Intraoperative anesthesia state monitoring equipmentTechnical Field
The invention relates to the technical field of medical equipment, in particular to an intraoperative anesthesia state monitoring device.
Background
Anesthesia, which is used to eliminate pain during surgery, has become one of the major problems that need to be considered during surgery. While actual anesthesia is in progress, assessing the depth of anesthesia during surgery is critical to patient pain and physical damage from excessive anesthesia. Although much work has been done on knowledge and measurement of the depth of anesthesia, the development of an index capable of objectively evaluating the depth of anesthesia is still insufficient in addition to the use of vital signs of a patient, and the monitoring result of a combination of an Electrocardiogram (ECG), blood pressure and an electroencephalogram (EEG) is a viable approach to evaluate the depth of anesthesia as an objective index of the depth of anesthesia and the health status of an anesthetized patient.
The anesthesiologist should be continuously monitored, and electronic monitors have been developed to enhance the visual perception of anesthesiologists to more accurately describe the patient's condition. The two most critical parameters monitored during anesthesia are electrocardiogram and arterial blood pressure. An electrocardiogram may give an image of the heart's microenvironment and blood pressure monitoring may help determine if blood flow to the organ is sufficient. An electrocardiogram is used as an indicator of cardiac electrical activity, which provides information about arrhythmias and myocardial environment during anesthesia. Heart rate can be categorized as normal, tachycardia or bradycardia, and many disease conditions, anesthetics and analgesics affect the patient's heart rate, and under normal conditions, the target heart rate of an anesthetized patient should be close to the normal heart rate of an animal when sleeping.
Arterial blood pressure measurement and monitoring is the most important technique for determining the status of an anesthetized patient. The average blood pressure is important because it determines the sufficiency of blood flow to and within the organs, and most organs within the human body have an intrinsic autoregulation system that maintains proper tissue perfusion over the average blood pressure range of 60-120 mmhg. When the average blood pressure is below 60 mmhg, blood flow is compromised and tissue and organ dysfunction may occur. Many analgesic/anesthetic drugs can affect blood pressure, most commonly lowering blood pressure, and should be analyzed specifically based on the case of the use of the anesthetic drug in the surgery.
Electroencephalograph monitors, such as the BISTM monitor (Covidien, boulder, CO USA), have been widely used in the past decades. The concentration-dependent changes in electroencephalogram waveforms between different drugs are very similar and they enhance gamma aminobutyric acid receptor type a (GABAA). Many factors, such as nociceptive stimuli, high carbon dioxide respiration, low carbon dioxide respiration, and hypothermia, also affect changes in the original brain wave shape. Among them, the nociceptive stimulus plays a very important role in the surgical procedure, such as the original brain wave shape when using isoflurane for anesthesia, and the amplitude is shallower and the frequency is higher when using mild anesthesia. When higher concentrations are given, the amplitude deepens and the brain wave frequency slows down. During deep anesthesia, a pattern of "bursts and suppressions" becomes apparent, possessing the property of extreme activity, manifested as high frequency, large amplitude waves (bursts), alternating with flat trajectories (suppressions). This pattern, excluding cerebral ischemia or other factors, indicates that the anesthesia is too deep. In addition, the planar trajectory becomes dominant and the final waveform is no longer apparent. The order of this pattern change is almost identical during the anesthesia of isoflurane, sevoflurane or propofol. The major difference in power between the volatile drug (isoflurane or sevoflurane) and the electroencephalogram of propofol in the theta wave range is evident. During propofol anesthesia, the θ -wave energy is low regardless of concentration, but during isoflurane or sevoflurane anesthesia, the θ -wave energy increases under surgical concentration anesthesia.
However, the existing anesthesia monitoring equipment can only monitor by a single technology, and no related monitoring equipment which has complete functions and can monitor the anesthesia degree and the physical state of an anesthetized patient simultaneously exists; meanwhile, the existing monitoring equipment monitors through single parameters, so that the monitoring result of the anesthesia degree is relatively single, and the effect that the multi-parameter monitoring is mutually corresponding to the evidence and the accuracy of the monitoring result is improved cannot be achieved; in addition, the existing electroencephalogram anesthesia monitoring equipment is complex in operation, intelligent, simple and easy to use cannot be achieved, and a doctor needs to accurately paste corresponding electrodes to corresponding positions in the use process, so that the operation is complex.
