CN118749998B

CN118749998B - A minimally invasive surgery nerve monitoring system

Info

Publication number: CN118749998B
Application number: CN202410788918.0A
Authority: CN
Inventors: 张富丽
Original assignee: Beijing Tiantan Hospital
Current assignee: Beijing Tiantan Hospital
Priority date: 2024-06-19
Filing date: 2024-06-19
Publication date: 2025-02-11
Anticipated expiration: 2044-06-19
Also published as: CN118749998A

Abstract

The invention relates to the field of artificial intelligence, in particular to a minimally invasive surgery nerve monitoring system, which comprises an acquisition module, an analysis module, a processing module and an adjustment module; the system comprises an acquisition module, an analysis module, a processing module, a pre-warning module, a processing module and a safety warning module, wherein the acquisition module monitors electroencephalogram signals and related neuroanatomical physiological parameters in minimally invasive surgery in real time, the analysis module calculates nerve path length and waveform time intervals, a three-dimensional nerve model is constructed, abnormal electroencephalogram signals are identified, the processing module integrates historical data to evaluate surgery risks, the pre-warning module outputs safety warning according to comparison between the risk degree and a preset threshold, and the system effectively monitors nerve conditions in minimally invasive surgery and improves surgery safety.

Description

Minimally invasive surgery nerve monitoring system

Technical Field

The invention relates to the field of artificial intelligence, in particular to a minimally invasive surgery nerve monitoring system.

Background

With the development of minimally invasive surgical techniques, nerve monitoring systems have also begun to find application in minimally invasive surgery. The system can monitor indexes such as nerve function, blood flow and the like in real time, and helps doctors to better control operation risks and protect nerve function.

Patent document publication CN105744887a discloses systems and methods for neurophysiologic assessment during surgical procedures. The health and status of the inferior motor nerve pathways is established in tissues having neural structures that, when contacted or infringed, cause nerve damage to the patient. Computer-implemented methods monitor the health of the underlying motor nerve pathways prior to and during spinal surgery. The method includes transmitting a first percutaneous, abdominal stimulation signal to a spinal nerve root above a surgical target site, and determining a neuromuscular response dataset based on transmission of the stimulation signal to muscles positioned below the surgical target site.

It follows that there is a problem in the prior art that the overall health of the patient, the risk of surgery and possible complications are underassessed prior to surgery, resulting in a low safety risk of surgery.

Disclosure of Invention

Therefore, the invention provides a minimally invasive surgery nerve monitoring system, which is used for solving the problem that the prior art has low surgery risk safety due to insufficient evaluation of the overall health condition, surgery risk and possible complications of a patient before surgery.

To achieve the above object, the present invention provides a minimally invasive surgical nerve monitoring system, comprising:

the acquisition module is used for acquiring electroencephalogram signals of minimally invasive surgery in real time and also used for acquiring nerve fiber length, myelin sheath thickness, nerve conduction speed, fiber bundle paths, synapse number and brain interval distance;

The analysis module is connected with the acquisition module and used for calculating nerve path length according to the brain interval distance, the synapse quantity and the fiber bundle path, calculating waveform time intervals according to the nerve path length, the bone sheath thickness and the nerve conduction speed, constructing a three-dimensional nerve model according to the synapse quantity, the nerve fiber length, the nerve conduction speed and the bone sheath thickness, and identifying abnormal phase signals of the electroencephalogram signals according to the three-dimensional nerve model and the waveform time intervals;

the processing module is connected with the analysis module and used for acquiring historical data of the minimally invasive surgery and evaluating the risk degree of the minimally invasive surgery according to the abnormal phase signals and the historical data;

and the early warning module is connected with the processing module and used for obtaining a comparison result according to the comparison of the risk degree and the preset safety risk degree and outputting safety warning information when the comparison result exceeds the safety risk degree.

Further, the analysis module includes:

A nerve length calculation unit for obtaining the nerve path length from a sum of the brain interval distances, a length synthesis of the fiber bundle paths, and a sum of the synapses;

And the waveform interval calculation unit is connected with the nerve length calculation unit and is used for obtaining the waveform time interval according to the quotient of the nerve path length and the nerve conduction speed plus the quotient of the bone sheath thickness and the nerve conduction speed.

Further, the analysis module further comprises:

A parameter input unit for taking the number of synapses, the nerve fiber length, the nerve conduction velocity and the bone sheath thickness as input parameters of a model;

the nerve model construction unit is connected with the parameter input unit and used for constructing a three-dimensional nerve model according to the input parameters;

The signal simulation unit is connected with the nerve model construction unit and is used for simulating the propagation process of nerve impulse according to the three-dimensional nerve model and the waveform time interval so as to calculate a standard phase signal;

the abnormal identification unit is connected with the signal simulation unit and is used for comparing the standard phase signal with the electroencephalogram signal to identify the abnormal phase signal.

