as to implement the regulation. When the interval signal falls within the interval threshold, the control module controls the driving module to adjust the anesthesia depth value to the anesthesia depth value corresponding to this special time point.

[0013] Preferably, the driving module includes an infusion pump for controlling the intravenous infusion of the anesthetic.

[0014] Preferably, the motion state recognition module includes a video processing unit and a convolutional neural network operation unit; wherein,

[0015] The video processing unit is used to receive the video collected by the video acquisition module, decompose it into a continuous sequence of image frames according to the number of frames, and send it to the convolutional neural network operation unit;

[0016] The convolutional neural network operation unit is used to perform frame-by-frame feature extraction on the image frame sequence to obtain feature vectors representing each frame image. A long short-term memory network is introduced into the convolutional neural network operation unit, and the feature vectors are used as the input sequence of the long short-term memory network to obtain the motion state of the person in the video collected by the video acquisition module. The convolutional neural network operation unit needs to pre-collect a large number of surgical videos of different types and different scenarios, covering various common surgeries, to ensure that the model has wide adaptability. Then, professional medical personnel annotate the collected surgical videos frame by frame to obtain the surgical steps corresponding to each time period.

[0017] Preferably, it further includes a pharmacokinetic prediction module. The pharmacokinetic prediction module uses a three-compartment model to describe the distribution and metabolism of anesthetic drugs, and the mathematical expression is as follows:

[0018]

[0019] In the formula: C1, C2, and C3 are the drug concentrations in each compartment, kij is the transfer rate constant, R(t) is the infusion rate, and V1 is the volume of the central compartment;

[0020] Calculate k dynamically according to the patient's age, weight, liver and kidney functions ij：

[0021]

[0022] In the formula: CL is the drug clearance rate, Q2 and Q3 are the tissue blood flows of the two peripheral compartments, and V2 and V3 are the volumes of the two peripheral compartments;

[0023] Obtain the predicted anesthesia depth value, and the mathematical expression is as follows:

[0024]

[0025] In the formula: BIS perd is the predicted anesthesia depth value, BIS awake is the anesthesia depth value in the awake state, BIS max is the decrease amplitude of the anesthesia depth value when the anesthetic reaches the maximum inhibition, C e is the drug concentration in the effect compartment, EC 50 is the median effective concentration, and n is the steepness coefficient describing the concentration-effect curve.

[0026] Preferably, the control module includes a delay compensation controller. The delay compensation controller generates the anesthetic delivery rate based on the anesthesia depth value predicted by the pharmacokinetic prediction module.

[0027] Preferably, the control module further includes a Smith predictor, which is used to further optimize the anesthetic delivery rate based on the PID control system.

[0028] Preferably, the control module further includes a feedforward compensation module, which uses a long short-term memory network to predict the surgical steps within a specific future time and adjusts the input rate in advance according to the predicted surgical steps.

[0029] An intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals provided by the present invention has the following beneficial effects:

[0030] By adding a video acquisition module for real-time input of the surgical process, a motion state recognition module for capturing the motion state of the people in the video captured by the video acquisition module, identifying the time point of the current motion state in the surgical process, and a comparison processing module for predicting the interval between this time point and a special time point, it is possible to make the patient enter an appropriate anesthesia depth in advance when performing surgery at a special time point. Based on the BIS monitoring module to monitor the anesthesia depth value, the control module can control the driving module to work, so that the patient reaches the anesthesia depth value corresponding to each special time point;

[0031] By setting a video processing unit and a convolutional neural network operation unit, importing the surgical video into the convolutional neural network operation unit of the VGG architecture for training, dividing the image frame data set processed by the video processing unit into a training set, a validation set, and a test set, and performing model training and model evaluation to improve the accuracy of motion state recognition. By introducing modules such as a pharmacokinetic prediction module and a delay compensation controller, the problem of overshoot of the driving module caused by anesthesia delay response can be compensated, making the system more perfect and practical. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only 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.

[0033] Figure 1 It is a system block diagram of an intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals described in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] The following further elaborates in detail on an intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals in combination with the accompanying drawings and specific embodiments. According to the following description, the advantages and features of the present invention will be clearer. It should be noted that the accompanying drawings are all in a very simplified form and use non-precise scales, only for the purpose of conveniently and clearly assisting in explaining the embodiments of the present invention.