Disclosure of Invention
In view of the above-mentioned problems, the present invention aims to provide an intraoperative anesthesia state monitoring apparatus combining an Electrocardiogram (ECG), a blood pressure, and an electroencephalogram (EEG), which can monitor the physical function state of an anesthesiologist while monitoring the actual anesthesia degree, and maximally secure the physical health of the anesthesiologist; the monitoring device can also verify the monitoring result of the electroencephalogram data through the change of the electrocardio and the blood pressure, is convenient for doctors to use, resists radiation and interference, has the advantages of portability and easy use, and is suitable for anesthesiologists with different operation conditions and different levels.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
the utility model provides an anesthesia status monitoring equipment in art, includes brain electricity monitoring module, its characterized in that: the system also comprises an electrocardio monitoring module, a blood pressure monitoring module and an MCU;
the electroencephalogram monitoring module is used for monitoring EEG (electroencephalogram) of an anesthetized patient;
the electrocardio monitoring module is used for monitoring electrocardio data ECG of an anesthetized patient;
the blood pressure monitoring module is used for monitoring instant blood pressure data of an anesthetized patient;
the MCU is used for collecting and processing the data of the electroencephalogram monitoring module, the electrocardiographic monitoring module and the blood pressure monitoring module, generating BIS index data and anesthesia depth indexing information and outputting the BIS index data and anesthesia depth indexing information.
Further, the brain electricity monitoring module comprises a three-lead brain electricity electrode of silver chloride and a brain electricity analog-to-digital converter, wherein the three-lead brain electricity electrode is in communication connection with the brain electricity analog-to-digital converter through an AFE analog front-end circuit, and the brain electricity analog-to-digital converter is in communication connection with the MCU through an SPI bus.
Furthermore, the electrocardio monitoring module comprises an electrocardio sensor and an electrocardio analog-to-digital converter, wherein the electrocardio sensor is in communication connection with the electrocardio analog-to-digital converter through an AFE analog front-end circuit, and the electrocardio analog-to-digital converter is in communication connection with the MCU through an SPI bus.
Further, the blood pressure monitoring module comprises a blood pressure sensor and a blood pressure analog-to-digital converter, wherein the blood pressure sensor is in communication connection with the blood pressure analog-to-digital converter through an AFE analog front-end circuit, and the blood pressure analog-to-digital converter is in communication connection with the MCU through an IIC bus.
Furthermore, a WIFI/Bluetooth connection module and a communication interface are embedded and arranged on the MCU, and keys and a screen are also connected to the MCU; the monitoring device also includes a battery management system that provides power to the entire monitoring device.
Furthermore, the MCU adopts an algorithm model based on a deep learning framework combining CNN and LSTM/GRU/RNN to generate anesthesia depth classification information;
the algorithm model comprises any one of three CNN convolutional neural networks and LSTM, GRU, RNN, wherein the three CNN convolutional neural networks respectively extract key features of EEG data, ECG data and blood pressure data, then LSTM or GRU or RNN is used for splicing the extracted key features of the EEG data, the ECG data and the blood pressure data and extracting the features to a deeper degree, and finally anesthesia depth classification is carried out.
Further, the anesthesia depth comprises three stages of anesthesia induction, anesthesia maintenance and anesthesia reviving.
Further, each CNN convolutional neural network includes an input layer, 4 convolutional layers, 2 max pooling layers and 1 dropout layer, and the ECG and EEG two characteristic parameters share weights in the convolutional layers.
Further, after key features of the extracted EEG data, ECG data and blood pressure data are spliced by the LSTM, GRU or RNN, time sequence features are extracted first, then the probability of each anesthesia depth classification is obtained through Softmax, and the category corresponding to the maximum probability is selected as the anesthesia depth classification.
