Intelligent Parameter Optimization Method for Endometrial Radiofrequency Ablation Based on Real-Time Impedance Feedback

Abstract

This invention relates to the field of endometrial radiofrequency ablation technology, specifically to an intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback. This method synchronously acquires the instantaneous total power and instantaneous total admittance of the radiofrequency circuit in real time and constructs a sliding time window sequence; it performs local extremum normalization to eliminate dimensional differences, obtaining a two-dimensional phase space point cloud set; it uses the Vietoris-Rips algorithm to extract one-dimensional continuous coherence features, obtaining closed-loop feature values ​​of the point cloud; based on the closed-loop feature values ​​of the point cloud, window energy consumption, and admittance drop, it obtains the tissue dehydration characteristic index; finally, based on the real-time instantaneous total admittance, combined with admittance drop and the dehydration characteristic index, it makes a decision and outputs energy processing instructions. This invention can accurately distinguish between pseudo-high impedance caused by air bubbles and true deep dehydration penetration, ensuring thorough ablation while avoiding the risk of thermal damage.

Description

Technical Field

[0001] This invention relates to the field of endometrial radiofrequency ablation technology, specifically to an intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback. Background Technology

[0002] Radiofrequency ablation of the endometrium is a commonly used minimally invasive technique for the clinical treatment of abnormal uterine bleeding. Its basic principle is to release a high-frequency alternating current into the uterine cavity through a vaginal electrode. The high-frequency electromagnetic field causes the polar molecules in the target tissue to vibrate and rub at high speed, thereby generating heat energy, which causes the endometrial tissue to be heated, dehydrated, and undergo coagulative necrosis.

[0003] In the closed-loop control process of radiofrequency ablation, real-time monitoring of the equivalent impedance of the circuit is the core means to determine the degree of ablation and ensure surgical safety. Conventional radiofrequency ablation equipment usually presets a fixed absolute impedance threshold. Once the real-time impedance rises to this threshold, the system considers that the tissue has completed dehydration and immediately cuts off the radiofrequency energy output. However, in the early and middle stages of ablation, the blood and secretions on the surface of the target area boil when heated, instantly generating a large number of steam bubbles. These bubbles cover the electrode surface, forming a thin gas layer, which causes a transient and sharp spike in equivalent impedance (i.e., a false impedance rise). Existing equipment often cannot identify this bubble illusion and mistakenly judges it as tissue drying, thus prematurely cutting off energy output and resulting in incomplete ablation. Secondly, as ablation progresses into the later stages, with the loss of superficial moisture, radiofrequency energy begins to conduct to the deep dense muscle layer and cause real deep dehydration. Due to the physical characteristics of deep tissue, the resulting impedance rise is sometimes less severe than that caused by the boiling of superficial liquid. This leads to the system failing to issue a power-off command for a long time when facing major medical risks such as irreversible thermal penetration and uterine perforation because the impedance has not reached a high threshold.

[0004] Therefore, existing impedance monitoring algorithms cannot accurately distinguish between false high impedance caused by superficial bubbles and true high impedance caused by deep penetration. This leads to the power-off control logic of conventional radiofrequency ablation equipment oscillating between two extreme errors: incomplete ablation (false positive and accidental shutdown) and transmural thermal damage (false negative and missed detection). This severely restricts the success rate and safety of endometrial radiofrequency ablation. Summary of the Invention

[0005] To address the technical problem that existing impedance monitoring algorithms struggle to accurately distinguish between false high impedance caused by superficial air bubbles and true high impedance caused by deep penetration, leading to both incomplete ablation and the risk of transmural thermal damage, this invention aims to provide an intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback. The specific technical solution adopted is as follows:

[0006] In a first aspect, one embodiment of the present invention provides a method for intelligent optimization of endometrial radiofrequency ablation parameters based on real-time impedance feedback, the method comprising the following steps:

[0007] Real-time synchronous acquisition of the instantaneous total power and instantaneous total admittance of the radiofrequency circuit during endometrial radiofrequency ablation, and construction of a sliding time window sequence;

[0008] Local extremum normalization is performed on the instantaneous total power and instantaneous total admittance in the sliding time window sequence to obtain a two-dimensional phase space point cloud set;

[0009] The Vietoris-Rips algorithm is used to extract one-dimensional continuous cohomology features from a two-dimensional phase space point cloud set, and the closed-loop feature values ​​of the point cloud representing the generation and rupture cycle amplitude of bubbles are obtained.

[0010] Based on the point cloud closed-loop feature value, the energy consumption in the sliding time window sequence, and the instantaneous total admittance drop, the tissue dehydration characteristic index is obtained;

[0011] Based on real-time monitored instantaneous total admittance, the decision-making logic is determined by combining the instantaneous total admittance drop in the sliding time window sequence and the tissue dehydration characteristic index, and an energy processing command is output.

[0012] Furthermore, the method for obtaining the two-dimensional phase space point cloud set is as follows:

[0013] Obtain the maximum and minimum values ​​of the instantaneous total power within the sliding time window sequence, and use them as the first power and the second power, respectively.

[0014] Obtain the maximum and minimum values ​​of the instantaneous total admittance within the sliding time window sequence, and use them as the first admittance and the second admittance, respectively.

[0015] When the difference between the first power and the second power is greater than or equal to the preset power fluctuation threshold, the difference between the total instantaneous power and the second power in each instantaneous time window sequence is divided by the sum of the difference between the first power and the second power and the first preset minimum positive number to obtain the normalized result of the total instantaneous power.

[0016] The difference between the total instantaneous admittance and the second admittance in each sliding time window sequence is divided by the sum of the difference between the first admittance and the second admittance and the second preset minimum positive number to obtain the normalized result of the total instantaneous admittance.

[0017] The normalized instantaneous total power and normalized instantaneous total admittance corresponding to each sampling moment in the sliding time window sequence are combined to generate two-dimensional coordinate points. Multiple consecutive two-dimensional coordinate points are then aggregated to form a two-dimensional phase space point cloud set.

[0018] Furthermore, the method for obtaining the point cloud closed-loop feature values ​​is as follows:

[0019] A gridded downsampling process is performed on a two-dimensional phase space point cloud set to obtain a simplified point cloud subset;

[0020] One-dimensional continuous homology features of a subset of simplified point clouds are calculated based on the Vietoris-Rips algorithm to obtain a one-dimensional continuous homology barcode.

[0021] Extract the dominant topological feature term with the longest lifespan from the one-dimensional continuous homology barcode, and obtain the generation filter distance corresponding to its birth time and the closure filter distance corresponding to its death time.