Further, the neural model construction unit includes:

A geometric modeling subunit, configured to create geometric structures of neurons and neural networks according to input parameters of the model to obtain a geometric model;

the electrophysiology simulation subunit is connected with the geometric modeling subunit and used for simulating the physiological characteristics of the neurons to obtain an electrophysiology model;

And the signal transmission modeling sub-unit is respectively connected with the geometric modeling sub-unit and the electrophysiological simulation sub-unit and is used for building a model for transmitting the electroencephalogram signals in the neural network according to the geometric model and the electrophysiological model so as to obtain a signal transmission model.

Further, the signal simulation unit includes:

a numerical simulation subunit, configured to simulate a propagation process of neural activity in the neural network by using a numerical method to obtain a propagation result;

the signal synthesis subunit is connected with the numerical simulation subunit and used for synthesizing corresponding electroencephalogram signals according to the transmission result to obtain synthesized signals;

and the standard phase signal calculation subunit is connected with the signal synthesis subunit and is used for calculating the standard phase signal according to the time sequence of the synthesized signal.

Further, the abnormality recognition unit includes:

a signal alignment subunit configured to align a time axis of the standard phase signal and the electroencephalogram signal;

The difference calculation subunit is connected with the signal alignment subunit and used for comparing waveforms of the aligned standard phase signals and electroencephalogram signals and calculating differences of the standard phase signals and the electroencephalogram signals in a time window so as to obtain a calculation result;

The threshold value determining subunit is connected with the difference calculating subunit and used for setting a difference threshold value according to the calculation result;

And the abnormality detection subunit is connected with the threshold value determination subunit and is used for comparing the calculated result with the difference threshold value to identify the abnormal stage signal.

Further, the processing module includes:

the characteristic extraction unit is connected with the data collection unit and used for extracting the historical data of the age of the patient, the gender of the patient, the disease state before the operation of the patient, the operation time and the bleeding amount to obtain a first characteristic result;

The model training unit is connected with the feature extraction unit and used for training a risk assessment model according to the first feature result and the historical data to obtain a trained risk assessment model;

The model verification unit is connected with the model training unit and used for verifying the trained model by using the test set to obtain a verification result;

And the risk assessment unit is respectively connected with the model training unit and the model verification unit and is used for assessing the risk degree of the minimally invasive surgery by combining the trained model with the verification result and the abnormal stage signal with the historical data so as to obtain an assessment result.

Further, the risk assessment unit includes:

a data integration subunit configured to integrate the training model, the verification model, the abnormal phase signal, and the historical data into a data set;

a feature selection subunit, coupled to the data integration subunit, for selecting features of the dataset using a recursive feature elimination method to obtain a second feature result;

the model fusion subunit is connected with the feature selection subunit and used for taking the second feature result as an input parameter, and fusing the training model and the verification model by using a weighted average method to obtain a target model;

And the risk calculation subunit is connected with the model fusion subunit and used for calculating the risk degree of the minimally invasive surgery according to the target model.

Further, the adjustment module includes:

The risk threshold management unit is used for managing a preset safety risk degree;

The real-time monitoring and comparing unit is connected with the risk threshold management unit and used for real-time monitoring and comparing the risk degree calculated by the risk assessment unit with the safety risk degree to obtain a comparison result;

and the risk warning output unit is connected with the real-time monitoring comparison unit and is used for outputting safety warning information according to the comparison result.

Further, the risk warning unit includes:

An acoustic warning subunit, configured to issue an acoustic warning when the risk level exceeds a preset security risk level;

and the visual warning subunit is used for sending out a visual warning when the risk level exceeds the preset safety risk level.

Compared with the prior art, the invention has the beneficial effects that the acquisition module can timely discover the operation condition which possibly affects the nerve function by monitoring the brain electrical activity in real time, thereby taking measures to avoid nerve damage. The analysis module facilitates predicting a transmission time of a neural signal by calculation of a neural path length and a waveform time interval, thereby reducing surgical complications due to neural conduction delays. The construction of the three-dimensional nerve model provides detailed nerve structure information for doctors, is helpful for optimizing operation paths and schemes and reduces damage to surrounding healthy tissues. By means of the model and calculation analysis, the abnormality in the electroencephalogram signal can be found in time. The processing module can evaluate the risk of the operation more accurately by analyzing the abnormal stage signals and the historical data, and help doctors to make better decisions. When the monitored risk degree exceeds the preset safety risk degree, the system can automatically output safety warning information to remind doctors to take measures, and nerve damage is avoided.

In particular, the nerve length calculation unit can obtain more accurate nerve path length by calculating the distance between brains, the path length of fiber bundles and the number of synapses, which is helpful for better understanding the physical basis of nerve signal transmission and provides reliable quantitative data for neuroscience research. The waveform interval calculation unit considers the influence of the bone sheath on the nerve conduction speed, so that the calculation result is closer to the actual situation. The nerve impulse conduction method is helpful for in-depth research on the conduction process of nerve impulses on nerve fibers, and provides theoretical basis for research and treatment of nerve diseases.