The beneficial effects of the invention are as follows: compared with the prior art, the invention has the advantages that,
1. the invention provides an intraoperative anesthesia state monitoring device combining Electrocardiogram (ECG), blood pressure and electroencephalogram (EEG), which can monitor the physical function state of an anesthetized patient and monitor the actual anesthesia degree, thereby maximally guaranteeing the physical health of the anesthetized patient; the monitoring device can also verify the monitoring result of the brain electrical data through the change of the electrocardio and the blood pressure, and realize that the multi-parameter monitoring is mutual evidence, thereby improving the accuracy of the monitoring result;
2. the monitoring equipment integrates the monitoring of an Electrocardiogram (ECG), blood pressure and an electroencephalogram (EEG) on one equipment, is convenient for doctors to use, resists radiation and interference, uses the blood pressure and the electrocardio electrode position as the basis to position the electroencephalogram electrode, and relatively enables the electroencephalogram electrode position to be easier to judge, thereby simplifying the operation flow of anesthesiologists, having the advantages of portability and easy use, and being suitable for anesthesiologists with different operation conditions and different levels.
3. According to the monitoring equipment, anesthesia depth classification information is generated through an algorithm model of a deep learning framework combining CNN and LSTM/GRU/RNN, when an MCU acquires data of an Electrocardiogram (ECG), blood pressure and an electroencephalogram (EEG), the data are firstly used for judging the anesthesia degree of a patient in the state (the system also supports an anesthesia mode and time of an active option of an anesthesiologist), different anesthesia analysis models are matched based on the anesthesia degree judgment, the anesthesia effect and the potential anesthesia operation damage are judged through algorithm comparison of the difference of the state of the anesthetized patient and the standard anesthesia state and the abnormal state in the real-time monitoring process, the whole process is more intelligent, and real-time monitoring of the anesthesia degree and the physical skill state of the patient using different anesthesia methods can be realized.
Drawings
FIG. 1 is a block diagram of the hardware system of the anesthesia status monitoring device of the present invention.
FIG. 2 is a schematic diagram of a deep learning-based multi-sign parameter anesthesia depth assessment architecture according to the present invention.
FIG. 3 is a schematic diagram of a deep learning-based multi-sign parameter anesthesia depth assessment algorithm architecture according to the present invention.
Detailed Description
In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is further described below with reference to the accompanying drawings and examples.
Embodiment one:
an intraoperative anesthesia state monitoring device is shown in figure 1, and comprises an electroencephalogram monitoring module, an electrocardiograph monitoring module, a blood pressure monitoring module and an MCU;
the electroencephalogram monitoring module is used for monitoring EEG (electroencephalogram) of an anesthetized patient;
the electrocardio monitoring module is used for monitoring electrocardio data ECG of an anesthetized patient;
the blood pressure monitoring module is used for monitoring instant blood pressure data of an anesthetized patient;
the MCU is used for collecting and processing the data of the electroencephalogram monitoring module, the electrocardiograph monitoring module and the blood pressure monitoring module, generating BIS index data and outputting the BIS index data;
the intra-operative anesthesia state monitoring device is arranged at the head position of a patient through a single-lead electrocardio electrode (an electrocardio sensor) to monitor electrocardio data of the patient under anesthesia and is used for monitoring the anesthesia degree and the health of the patient; the three-lead brain electrical electrodes of silver chloride are arranged at the positions of the heads FP1 and FP2 of the patient to monitor brain electrical data of the anesthetized patient, and are used for monitoring the anesthesia degree of the anesthetized patient; the instantaneous blood pressure data of an anesthetized patient is monitored by placing a blood pressure sensor in the patient's head position. The sensors can be used for judging the anesthesia degree and physical skill status of an anesthetized patient in the anesthetized state of an anesthetized operation.
