High-frequency ablation hemostatic forceps closed-loop self-adaptive control system integrated with tissue monitoring
Through the closed-loop adaptive control system of high-frequency ablation hemostasis tweezers with integrated tissue monitoring, the machine learning model is used to adjust the ablation parameters in real time, solving the problem that traditional equipment is difficult to dynamically adjust the ablation parameters, and improving the accuracy and safety of ablation hemostasis.
Patent Information
- Application Number
- CN202411106191.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-13
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional high-frequency ablation equipment is difficult to dynamically adjust the ablation parameters according to individual differences between different patients, and lacks real-time tissue status monitoring and automated feedback control, resulting in insufficient surgical effectiveness and safety.
The high-frequency ablation hemostasis tweezer closed-loop adaptive control system is adopted with integrated tissue monitoring, including data acquisition, model training, data analysis, risk prediction and risk remediation modules, and the ablation parameters are predicted through machine learning models and bleeding risk coefficients, and the ablation parameters are adjusted in real time. Generate optimization and remediation instructions.
It realizes the personalized setting of ablation parameters according to the individual characteristics of the patient, improves the accuracy and safety of ablation hemostasis, reduces the risk of excessive or insufficient ablation, and improves surgical efficiency and control accuracy.
Smart Images

Figure CN119007984A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of high-frequency ablation hemostasis, and in particular relates to a high-frequency ablation hemostatic forceps closed-loop adaptive control system with integrated tissue monitoring. Background Art
[0002] High-frequency ablation (HFA) technology is widely used in medical procedures to cut tissue, coagulate blood and ablate tumors. High-frequency current generates heat through tissue, which can be used to cut tissue or coagulate blood vessels to stop bleeding. This technology is particularly important in minimally invasive surgery because it can reduce bleeding and improve surgical precision.
[0003] For example, the Chinese patent with publication number CN116712163A discloses a robot system for performing radiofrequency ablation surgery, including: a radiofrequency generator for generating high-frequency current; a radiofrequency electrode for transmitting high-frequency current to the target tissue; a robotic arm for driving the radiofrequency electrode to move in space; a navigation system for obtaining the patient's heart structure and abnormal point information, and automatically planning the best ablation path and parameters based on the information; an image processing system for real-time acquisition and processing of the position information of the radiofrequency electrode in the body, and matching and correcting it with the navigation system; a monitoring system for real-time monitoring of current, temperature, impedance and other indicators during the ablation process, and timely adjusting the ablation strategy based on the indicators; and a control system for controlling the coordination and communication between the radiofrequency generator, the robotic arm, the navigation system, the image processing system and the monitoring system. The robot system of this invention can improve the accuracy, safety and efficiency of radiofrequency ablation surgery.
[0004] The above patents also have the following problems:
[0005] Traditional high-frequency ablation equipment cannot determine ablation parameters and dynamically adjust surgical plans based on the individual differences of different patients. In addition, the tissue characteristics and lesion locations of different patients are different. Existing equipment is difficult to adaptively adjust ablation parameters, which may lead to unsatisfactory surgical results. It lacks real-time tissue status monitoring function and cannot effectively judge tissue changes during the ablation process, which may lead to excessive or insufficient ablation, affecting surgical results and safety. It lacks an automated feedback control mechanism and has a low level of intelligence, which needs to be improved. Summary of the invention
[0006] The object of the present invention is to provide a closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring, so as to solve the technical problems raised in the existing background technology.
[0007] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring, comprising:
[0008] The data acquisition module is used to collect the historical training data set of the ablation hemostasis area, the historical training data set includes the ablation comprehensive impact data and the ablation parameter set, and converts the ablation parameter set into an ablation number, the ablation number is i=1, 2, 3...n, and n represents the total number of ablation numbers.
[0009] A model training module, based on a historical training data set, trains a machine learning model for predicting ablation numbers, collects real-time comprehensive ablation impact data, predicts ablation numbers based on the trained machine learning model, and then obtains ablation parameter sets corresponding to the ablation numbers;
[0010] The data analysis module monitors the ablation effect of the ablation hemostasis area in real time, obtains the real-time ablation standard coefficient, and determines whether to generate an optimization instruction based on the real-time ablation standard coefficient;
[0011] A data acquisition module, which acquires tissue characteristic data during the ablation hemostasis process and calculates a real-time bleeding risk coefficient based on the tissue characteristic data;
[0012] The risk prediction module inputs the tissue characteristic data acquired in real time into a pre-built machine learning model for predicting the bleeding risk coefficient at a future moment, and outputs the bleeding risk coefficient at a future moment;
[0013] The risk remediation module generates corresponding remediation level instructions based on the predicted bleeding risk coefficient at a future moment.