[0022] The result of subtracting the generated filter distance from the closed filter distance is used as the point cloud closed-loop feature value characterizing the magnitude of bubble generation and rupture cycles.

[0023] Furthermore, the method for obtaining the tissue dehydration characteristic index is as follows:

[0024] The sum of the instantaneous total power at each sampling moment in the sliding time window sequence is divided by the preset capacity to obtain the average energy consumption of the window.

[0025] Based on the instantaneous total admittance drop in the sliding time window sequence, obtain the smoothed admittance drop value;

[0026] The negative product of the point cloud closed-loop feature value and the preset sensitivity coefficient is used as the exponent of the natural constant to construct an exponential function and obtain the nonlinear penalty factor.

[0027] When the smoothed admittance drop value is greater than the preset significant drop threshold, the product of the window average energy consumption and the nonlinear penalty factor is divided by the sum of the smoothed admittance drop value and the third preset minimum positive number to obtain the tissue dehydration characteristic index.

[0028] When the smooth admittance drop value is less than or equal to the preset significant drop threshold, 0 is used as the tissue dehydration characteristic index.

[0029] Furthermore, the method for obtaining the smoothed admittance drop value is as follows:

[0030] The average of the earliest stored, preset number of instantaneous total admittances in the sliding time window sequence is used as the early average admittance.

[0031] The average of the most recently stored, preset number of instantaneous total admittances in the sliding time window sequence is used as the recent average admittance.

[0032] The difference between the early average admittance and the recent average admittance is used as the smoothed admittance drop value.

[0033] Furthermore, the method for determining the energy processing command based on real-time monitored instantaneous total admittance, combined with the instantaneous total admittance drop in the sliding time window sequence and the tissue dehydration characteristic index, is as follows:

[0034] Calculate the reciprocal of the instantaneous total admittance at the current sampling moment, and use it as the current instantaneous equivalent total impedance;

[0035] When the current instantaneous equivalent total impedance is less than the preset absolute impedance monitoring threshold, an energy maintenance command is sent to the RF generator.

[0036] When the current instantaneous equivalent total impedance is greater than or equal to the preset absolute impedance monitoring threshold, and the smooth admittance drop value is less than or equal to zero, an energy maintenance command is sent to the RF generator.

[0037] When the current instantaneous equivalent total impedance is greater than or equal to the preset absolute impedance monitoring threshold, and the smooth admittance drop value is greater than zero, compare the tissue dehydration characteristic index with the preset deep penetration judgment threshold:

[0038] If the tissue dehydration characteristic index is less than the preset deep penetration judgment threshold, the current impedance increase is determined to be a false high impedance caused by bubble isolation, and an energy maintenance command is sent to the radio frequency generator.

[0039] If the tissue dehydration characteristic index is greater than or equal to the preset deep penetration judgment threshold, it is determined that the target tissue has undergone substantial dehydration and thermal penetration, and an energy cutoff command is sent to the radio frequency generator.

[0040] Furthermore, the method for obtaining the instantaneous total power is as follows:

[0041] In the radio frequency circuit, the instantaneous effective values ​​of voltage and current are synchronously sampled and processed by envelope detection or root mean square conversion.

[0042] The product of the instantaneous effective value of voltage and the instantaneous effective value of current at each sampling time is taken as the instantaneous total power at each sampling time.

[0043] Furthermore, the method for obtaining the instantaneous total admittance is as follows:

[0044] Divide the effective value of the instantaneous current at each sampling moment by the effective value of the instantaneous voltage to obtain the instantaneous total admittance at each sampling moment.

[0045] Furthermore, the method for constructing the sliding time window sequence is as follows:

[0046] The data pairs consisting of instantaneous total power and instantaneous total admittance are stored using a first-in-first-out queue with a preset capacity;

[0047] Once the total number of data pairs stored in the first-in-first-out queue reaches a preset capacity, whenever a new data pair is stored, the oldest historical data pair in the first-in-first-out queue is removed to obtain a sliding time window sequence consisting of a preset capacity of data pairs.

[0048] Secondly, another embodiment of the present invention provides an intelligent optimization system for endometrial radiofrequency ablation parameters based on real-time impedance feedback. The system includes: a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above methods.

[0049] The present invention has the following beneficial effects:

[0050] This invention first synchronously acquires the instantaneous total power and instantaneous total admittance of the radiofrequency circuit during endometrial radiofrequency ablation in real time. This ensures that the extracted underlying electrical data can quantify the system's thermal energy injection intensity and the overall conductivity of the target area in real time and with high fidelity. Furthermore, it constructs a sliding time window sequence, which is beneficial for establishing a smooth and statistically significant historical reference interval for subsequent identification of the alternating trajectories of bubble boiling and backfilling, completely eliminating algorithm deadlock caused by single-point spike noise. To eliminate the huge dimensional gap between instantaneous total power and instantaneous total admittance, and to prevent the tiny values ​​of the admittance dimension from being misinterpreted... The fluctuations are masked by the absolute values ​​of the power dimension. Therefore, local extremum normalization is performed on the instantaneous total power and instantaneous total admittance in the sliding time window sequence to obtain a two-dimensional phase space point cloud set, mapping the disordered one-dimensional temporal electrical fluctuations to a scale-equivalent two-dimensional geometric feature space. To accurately identify the periodic closed trajectories formed by bubble vaporization and liquid backfilling from complex electromagnetic and physiological noise, the Vietoris-Rips algorithm is used to extract one-dimensional continuous cohomology features from the two-dimensional phase space point cloud set, obtaining a point cloud characterizing the amplitude of the bubble generation and rupture cycle. Closed-loop eigenvalues ​​eliminate misjudgments caused by non-closed impedance rises due to unidirectional dehydration of deep dense tissues at the root of data distribution. To remove the non-essential energy weights wasted by the superficial bubble boiling phenomenon and avoid false triggering of deep penetration warnings due to overestimation of original thermal energy, a tissue dehydration characteristic index is obtained based on the point cloud closed-loop eigenvalues, energy consumption in the sliding time window sequence, and instantaneous total admittance drop. This precisely quantifies the effective thermal energy cost that drives substantial irreversible dehydration of deep dense tissues. Finally, based on real-time monitored instantaneous total admittance, the system... After the current resistance reaches the high resistance warning switch, the decision-making logic is made by combining the instantaneous total admittance drop in the sliding time window sequence and the tissue dehydration characteristic index. This successfully breaks through the blind spot of traditional equipment that only relies on the absolute threshold of impedance for control. It accurately outputs energy processing instructions to ensure that heat energy is injected during the false high impedance peak period caused by bubbles to ensure the thoroughness of ablation. When encountering the risk of real deep thermal penetration, the power supply circuit is precisely cut off. This is conducive to significantly improving the clinical success rate and safety of endometrial radiofrequency ablation without adding additional hardware sensors. Attached Figure Description

[0051] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1A schematic flowchart illustrating an intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback, provided as an embodiment of the present invention.