Specifically, the electroencephalogram monitoring module comprises a three-lead electroencephalogram electrode of silver chloride and an electroencephalogram analog-to-digital converter, the three-lead electroencephalogram electrode inputs an acquired analog signal to the ADS1299 through an AFE analog front-end circuit, the ADS1299 is a 24-bit analog-to-digital converter, the analog-to-digital converter converts an EEG analog signal into a digital signal finally, and then the digital signal is input to the MCU through an SPI bus, and the MCU refers to a Micro-controller-unit.
The electrocardio monitoring module comprises an electrocardio sensor and an electrocardio analog-to-digital converter, wherein the electrocardio sensor firstly carries out signal conditioning such as filtering amplification through a signal conditioning circuit, then sends the conditioned signal to the analog-to-digital converter in the MCU, and the analog-to-digital converter converts the conditioned signal into a digital signal and then processes the digital signal in the MCU; the electrocardiosignal is also a silver chloride electrode in fact, and similar to EEG electroencephalogram, the electrocardiosignal of the patient is acquired by the electrode, and is processed through an AFE analog front end, then is subjected to analog-to-digital conversion, and finally is transmitted to the MCU through an SPI bus.
The blood pressure monitoring module comprises a blood pressure sensor and a blood pressure analog-to-digital converter, wherein the blood pressure sensor firstly carries out signal conditioning such as filtering amplification through a signal conditioning circuit, then sends the conditioned signal to the analog-to-digital converter in the MCU, and the analog-to-digital converter converts the conditioned signal into a digital signal and then processes the digital signal in the MCU; the blood pressure sensor is in communication connection with the blood pressure analog-to-digital converter through the AFE analog front-end circuit, and the blood pressure analog-to-digital converter is in communication connection with the MCU through the IIC bus.
The MCU is embedded with a WIFI/Bluetooth connection module and a communication interface, and is also connected with a key and a screen; the monitoring device also includes a battery management system that provides power to the entire monitoring device.
After receiving the data sent by the 3 sensors (the three-lead electroencephalogram electrode, the electrocardio sensor and the blood pressure sensor), the MCU firstly performs basic processing such as filtering and then obtains the final value of BIS index (brain electricity double-frequency index) after data fusion
The BIS index value calculated by the MCU can be given to the host through a corresponding communication line, data can be sent out through wireless communication media such as wifi/BLE Bluetooth, and the like, and the data are sent to the cloud for processing so as to obtain a more accurate BIS index value; the changing trend of BIS index value and the like can be displayed on the screen in real time; the key can realize interaction of the whole system in different working states and the like in different working processes; the BMS battery management system mainly realizes energy management of the system.
Embodiment two:
on the basis of the first embodiment, the monitoring equipment can also calculate and output anesthesia depth classification information of an anesthetized patient through the MCU, and runs an algorithm model of a trained deep learning framework combining CNN and GRU/RNN/LSTM through various vital sign parameters (EEG, ECG and blood pressure) of the anesthetized patient monitored in real time, so that anesthesia depth information can be calculated quantitatively finally, and the workload of anesthetizing doctors in operation is further reduced.
The deep learning-based anesthesia depth system framework is shown in fig. 2, the core is a deep learning algorithm based on a convolutional neural network and a long-term and short-term memory network, input data of the algorithm are ECG (electrocardiogram) data, ECG (electrocardiogram) data and blood pressure data after acquisition and analysis in the first embodiment, and algorithm output is anesthesia depth classification, wherein the algorithm output comprises 3 stages of anesthesia induction, anesthesia maintenance and anesthesia recovery.
The algorithm core process of the deep learning framework based on the combination of CNN and LSTM/GRU/RNN is that key features of EEG, ECG and blood pressure are respectively extracted through CNN convolutional neural network, and the algorithm is similar to data filtering. And then splicing the key data of the three, extracting features of a deeper degree by using LSTM (LSTM), GRU (GRU) or common RNN, and finally classifying the anesthesia depth.