In particular, the neural model construction unit may precisely construct a three-dimensional neural model to improve accuracy of simulation by inputting parameters of the parameter input unit. The analog signal unit obtains standard phase signals through the propagation process of the analog nerve impulses, and can more accurately identify abnormal signals which are inconsistent with the modes, so that the diagnosis accuracy is improved. The abnormality recognition unit is capable of detecting signal abnormality at an early stage, which contributes to early diagnosis and intervention of a disease.

In particular, by means of the geometric modeling subunit, the geometric structures of the neurons and the neural network can be accurately created according to the input parameters, so that the model can truly reflect the physical layout of the nervous system. The electrophysiological simulation subunit is capable of simulating the physiological properties of the neuron, such that the model is capable of more accurately reflecting the electrical activity of the neuron. The signal transmission modeling subunit can monitor and simulate the transmission process of the electroencephalogram signals in the neural network in real time, so that researchers are helped to know how the signals propagate and interact in the complex neural network.

In particular, the numerical simulation subunit can accurately simulate the dynamic response of neurons by using a numerical method. The electroencephalogram signals synthesized by the synthesis signal subunit reflect the electrophysiological characteristics of neuron activities and are close to the actually recorded electroencephalogram signals. The standard phase signal calculation unit can accurately identify delay and synchronicity change of nerve impulse propagation by calculating the standard phase signal, which is important for detecting abnormal activity in an electroencephalogram.

In particular, the signal alignment subunit can ensure that the standard phase signal and the electroencephalogram signal are compared at the same time point by aligning the time axes of the two signals, so that the difference between the signals can be accurately calculated. The difference calculation subunit can calculate the difference between the standard phase signal and the electroencephalogram signal in a specific time window by comparing the waveforms of the standard phase signal and the electroencephalogram signal after alignment. This helps to identify abnormal patterns in the electroencephalogram signal. The threshold value determination subunit uses the difference threshold value set by the calculation result as a judgment standard to identify which differences are significant, so as to determine which signals belong to an abnormal stage. The abnormality detection subunit can identify abnormal phase signals by comparing the calculation result with the difference threshold value, which is helpful for timely finding brain function abnormalities possibly occurring in the operation process.

In particular, the feature extraction unit can reduce the dimension of the data by extracting key clinical features from the historical data, improve the efficiency of data processing and provide more information support for risk assessment. The simulation training unit can effectively predict the operation risk by establishing an accurate risk assessment model. The model verification unit verifies the training model by using the test set to ensure the performance of the risk assessment model on unknown data, and the reliability and the practicability of the model are improved. The risk assessment unit provides accurate, real-time and reliable risk assessment by combining training models, verification models, abnormal phase signals and historical data.

In particular, the data integration subunit may ensure the integrity and consistency of data by integrating data from different sources into one unified data set. The feature selection subunit can effectively reduce feature dimensions, avoid overfitting and improve the interpretation of the model by using a recursive feature elimination method. The model fusion subunit fuses predictions of a plurality of models through a weighted average method, so that stability and robustness of risk assessment can be improved. The risk computing sub-unit computes the risk degree according to the fused target model, and can provide a comprehensive risk assessment result to help medical professionals make more intelligent decisions.

In particular, the risk threshold management unit is responsible for managing a preset safety risk degree, and the sensitivity and the accuracy of the early warning system can be ensured by setting a reasonable threshold. The real-time monitoring comparison unit can continuously monitor the risk degree calculated by the risk assessment unit, which enables an organization to dynamically adjust the risk threshold to adapt to a continuously changing safety environment. The risk warning output unit can remarkably improve the safety of the minimally invasive surgery by outputting the safety warning.

In particular, by providing both an audio and visual warning, the efficiency and reliability of the transmission of warning information can be improved. Different people may be more inclined to pay attention to one of the warning modes, and the multi-mode warning can ensure that all people can receive warning information in time.

Drawings

Fig. 1 is a schematic structural diagram of a minimally invasive surgery nerve monitoring system according to an embodiment of the present invention;

fig. 2 is a schematic structural diagram of an analysis module in a minimally invasive surgery nerve monitoring system according to an embodiment of the present invention;

fig. 3 is a schematic structural diagram of a processing module in the minimally invasive surgery nerve monitoring system according to an embodiment of the present invention;

Fig. 4 is a schematic structural diagram of an early warning module in a minimally invasive surgery nerve monitoring system according to an embodiment of the present invention.

Detailed Description

The invention will be further described with reference to examples for the purpose of making the objects and advantages of the invention more apparent, it being understood that the specific examples described herein are given by way of illustration only and are not intended to be limiting.

Preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are merely for explaining the technical principles of the present invention, and are not intended to limit the scope of the present invention.