[0014] Preferably, the training method of the machine learning model for predicting the ablation number is as follows:
[0015] Converting the collected comprehensive ablation impact data into a corresponding set of feature vectors;
[0016] Each group of feature vectors is used as the input of the machine learning model, the machine learning model takes the ablation number corresponding to each group of ablation comprehensive impact data as the output, the ablation number actually corresponding to each group of ablation comprehensive impact data is used as the prediction target, and minimizing the loss function value of the machine learning model is used as the training goal; training is stopped when the loss function value of the machine learning model is less than or equal to the preset target loss value.
[0017] Preferably, the related parameters of the ablation standard coefficient include: electrical impedance, reflectivity, projection rate and absorptivity of the tissue; weighted operation is performed on the electrical impedance, reflectivity, projection rate and absorptivity of the tissue to obtain the ablation standard coefficient.
[0018] Preferably, the method of determining whether to generate an optimization instruction based on the real-time ablation standard coefficient is as follows:
[0019] The real-time ablation standard coefficient is compared with the preset ablation standard coefficient threshold to obtain the ablation standard coefficient difference, as follows:
[0020] Xc=Sj-YSj;
[0021] In the formula, Xc is the difference of the ablation standard coefficient, Sj is the real-time ablation standard coefficient, and YSj is the ablation standard coefficient threshold;
[0022] When Xc>0, it means that the real-time ablation standard coefficient is greater than or equal to the preset ablation standard coefficient threshold, and the ablation effect meets the standard, and no optimization instruction is generated at this time;
[0023] When Xc≤0, it means that the real-time ablation standard coefficient is less than the preset ablation standard coefficient threshold, and the ablation effect does not meet the standard, and an optimization instruction is generated at this time.
[0024] Preferably, the optimization instruction includes:
[0025] The maximum thresholds corresponding to the ablation power, ablation frequency, and ablation time are recorded as the maximum ablation power, the maximum ablation frequency, and the maximum ablation time, respectively;
[0026] Increasing the ablation power has a higher priority than increasing the ablation frequency, and increasing the ablation frequency has a higher priority than increasing the ablation time;
[0027] When the ablation power, ablation frequency, and ablation time have not reached their respective maximum thresholds, the ablation power is first increased until the ablation effect reaches the target, and then the ablation power is stopped;
[0028] When the ablation power is increased to the maximum ablation power and the ablation effect does not meet the standard, the ablation frequency is increased until the ablation effect meets the standard, and then the ablation frequency is stopped;
[0029] When the ablation frequency is increased to the maximum ablation frequency and the ablation effect does not meet the standard, the ablation time is increased until the ablation effect meets the standard, and then the increase of the ablation time is stopped.
[0030] Preferably, the tissue characteristic data include tissue temperature difference, tissue electrical impedance difference, tissue reflectivity difference, tissue transmittance difference and blood flow; weighted operations are performed on the tissue temperature difference, tissue electrical impedance difference, tissue reflectivity difference, tissue transmittance difference and blood flow to obtain a bleeding risk coefficient.
[0031] Preferably, the training method of the machine learning model for predicting the bleeding risk coefficient at a future moment includes:
[0032] Preset the prediction time step S, sliding step H and sliding window length C; convert the historical tissue characteristic data into multiple training samples using the sliding window method, one training sample corresponds to a label and constitutes a group of training data, and multiple groups of training data constitute a training set. The training set is used as the input of the machine learning model, and the bleeding risk coefficient at future moments after the predicted time step S is used as the output. The subsequent bleeding risk coefficient of each training set is used as the prediction target, and the prediction accuracy is used as the training target to train the machine learning; generate a machine learning model that predicts the bleeding risk coefficient at future moments based on the tissue characteristic data.
[0033] Preferably, the remediation level instruction includes a primary remediation level, a secondary remediation level and a tertiary remediation level, and the severity of the primary remediation level, the secondary remediation level and the tertiary remediation level increases in sequence.
[0034] Preferably, the method of generating the corresponding remedial level instruction based on the predicted bleeding risk coefficient at a future moment is as follows:
[0035] The predicted bleeding risk coefficient is marked as YXf, and the preset bleeding risk coefficient thresholds are Xf1, Xf2, and Xf3, where Xf1<Xf2<Xf3;
[0036] When YXf≤Xf1, no remedial level instruction is generated;
[0037] When Xf2≥YXf>Xf1, at this time, a first-level remediation level instruction is generated;
[0038] When Xf3≥YXf>Xf2, at this time, a secondary remediation level instruction is generated;
[0039] When YXf>Xf3, at this time, a third-level relief level instruction is generated.