[0053] Figure 2 A flowchart illustrating a method for obtaining tissue dehydration characteristic indices according to an embodiment of the present invention;

[0054] Figure 3 The diagram shows a structural diagram of an intelligent optimization system for endometrial radiofrequency ablation parameters based on real-time impedance feedback, provided in one embodiment of the present invention.

[0055] Figure 4 This is a schematic diagram of a computer device provided according to an embodiment of the present invention. Detailed Implementation

[0056] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback proposed by the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0058] The following description, in conjunction with the accompanying drawings, details the specific scheme of the intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback provided by this invention.

[0059] Example 1:

[0060] This invention proposes an intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback. Please refer to [link / reference]. Figure 1 The diagram illustrates a schematic flowchart of an intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback, according to an embodiment of the present invention. The method includes the following steps:

[0061] Step S1: Real-time synchronous acquisition of the instantaneous total power and instantaneous total admittance of the radiofrequency circuit during endometrial radiofrequency ablation, and construction of a sliding time window sequence.

[0062] Specifically, considering that the equivalent resistance between the electrode and the target tissue interface will dynamically change as the local tissue is heated and dehydrated during radiofrequency ablation, in order to directly quantify the current injection intensity of radiofrequency energy and the energizing capacity of the target tissue, this embodiment synchronously acquires the instantaneous total power and instantaneous total admittance of the radiofrequency circuit during the endometrial radiofrequency ablation process in real time, as the underlying data source for subsequent intelligent identification of bubble artifacts and judgment of deep penetration.

[0063] The instantaneous total power and instantaneous total admittance are obtained by: synchronously sampling the instantaneous effective values ​​of voltage and current in the radio frequency circuit through an internal unified clock driving mechanism, which are obtained by envelope detection or root mean square conversion, to avoid the voltage dropping to zero instantaneously when the high-frequency alternating current crosses the zero point, which would cause subsequent division operations to crash; envelope detection and root mean square conversion are well known and will not be described in detail; then the product of the instantaneous effective value of voltage and the instantaneous effective value of current at each sampling moment is used as the instantaneous total power at each sampling moment, which accurately reflects the intensity of macroscopic thermal energy injected into the uterine cavity by the system at each sampling moment;

[0064] Because admittance, as the reciprocal of resistance, can more linearly characterize the retention state of the conductive medium, this embodiment divides the effective value of the instantaneous current at each sampling moment by the effective value of the instantaneous voltage to obtain the instantaneous total admittance at each sampling moment. The larger the instantaneous total admittance, the more unobstructed the overall conductive channel between the tissue and fluid in the uterine cavity at the corresponding sampling moment; conversely, the smaller the admittance, the greater the electrical resistance caused by tissue moisture loss, bubble isolation, or tissue drying. This embodiment sets the time interval between two adjacent sampling moments to 10 milliseconds to ensure that the millisecond-level transient trajectory of the superficial fluid in the target area being heated and boiling and the bubble bursting and backfilling can be captured with high fidelity, while preventing the data dimension accumulated within a unit time window from being too large, which would overload the computing power of the microprocessor when calculating the topological features later. Implementers can set the time interval between two adjacent sampling moments according to the hardware sampling capability of the underlying analog-to-digital converter and the computing performance indicators of the microprocessor; no limitation is imposed here.

[0065] Considering that isolated electrical data under a single sampling cycle is highly susceptible to disturbances from local transient electromagnetic spikes, and that the subsequent bubble recognition algorithm must rely on a continuous time series of data to delineate the alternating trajectory of liquid boiling and backfilling, in order to establish a smooth and safe historical data reference interval, this embodiment constructs a sliding time window sequence to ensure that the data delivered to subsequent modules has statistical significance. The method for constructing the sliding time window sequence is as follows: First, a first-in-first-out (FIFO) queue with a preset capacity is used to store data pairs consisting of instantaneous total power and instantaneous total admittance. When the total number of data pairs stored in the FIFO queue is less than the preset capacity, it is determined that the device is currently in the data accumulation period. At this time, the system actively intercepts the subsequent feature extraction and decision logic calculations, directly issues an energy maintenance command to the RF generator, and skips the subsequent feature extraction and decision steps to avoid algorithm out-of-bounds or division-by-zero deadlock caused by blank historical data. Then, after the total number of data pairs stored in the FIFO queue reaches the preset capacity, it is determined that the data accumulation period has ended. At this time, whenever a new data pair is stored from the tail of the queue, the oldest historical data pair at the head of the FIFO queue is simultaneously removed to ensure that the queue always contains only the latest continuous segment of data of equal length, and finally obtains a sliding time window sequence consisting of a preset capacity of data pairs.

[0066] This embodiment sets the preset capacity to 100, combined with a 10-millisecond sampling interval, to ensure that the sliding time window precisely covers the electrical changes within the past 1.0 second. Considering that the complete physical evolution cycle of the liquid on the target tissue surface from heating and vaporization to the formation of a thin bubble layer and the subsequent bubble rupture and backfilling is typically between 0.1 and 0.5 seconds, this 1.0-second time span is sufficient to cover at least one complete liquid boiling and steam bubble rupture cycle, thus ensuring the formation of a complete closed trajectory in the two-dimensional phase space. The implementer can flexibly calculate and set the preset capacity and sampling period values ​​within a macroscopic time span of 0.5 to 2.0 seconds based on the actual bubble survival and rupture physical cycles calibrated in ex vivo endometrial ablation experiments at different radiofrequency powers, to ensure that the VR algorithm can stably capture closed-loop features; no limitations are imposed here. It should be noted that the output sliding time window sequence is a secure data source that is strictly aligned along the time axis and has undergone capacity verification, providing a reliable basis for subsequent extreme value normalization mapping in the two-dimensional phase space.

[0067] Step S2: Perform local extremum normalization on the instantaneous total power and instantaneous total admittance in the sliding time window sequence to obtain a two-dimensional phase space point cloud set.