It should be noted that, in the description of the present invention, terms such as "upper," "lower," "left," "right," "inner," "outer," and the like indicate directions or positional relationships based on the directions or positional relationships shown in the drawings, which are merely for convenience of description, and do not indicate or imply that the apparatus or elements must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention.

In addition, it should be noted that, in the description of the present invention, unless explicitly specified and limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, integrally connected, mechanically connected, electrically connected, directly connected, indirectly connected through an intermediate medium, or in communication between two elements. The specific meaning of the above terms in the present invention can be understood by those skilled in the art according to the specific circumstances.

Referring to fig. 1, a minimally invasive surgery nerve monitoring system provided in an embodiment of the present invention includes:

the acquisition module 10 is used for acquiring electroencephalogram signals of minimally invasive surgery in real time and also used for acquiring nerve fiber length, myelin sheath thickness, nerve conduction speed, fiber bundle paths, synapse number and brain interval distance;

an analysis module 20 connected to the acquisition module 10, for calculating a nerve path length according to the brain interval distance, the synapse number and the fiber bundle path, calculating a waveform time interval according to the nerve path length, the bone sheath thickness and the nerve conduction velocity, constructing a three-dimensional nerve model according to the synapse number, the nerve fiber length, the nerve conduction velocity and the bone sheath thickness, and identifying abnormal phase signals of the electroencephalogram signals according to the three-dimensional nerve model and the waveform time interval;

The processing module 30 is connected with the analysis module 20 and used for acquiring historical data of the minimally invasive surgery and evaluating the risk degree of the minimally invasive surgery according to the abnormal phase signals and the historical data;

The early warning module 40 is connected to the processing module 30, and is configured to obtain a comparison result according to the comparison between the risk level and a preset safety risk level, and output a safety warning message when the comparison result exceeds the safety risk level.

In particular, the acquisition module is equipped with a plurality of electroencephalogram electrodes for capturing brain electrical activity of scalp or in-vivo electrodes in real time. The length, myelin thickness, and inter-brain distance of nerve fibers are determined by Magnetic Resonance Imaging (MRI), and measured by electrophysiological tests, such as nerve conduction velocity tests. This typically involves placing electrodes on the skin surface and measuring the time between stimulus and response to calculate nerve conduction velocity. The number of synapses was measured by fMRI (functional magnetic resonance imaging) and the fiber bundle path was determined by Diffusion Tensor Imaging (DTI). The analysis module uses mathematical models and computational methods to calculate nerve path lengths, and calculates the time required for signals to travel between different brain regions based on nerve conduction velocity and nerve path length. The nerve fiber length, myelin sheath thickness, nerve conduction velocity, and fiber bundle path data are integrated to create a three-dimensional model reflecting the actual nerve structure. And combining the electroencephalogram signals acquired in real time with the three-dimensional nerve model, analyzing delay and change in signal transmission, and identifying possible abnormal signals. The processing module gathers and collates past minimally invasive surgery data, including electroencephalogram signals, surgery results, possible complications and the like in the surgery process. And combining the abnormal stage signals and the historical data, and evaluating the risk degree of the current operation by using a statistical method and a machine learning algorithm. The adjustment module compares the risk assessment result with a preset safety risk degree to determine whether potential risks exist. And when the risk exceeds a preset threshold, automatically outputting a safety warning message to remind a doctor to take corresponding measures. An interface is provided that allows a physician to adjust surgical protocols or take other precautions based on information provided by the system.

Specifically, the acquisition module can timely discover the operation condition which possibly affects the nerve function by monitoring the brain electrical activity in real time, so that measures are taken to avoid nerve damage. The analysis module facilitates predicting a transmission time of a neural signal by calculation of a neural path length and a waveform time interval, thereby reducing surgical complications due to neural conduction delays. The construction of the three-dimensional nerve model provides detailed nerve structure information for doctors, is helpful for optimizing operation paths and schemes and reduces damage to surrounding healthy tissues. By means of the model and calculation analysis, the abnormality in the electroencephalogram signal can be found in time. The processing module can evaluate the risk of the operation more accurately by analyzing the abnormal stage signals and the historical data, and help doctors to make better decisions. When the monitored risk degree exceeds the preset safety risk degree, the system can automatically output safety warning information to remind doctors to take measures, and nerve damage is avoided.

Specifically, as shown in fig. 2, the analysis module 20 includes:

A nerve length calculation unit 21 for obtaining the nerve path length from the sum of the distances between the brains, the length synthesis of the fiber bundle paths, and the sum of the number of synapses;

A waveform interval calculation unit 22 connected to the nerve length calculation unit 21 for adding the quotient of the bone sheath thickness and the nerve conduction velocity to the quotient of the nerve path length and the nerve conduction velocity to obtain the waveform time interval.