[0040] Preferably, the ablation parameter set is {(a1, b1, c1), (a2, b2, c2) ..., (a n , b n , c n )} When the ablation parameter set is numbered, a, b, and c correspond to ablation power, ablation time, and ablation frequency, respectively. n , b n , c n Indicates the values of ablation power, ablation time and ablation frequency corresponding to the ablation number n.
[0041] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:
[0042] 1. The present invention trains a machine learning model based on ablation comprehensive impact data to predict ablation parameter sets. It can personalize ablation parameters according to the specific characteristics of each patient, more accurately determine parameters such as ablation power, ablation frequency and time, reduce the risk of excessive or insufficient ablation, accurately control ablation parameters, reduce damage to surrounding healthy tissues, and achieve good ablation hemostasis effect.
[0043] 2. The present invention can timely discover areas with unsatisfactory ablation effects by real-time monitoring of key parameters of the ablation area during the ablation hemostasis process, and adjust parameters to ensure the consistency and accuracy of the ablation effect. It can also automatically adjust ablation parameters based on monitoring data to improve control accuracy, thereby improving the accuracy, safety and efficiency of ablation hemostasis.
[0044] 3. The present invention can discover possible safety hazards in the ablation area in advance by predicting the bleeding risk factor at future times, and determine different remedial measures through different remedial level instructions, so as to accurately intervene, improve the safety of the ablation process, and prevent excessive or insufficient ablation. Through the prediction of the bleeding risk factor and the corresponding level of remedial measures, the trial and error time and the time for handling emergencies can be effectively reduced, and the efficiency and level of ablation hemostasis can be improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0046] Figure 1 The module structure diagram of the embodiment 1 of the present invention is shown;
[0047] Figure 2 The module structure diagram of the second embodiment of the present invention is shown; DETAILED DESCRIPTION
[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0049] Example 1
[0050] Reference Figure 1As shown, a closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring includes: a data acquisition module, a model training module and a data analysis module, and each module is connected by wired / wireless means to realize data transmission;
[0051] A data acquisition module is used to acquire a historical training data set of an ablation hemostasis area, wherein the historical training data set includes ablation comprehensive impact data and an ablation parameter set, and convert the ablation parameter set into an ablation number, wherein the ablation number is i=1, 2, 3....n, and n represents the total number of ablation numbers;
[0052] Specifically, the comprehensive ablation impact data include the patient's age, B lM index, tissue characteristics and ambient temperature; wherein the tissue characteristics include tissue type, blood flow, tissue temperature, tissue electrical impedance, and tissue optical properties;
[0053] The BMI index is used to assess whether a person is underweight, normal, overweight or obese; the BMI calculation formula is weight (kg) divided by height (m) squared, the specific calculation formula is:
[0054]
[0055] Where M is weight and G is height. The BMI index can be calculated by obtaining the user's height and weight through a smart scale. During the data collection process, the BMI index error can be within the range of ±0.5.
[0056] Tissue types include muscle, fat, blood vessels, tumors, etc. Different tissue types have different absorption and scattering characteristics of high-frequency ablation energy, which affects the ablation power and time;
[0057] Blood flow affects the cooling rate of tissues. Blood flow will take away some heat, affecting the ablation effect. Areas with large blood flow require higher power and longer time. Blood flow is acquired in real time through the laser Doppler flowmeter integrated at the front end of the high-frequency ablation hemostatic forceps.
[0058] Tissue temperature directly affects the ablation effect. Too high a temperature will lead to carbonization, while too low a temperature will not lead to effective ablation. Real-time temperature feedback helps to dynamically adjust the ablation power and time. Tissue temperature is acquired in real time through the optical temperature sensor integrated at the front end of the high-frequency ablation hemostatic forceps.
[0059] The electrical impedance of the tissue reflects the electrical conductivity of the tissue and affects the conduction of high-frequency current and the ablation effect. Tissues with high resistance require higher power, while tissues with low resistance require lower power. The electrical impedance sensor integrated at the front end of the high-frequency ablation hemostatic forceps measures the changes in the electrical impedance of the tissue in real time.
[0060] The optical properties of tissues include: reflectivity, transmittance, absorptivity and scattering coefficient of tissues; the optical properties of the ablation area are acquired by real-time monitoring of the optical sensor integrated in the front end of the high-frequency ablation hemostatic forceps;
[0061] Reflectivity of tissue:
[0062] Reflectivity refers to the ratio of incident light reflected by the tissue surface. High reflectivity means that more light energy is reflected by the tissue surface, resulting in less energy actually entering the tissue and lowering the ablation efficiency. High reflectivity will increase the surface temperature, which may cause surface burns while deep tissues fail to be effectively ablated.
[0063]
[0064] Where Fs represents the reflectivity of the tissue, I r Represents the reflected light intensity, I o Represents the incident light intensity.