The convolutional neural network mainly comprises five layers, namely a data Input Layer (Input Layer), a convolutional calculation Layer (CONV Layer), a ReLU excitation Layer (ReLU Layer), a Pooling Layer (Pooling Layer) and a full-connection Layer (FC Layer), which are sequentially connected, and each Layer receives output characteristic data of the previous Layer and provides the output characteristic data for the next Layer. The data input layer extracts the characteristics of input data; the convolution calculation layer carries out convolution mapping on the features; the excitation layer is used for exciting the neurons by using a nonlinear excitation function to achieve the condition and transmitting the characteristic information to the next neurons; the pooling layer is used for compressing the data and parameters, so that the over-fitting condition is reduced; the full connection layer is used for connecting the characteristic information of all the output layers and collecting and finishing the information to finish the output.
A recurrent neural network (Recurrent Neural Network, RNN) is a neural network for processing sequence data. Compared to a general neural network, he can process data of a sequence variation. For example, RNNs can solve such problems well, because the meaning of a word may be different from what is mentioned above. Long short-term memory (LSTM) is a special RNN, mainly to solve the problems of gradient extinction and gradient explosion in the Long sequence training process. In short, LSTM is able to perform better in longer sequences than normal RNNs. Long short-term memory (LSTM) is a special RNN, mainly to solve the problems of gradient extinction and gradient explosion in the Long sequence training process. In short, LSTM is able to perform better in longer sequences than normal RNNs.
Gate cycle unit (GRU): like LSTM, GRU solves the problem of gradient disappearance of simple RNNs. However, a difference from LSTM is that the GRU uses fewer gates and has no separate internal memory, i.e. cell state. Thus, the GRU relies entirely on hidden states as memory, resulting in a simpler architecture. The reset gate is responsible for short term memory because it decides how much past information to retain and ignore. The refresh gate is responsible for long-term memory, comparable to the forgetting gate of LSTM.
The algorithm model in the invention comprises any one of three CNN convolutional neural networks and LSTM, GRU, RNN, wherein the three CNN convolutional neural networks respectively extract key features of EEG data, ECG data and blood pressure data, then the key features of the extracted EEG data, ECG data and blood pressure data are spliced and extracted by using LSTM (global system for mobile unit) or GRU (global system for mobile unit) or RNN (global system for mobile unit), and finally the anesthesia depth is classified.
More specifically, the former part of the algorithm acquires key features through CNN convolutional neural drop networks, then reduces data volume, and preferably performs early data processing, and the deep learning models of the branches corresponding to the 3 key sign parameters of the part have the same structure and all comprise 4 convolutional layers, 2 maximum pooling layers and 1 dropout layer. Since the two key parameters of the ECG and EEG are similar, the ECG and EEG branches employ the same CNN convolutional neural network key parameters. Also to provide efficiency-reducing training and model runtime after training is completed, both the ECG and EEG feature parameters share weights in the convolutional neural network part, i.e. the same weights are used. The critical features of the signals of blood pressure have some unique features with the ECG and EEG, so the parts requiring blood pressure need to be modeled using different critical parameters from the ECG and EEG. Essentially the purpose of using CNN convolutional neural networks for these 3 physical features is to reduce the amount of data to obtain key features.
The latter part of the whole algorithm adopts LSTM or GRU or RNN to splice and fuse the simplified key characteristic data acquired through CNN convolutional neural network, because the 3 characteristic parameters of EEG, ECG and blood pressure have great relation with time sequence, after the 3 characteristics are spliced according to channels, the time sequence characteristic extracted by LSTM is adopted, finally the probability of each anesthesia depth class is acquired through Softmax, and the class corresponding to the maximum probability is selected as the anesthesia depth class. In 3 anesthesia depth classifications, the accuracy of anesthesia wake prediction is of paramount importance, because patients may have intra-operative awareness, affecting post-operative recovery, whereas misclassification errors between anesthesia induction and anesthesia maintenance do not have a major impact on the patient. Therefore, the loss function in model training adopts a weighted Softmax function to give different weights to the 3 anesthesia depth classification results, namely, the punishment of anesthesia wakeup prediction errors is increased.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made without departing from the spirit and scope of the invention, which is defined in the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.