Specifically, the calculation formula of the nerve length is:

Where L is the total nerve path length, d _i is the ith brain interval distance, p _j is the jth fiber bundle path length, N is the total number of brain intervals, m is the total number of fiber bundles, k is a coefficient related to the number of synapses N, which is used to adjust the total length to reflect the effect of the synapses on signal propagation.

In neuroscience, synapses are points of attachment between neurons that transmit signals, and their efficiency and function can be affected by a variety of factors, such as the distance of the synaptic cleft, the responsiveness of the postsynaptic membrane, the release and reuptake of neurotransmitters, and the like. The purpose of introducing the k-factor is to take into account the effect of these additional factors of synapses on signal propagation velocity when calculating nerve path length. Different synapses may have different transfer efficiencies, and some may be slower or less stable than others. The k-factor can be used to adjust the overall length in order to more accurately estimate the delay of the signal propagation throughout the neural network. In practical applications, the value of k may need to be determined from experimental data. For example, if some synapses are known to be slower than others, the transit times of different types of synapses may be experimentally measured and these data used to set the value of k. Overall, the k-factor is an adjustment factor for better reflecting the calculated nerve path length to the delay of actual signal propagation, thereby providing a more accurate model and understanding in neuroscience and neurophysiology studies.

The calculation formula of the waveform time interval is as follows:

Where Δt is the waveform time interval, t is the thickness of the myelin sheath, and v is the conduction velocity of nerve impulses on the nerve fibers.

Specifically, the nerve length calculation unit can obtain more accurate nerve path length by calculating the distance between brains, the fiber bundle path length and the synapse number, which is helpful for better understanding the physical basis of nerve signal transmission and provides reliable quantitative data for neuroscience research. The waveform interval calculation unit considers the influence of the bone sheath on the nerve conduction speed, so that the calculation result is closer to the actual situation. The nerve impulse conduction method is helpful for in-depth research on the conduction process of nerve impulses on nerve fibers, and provides theoretical basis for research and treatment of nerve diseases.

Specifically, the analysis module further includes:

Specifically, the parameter input unit receives parameters such as the number of synapses, the length of nerve fibers, the nerve conduction velocity, and the thickness of myelin sheath through the parameter input unit. These parameters are considered as input data for constructing a three-dimensional neural model. The neural model building unit receives the input parameters from the parameter input unit. These input parameters are used to construct a three-dimensional neural model that is capable of modeling the structure and function of the neural fibers. And simulating the propagation process of nerve impulses in the neural network according to the constructed three-dimensional neural model and the calculated waveform time interval. Simulations include potential changes, neurotransmitter release, post-synaptic potential generation, and the like. Dynamic processes of nerve impulse propagation are calculated using numerical simulation methods, such as Finite Element Analysis (FEA) or Finite Difference Method (FDM). The signals generated by the nerve impulse propagation process obtained by simulation are compared with actual electroencephalogram signals. The abnormal phase signal is identified using a pattern recognition algorithm, such as a Support Vector Machine (SVM). Whether the signal is abnormal is determined by setting a threshold or using a classification performance index (e.g., accuracy, recall, F1 score) in machine learning.

Specifically, the neural model construction unit may precisely construct a three-dimensional neural model to improve accuracy of simulation by inputting parameters of the parameter input unit. The analog signal unit obtains standard phase signals through the propagation process of the analog nerve impulses, and can more accurately identify abnormal signals which are inconsistent with the modes, so that the diagnosis accuracy is improved. The abnormality recognition unit is capable of detecting signal abnormality at an early stage, which contributes to early diagnosis and intervention of a disease.

Specifically, the neural model construction unit includes:

Specifically, input parameters (e.g., number of synapses, length of nerve fibers, etc.) are used to determine the size and shape of the neuron. And constructing the geometric structure of the neural network according to the input parameters, including the layout and connection modes of the neurons. A geometric model is created that can visualize the structure of the neural network. The electrophysiological simulation subunit simulates physiological characteristics of neurons, such as membrane potential, generation and propagation of action potential, and the like, by using electrophysiological theory and formula. And comparing the simulation result with experimental data, and adjusting model parameters to improve the accuracy of the model. Outputting the result of the electrophysiological model, such as the change of the membrane potential with time, and the like. The signal transmission building module is used for building a mathematical model for transmitting the electroencephalogram signals in the neural network according to the geometric model and the electrophysiological model. And a calculation algorithm of a signal transmission model, such as a differential equation, a differential equation and the like, is realized. And analyzing the stability, convergence and accuracy of the signal transmission model.

In the present embodiment, it is assumed that the parameter inputs are as follows:

The number of synapses 1000 (number of connections between neurons);

nerve fiber length 1mm (length of each neuron axon);

nerve conduction velocity 120m/s (propagation velocity of signal in axon);

the thickness of the bone sheath is 0.1mm (the thickness of the bone sheath outside the nerve fiber).