[0068] Specifically, in order to address the hundreds of orders of magnitude difference between instantaneous total power (typically in the tens of watts range) and instantaneous total admittance (typically in the fraction of a Siemens range), and to prevent the tiny features of the admittance dimension from being completely swallowed up by the huge values ​​of the power dimension when calculating the Euclidean distance, this embodiment performs local extremum normalization processing on the instantaneous total power and instantaneous total admittance in the sliding time window sequence to obtain a two-dimensional phase space point cloud set, thus mapping the originally dimensionally separated one-dimensional time-series electrical data to a two-dimensional geometric feature space with equal scale.

[0069] Preferably, in one feasible embodiment of this method, the method for obtaining the two-dimensional phase space point cloud set is as follows: In order to determine the absolute boundary of electrical data fluctuations within the current time window, the maximum and minimum values ​​of the instantaneous total power within the sliding time window sequence are first obtained, and used as the first power and the second power respectively; the maximum and minimum values ​​of the instantaneous total admittance within the sliding time window sequence are obtained, and used as the first admittance and the second admittance respectively; considering that the power extreme difference approaches zero in the extremely stable constant power output mode of the radio frequency ablation device, which would lead to a very small denominator in the normalization formula, this embodiment infinitely amplifies the background white noise of the underlying hardware in the division operation, in order to prevent To prevent such erroneous spatial features generated by meaningless noise from interfering with subsequent decisions, the difference between the first power and the second power is obtained and compared with a preset power fluctuation threshold. In this embodiment, the preset power fluctuation threshold is set to one percent of the average instantaneous total power within the current sliding time window sequence. This ensures that the white noise of the RF generator hardware background and minor electromagnetic crosstalk are effectively filtered out, while avoiding the mistaken interception of energy fluctuations caused by the actual vaporization of shallow surface liquid. The implementer can set the size of the preset power fluctuation threshold according to the rated output capability, load response sensitivity, and effective resolution of the analog-to-digital conversion circuit of the selected RF generator. No limitation is imposed here.

[0070] When the difference between the first power and the second power is greater than or equal to the preset power fluctuation threshold, it indicates that the current RF circuit has sufficient signal fluctuations to construct a true two-dimensional phase space, meaning that the intensity of the injected thermal energy has a physically dynamic adjustment space. It should be noted that the two-dimensional phase space feature extraction mechanism constructed in this embodiment is not only applicable to variable power RF modes with adaptive impedance adjustment mechanisms, but also has a solid physical basis in conventional constant power setting modes. In constant power mode, the underlying hardware typically uses pulse width modulation or high-frequency switching to maintain constant macroscopic power. When faced with sudden load changes caused by violent boiling of shallow bubbles in the target area, the closed-loop feedback adjustment of the hardware will inevitably generate microscopic power response ripple and adjustment overshoot within the control cycle. This embodiment utilizes the coupling relationship between this underlying microscopic adjustment ripple and admittance change to transform the macroscopic physical characterization into a feature mapping of the microscopic control algorithm, ensuring that an effective two-dimensional phase space point cloud set can be extracted in any RF output mode.

[0071] At this point, the difference between the instantaneous total power and the second power in each instantaneous period of the sliding time window sequence is divided by the sum of the difference between the first power and the second power and the first preset minimum positive number to obtain the normalized result of the instantaneous total power, ensuring that the instantaneous total power is proportionally shifted and scaled to the range of 0 to 1; simultaneously, the difference between the instantaneous total admittance and the second admittance in each instantaneous period of the sliding time window sequence is divided by the sum of the difference between the first admittance and the second admittance and the second preset minimum positive number to obtain the normalized result of the instantaneous total admittance, ensuring that the instantaneous total admittance is also shifted and scaled to the range of 0 to 1; in this embodiment, the first preset minimum positive number and the second preset minimum positive number are both set to 1. As a fallback mechanism to completely prevent hardware division-by-zero interrupt crashes caused by a completely zero denominator, implementers can set a first preset minimum positive number and a second preset minimum positive number according to the processor's floating-point operation precision, without any restrictions here;

[0072] To intuitively reconstruct the physical cycle trajectory of bubble boiling and backfilling, the normalized instantaneous total power and normalized instantaneous total admittance corresponding to each sampling moment in the sliding time window sequence are combined to generate two-dimensional coordinate points. This helps to transform isolated electrical values ​​into a geometrically distributed form with sequential correlation. Finally, multiple consecutive two-dimensional coordinate points are summarized to form a two-dimensional phase space point cloud set, providing a standardized high-quality data source for subsequent identification of topological hole features.

[0073] It should be noted that when the difference between the first power and the second power is less than the preset power fluctuation threshold, it indicates that the system is in a stable state with strictly constant output and no significant electrical disturbances. In this state, the intensity of the injected thermal energy remains almost constant, and there is no physical basis for a coupling between dynamic energy adjustment and admittance changes. Therefore, the system directly assigns zero to the closed-loop feature values ​​of the point cloud used to characterize the amplitude of bubble generation and rupture cycles, and actively skips the generation of two-dimensional coordinate points and the extraction of one-dimensional continuous coherence features. This forced zeroing fallback logic not only significantly saves computing resources, but also fundamentally eliminates false positives caused by noise amplification during the bubble-free boiling period.

[0074] Step S3: Use the Vietoris-Rips algorithm to extract one-dimensional continuous cohomology features from the two-dimensional phase space point cloud set, and obtain the point cloud closed-loop feature values ​​that characterize the cycle amplitude of bubble generation and rupture.

[0075] Specifically, considering that in the early stage of ablation, the shallow liquid vaporizes upon heating, causing a sharp drop in admittance, followed by a surge in admittance due to bubble rupture or the refilling of surrounding liquid, this recurring physical phenomenon will inevitably form a closed loop trajectory in the scatter plot of the two-dimensional phase space. However, the simple thermal dehydration of the deep, dense muscle layer leads to irreversible water loss, resulting in a unidirectional, long-term downward trend in admittance, and its scatter trajectory will never backtrack or close. To accurately identify the periodic closed loop representing bubble fluctuations from the disordered electrical scatter plot, this embodiment utilizes the Vietoris-Rips algorithm in computational topology to extract the one-dimensional continuous homology features of the two-dimensional phase space point cloud set. This transforms the invisible pseudo-high impedance phenomenon into a quantitative geometric topological index, thereby obtaining the point cloud closed-loop feature value characterizing the amplitude of the bubble generation and rupture cycle. This eliminates the misjudgment of impedance rise caused by the unidirectional drying of deep tissue from the root of the data distribution. The larger the point cloud closed-loop feature value, the larger the volume of the generated bubble and the more hollow the cycle trajectory, indicating a more significant pseudo-high impedance phenomenon caused by bubble isolation. Among them, the Vietoris-Rips algorithm and the solution process of one-dimensional continuous homology are well-known techniques in the field of computational topology, and the specific iterative process of calculating the simple complex will not be described in detail; however, in this embodiment, in order to avoid memory overflow and computational overload, the system limits the Vietoris-Rips algorithm to only calculate up to the one-dimensional simple complex, and sets the upper limit of the distance parameter to the maximum effective geometric diameter of the normalized space (i.e., 1.0).