[0065] Transmittance of tissue:
[0066] Transmittance is the proportion of incident light that is detected after passing through tissue. High transmittance means that light can penetrate deeper into tissue and the energy distribution is more uniform, which is helpful for ablation of deep tissue. However, too high transmittance may cause energy to diffuse outside the target area, reducing the local ablation effect.
[0067]
[0068] Where Ts represents the transmittance of the tissue, I t Represents the intensity of transmitted light, I o Represents the incident light intensity.
[0069] Absorption rate of tissue:
[0070] Absorption rate is the ratio of incident light absorbed by tissue, which is calculated by known reflectivity and transmittance. High absorption rate means more light energy is absorbed by tissue and converted into heat energy, thus improving ablation efficiency. Too high absorption rate may lead to excessive local temperature, causing carbonization or excessive damage to tissue.
[0071] Absorption is the portion of incident light that is not reflected or transmitted and is calculated as:
[0072] Xs=1-Fs-Ts;
[0073] Where Xs represents the absorptivity of the tissue, Fs represents the reflectivity of the tissue, and Ts represents the transmittance of the tissue.
[0074] Scattering coefficient of tissue:
[0075] The scattering coefficient indicates the degree to which light is scattered in the tissue. A high scattering coefficient causes the light energy to be scattered multiple times in the tissue, resulting in energy dispersion and reduced local ablation efficiency.
[0076] The scattering coefficient is calculated as follows:
[0077] μ S =μ t -μ a ;
[0078] In the formula, μ s represents the scattering coefficient, μ t Represents the total attenuation coefficient, μ a represents the absorption coefficient, where the absorption coefficient can be obtained from reference data when the tissue is known.
[0079] The total attenuation coefficient is calculated as follows:
[0080]
[0081] Where I(s) represents the light intensity measured at a depth of s, I r represents the incident light intensity, and s represents the propagation depth of light in the tissue.
[0082] The ambient temperature affects the initial temperature and thermal conductivity of the tissue. Lower power and shorter time may be required in a high temperature environment. The ambient temperature is obtained by monitoring the ambient temperature of the surgical area in real time through a temperature sensor.
[0083] Ablation parameters include ablation power, ablation time, and ablation frequency;
[0084] The ablation parameter set is {(a1, b1, c1), (a2, b2, c2)…, (a n , b n , c n )};
[0085] For example, when the ablation parameter sets are numbered, i=1, 2, ... n, where a, b, c correspond to ablation power, ablation time, and ablation frequency, respectively. n , b n , c n Indicates the values of ablation power, ablation time and ablation frequency corresponding to the ablation number n.
[0086] It should be noted that the patient's age and tissue type can be input through a user-interactive screen page paired with the ablation hemostatic forceps.
[0087] A model training module, based on a historical training data set, trains a machine learning model for predicting ablation numbers, collects real-time comprehensive ablation impact data, predicts ablation numbers based on the trained machine learning model, and then obtains ablation parameter sets corresponding to the ablation numbers;
[0088] Specifically, the training method of the machine learning model for predicting the ablation number is as follows:
[0089] Converting the collected comprehensive ablation impact data into a corresponding set of feature vectors;
[0090] Each group of feature vectors is used as the input of the machine learning model, the machine learning model takes the ablation number corresponding to each group of ablation comprehensive impact data as the output, the ablation number actually corresponding to each group of ablation comprehensive impact data is used as the prediction target, and minimizing the loss function value of the machine learning model is used as the training goal; training is stopped when the loss function value of the machine learning model is less than or equal to the preset target loss value.
[0091] The machine learning model may be one of SVM regression, random forest regression or neural network regression models;
[0092] The machine learning model loss function value is the mean square error;
[0093] Mean square error is one of the commonly used loss functions. By transforming the loss function formula Minimization is used as the goal to train the model so that the machine learning model better fits the data, thereby improving the performance and accuracy of the model;
[0094] In the loss function, MSE is the loss function value of the machine learning model, x is the feature vector group number; m is the number of feature vector groups; y is the number of feature vector groups; x is the ablation number predicted by the x-th group of feature vectors, is the ablation number actually corresponding to the ×th group of feature vectors;
[0095] Other model parameters of the machine learning model, target loss value, optimization algorithm, training set test set validation set ratio, and loss function optimization are all achieved through actual engineering implementation and continuous experimental tuning.
[0096] In this embodiment, the corresponding ablation parameter set is obtained according to the predicted ablation number, and the ablation operation is performed according to the ablation power and ablation time in the ablation parameter set. The ablation power and ablation time can be determined in combination with the individual characteristics and tissue characteristics of the patient to more effectively achieve hemostasis.