Depending on the number of synapses and the length of the nerve fibers, a network can be constructed comprising 1000 nodes, each representing a neuron. The neurite length of neurons is 1mm, which will be one dimension in the model. The membrane potential model of the neuron can be used for simulating the membrane potential change of the neuron in a resting state and an excited state. Assuming that the threshold potential of the neuron is-55 mV, when the membrane potential reaches this value, the neuron will generate an action potential. The width of the action potential was 2ms, and the peak potential was +30mV. Based on the geometric model and the electrophysiological model, a differential equation can be established to model the propagation of signals in the neural network.

For example, for a simple integration-and-fi re model, the equation for the signaling model may be as follows:

V is membrane potential, V _rest is resting potential, I _syn is input current from synapses, τ is the time constant of the membrane. Using this equation, the propagation of signals in the neural network can be simulated and the change in membrane potential over time observed.

Specifically, through the geometric modeling subunit, the geometric structures of the neurons and the neural network can be accurately created according to the input parameters, so that the model can truly reflect the physical layout of the nervous system. The electrophysiological simulation subunit is capable of simulating the physiological properties of the neuron, such that the model is capable of more accurately reflecting the electrical activity of the neuron. The signal transmission modeling subunit can monitor and simulate the transmission process of the electroencephalogram signals in the neural network in real time, so that researchers are helped to know how the signals propagate and interact in the complex neural network.

Specifically, the signal simulation unit includes:

Specifically, a suitable numerical method (finite difference method, finite element analysis, etc.) is selected to simulate the propagation of neural activity in a neural network. The initial conditions and boundary conditions of the simulation are set according to parameters of the neural model (such as the number of synapses, the length of nerve fibers, the nerve conduction velocity, the thickness of bone sheath, etc.) and the waveform time interval. The propagation of nerve impulses in the neural network was calculated using a numerical simulation method, and the potential changes and propagation time were recorded. The signal synthesis unit synthesizes corresponding electroencephalogram signals according to the propagation result output by the numerical simulation subunit. The composite signal should reflect the propagation of nerve impulses in different areas and at different points in time. The potential change of the neurons is simulated using an integral function or a differential function, and a continuous electroencephalogram signal is synthesized. The standard phase signal calculation subunit calculates a standard phase signal according to the time sequence of the synthesized signal. The standard phase signal is a reflection of the propagation delay of nerve impulses on the different synapses and nerve fibers. The calculation subunit uses a time analysis algorithm to identify a particular phase of the signal and calculates the value of the standard phase signal.

In this embodiment, it is assumed that there is a simple neural network consisting of two neurons, wherein a first neuron is connected to a second neuron by a synapse. A one-dimensional neuron model will be used in which the membrane potential of neurons varies over time, described by the following equation:

Where V (t) is the membrane potential of the neuron at time t, V _m is the resting potential of the neuron, τ is the time constant of the neuron, and I _input (t) is the input current, which in this example mimics the synaptic input.

The model parameters V _m = -50mV, τ = 2ms, and the synaptic input I _input (t) (t) is a unit step function, indicating that the synapse releases neurotransmitters at t = 0.

Numerical integration was performed using the euler method, simulating the change in membrane potential of neurons over time. For example, a time step Δt=0.1 ms may be selected and simulated for a period of time t from 0 to 10ms.

After the simulation is completed, the membrane potential data of the neurons are synthesized into one signal. In this example, the analog data may be used directly to synthesize the signal.

The standard phase signal is calculated from the time series of the composite signal. In this example, the standard phase signal may refer to the time delay of the propagation of the neuronal excitation.

Specifically, the abnormality recognition unit includes:

Specifically, the signal alignment subunit collects time-series data of the standard phase signal and the actual phase signal. Signal processing techniques (such as cross-correlation analysis) are used to find the optimal time alignment of the two signal sequences. And according to the found alignment point, adjusting the time axis of the actual stage signal to align with the standard stage signal. The difference computation subunit compares waveforms of the aligned standard phase signal and actual phase signal. The difference between the two signals over a time window is calculated, including amplitude differences, latency differences, etc. These differences are quantified, for example, by calculating the absolute or relative value of the differences. The threshold determination subunit determines a base threshold according to statistical properties (such as mean value, standard deviation) of the signal. Statistical tests (e.g., t-test, chi-square test) are used to determine the significance threshold of the differences. The expert knowledge and the historical data are combined to set an experience threshold. The abnormality detection subunit compares the output of the difference calculation subunit with the threshold set by the threshold determination subunit. If the calculated difference exceeds a threshold, the signal is marked as abnormal.