[0076] Preferably, in one feasible method of this embodiment, the method for obtaining the closed-loop feature value of the point cloud is as follows: First, considering the computing power limitation of conventional medical device microprocessors, directly performing simple complex calculations on 100 scattered points is very likely to exceed the 10-millisecond control cycle, causing system lag. Therefore, this embodiment performs gridded downsampling processing on the two-dimensional phase space point cloud set. Specifically, the two-dimensional normalized space from 0 to 1 is divided into uniform grids with a preset step size (e.g., 0.05). The center point or the real scattered point closest to the center in each grid containing scattered points is retained, and the remaining redundant points in the grid are removed to obtain a simplified subset of the point cloud. This operation effectively removes the densely overlapping points inside the bubble loop trajectory, and exponentially compresses the number of points involved in the calculation without destroying the macroscopic topological loop features. Then, based on Vietoris-Rips The algorithm calculates the one-dimensional persistent homology features of a simplified subset of point clouds to obtain one-dimensional persistent homology barcodes recording the lifecycle of each topological hole feature. Since the scatter plot may contain tiny, short-lived holes caused by local discrete noise, to pinpoint the core dominant trajectory reflecting macroscopic bubble vaporization and backfilling phenomena, the algorithm extracts the longest-lived dominant topological feature term from the one-dimensional persistent homology barcode. This yields the generation filter distance at its birth time and the closing filter distance at its death time. The generation and closing filter distances quantify the geometric scale of the largest annular hole when it just forms a closed loop and when it is completely filled by intersecting connected edges, respectively. To intuitively quantify the intensity and scope of the bubble alternation phenomenon, the result of subtracting the generation filter distance from the closing filter distance is used as the point cloud closed-loop feature value characterizing the amplitude of the bubble generation and bursting cycle. It should be noted that if no closed hole features are extracted, the point cloud closed-loop feature value is directly assigned to zero, indicating that the current state is purely unidirectional dehydration and heating.

[0077] Step S4: Obtain the tissue dehydration characteristic index based on the point cloud closed-loop feature value, the energy consumption in the sliding time window sequence, and the instantaneous total admittance drop.

[0078] Specifically, considering that traditional radio frequency control logic directly correlates total energy consumption with impedance changes, it is easy to overestimate the actual absorbed energy of deep tissues. In order to remove the non-substantial energy weights that are wasted by the bubble-maintaining phenomenon from the original thermal energy injected by the system, this embodiment obtains the tissue dehydration characteristic index based on the point cloud closed-loop feature value, the energy consumption in the sliding time window sequence, and the instantaneous total admittance drop. This accurately reflects the thermal energy cost of the system causing irreversible substantial dehydration of deep dense tissues. The larger the tissue dehydration characteristic index, the more significant the tendency of radio frequency energy to penetrate into the high-impedance deep muscle layer, even after eliminating the interference of bubble boiling.

[0079] Preferably, in one feasible embodiment, the method for obtaining the tissue dehydration characteristic index is described in [reference needed]. Figure 2 The document presents a flowchart of a method for obtaining tissue dehydration characteristic indices, as provided in this embodiment. The method includes the following steps:

[0080] Step S201: Obtain the average energy consumption of the window.

[0081] Considering that directly summing the total energy consumption within the sliding time window would lead to instability of the dimensional reference due to changes in the sliding time window length or sampling rate, this embodiment quantifies the standardized average intensity of the thermal energy injected into the uterine cavity by the system over a fixed period of time. Therefore, this embodiment divides the sum of the instantaneous total power at each sampling moment in the sliding time window sequence by a preset capacity to obtain the window average energy consumption, effectively eliminating the interference of time span on subsequent energy weight calculations. A larger window average energy consumption indicates a larger total scale of macroscopic thermal energy continuously injected into the uterine cavity by the radio frequency generator within that sliding time window, and a more significant energy compensation by the system in attempting to overcome the equivalent resistance at the electrode-tissue interface to maintain the set output intensity.

[0082] Step S202: Obtain the smoothed admittance drop value.

[0083] In order to quantify the decreasing trend of the overall conductivity of the target area and avoid drastic distortion of the admittance difference caused by unpredictable transient electromagnetic spikes (single-point noise) in a single sampling, this embodiment obtains the smoothed admittance drop value based on the instantaneous total admittance drop in the sliding time window sequence. The specific method for obtaining the smoothed admittance drop value is as follows: the average value of the earliest stored preset number of instantaneous total admittances in the sliding time window sequence is used as the early average admittance to characterize the initial electrical conductivity at the beginning of the sliding time window; then, the average value of the latest stored preset number of instantaneous total admittances in the sliding time window sequence is used as the recent average admittance to characterize the residual electrical conductivity at the end of the sliding time window; in order to eliminate the single-point peak interference at the beginning and end and to truly reflect the downward slope of tissue conductivity, this embodiment uses the difference between the early average admittance and the recent average admittance as the smoothed admittance drop value, which accurately reflects the unidirectional decay trend of the overall conductivity of the target area tissue within the sliding time window span; the larger the smoothed admittance drop value, the more severe the water loss of the tissue on a macroscopic scale, and the more rapid the dehydration process of local densification. In this embodiment, the preset quantity is set to 4 to ensure that minor glitch can be smoothed out without causing a significant time delay effect. The implementer can set the size of the preset quantity according to the sampling frequency and the signal-to-noise ratio of the underlying electrical signal, which is not limited here.

[0084] Step S203: Obtain the nonlinear penalty factor.

[0085] Considering that although the boiling cycle of bubbles consumes a lot of energy, this energy is not used to damage the deep tissue muscle layer. In order to mathematically subtract the share of ineffective thermal energy used to maintain the vaporization of superficial fluid from the total energy consumption, this embodiment takes the negative of the product of the point cloud closed-loop feature value and the preset sensitivity coefficient, and uses it as the exponent of the natural constant to construct an exponential function to obtain a nonlinear penalty factor. This accurately reflects the proportion of effective energy retained per unit of energy consumption that is actually converted into irreversible dehydration of tissue. The larger the nonlinear penalty factor, the less the current radio frequency energy is vaporized superficially in the target area, and the higher the weight of the energy substantially absorbed by the deep dense tissue and converted into thermal damage.