[0035] In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or

[0036] implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise specifically defined.

[0037] It should be noted that the illustrations provided in the following embodiments only schematically illustrate the basic concept of the present invention. Therefore,

[0038] the illustrations only show the components related to the present invention, rather than being drawn according to the number, shape, and size of the components in actual implementation. The actual form, quantity, and ratio of each component during actual implementation may be arbitrarily changed, and the layout form of its components may also be more complex.

[0039] An intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals, as Figure 1 shown, includes a data acquisition module, a BIS monitoring module, a video acquisition module, a motion state recognition module, a comparison and processing module, a control module, and a driving module;

[0040] The data acquisition module is used to acquire the patient's body information, including electroencephalogram signals, respiratory rate, blood oxygen content, and patient identity information;

[0041] The BIS monitoring module is used to receive the body information sent by the data acquisition module, monitor the patient's anesthesia depth value in real time according to the body information, and is electrically connected to the control module to send the anesthesia depth value to the control module;

[0042] The video acquisition module is used to record the surgical process in real time and send it to the motion state recognition module for processing;

[0043] The motion state recognition module is used to capture the motion state of the person in the video captured by the video acquisition module and send the motion state to the comparison and processing module;

[0044] A comparison processing module, configured to identify, according to a motion state, a time point of the current motion state during a surgical procedure, and predict an interval between the time point and a special time point, where the special time point is a time point during the surgical procedure when the anesthesia depth needs to be changed, and the comparison processing module sends a signal of the interval between the time point and the special time point to a control module;

[0045] A control module, configured to set an anesthesia depth value and an interval threshold. The control module receives the anesthesia depth value and the interval signal sent by the comparison processing module. When the anesthesia depth value is higher than a preset anesthesia depth value, the control module controls a driving module to increase the dosage of anesthetic. When the anesthesia depth value is lower than the preset anesthesia depth value or the decreasing speed of the anesthesia depth value exceeds a preset anesthesia depth threshold, the control module controls the driving module to reduce the dosage of anesthetic, so as to implement regulation. When the interval signal falls within the interval threshold, the control module controls the driving module to adjust the anesthesia depth value to the anesthesia depth value corresponding to the special time point.

[0046] In a specific embodiment, a data acquisition module acquires the physical information of a patient and sends it to a BIS monitoring module. The BIS monitoring module monitors the anesthesia depth value of the patient in real time according to the received physical information. A video acquisition module records the surgical procedure in real time. A motion state recognition module captures the motion state of a person in the video captured by the video acquisition module and sends it to the comparison processing module for motion state comparison. The comparison processing module identifies a time point of the current motion state during the surgical procedure, and predicts that the interval between the time point and a special time point is 1 minute. The special time point requires an increase in the anesthesia depth value, and the interval threshold is preset to 1 minute. At this time, the control module controls the driving module to increase the input amount of anesthetic, so that when performing the surgery at the special time point, the patient can enter an appropriate anesthesia depth in advance. Based on the monitoring of the anesthesia depth value by the BIS monitoring module, the control module can control the driving module to work, so that the patient reaches the anesthesia depth value corresponding to each special time point.

[0047] As an implementation manner, the driving module includes an infusion pump, configured to precisely control the intravenous input of anesthetic.

[0048] As an implementation, the motion state recognition module includes a video processing unit and a convolutional neural network operation unit. The video processing unit is used to receive the video collected by the video acquisition module, decompose it into a continuous sequence of image frames by frame number, and send it to the convolutional neural network operation unit. If the operation ability of the comparison processing module used in the operation is weak, then one frame is extracted every fixed number of frames as a representative frame to form an image key frame sequence including key information, and then sent to the convolutional neural network operation unit to reduce the amount of computation and reduce the operation delay. The convolutional neural network operation unit is used to perform frame-by-frame feature extraction on the image frame sequence or the image key frame sequence to obtain the feature vector of each representative frame image. A long short-term memory network is introduced into the convolutional neural network operation unit, and the feature vector is used as the input sequence of the long short-term memory network to obtain the motion state of the person in the video collected by the video acquisition module. The convolutional neural network operation unit needs to pre-collect a large number of surgical videos of different types and different scenarios, covering various common surgeries to ensure that the model has wide adaptability. Then, professional medical personnel annotate the collected surgical videos frame by frame to clarify the surgical steps corresponding to each time period, such as "preparation before skin incision", "skin incision", "preparation before tissue separation", "tissue separation", "preparation before wound suture", "wound suture", "preparation before exploration", "exploration", etc., and mark the surgical operations that require adjustment of the anesthesia depth value, such as "skin incision", "tissue separation", "wound suture", "exploration", etc. as special time points. The surgical video is imported into the convolutional neural network operation unit of the VGG architecture for training. The image frame data set processed by the video processing unit is divided into a training set, a validation set, and a test set for model training and model evaluation to improve the accuracy rate.

[0049] Since it takes 1-2 minutes for the anesthesia to take effect after injection, there is asynchrony with the anesthesia depth value output by the BIS monitoring module, which will cause overshoot of the driving module. For this reason, in one implementation, the system further includes a pharmacokinetic prediction module. The pharmacokinetic prediction module uses a three-compartment model to describe the distribution and metabolism of anesthetic drugs, and the mathematical expression is as follows:

[0050]

[0051] In the formula: C1, C2, and C3 are the drug concentrations in each compartment, k ij is the transfer rate constant, R(t) is the infusion rate, and V1 is the volume of the central compartment;

[0052] Dynamically calculate k according to the patient's age, weight, liver and kidney function ij：

[0053]

[0054] Where: CL is the drug clearance rate, Q2 and Q3 are the tissue blood flows of two peripheral compartments, and V2 and V3 are the volumes of two peripheral compartments;

[0055] Obtain the predicted anesthesia depth value, and the mathematical expression is as follows:

[0056]

[0057] Where: BIS perd is the predicted anesthesia depth value, BIS awake is the anesthesia depth value in the awake state, BIS max is the decrease amplitude of the anesthesia depth value when the anesthetic reaches the maximum inhibition, C e is the drug concentration in the effect compartment, EC 50 is the median effective concentration, and n is the steepness coefficient describing the concentration-effect curve;

[0058] The control module includes a delay compensation controller, and the delay compensation controller generates an anesthetic delivery rate based on the anesthesia depth value predicted by the pharmacokinetic prediction module; specifically,

[0059] Based on the pharmacokinetic prediction module, predict the anesthesia depth value in the next T minutes, and the mathematical expression is as follows:

[0060] BIS(t + τ) = f(C e (t), R(t + τ)) (τ = 1, 2,..., T)

[0061] Optimize the infusion rate R(t) to make the predicted anesthesia depth value track the target value, and the mathematical expression is:

[0062]

[0063] Where: BIS(t + τ) is the future BIS value predicted by the pharmacokinetic prediction module, R(t) is the infusion rate to be optimized, τ is the time step within the prediction horizon, T is the length of the prediction horizon, BIS target is the target BIS value;

[0064] Introduce a Smith predictor to compensate for the delay, and the mathematical expression is:

[0065]

[0066] Where the error signal e(t) = BIS target - BIS perd (t),

[0067] Where: u(t) is the adjusted infusion rate, K p 、K i 、K dare the proportional, integral, and derivative gain coefficients.

[0068] It also includes a feedforward compensation module. The feedforward compensation module uses a long short-term memory network to predict the surgical steps within a specific future time. The mathematical expression is:

[0069] Step(t + 1) = LSTM(Step(t), Step(t - 1), Step(t - 2),...)

[0070] And it adjusts the input rate in advance according to the predicted surgical steps. The mathematical expression is:

[0071]

[0072] where: ΔR i is the preset increment for step i, and w i is the confidence weight.

[0073] Through the above delay compensation controller and feedforward compensation module, the problem of overshoot of the drive module caused by the anesthetic delay response can be compensated for.

[0074] It should be noted that all the above operations and data will be displayed on the monitor, and the anesthesiologist can intervene at any time through the control module.

[0075] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals, characterized in that, It includes a data acquisition module, a BIS monitoring module, a video acquisition module, a motion state recognition module, a comparison and processing module, a control module, and a driving module. Among them, The data acquisition module is used to acquire the physical information of the patient. The BIS monitoring module is used to receive the physical information sent by the data acquisition module, monitor the anesthesia depth value of the patient in real time according to the physical information, and is electrically connected to the control module to send the anesthesia depth value to the control module. The video acquisition module is used to record the surgical process in real time and send it to the motion state recognition module for processing. The motion state recognition module is used to capture the motion state of the person in the video captured by the video acquisition module and send the motion state to the comparison and processing module. The comparison and processing module is used to identify the time point of the current motion state in the surgical process according to the motion state, and judge the interval between this time point and the special time point. The special time point is the time point when the anesthesia depth needs to be changed during the surgical process. The comparison and processing module sends the interval signal between this time point and the special time point to the control module. The control module is used to set the anesthesia depth value and the interval threshold. The control module receives the anesthesia depth value and the interval signal sent by the comparison and processing module. When the anesthesia depth value is higher than the preset anesthesia depth value, the control module controls the driving module to increase the dosage of the anesthetic. When the anesthesia depth value is lower than the preset anesthesia depth value or the decreasing speed of the anesthesia depth value exceeds the preset anesthesia depth threshold, the control module controls the driving module to reduce the dosage of the anesthetic, so as to implement the regulation. When the interval signal falls within the interval threshold, the control module controls the driving module to adjust the anesthesia depth value to the anesthesia depth value corresponding to this special time point.

2. The intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals according to claim 1, characterized in that The driving module includes an infusion pump, which is used to control the intravenous infusion of the anesthetic.

3. An intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals according to claim 1, characterized in that, The motion state recognition module includes a video processing unit and a convolutional neural network operation unit. Among them, The video processing unit is used to receive the video collected by the video acquisition module, decompose it into a continuous sequence of image frames according to the number of frames, and send it to the convolutional neural network operation unit. The convolutional neural network operation unit is used to extract the features of each frame of the image frame sequence to obtain the feature vector of each representative frame image. A long short-term memory network is introduced into the convolutional neural network operation unit. The feature vector is used as the input sequence of the long short-term memory network to obtain the motion state of the person in the video collected by the video acquisition module. The convolutional neural network operation unit needs to pre-collect a large number of surgical videos of different types and different scenarios, covering various common surgeries to ensure that the model has wide adaptability. Then, professional medical personnel annotate each frame of the collected surgical videos to obtain the surgical steps corresponding to each time period.

4. An intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals according to claim 3, characterized in that, It also includes a pharmacokinetic prediction module. The pharmacokinetic prediction module uses a three-compartment model to describe the distribution and metabolism of anesthetic drugs. The mathematical expression is as follows: In the formula: C1, C2, and C3 are the drug concentrations in each compartment, kij is the transport rate constant, R(t) is the infusion rate, and V1 is the volume of the central compartment. Dynamically calculate k based on the patient's age, weight, liver and kidney functions ij： Where: CL is the drug clearance rate, Q2 and Q3 are the tissue blood flows of the two peripheral compartments, and V2 and V3 are the volumes of the two peripheral compartments; Obtain the predicted anesthesia depth value, and the mathematical expression is as follows: Where: BIS perd is the predicted anesthesia depth value, BIS awake is the anesthesia depth value in the awake state, BIS max is the decline amplitude of the anesthesia depth value when the anesthetic reaches the maximum inhibition, C e is the drug concentration in the effect compartment, EC 50 is the median effective concentration, and n is the steepness coefficient describing the concentration-effect curve.

5. An intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals according to claim 4, characterized in that, The control module includes a delay compensation controller, and the delay compensation controller generates an anesthetic delivery rate based on the anesthesia depth value predicted by the pharmacokinetic prediction module.

6. The intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals according to claim 5, characterized in that, The control module further includes a Smith predictor, and the Smith predictor is used to further optimize the anesthetic delivery rate based on the PID control system.

7. An intelligent anesthesia depth regulation system based on closed-loop feedback of electroencephalogram signals according to claim 6, characterized in that, The control module also includes a feed-forward compensation module, and the feed-forward compensation module uses a long short-term memory network to predict the surgical steps within a specific future time and adjusts the input rate in advance according to the predicted surgical steps.

Images