Specifically, the signal alignment subunit can ensure that the standard phase signal and the electroencephalogram signal are compared at the same time point by aligning the time axes of the two signals, thereby accurately calculating the difference between the signals. The difference calculation subunit can calculate the difference between the standard phase signal and the electroencephalogram signal in a specific time window by comparing the waveforms of the standard phase signal and the electroencephalogram signal after alignment. This helps to identify abnormal patterns in the electroencephalogram signal. The threshold value determination subunit uses the difference threshold value set by the calculation result as a judgment standard to identify which differences are significant, so as to determine which signals belong to an abnormal stage. The abnormality detection subunit can identify abnormal phase signals by comparing the calculation result with the difference threshold value, which is helpful for timely finding brain function abnormalities possibly occurring in the operation process.

Specifically, as shown in fig. 3, the processing module 30 includes:

A feature extraction unit 31 connected to the data collection unit for extracting the historical data for patient age, patient gender, patient pre-operative disease state, operative time and bleeding amount to obtain a first feature result;

A model training unit 32, connected to the feature extraction unit 31, for training a risk assessment model according to the first feature result and the history data to obtain a training model;

A model verification unit 33 connected to the model training unit 32 for verifying the training model using a test set to obtain a verification model;

And a risk assessment unit 34 connected to the model training unit 32 and the model verification unit 33, respectively, for using a training model and a verification model to assess the risk level of the minimally invasive surgery in combination with the abnormal phase signal and the history data to obtain an assessment result.

Specifically, age information of the patient is extracted from the history data, and the ages are converted into continuous variables or classified variables. Extracting the sex information of the patient from the historical data is converted into binary classification variables (such as 0 for female and 1 for male). The pre-operative disease state of the patient is extracted from the historical data, possibly using a code (e.g., 0 for no disease, 1 for disease). Extracting the duration of the procedure from the historical data may convert the time into minutes or hours. Extraction of the amount of bleeding during the procedure from the historical data may convert the amount of bleeding into milliliters or units. The model training unit selects an appropriate machine learning model, such as a logistic regression, decision tree, random forest, or deep learning model, based on the problem characteristics and data type. The selected model is trained using the first feature results and the historical data provided by the feature extraction unit. This process may include adjusting model parameters such as learning rate, batch size, number of layers, etc. Cross-validation techniques are applied to evaluate the generalization ability of the model, such as K-fold cross-validation. One or more test sets are separated from the historical data for evaluating the performance of the model. And verifying the training model by using the test set, and calculating evaluation indexes such as accuracy, recall rate, F1 score and the like of the model. And adjusting model parameters or selecting different model architectures according to the verification model to improve the model performance. The risk assessment unit predicts the risk level of the minimally invasive surgery using the trained risk assessment model and the verification result. The anomaly phase signals are combined with historical data and model predictions to identify and evaluate risk events. Based on the model output and the abnormal phase signal, the surgical risk is quantitatively assessed, as represented using a risk score or risk level.

Specifically, the feature extraction unit can reduce the dimension of the data by extracting key clinical features from the historical data, improve the efficiency of data processing, and provide more information support for risk assessment. The simulation training unit can effectively predict the operation risk by establishing an accurate risk assessment model. The model verification unit verifies the training model by using the test set to ensure the performance of the risk assessment model on unknown data, and the reliability and the practicability of the model are improved. The risk assessment unit provides accurate, real-time and reliable risk assessment by combining training models, verification models, abnormal phase signals and historical data.

Specifically, the risk assessment unit includes:

Specifically, the data integration subunit collects and collates training models, verification models, abnormal phase signals, and historical data. Ensuring that all data uses the same format and normalization method for subsequent processing. If the data is from a different source, data cleansing and preprocessing are required to eliminate the differences. The feature selection subunit uses a Recursive Feature Elimination (RFE) method to select features of the dataset. An initial model is trained on the training dataset and feature importance is calculated. Descending order according to the importance of the features, and gradually removing the features with lower importance. This process is repeated until a predetermined number of features is reached or the performance of the model is no longer improved. The predetermined number is set by those skilled in the art based on the actual implementation. And obtaining a second characteristic result. The model fusion subunit takes the second feature result output by the feature selection subunit as an input parameter. The training model and the verification model are fused by using a weighted average method. Each model is assigned a weight, which is typically determined based on the cross-validation performance of the model. And calculating a weighted average value of the predictions of each model to obtain a final prediction result. The risk calculation subunit uses the fused model to calculate the risk degree of the minimally invasive surgery. The risk level may be determined based on a predictive probability or score of the model.

Specifically, as shown in fig. 4, the early warning module 40 includes:

a risk threshold management unit 41 for managing a preset security risk level;

the real-time monitoring and comparing unit 42 is connected with the risk threshold management unit 41 and is used for real-time comparing the risk degree calculated by the real-time monitoring and risk assessment unit with the safety risk degree to obtain a comparison result;

and the risk warning output unit 43 is connected with the real-time monitoring and comparing unit 42 and is used for outputting safety warning information according to the comparison result.