[0097] The data analysis module monitors the ablation effect of the ablation area in real time, obtains the real-time ablation standard coefficient, and determines whether to generate an optimization instruction based on the real-time ablation standard coefficient;
[0098] Specifically, the parameters that affect the real-time ablation standard coefficient include: electrical impedance, reflectivity, projection rate and absorption rate of the tissue;
[0099]
[0100] In the formula, Sj represents the real-time ablation standard coefficient, Dz represents the electrical impedance of the tissue, Fs represents the reflectivity of the tissue, Xs represents the absorptivity of the tissue, and Ts represents the transmittance of the tissue; represents the weight coefficient, Both are greater than 0.
[0101] It should be noted that, if the electrical impedance Dz, reflectivity Fs and absorptivity Xs of the tissue increase, and the transmittance Ts of the tissue decreases, the ablation standard coefficient Sj will also increase relatively, indicating that the ablation effect is better; conversely, if the electrical impedance Dz, reflectivity Fs and absorptivity Xs of the tissue decrease, and the transmittance Ts of the tissue increases, the ablation standard coefficient Sj will also decrease relatively, indicating that the ablation effect is poor.
[0102] The method of determining whether to generate an optimization instruction based on the real-time ablation standard coefficient is as follows:
[0103] The real-time ablation standard coefficient is compared with the preset ablation standard coefficient threshold to obtain the ablation standard coefficient difference, as follows:
[0104] Xc=Sj-YSj;
[0105] In the formula, Xc is the difference of the ablation standard coefficient, Sj is the real-time ablation standard coefficient, and YSj is the ablation standard coefficient threshold;
[0106] When Xc>0, it means that the real-time ablation standard coefficient is greater than or equal to the preset ablation standard coefficient threshold, and the ablation effect meets the standard, and no optimization instruction is generated at this time;
[0107] When Xc≤0, it means that the real-time ablation standard coefficient is less than the preset ablation standard coefficient threshold, and the ablation effect does not meet the standard, and an optimization instruction is generated at this time.
[0108] Specifically, the optimization instructions include:
[0109] The maximum thresholds corresponding to the ablation power, ablation frequency, and ablation time are recorded as the maximum ablation power, the maximum ablation frequency, and the maximum ablation time, respectively;
[0110] Increasing the ablation power has a higher priority than increasing the ablation frequency, and increasing the ablation frequency has a higher priority than increasing the ablation time;
[0111] When the ablation power, ablation frequency, and ablation time have not reached their respective maximum thresholds, the ablation power is first increased until the ablation effect reaches the target, and then the ablation power is stopped;
[0112] When the ablation power is increased to the maximum ablation power and the ablation effect does not meet the standard, the ablation frequency is increased until the ablation effect meets the standard, and then the ablation frequency is stopped;
[0113] When the ablation frequency is increased to the maximum ablation frequency and the ablation effect does not meet the standard, the ablation time is increased until the ablation effect meets the standard, and then the increase of the ablation time is stopped.
[0114] Through the above-mentioned priority setting, the ablation effect can be guaranteed while minimizing the increase in ablation time.
[0115] In this embodiment, by training and predicting a machine learning model for ablation parameter sets based on the patient's age, BMI index, tissue type, blood flow, tissue temperature, tissue electrical impedance, optical properties of tissue, and ambient temperature, the ablation parameters can be set individually according to the specific characteristics of each patient, and the parameters such as ablation power, ablation frequency, and time can be determined more accurately, thereby reducing the risk of excessive or insufficient ablation, accurately controlling the ablation parameters, reducing damage to surrounding healthy tissues, and achieving good ablation hemostasis effects;
[0116] By real-time monitoring of key parameters of the ablation area during the ablation hemostasis process, areas with unsatisfactory ablation effects can be discovered in time, and parameter adjustments can be made to ensure the consistency and accuracy of the ablation effect. By real-time monitoring of parameters, signs of insufficient ablation can be discovered in time, and the ablation power, ablation frequency, or ablation time can be increased to ensure complete hemostasis. The ablation parameters can be automatically adjusted according to the monitoring data to improve control accuracy, thereby improving the accuracy, safety, and efficiency of ablation hemostasis.
[0117] Example 2
[0118] Reference Figure 2 As shown, the present invention further includes, based on Example 1:
[0119] A data acquisition module, which acquires tissue characteristic data during the ablation hemostasis process and calculates a real-time bleeding risk coefficient based on the tissue characteristic data;
[0120] Specifically, the tissue characteristic data includes tissue temperature difference, tissue electrical impedance difference, tissue reflectivity difference, tissue transmittance difference and blood flow;
[0121] The tissue temperature difference refers to the difference between the tissue temperature during the ablation process and the effective ablation temperature. The effective ablation temperature is usually between 60°C and 100°C. Too low a temperature may not effectively stop bleeding, increasing the risk of bleeding. Too high a temperature may cause tissue carbonization and damage to surrounding blood vessels, increasing the risk of bleeding. The greater the tissue temperature difference, the greater the risk of bleeding.
[0122] Tissue impedance difference refers to the difference between tissue impedance during ablation and before ablation. Impedance changes can reflect tissue status. If impedance changes are not obvious during ablation, effective hemostasis may not be achieved, increasing the risk of bleeding. The smaller the tissue impedance difference, the greater the risk of bleeding.
[0123] Tissue reflectivity difference refers to the difference between tissue reflectivity during ablation and tissue reflectivity before ablation. If the reflectivity of the tissue does not change significantly after ablation, there may be blood vessels that have not been effectively ablated, increasing the risk of bleeding. The smaller the tissue reflectivity difference, the greater the risk of bleeding.
[0124] Tissue transmittance difference refers to the difference between tissue transmittance during ablation and tissue transmittance before ablation. If the tissue transmittance does not change significantly after ablation, there may be blood vessels that have not been effectively ablated, increasing the risk of bleeding. The smaller the tissue transmittance difference, the greater the risk of bleeding.
[0125] Blood flow refers to the blood flow in the ablation area. During the ablation process, the blood flow in the ablation area should be significantly reduced. If the blood flow is not significantly reduced, there may be a risk of continued bleeding.
[0126] The real-time bleeding risk factor is generated as follows:
[0127]
[0128] Wherein, Xf represents the real-time bleeding risk coefficient, Wc represents the tissue temperature difference, Xl represents the blood flow, Dc represents the tissue electrical impedance difference, Fc represents the tissue reflectivity difference, and Tc represents the tissue transmittance difference; represents the weight coefficient, Both are greater than 0.
[0129] It should be noted that the larger the tissue temperature difference Wc and blood flow Xl, and the smaller the tissue electrical impedance difference Dc, tissue reflectivity difference Fc and tissue transmittance difference Tc, the larger the corresponding bleeding risk coefficient Xf, indicating that the bleeding risk in the ablation hemostasis area is higher; otherwise, it means that the bleeding risk in the ablation hemostasis area is lower.
[0130] It should be noted that the formulas involved in the above-mentioned are all calculated by removing dimensions and taking numerical values. They are a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The weight coefficients in the formulas and the various preset thresholds in the analysis process are set by technical personnel in this field according to actual conditions or obtained by simulating a large amount of data; the size of the weight coefficient is to quantify each parameter to obtain a specific value for subsequent comparison. The size of the weight coefficient depends on the amount of sample data and the corresponding processing coefficients initially set by technical personnel in this field for each group of sample data; as long as it does not affect the proportional relationship between the parameters and the quantized values.
[0131] The risk prediction module inputs the tissue characteristic data acquired in real time into a pre-built machine learning model for predicting the bleeding risk coefficient at a future moment, and outputs the bleeding risk coefficient at a future moment;
[0132] The training method of the machine learning model for predicting the bleeding risk coefficient at a future moment includes;
[0133] The tissue feature data is used as the input of the multi-feature time series prediction machine learning model, and the bleeding risk coefficient at future moments is used as the output of the multi-feature time series prediction machine learning model. The bleeding risk coefficient at future moments is used as the prediction target, and minimizing the sum of the prediction accuracies of all training data is used as the training target. The multi-feature time series prediction machine learning model is trained until the sum of the prediction accuracies converges and the training is stopped.
[0134] The loss function value of the machine learning model is the mean square error.
[0135] Mean square error is one of the commonly used loss functions. The model is trained with minimization as the goal, so that the machine learning model can better fit the data, thereby improving the performance and accuracy of the model.
[0136] In the loss function, MSE is the loss function value of the machine learning model, r is the feature vector group number; t is the number of feature vector groups; w r is the bleeding risk coefficient at the future moment corresponding to the rth group of eigenvectors, is the bleeding risk coefficient at the future moment actually corresponding to the rth group of eigenvectors.
[0137] Other model parameters of the machine learning model, target loss value, optimization algorithm, training set test set validation set ratio, and loss function optimization are all achieved through actual engineering implementation and continuous experimental tuning.
[0138] Methods for predicting bleeding risk at a future time include:
[0139] Preset the prediction time step S, sliding step H and sliding window length C; convert the historical tissue characteristic data into multiple training samples using the sliding window method, one training sample corresponds to a label and constitutes a group of training data, and multiple groups of training data constitute a training set. The training set is used as the input of the machine learning model, and the bleeding risk coefficient at future moments after the predicted time step S is used as the output. The subsequent bleeding risk coefficient of each training set is used as the prediction target, and the prediction accuracy is used as the training target to train the machine learning; generate a machine learning model that predicts the bleeding risk coefficient at future moments based on the tissue characteristic data.
[0140] Specifically, a simple example of the sliding window method is as follows: assuming that you want to use historical organizational feature data packets 1, 2, 3, 4, 5, 6, 7 to train a time prediction model to predict the value of one time step in the future, you can use a sliding window of length 4 and a sliding step of length 1 to generate a predicted future training set and prediction target. For example, the training sets are 1, 2, 3, 4, 2, 3, 4, 5 and 3, 4, 5, 6, and the prediction targets are 5, 6 and 7.
[0141] The sliding window method can be used to accurately and in advance predict the bleeding risk coefficient at a future moment based on the existing historical tissue characteristic data packets, so that when summarizing the ablation hemostasis process and monitoring the tissue characteristics of the ablation area, possible safety hazards in the ablation area can be discovered in advance, and accurate data support can be provided for subsequent processing without lags in ablation hemostasis monitoring.
[0142] The risk remediation module generates corresponding remediation level instructions based on the predicted bleeding risk coefficient at a future moment.
[0143] Specifically, the remediation level instructions include a first-level remediation level, a second-level remediation level, and a third-level remediation level, and the severity of the first-level remediation level, the second-level remediation level, and the third-level remediation level increases in sequence;
[0144] Based on the predicted bleeding risk coefficient at a future time, the corresponding remediation level instruction is generated as follows:
[0145] The predicted bleeding risk coefficient is marked as YXf, and the preset bleeding risk coefficient thresholds are Xf1, Xf2, and Xf3, Xf1<Xf2<Xf3;
[0146] When YXf≤Xf1, no remedial level instruction is generated;
[0147] When Xf2≥YXf>Xf1, at this time, a first-level remediation level instruction is generated;
[0148] When Xf3≥YXf>Xf2, at this time, a secondary remediation level instruction is generated;
[0149] When YXf>Xf3, at this time, a third-level remedial level instruction is generated;
[0150] For example, the first-level remedial level can be a 10% increase in ablation power, a 5% increase in ablation time, and a 10% increase in ablation frequency; the second-level remedial level can be a 20% increase in ablation power, a 10% increase in ablation time, and a 15% increase in ablation frequency; the third-level remedial level can be a 30% increase in ablation power, a 15% increase in ablation time, and a 20% increase in ablation frequency; the ablation parameters corresponding to the remedial level can also be customized according to the specific situation.
[0151] In this embodiment, by predicting the bleeding risk coefficient at a future moment, the potential safety hazards in the ablation area can be discovered in advance, and then corresponding remedial measures can be taken according to the prediction results, which can effectively prevent intraoperative bleeding, improve surgical safety, reduce the suddenness of intraoperative bleeding, and reduce the risks brought by emergency treatment;
[0152] By determining different remedial measures through different remedial level instructions, precise intervention can be made to improve the safety of the ablation process and prevent excessive or insufficient ablation. By predicting the bleeding risk factor and taking remedial measures of corresponding levels, the trial and error time and the time for dealing with emergencies can be effectively reduced, thereby improving the efficiency and level of ablation hemostasis.
[0153] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
[0154] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.
Claims
1. A closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring, characterized in that: include: A data acquisition module is used to acquire a historical training data set of an ablation hemostasis area, wherein the historical training data set includes ablation comprehensive impact data and an ablation parameter set, and convert the ablation parameter set into an ablation number, wherein the ablation number is i=1, 2, 3....n, and n represents the total number of ablation numbers; A model training module, based on a historical training data set, trains a machine learning model for predicting ablation numbers, collects real-time comprehensive ablation impact data, predicts ablation numbers based on the trained machine learning model, and then obtains ablation parameter sets corresponding to the ablation numbers; The data analysis module monitors the ablation effect of the ablation hemostasis area in real time, obtains the real-time ablation standard coefficient, and determines whether to generate an optimization instruction based on the real-time ablation standard coefficient; A data acquisition module, which acquires tissue characteristic data during the ablation hemostasis process and calculates a real-time bleeding risk coefficient based on the tissue characteristic data; The risk prediction module inputs the tissue characteristic data acquired in real time into a pre-built machine learning model for predicting the bleeding risk coefficient at a future moment, and outputs the bleeding risk coefficient at a future moment; The risk remediation module generates corresponding remediation level instructions based on the predicted bleeding risk coefficient at a future moment.
2. A closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring as claimed in claim 1, characterized in that: The training method of the machine learning model for predicting the ablation number is as follows: Converting the collected comprehensive ablation impact data into a corresponding set of feature vectors; Each group of feature vectors is used as the input of the machine learning model. The machine learning model uses the ablation number corresponding to each group of ablation comprehensive impact data as the output, uses the ablation number actually corresponding to each group of ablation comprehensive impact data as the prediction target, and uses minimizing the loss function value of the machine learning model as the training target; Stop training when the machine learning model loss function value is less than or equal to the preset target loss value.
3. A closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring as claimed in claim 1 or 2, characterized in that: The related parameters of the ablation standard coefficient include: electrical impedance, reflectivity, projection rate and absorption rate of the tissue; the electrical impedance, reflectivity, projection rate and absorption rate of the tissue are weighted to obtain the ablation standard coefficient.
4. A closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring as claimed in claim 3, characterized in that: The method for determining whether to generate an optimization instruction based on the real-time ablation standard coefficient is as follows: The real-time ablation standard coefficient is compared with the preset ablation standard coefficient threshold to obtain the ablation standard coefficient difference, as follows: Xc=Sj-YSj; In the formula, Xc is the difference of the ablation standard coefficient, Sj is the real-time ablation standard coefficient, and YSj is the ablation standard coefficient threshold; When Xc>0, it means that the real-time ablation standard coefficient is greater than or equal to the preset ablation standard coefficient threshold, and the ablation effect meets the standard, and no optimization instruction is generated at this time; When Xc≤0, it means that the real-time ablation standard coefficient is less than the preset ablation standard coefficient threshold, and the ablation effect does not meet the standard, and an optimization instruction is generated at this time.
5. A closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring as claimed in claim 4, characterized in that: The optimization instructions include: The maximum thresholds corresponding to the ablation power, ablation frequency, and ablation time are recorded as the maximum ablation power, the maximum ablation frequency, and the maximum ablation time, respectively; Increasing the ablation power has a higher priority than increasing the ablation frequency, and increasing the ablation frequency has a higher priority than increasing the ablation time; When the ablation power, ablation frequency, and ablation time have not reached their respective maximum thresholds, the ablation power is first increased until the ablation effect reaches the target, and then the ablation power is stopped; When the ablation power is increased to the maximum ablation power and the ablation effect does not meet the standard, the ablation frequency is increased until the ablation effect meets the standard, and then the ablation frequency is stopped; When the ablation frequency is increased to the maximum ablation frequency and the ablation effect does not meet the standard, the ablation time is increased until the ablation effect meets the standard, and then the increase of the ablation time is stopped.
6. A closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring as claimed in claim 1, characterized in that: The tissue characteristic data include tissue temperature difference, tissue electrical impedance difference, tissue reflectivity difference, tissue transmittance difference and blood flow; The tissue temperature difference, tissue electrical impedance difference, tissue reflectivity difference, tissue transmittance difference and blood flow are weighted to obtain the bleeding risk coefficient.
7. A closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring as claimed in claim 1, characterized in that: The training method of the machine learning model for predicting the bleeding risk coefficient at a future moment includes; Preset the prediction time step S, sliding step H and sliding window length C; convert the historical tissue feature data into multiple training samples using the sliding window method, one training sample corresponds to one label and constitutes a group of training data, and multiple groups of training data constitute a training set, and use the training set as the input of the machine learning model, and predict the bleeding risk coefficient at the future moment after the time step S as the output, and use the subsequent bleeding risk coefficient of each training set as the prediction target, and use the prediction accuracy as the training target to train the machine learning; Generate a machine learning model that predicts bleeding risk factors at future moments based on tissue characteristic data.
8. The closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring as claimed in claim 1, characterized in that: The remediation level instructions include a primary remediation level, a secondary remediation level and a tertiary remediation level, and the severity of the primary remediation level, the secondary remediation level and the tertiary remediation level increases in sequence.
9. A closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring as claimed in claim 8, characterized in that: The method for generating the corresponding remedial level instruction based on the predicted bleeding risk coefficient at a future moment is as follows: The predicted bleeding risk coefficient is marked as YXf, and the preset bleeding risk coefficient thresholds are Xf1, Xf2, and Xf3, where Xf1<Xf2<Xf3; When YXf≤Xf1, no remedial level instruction is generated; When Xf2≥YXf>Xf1, at this time, a first-level remediation level instruction is generated; When Xf3≥YXf>Xf2, at this time, a secondary remediation level instruction is generated; When YXf>Xf3, at this time, a third-level relief level instruction is generated.
10. The closed-loop adaptive control system for high-frequency ablation hemostatic forceps with integrated tissue monitoring as claimed in claim 1, characterized in that: The ablation parameter set is {(a1, b1, c1), (a2, b2, c2) ..., (a n , b n , c n )}, when the ablation parameter set is numbered, a, b, c correspond to ablation power, ablation time, and ablation frequency, respectively. n , b n , c n Indicates the values of ablation power, ablation time and ablation frequency corresponding to the ablation number n.