[0086] It is known that the more intense the bubble cycle, the larger the point cloud closed-loop feature value. According to the characteristics of the exponential function, the nonlinear penalty factor will decay exponentially and approach zero, thereby strongly reducing the energy consumption during the period of severe bubble interference. This ensures that the tissue dehydration feature index calculated subsequently will not be falsely overestimated due to the violent boiling of the surface liquid. In this embodiment, a preset sensitivity coefficient of 3 is set to adjust the penalty intensity of the algorithm on the bubble interference feature. This ensures that it can significantly suppress the high energy consumption illusion caused by large-size closed loops, while not over-penalizing the trace particles of bubbles that accompany normal dehydration, which would lead to an underestimation of the true damage. The implementer can set the preset sensitivity coefficient according to the bubble aggregation rate caused by the surface morphology of the radio frequency electrode and the sensitivity of the system sampling frequency to loop capture. This is not limited here.

[0087] Step S204: Obtain tissue dehydration characteristic index.

[0088] It is known that in the very early stages of ablation, due to the extremely abundant tissue moisture, the admittance hardly decreases or even slightly increases with heat. Forcibly dividing the penalized, minimal energy consumption by a smooth admittance drop value approaching zero would lead to an uncontrollable mathematical spike in the final value. To prevent false triggering of an extremely high index alarm and deadlock during a period when there is no risk of tissue penetration, this embodiment compares the smooth admittance drop value with a preset significant drop threshold. The division operation is only initiated when a substantial physical drop in admittance occurs. This embodiment sets the preset significant drop threshold to 5% of the average admittance in the early part of the current time window, ensuring a strict boundary between the moisture-filled period and the actual dehydration period. The implementer can set the preset significant drop threshold based on the initial conductivity of the specific tissue; no limitation is imposed here. When the smooth admittance drop value exceeds the preset significant drop threshold, it indicates that the tissue conductivity has begun to irreversibly decrease. At this point, the product of the window's average energy consumption and the nonlinear penalty factor is divided by the sum of the smooth admittance drop value and a third preset minimum positive number to obtain the tissue dehydration characteristic index. This embodiment sets the third preset minimum positive number to... To ensure that even when the smooth admittance drop value approaches zero or falls below the floating-point arithmetic threshold in a state of extreme water abundance or slight impedance change in the tissue, the hardware system will not be interrupted by division by zero and thus crash. The implementer can set a third preset minimum positive number based on the floating-point data type width of the microprocessor and the division protection mechanism of the underlying arithmetic library, without any restrictions here. When the smooth admittance drop value is less than or equal to the preset significant drop threshold, it indicates that the current resistance has not increased or the tissue is in a safe conduction state with abundant water. At this time, 0 is used as the tissue dehydration characteristic index, which effectively avoids the misjudgment of characteristic spike caused by an excessively small denominator.

[0089] Step S5: Based on the real-time monitored instantaneous total admittance, and combined with the instantaneous total admittance drop in the sliding time window sequence and the tissue dehydration characteristic index, a decision logic is made to determine the energy processing command.

[0090] Specifically, it is known that in conventional radiofrequency ablation equipment, the control system typically only monitors the overall impedance. Once the impedance spikes, it blindly cuts off the power, which can easily lead to premature termination of ablation when superficial bubbles are generated. In order to completely solve the high misjudgment and shutdown rate caused by single-dimensional impedance threshold control, this embodiment first analyzes whether the current target area has reached the traditional high impedance warning boundary based on real-time monitored instantaneous total admittance. In order to accurately distinguish the physical authenticity of the impedance increase after the warning is triggered, the decision logic is made by combining the instantaneous total admittance drop in the sliding time window sequence and the tissue dehydration characteristic index, and outputting energy processing instructions. This realizes the intelligent control leap from blind power-off to adaptive joint adjudication, which is conducive to adaptively distinguishing between low-energy-consuming, high-impedance bubble artifacts and high-energy-consuming, high-impedance real thermal penetration without adding additional hardware sensors. This significantly reduces the clinical misjudgment and shutdown rate to ensure the thoroughness of a single ablation, while effectively avoiding the major risk of transmural thermal damage to deep tissues.

[0091] Preferably, in one feasible manner of this embodiment, the method for performing decision logic judgment and outputting energy processing instructions is as follows: First, calculate the reciprocal of the instantaneous total admittance at the current sampling moment as the current instantaneous equivalent total impedance. Although admittance has excellent stability when calculating characteristics, impedance values ​​are more in line with existing standards and intuitive cognitive habits in the final clinical safety monitoring stage. The larger the current instantaneous equivalent total impedance, the greater the overall electrical resistance encountered by the alternating current as it passes through the uterine cavity, and there may be local bubble isolation or tissue desiccation. Conversely, the smaller the current instantaneous equivalent total impedance, the more sufficient the target area moisture and the better the electrode contact, and the radio frequency energy transmission is in the conventional benign heating stage.

[0092] To avoid frequent and ineffective screening by calling complex characteristic indices during the safe phase of stable heating of large tissue areas in the early stages of ablation, this embodiment sets a preset absolute impedance monitoring threshold of 180 ohms (a warning trigger switch embedded in the system). This ensures that the system remains in a silent monitoring state in the normal low impedance range to save computing power. The implementer can set the preset absolute impedance monitoring threshold according to the geometric surface area of ​​the radiofrequency electrode used and the clinical calibration specifications. It is usually set to 2.5 to 3 times the equivalent impedance of the basic normal tissue, and no limit is imposed here. When the current instantaneous equivalent total impedance is less than the preset absolute impedance monitoring threshold, it indicates that the target area tissue has sufficient moisture and good conductivity. At this time, an energy maintenance command is sent to the radiofrequency generator to ensure that the radiofrequency energy is continuously and stably injected.

[0093] When the current instantaneous equivalent total impedance is greater than or equal to the preset absolute impedance monitoring threshold, and the smoothed admittance drop value is less than or equal to zero, it indicates that although the current instantaneous impedance has touched the red line, the recent average admittance is greater than or equal to the earlier average admittance within the past time window. This means that the local conductive path is recovering (e.g., a large amount of conductive liquid overflows from heated tissue, or the electrode contact surface becomes tighter), and the overall impedance actually shows a benign trend of stabilization or even decrease. At this time, an energy maintenance command is issued to the RF generator to ensure that the system will not erroneously shut down due to the spike in a single impedance sampling. When the current instantaneous equivalent total impedance is greater than or equal to the preset absolute impedance monitoring threshold, and the smoothed admittance drop value is greater than zero, it indicates that the current equivalent impedance is recovering. The impedance is approaching the power-off condition of conventional equipment, and the impedance is indeed in a deteriorating trend of continuous increase. In order to further determine whether this deterioration is caused by superficial bubble isolation or by deep myometrial dehydration, the tissue dehydration characteristic index is compared with the preset deep penetration judgment threshold. In this embodiment, the preset deep penetration judgment threshold is set as the critical penetration thermal energy cost constant (e.g., set to 400) pre-calibrated through in vitro dense myometrial experiments. This ensures that the preset deep penetration judgment threshold can accurately correspond to the energy boundary when the radiofrequency energy just begins to penetrate the dense myometrium and there is a danger of thermal penetration. The implementer can set the preset deep penetration judgment threshold according to the in vitro tissue ablation calibration experimental data of equipment with different power levels. No limitation is made here.

[0094] If the tissue dehydration characteristic index is less than the preset deep penetration threshold, it indicates that the effective thermal energy cost of overcoming the current impedance increase is low. The current impedance increase is mostly caused by vapor bubbles generated by the vaporization of the superficial liquid isolating the electrode. The dense muscle layer at the bottom has not been completely destroyed. Therefore, the current impedance increase is determined to be a false high impedance caused by bubble isolation. An energy maintenance command is sent to the radio frequency generator, enabling the radio frequency system to continue to heat the target tissue by crossing the false high impedance peak generated by bubble isolation, ensuring the thoroughness of ablation. If the tissue dehydration characteristic index is greater than or equal to the preset deep penetration threshold, it indicates that the effective thermal energy invested by the system to overcome the same impedance increase is extremely large, reaching or even exceeding the energy cost of penetrating the dense muscle layer. The surface water of the target area has been completely lost, and the radio frequency energy is being conducted to the dangerous deep muscle layer in large quantities. Therefore, it is determined that substantial dehydration thermal penetration has occurred in the target tissue. An energy cutoff command is sent to the radio frequency generator, forcibly disconnecting the power supply circuit of the radio frequency generator, thereby safely and timely ending the radio frequency ablation process.

[0095] In summary, this embodiment synchronously acquires the instantaneous total power and instantaneous total admittance of the RF circuit in real time and constructs a sliding time window sequence; it performs local extremum normalization processing to eliminate dimensional differences, obtaining a two-dimensional phase space point cloud set; it uses the Vietoris-Rips algorithm to extract one-dimensional continuous coherence features, obtaining the point cloud closed-loop feature value; based on the point cloud closed-loop feature value, window energy consumption, and admittance drop, it obtains the tissue dehydration characteristic index; finally, based on the real-time instantaneous total admittance, combined with the admittance drop and dehydration characteristic index, it makes a decision and outputs an energy processing command. This invention can accurately distinguish between the pseudo-high impedance of bubbles and true deep dehydration penetration, ensuring thorough ablation while avoiding the risk of thermal damage.

[0096] Example 2:

[0097] This invention also proposes an intelligent optimization system for endometrial radiofrequency ablation parameters based on real-time impedance feedback. Please refer to [link to relevant documentation]. Figure 3 The diagram illustrates a structural diagram of an intelligent optimization system for endometrial radiofrequency ablation parameters based on real-time impedance feedback, provided by an embodiment of the present invention. The system includes: a data acquisition module 10, a two-dimensional phase space point cloud set acquisition module 20, a point cloud closed-loop feature value acquisition module 30, a tissue dehydration feature index acquisition module 40, and a data processing module 50.

[0098] The data acquisition module 10 is used to synchronously acquire the instantaneous total power and instantaneous total admittance of the radiofrequency circuit during the endometrial radiofrequency ablation process in real time, and to construct a sliding time window sequence.

[0099] The two-dimensional phase space point cloud set acquisition module 20 is used to perform local extremum normalization processing on the instantaneous total power and instantaneous total admittance in the sliding time window sequence to acquire the two-dimensional phase space point cloud set.

[0100] The point cloud closed-loop feature value acquisition module 30 is used to extract one-dimensional continuous cohomology features of a two-dimensional phase space point cloud set using the Vietoris-Rips algorithm, and to obtain point cloud closed-loop feature values ​​that characterize the cycle amplitude of bubble generation and rupture.

[0101] The tissue dehydration characteristic index acquisition module 40 is used to acquire the tissue dehydration characteristic index based on the point cloud closed-loop characteristic value, the energy consumption in the sliding time window sequence, and the instantaneous total admittance drop.

[0102] The data processing module 50 is used to make decision-making logic based on the instantaneous total admittance monitored in real time, combined with the instantaneous total admittance drop in the sliding time window sequence and the tissue dehydration characteristic index, and output energy processing instructions.

[0103] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the intelligent optimization system for endometrial radiofrequency ablation parameters based on real-time impedance feedback and the intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiments and will not be repeated here.

[0104] Example 3:

[0105] This invention also proposes an intelligent optimization device for endometrial radiofrequency ablation parameters based on real-time impedance feedback. The device includes a memory and a processor. The memory stores executable program code, and the processor calls and executes this executable program code to perform the intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback provided in the embodiments of this application. Specifically, the device may be a chip, component, or module. The chip may include a connected processor and memory; the memory stores instructions, and when the processor calls and executes the instructions, the chip can perform the intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback provided in the above embodiments.

[0106] In addition, this embodiment also protects a computer device; please refer to [link to relevant documentation]. Figure 4The computer device includes a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402. When the processor 402 executes the computer program 403, the computer device can execute any of the aforementioned intelligent optimization methods for endometrial radiofrequency ablation parameters based on real-time impedance feedback.

[0107] Example 4:

[0108] The present invention also provides a computer-readable storage medium storing computer program code, which, when run on a computer, causes the computer to execute the above-mentioned method steps to implement the intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback provided in the above embodiments.

[0109] Example 5:

[0110] The present invention also provides a computer program product that, when run on a computer, causes the computer to perform the above-mentioned related steps to realize the intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback provided in the above embodiments.

[0111] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0112] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0113] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for intelligent optimization of endometrial radiofrequency ablation parameters based on real-time impedance feedback, characterized in that, The method includes the following steps: Real-time synchronous acquisition of the instantaneous total power and instantaneous total admittance of the radiofrequency circuit during endometrial radiofrequency ablation, and construction of a sliding time window sequence; Local extremum normalization is performed on the instantaneous total power and instantaneous total admittance in the sliding time window sequence to obtain a two-dimensional phase space point cloud set; The Vietoris-Rips algorithm is used to extract one-dimensional continuous cohomology features from a two-dimensional phase space point cloud set, and the closed-loop feature values ​​of the point cloud representing the generation and rupture cycle amplitude of bubbles are obtained. Based on the point cloud closed-loop feature value, the energy consumption in the sliding time window sequence, and the instantaneous total admittance drop, the tissue dehydration characteristic index is obtained; Based on real-time monitored instantaneous total admittance, the decision-making logic is determined by combining the instantaneous total admittance drop in the sliding time window sequence and the tissue dehydration characteristic index, and an energy processing command is output.

2. The intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback as described in claim 1, characterized in that, The method for obtaining the two-dimensional phase space point cloud set is as follows: Obtain the maximum and minimum values ​​of the instantaneous total power within the sliding time window sequence, and use them as the first power and the second power, respectively. Obtain the maximum and minimum values ​​of the instantaneous total admittance within the sliding time window sequence, and use them as the first admittance and the second admittance, respectively. When the difference between the first power and the second power is greater than or equal to the preset power fluctuation threshold, the difference between the total instantaneous power and the second power in each instantaneous time window sequence is divided by the sum of the difference between the first power and the second power and the first preset minimum positive number to obtain the normalized result of the total instantaneous power. The difference between the total instantaneous admittance and the second admittance in each sliding time window sequence is divided by the sum of the difference between the first admittance and the second admittance and the second preset minimum positive number to obtain the normalized result of the total instantaneous admittance. The normalized instantaneous total power and normalized instantaneous total admittance corresponding to each sampling moment in the sliding time window sequence are combined to generate two-dimensional coordinate points. Multiple consecutive two-dimensional coordinate points are then aggregated to form a two-dimensional phase space point cloud set.

3. The intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback as described in claim 1, characterized in that, The method for obtaining the closed-loop feature values ​​of the point cloud is as follows: A gridded downsampling process is performed on a two-dimensional phase space point cloud set to obtain a simplified point cloud subset; One-dimensional continuous homology features of a subset of simplified point clouds are calculated based on the Vietoris-Rips algorithm to obtain a one-dimensional continuous homology barcode. Extract the dominant topological feature term with the longest lifespan from the one-dimensional continuous homology barcode, and obtain the generation filter distance corresponding to its birth time and the closure filter distance corresponding to its death time. The result of subtracting the generated filter distance from the closed filter distance is used as the point cloud closed-loop feature value characterizing the magnitude of bubble generation and rupture cycles.

4. The intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback as described in claim 1, characterized in that, The method for obtaining the tissue dehydration characteristic index is as follows: The sum of the instantaneous total power at each sampling moment in the sliding time window sequence is divided by the preset capacity to obtain the average energy consumption of the window. Based on the instantaneous total admittance drop in the sliding time window sequence, obtain the smoothed admittance drop value; The negative product of the point cloud closed-loop feature value and the preset sensitivity coefficient is used as the exponent of the natural constant to construct an exponential function and obtain the nonlinear penalty factor. When the smoothed admittance drop value is greater than the preset significant drop threshold, the product of the window average energy consumption and the nonlinear penalty factor is divided by the sum of the smoothed admittance drop value and the third preset minimum positive number to obtain the tissue dehydration characteristic index. When the smooth admittance drop value is less than or equal to the preset significant drop threshold, 0 is used as the tissue dehydration characteristic index.

5. The intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback as described in claim 4, characterized in that, The method for obtaining the smoothed admittance drop value is as follows: The average of the earliest stored, preset number of instantaneous total admittances in the sliding time window sequence is used as the early average admittance. The average of the most recently stored, preset number of instantaneous total admittances in the sliding time window sequence is used as the recent average admittance. The difference between the early average admittance and the recent average admittance is used as the smoothed admittance drop value.

6. The intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback as described in claim 5, characterized in that, The method for determining the energy processing command based on real-time monitored instantaneous total admittance, combined with the instantaneous total admittance drop in the sliding time window sequence and the tissue dehydration characteristic index, is as follows: Calculate the reciprocal of the instantaneous total admittance at the current sampling moment, and use it as the current instantaneous equivalent total impedance; When the current instantaneous equivalent total impedance is less than the preset absolute impedance monitoring threshold, an energy maintenance command is sent to the RF generator. When the current instantaneous equivalent total impedance is greater than or equal to the preset absolute impedance monitoring threshold, and the smooth admittance drop value is less than or equal to zero, an energy maintenance command is sent to the RF generator. When the current instantaneous equivalent total impedance is greater than or equal to the preset absolute impedance monitoring threshold, and the smooth admittance drop value is greater than zero, compare the tissue dehydration characteristic index with the preset deep penetration judgment threshold: If the tissue dehydration characteristic index is less than the preset deep penetration judgment threshold, the current impedance increase is determined to be a false high impedance caused by bubble isolation, and an energy maintenance command is sent to the radio frequency generator. If the tissue dehydration characteristic index is greater than or equal to the preset deep penetration judgment threshold, it is determined that the target tissue has undergone substantial dehydration and thermal penetration, and an energy cutoff command is sent to the radio frequency generator.

7. The intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback as described in claim 1, characterized in that, The method for obtaining the instantaneous total power is as follows: In the radio frequency circuit, the instantaneous effective values ​​of voltage and current are synchronously sampled and processed by envelope detection or root mean square conversion. The product of the instantaneous effective value of voltage and the instantaneous effective value of current at each sampling time is taken as the instantaneous total power at each sampling time.

8. The intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback as described in claim 7, characterized in that, The method for obtaining the instantaneous total admittance is as follows: Divide the effective value of the instantaneous current at each sampling moment by the effective value of the instantaneous voltage to obtain the instantaneous total admittance at each sampling moment.

9. The intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback as described in claim 1, characterized in that, The method for constructing the sliding time window sequence is as follows: The data pairs consisting of instantaneous total power and instantaneous total admittance are stored using a first-in-first-out queue with a preset capacity; Once the total number of data pairs stored in the first-in-first-out queue reaches a preset capacity, whenever a new data pair is stored, the oldest historical data pair in the first-in-first-out queue is removed to obtain a sliding time window sequence consisting of a preset capacity of data pairs.

10. A smart optimization system for endometrial radiofrequency ablation parameters based on real-time impedance feedback, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent optimization method for endometrial radiofrequency ablation parameters based on real-time impedance feedback as described in any one of claims 1-9.

Images