Specifically, the risk threshold management unit sets a risk threshold according to the type of surgery, the condition of the patient, and clinical experience. The set risk threshold is stored in the system to monitor the access to the comparison unit in real time. During surgery, the risk threshold is adjusted as needed. Multiple levels of risk thresholds are set to accommodate different surgical phases and risk situations. And the real-time monitoring and comparing unit is used for receiving the risk degree data calculated by the risk assessment unit in real time. And comparing and analyzing the real-time risk degree with a preset safety risk threshold value. A comparison is generated including whether the degree of risk exceeds a safety threshold. Dynamically monitoring the change of the risk degree and ensuring timely response. The risk warning output unit triggers a warning condition when the comparison result shows that the risk degree exceeds the safety threshold. And generating specific warning information according to the risk degree and the degree exceeding the threshold value.

Specifically, the risk threshold management unit is responsible for managing a preset safety risk degree, and the sensitivity and accuracy of the early warning system can be ensured by setting a reasonable threshold. The real-time monitoring comparison unit can continuously monitor the risk degree calculated by the risk assessment unit, which enables an organization to dynamically adjust the risk threshold to adapt to a continuously changing safety environment. The risk warning output unit can remarkably improve the safety of the minimally invasive surgery by outputting the safety warning.

Specifically, the risk warning unit includes:

Specifically, the audible warning subunit is activated when the risk level calculated by the risk assessment unit exceeds a preset safety risk level. The audible alert subunit will emit an audible signal of a certain frequency and intensity to ensure that it can be heard by operating room personnel in a noisy surgical environment. It may be configured to stop or adjust other non-emergency sounds while sounding an audible alert to ensure clarity and recognition of the alert sounds. A delay function may also be included to automatically stop the warning sounds after the risk level is reduced, reducing unnecessary interference. The visual warning subunit is activated when the risk level calculated by the risk assessment unit exceeds a preset security risk level. The visual alert subunit may display a clear alert signal, such as a red flashing warning light or a text prompt, on the display screen of the operating room. The visual alert subunit may also be provided with LED lights or other light emitting devices in critical locations of the operating room, such as near the operating table or in the line of sight of the doctor, to provide visual alerts.

Specifically, by providing both audio and visual warnings, the efficiency and reliability of the transmission of warning information can be improved. Different people may be more inclined to pay attention to one of the warning modes, and the multi-mode warning can ensure that all people can receive warning information in time.

Thus far, the technical solution of the present invention has been described in connection with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the scope of protection of the present invention is not limited to these specific embodiments. Equivalent modifications and substitutions for related technical features may be made by those skilled in the art without departing from the principles of the present invention, and such modifications and substitutions will be within the scope of the present invention.

The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, and various modifications and variations of the present invention will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A minimally invasive surgical nerve monitoring system, comprising:

The analysis module is connected with the acquisition module and used for calculating nerve path length according to the brain interval distance, the synapse number and the fiber bundle path, calculating waveform time intervals according to the nerve path length, the myelin sheath thickness and the nerve conduction speed, constructing a three-dimensional nerve model according to the synapse number, the nerve fiber length, the nerve conduction speed and the myelin sheath thickness, and identifying abnormal phase signals of the electroencephalogram signals according to the three-dimensional nerve model and the waveform time intervals;

The early warning module is connected with the processing module and used for obtaining a comparison result according to the comparison of the risk degree and the preset safety risk degree, and outputting safety warning information when the comparison result exceeds the safety risk degree;

the analysis module comprises:

a waveform interval calculation unit, coupled to the nerve length calculation unit, for adding the quotient of the myelin sheath thickness and the nerve conduction velocity to the quotient of the nerve path length and the nerve conduction velocity to obtain the waveform time interval;

The analysis module further includes:

a parameter input unit for taking the number of synapses, the nerve fiber length, the nerve conduction velocity and the myelin sheath thickness as input parameters of a model;

The abnormal identification unit is connected with the signal simulation unit and is used for comparing the standard phase signal with the electroencephalogram signal to identify the abnormal phase signal;

The neural model construction unit includes:

The signal transmission modeling sub-unit is respectively connected with the geometric modeling sub-unit and the electrophysiological simulation sub-unit and is used for building a model for transmitting the electroencephalogram signals in the neural network according to the geometric model and the electrophysiological model so as to obtain a signal transmission model;

The signal simulation unit includes:

2. The minimally invasive surgical nerve monitoring system according to claim 1, wherein the abnormality identification unit comprises:

3. The minimally invasive surgical nerve monitoring system of claim 2, wherein the processing module comprises:

4. The minimally invasive surgical nerve monitoring system of claim 3, wherein the risk assessment unit comprises:

a data integration subunit for integrating the training model, the verification model, the abnormal phase signal and the historical data into a data set;

5. The minimally invasive surgical nerve monitoring system of claim 4, wherein the pre-warning module comprises:

6. The minimally invasive surgical nerve monitoring system according to claim 5, wherein the risk alert output unit comprises: