Cervical chemotherapy injection intelligent positioning method and system

Abstract

The invention discloses an intelligent positioning method and system for chemotherapy injection for cervix uteri, and relates to the technical field of intelligent positioning, the method comprises the following steps: scanning through CT equipment to obtain a cervix uteri image, transmitting the image back to a processor, and combining a static positioning branch of an intelligent positioning module to make a positioning decision based on multi-scale triangular mesh division and target injection, and determining an injection strategy. The injection mode is single-needle or multi-needle cooperative injection. And transmitting the positioning strategy to an injection navigation system to guide and control the injector. Under auxiliary CT scanning monitoring, when motion deformation occurs in a target area, a new scanning image is returned, and strategy adjustment is carried out in combination with the dynamic adjustment branch. And feeding back the adjusted injection strategy to an injection navigation system, and carrying out positioning feedback regulation and control. The technical problem of injection deviation caused by inaccurate positioning and deformation of the target region in the cervical chemotherapy injection process in the prior art is solved, and the technical effects of accurately positioning the cervical target region and adapting to deformation of the target region in real time are achieved.

Description

Technical Field

[0001] The present application relates to the field of intelligent positioning technology, and in particular to an intelligent positioning method and system for cervical chemotherapy injection. Background Art

[0002] With the continuous development of medical imaging technology, especially CT scanning technology, the early diagnosis rate and treatment of gynecological diseases such as cervical cancer have been significantly improved. However, during the cervical chemotherapy injection process, injection deviation caused by inaccurate target positioning and target deformation remains a challenge. The traditional cervical chemotherapy injection method relies on the doctor's experience to determine the injection point, but due to the complex morphology of the cervical area and changes in patient position, positioning errors often occur, affecting the treatment effect. In addition, the cervical target area may be displaced when the patient's posture changes or the tumor deforms, which makes accurate injection difficult, and lacks an effective dynamic adjustment mechanism to cope with these changes. To this end, a new technical solution is urgently needed to accurately locate and dynamically adjust the target area through intelligent means to ensure that chemotherapy drugs can be accurately injected into the target area, thereby improving the accuracy and effect of treatment.

[0003] At present, relevant technologies still have technical problems such as inaccurate positioning and target area deformation during cervical chemotherapy injection, leading to injection deviation. Summary of the invention

[0004] The present application solves the technical problems of inaccurate positioning and target area deformation leading to injection deviation during cervical chemotherapy injection in the prior art by providing an intelligent positioning method and system for cervical chemotherapy injection.

[0005] The present application provides an intelligent positioning method for chemotherapy injection in the cervix, comprising:

[0006] The cervical scan image of the target user is obtained by scanning with a CT device; the cervical scan image is transmitted back to the processor, and combined with the static positioning branch of the intelligent positioning module, the cervical scan image is subjected to multi-scale triangulated mesh division and target injection positioning decision-making to determine the injection positioning strategy, wherein the intelligent positioning module is built into the processor, and includes a static positioning branch and a dynamic adjustment branch with lateral interaction, and the injection method is single-needle injection or multi-needle collaborative injection; a connection is established between the intelligent positioning module and the injection navigation system, and between the injection navigation system and the syringe, and the injection positioning strategy is transmitted to the injection navigation system to guide and control the syringe; auxiliary CT scanning monitoring, when there is target area motion deformation, the target area scan image is transmitted back, and combined with the dynamic adjustment branch of the intelligent positioning module, an injection strategy adjustment decision is made to determine the injection adjustment strategy; the injection adjustment strategy is fed back to the injection navigation system to perform positioning feedback control on the syringe.

[0007] The present application provides an intelligent positioning system for chemotherapy injection for cervix, comprising:

[0008] A cervical scanning image acquisition module, the cervical scanning image acquisition module is used to scan through a CT device to obtain a cervical scanning image of a target user; an injection positioning strategy determination module, the injection positioning strategy determination module is used to transmit the cervical scanning image back to the processor, and in combination with the static positioning branch of the intelligent positioning module, the cervical scanning image is subjected to multi-scale triangulated mesh division and target injection positioning decision-making to determine the injection positioning strategy, wherein the intelligent positioning module is built into the processor, and includes a static positioning branch and a dynamic adjustment branch with lateral interaction, and the injection method is single-needle injection or multi-needle coordinated injection; a guidance control module, the The guidance control module is used to establish the connection between the intelligent positioning module and the injection navigation system, and the connection between the injection navigation system and the syringe, transmit the injection positioning strategy to the injection navigation system, and guide and control the syringe; the injection adjustment strategy determination module, the injection adjustment strategy determination module is used to assist CT scanning monitoring, when there is target area motion deformation, the target area scanning image is returned, combined with the dynamic adjustment branch of the intelligent positioning module, the injection strategy adjustment decision is made, and the injection adjustment strategy is determined; the feedback control module, the feedback control module is used to feed back the injection adjustment strategy to the injection navigation system, and perform positioning feedback control on the syringe.

[0009] The intelligent positioning method and system for chemotherapy injection for the cervix proposed in this application first obtains cervical images through scanning with a CT device, and transmits the images back to the processor, and combines the static positioning branch of the intelligent positioning module to make decisions based on multi-scale triangular mesh division and target injection positioning to determine the injection strategy. The intelligent positioning module is built into the processor, and includes static positioning and dynamic adjustment branches. The injection method is single-needle or multi-needle collaborative injection. The intelligent positioning module is connected to the injection navigation system, and the positioning strategy is transmitted to the injection navigation system to guide and control the syringe. Under the auxiliary CT scanning monitoring, when the target area undergoes motion deformation, the new scanning image is transmitted back and the strategy is adjusted in combination with the dynamic adjustment branch. Finally, the adjusted injection strategy is fed back to the injection navigation system for positioning feedback regulation. By combining CT scanning and dynamic adjustment mechanism, the technical effect of accurately positioning the cervical target area and adapting to the deformation of the target area in real time is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solution of the embodiment of the present disclosure, the accompanying drawings of the embodiment of the present disclosure will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0011] Figure 1 A schematic diagram of a process flow of an intelligent positioning method for cervical chemotherapy injection provided in an embodiment of the present application;

[0012] Figure 2 A schematic diagram of the structure of an intelligent positioning system for cervical chemotherapy injection provided in an embodiment of the present application.

[0013] Explanation of the reference numerals: cervical scanning image acquisition module 10 , injection positioning strategy determination module 20 , guidance control module 30 , injection adjustment strategy determination module 40 , feedback control module 50 . DETAILED DESCRIPTION

[0014] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0015] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0016] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0017] The present application embodiment provides an intelligent positioning method for chemotherapy injection in the cervix, such as Figure 1 As shown, the method includes:

[0018] Step S100, obtain the target user's cervical scan image by scanning with a CT device. Specifically, using a CT device to obtain the target user's cervical scan image is a key starting step in the intelligent positioning scheme for cervical chemotherapy injection. Before scanning, medical staff need to fully communicate with the patient, explain the process and precautions in detail, and guide the patient to take a proper position with the legs bent and abducted and remain stable in the supine position. At the same time, the CT equipment is comprehensively inspected and debugged, and the scanning range, layer thickness, layer spacing, voltage, current and other parameters are reasonably set according to the cervical scanning requirements. It is also necessary to evaluate whether to use contrast agents. If used, explain the relevant situation to the patient and inject according to the procedures. During scanning, start the scanning program, and the equipment rotates and scans around the cervix according to the preset parameters. The operator monitors the patient's status and equipment operation in real time, and suspends processing in time if there is any abnormality. After the scan is completed, the collected raw data is transmitted to the computer system, reconstructed into a visual image through a specific algorithm, and stored in the image storage system. The operator preliminarily evaluates the image quality and handles any problems in a timely manner. Finally, the image is quickly and accurately transmitted back to the processor with a built-in intelligent positioning module through the network and other means, providing data support for subsequent multi-scale triangulated mesh division and target injection positioning decision-making in combination with the static positioning branch, ensuring data integrity and security during transmission.

[0019] Step S200, the cervical scan image is transmitted back to the processor, and combined with the static positioning branch of the intelligent positioning module, the cervical scan image is subjected to multi-scale triangular mesh division and target injection positioning decision, and the injection positioning strategy is determined, wherein the intelligent positioning module is built into the processor, and includes a static positioning branch and a dynamic adjustment branch with lateral interaction, and the injection method is single-needle injection or multi-needle collaborative injection. Specifically, after receiving the cervical scan image, the processor first pre-processes the image data, removes noise, enhances contrast and standardizes it to meet the calculation requirements of the intelligent positioning module. Then, the static positioning branch of the intelligent positioning module is activated, and the injection positioning sample library is called. The visual features of the tumor are mined from the sample to construct the first feature library, and the geometric features of the tumor and the injection parameters are mined to construct the second feature library. The first and second grid sizes of different precisions are set, and the tumor target area boundary is used as the initial position. The tumor target area is divided in detail with the first grid size, and the non-tumor target area is divided with the second grid size, and the image triangular grid is determined. Then, the first feature library is used as the recognition target and the multi-scale triangular mesh is used as the segmentation target for one-step training to optimize the model's ability to identify tumor visual features; the second feature library is used as the decision target and the image triangular mesh is used as the parameter generation standard for two-step training to learn and generate appropriate injection parameters. Finally, the injection targeting characteristics are determined based on the injection method and the physical and chemical properties of the injection agent, and the injection conditions are clarified. With this as a constraint, the comprehensive and accurate injection positioning strategy is determined in combination with the training results, and the parameters such as syringe positioning, angle, depth, and multi-needle injection needle spacing are clarified to provide guidance for the injection navigation system.

[0020] In one possible implementation, the cervical scan image is transmitted back to the processor, and combined with the static positioning branch of the intelligent positioning module, the cervical scan image is subjected to multi-scale triangulated mesh division and target injection positioning decision-making to determine the injection positioning strategy, wherein the intelligent positioning module is built into the processor, and includes a static positioning branch and a dynamic adjustment branch with lateral interaction, and the injection method is single-needle injection or multi-needle collaborative injection, and step S200 further includes step S210, calling injection positioning samples, mining tumor visual features, constructing a first feature library, mining tumor geometric features-injection parameters, and constructing a second feature library, wherein the tumor visual features include the tumor geometric features. Specifically, in order to achieve precise positioning of cervical chemotherapy injections, firstly, injection positioning samples of patients of different age groups, disease severity, and tumor morphology are widely collected, and patient information, tumor data, and injection operation records are annotated in detail to ensure that the samples are diverse and complete. Next, the image analysis algorithm is used to process the cervical images in the sample. After enhancing the contrast and clarity, the visual features of the tumor are mined through edge detection, shape and texture extraction techniques, including shape, texture and geometric features as part of it (such as size, position, and boundaries). These features are sorted and classified and stored in a specific data structure to build a first feature library for identifying unknown image tumors. Then, the relationship between the geometric features of the tumor and the injection parameters is deeply analyzed to accurately determine the three-dimensional spatial position of the tumor, measure the volume shape, and associate the injection parameters (the three-dimensional position, angle, and burial depth of the target, and the target spacing in the case of multiple injections). In order to solve the problem of inconsistent reference surfaces for injection parameters, the cervical area is gridded, and a unified reference baseline is established with a fixed anatomical structure as the origin and the grid coordinate axis as the reference direction. After analyzing a large number of samples, the mapping relationship between the two is summarized and organized into data pairs to build a second feature library that provides a basis for determining injection parameters, laying a solid data and analysis foundation for subsequent intelligent positioning decisions.

[0021] Step S220, wherein the injection parameters at least include the three-dimensional position of the target, the three-dimensional angle of the target, and the burial depth. If it is a multi-needle collaborative injection, the injection parameters also include the target spacing. Specifically, in cervical chemotherapy injection, accurately determining the injection parameters plays a key role in the treatment effect. When determining the three-dimensional position of the target, first obtain the cervical image through CT scanning and send it back to the processor for pre-processing, combine the static positioning branch of the intelligent positioning module, and use the first and second feature libraries constructed by the injection positioning sample to outline the tumor contour through image processing technology, and accurately locate the tumor based on the geometric features of the tumor. To determine the three-dimensional angle of the target, it is necessary to comprehensively consider the relationship between the tumor and the surrounding tissues, use the relationship between the geometric features and the injection angle in the second feature library, and combine the current tumor geometric feature analysis to determine the precise angle within the appropriate range, and also need to be fine-tuned through computer simulation. Determine the burial depth based on the depth of the tumor in the cervical tissue and the treatment needs of chemotherapy drugs, combine the characteristics of the drug to determine the initial depth, and then refer to the second feature library and consider the physiological differences of the patient's cervical tissue for adjustment. If multi-needle coordinated injection is used to determine the target spacing, the tumor volume and shape should be accurately assessed first, and the spacing should be calculated based on the tumor condition, drug diffusion characteristics, and multi-needle injection information from the second feature library. This should be optimized through computer simulation and verified with reference to successful cases, thereby providing precise parameter support for cervical chemotherapy injections and improving the accuracy and effectiveness of treatment.

[0022] Step S230, taking the first feature library as the identification target, taking the multi-scale triangular mesh based on the tumor target area as the initial position as the division target, and performing one-step training based on the injection positioning sample. Specifically, when constructing the intelligent positioning system for cervical chemotherapy injection, firstly, comprehensively comb through the collected injection positioning samples to ensure that all kinds of cervical tumor cases are covered and relevant information is marked in detail, and at the same time, check and optimize the first feature library. Then, according to the complexity and accuracy requirements of the tumor target area, set the high-precision first grid size and the relatively low-precision second grid size, select the representative boundary position of the tumor target area as the initial position, and use the professional grid division algorithm to perform three-dimensional multi-scale triangular grid division on the tumor target area and the surrounding area to ensure the grid quality. Then, select a suitable machine learning training algorithm such as convolutional neural network (CNN), set the training parameters such as learning rate, number of iterations, batch size, etc., input the multi-scale triangular mesh data and the first feature library into the model for training, and the model continuously adjusts the parameters during training to learn and identify the visual features of the tumor. After each iteration, the performance is evaluated and the parameters are adjusted according to the loss function. During the training period, validation set samples are also used regularly to evaluate the model's accuracy, recall rate and other indicators. If problems such as overfitting or underfitting occur, the training parameters are adjusted or the model structure is improved in a timely manner, thereby effectively completing one step of training and providing accurate model support for subsequent cervical chemotherapy injection positioning decisions.

[0023] Step S240, taking the second feature library as the decision target, taking the image triangular mesh as the parameter generation standard, and performing two-step training based on the injection positioning sample to determine the static positioning branch. Specifically, in order to determine the static positioning branch of the intelligent positioning of cervical chemotherapy injection, it is necessary to rigorously carry out two-step training with the second feature library as the decision target and the image triangular mesh as the parameter generation standard. First, make preliminary preparations, sort out the second feature library, ensure that the data associated with the geometric features of the tumor and the injection parameters are complete and accurate, and timely supplement and correct missing or erroneous data; clarify the image triangular mesh, review and optimize its division accuracy and range, so that it accurately reflects the spatial structure; organize the injection positioning samples, ensure that the samples are diverse and representative, and group them according to tumor characteristics and injection methods. Then implement two-step training, select a suitable machine learning algorithm model according to the characteristics of the task and data, such as a decision tree, support vector machine or deep learning model, set the key training parameters such as learning rate, number of iterations, regularization parameters, and input the samples, feature library and triangular mesh into the model training, so that it learns to generate injection parameters from the geometric features of the tumor, and continuously iterates and adjusts its own parameters to improve accuracy. Finally, the training results are evaluated and optimized, and scientific evaluation indicators such as mean square error, accuracy, and recall rate are selected. The model is regularly evaluated with validation set samples. Based on the evaluation results, if it is overfitting, regularization is enhanced and model complexity is reduced. If it is underfitting, training data is increased, parameters are adjusted, or the model is replaced. After multiple optimizations, a reliable static positioning branch is determined to provide strong support for the precise positioning of cervical chemotherapy injections.

[0024] In a possible implementation, the first feature library is used as the recognition target, the multi-scale triangular grid based on the tumor target area as the initial position is used as the division target, and one-step training is performed based on the injection positioning sample, and step S230 further includes step S231, setting the first grid size and the second grid size, wherein the accuracy of the first grid size is higher than that of the second grid size. Specifically, when constructing an intelligent positioning system for cervical chemotherapy injection, setting a reasonable grid size is extremely critical for accurate analysis and subsequent treatment. First, in-depth research on clinical needs and positioning goals is conducted to clarify that the tumor target area requires high accuracy to capture subtle features, and the accuracy requirements of non-tumor target areas are relatively low. Subsequently, relevant medical image processing, tumor positioning research and other materials and past cases are consulted, and the methods and effects of similar grid division applications are referred to. Based on this information, the first grid size is preliminarily set to the millimeter level, such as 0.5 mm, for accurate presentation of tumor details; the second grid size is set to the centimeter level, such as 1 cm, taking into account the spatial structure analysis of non-tumor target areas and improving computational efficiency. Next, simulation analysis is performed based on the initially set size using simulation software or sample data to observe the effect of dividing the tumor target area and the non-tumor target area, and the rationality is quantified by calculating evaluation indicators such as the tumor boundary fitting error and the completeness of the spatial description of the non-tumor area. According to the evaluation results, if the tumor boundary fitting error is large, the first grid size is reduced, and if the non-tumor area description is rough, the second grid size is increased, and the optimization is repeated. After multiple rounds of adjustments to achieve the ideal effect, the final first and second grid sizes are determined and applied to subsequent work.

[0025] Step S232, taking any boundary position of the tumor target area as the initial position, dividing the tumor target area based on the first grid size, dividing the non-tumor target area in three dimensions with the second grid size, and determining the image triangular grid. Specifically, in the intelligent positioning of cervical chemotherapy injection, determining the image triangular grid is a key step. First, comprehensively analyze the overall morphology of the tumor and its relative position relationship with the surrounding tissues, and select an initial position on the boundary of the tumor target area that can represent the main features and is convenient for expansion and division, such as selecting a circumferential protrusion for a circular tumor, and selecting a boundary point that reflects the contour change for an irregular tumor. Then, taking the initial position as the starting point, using professional algorithms such as the Delaunay triangulation algorithm or the advancing wavefront method, the tumor target area is divided according to the high-precision first grid size. Pay close attention to the boundary details and internal structure during division. If a blood vessel is encountered, adjust the grid appropriately to provide accurate data for analyzing the geometric characteristics of the tumor. After the tumor target area is divided, the non-tumor target area is divided three-dimensionally using a suitable algorithm with the second grid size starting from its boundary. The overall spatial structure and the relative position with the tumor target area are emphasized during the division. The grid is appropriately encrypted for non-tumor tissues close to the tumor, and conventional division is performed for areas far away. Finally, the grids of the tumor target area and non-tumor target area are integrated, the connection parts are checked to ensure a natural transition, and then optimized to remove redundant nodes and adjust the grid shape and size, so as to determine the complete image triangular grid, providing comprehensive and accurate spatial information for subsequent injection parameter generation and positioning decisions.

[0026] Step S233, wherein the injection parameters are converted and generated with the image triangular grid as a reference baseline. Specifically, in cervical chemotherapy injection, the injection parameters generated with the image triangular grid as a reference baseline are the key to ensuring accurate injection of chemotherapy drugs and determining the treatment effect. First, based on tumor analysis and clinical needs, the target position is determined in the tumor area by integrating the size, shape, internal structure and relationship with normal tissues of the tumor, and its coordinates in the grid are clarified with the help of the image triangular grid. Then, the injection angle is generated with the grid line where the target is located as a reference. The grid unit coordinate system where the target is located is first determined and associated with the whole. Then, according to the situation of the tumor and surrounding tissues, the spatial vector relationship between the target and the surrounding grid points is analyzed, and the angle is adjusted in combination with clinical experience and similar case data to ensure accurate insertion and reduce damage to normal tissues. Then, the relationship between the target in the grid coordinates and the grid space position is used, and the accuracy and splicing relationship are considered to calculate the accurate three-dimensional position of the target. The burial depth is determined based on the depth of the target in the tumor and the surrounding tissue conditions, combined with tumor characteristics and past case data. If multi-needle coordinated injection is used, the distribution of multiple targets is planned in the image triangle grid according to the shape and size of the tumor, and the grid unit spacing where the targets are located is calculated. Combined with drug diffusion and synergy requirements, the target spacing is determined and optimized with reference to successful cases and research data, providing precise parameters for cervical chemotherapy injections and improving the accuracy and effectiveness of treatment.

[0027] In a possible implementation, the cervical scan image is transmitted back to the processor, and the cervical scan image is subjected to multi-scale triangulated mesh division and target injection positioning decision based on the static positioning branch of the intelligent positioning module to determine the injection positioning strategy, wherein the intelligent positioning module is built in the processor, and includes a static positioning branch and a dynamic adjustment branch with lateral interaction, and the injection method is single-needle injection or multi-needle collaborative injection. Step S200 further includes step S250, according to the injection positioning sample, mining a third feature library based on motion deformation trend-injection parameter adjustment trend. Specifically, mining the third feature library based on motion deformation trend-injection parameter adjustment trend requires close cooperation of multiple steps. First, widely collect injection positioning samples, covering different patients and tumor conditions, record injection parameters, tumor images at different time points and physiological states in detail, and treatment feedback data, and store them in categories after sorting. Then carry out tumor motion deformation trend analysis, use professional software and algorithms to reduce image noise, enhance contrast, and align, extract feature points to track position changes, quantify parameters such as displacement, rotation, and volume change, and classify motion modes according to motion amplitude, frequency, etc. Then, we conducted an analysis of the injection parameter adjustment trend, sorted out the parameter adjustment records, associated tumor motion deformation with parameter adjustment, established a mathematical model to find the internal connection, and summarized the parameter adjustment strategies under different motion deformation modes. Finally, we built a third feature library, designed a reasonable storage structure, entered data and strictly verified it, updated and maintained it regularly, added new motion deformation modes and parameter adjustment strategies, and improved the existing data, thus providing strong support for the dynamic adjustment of cervical chemotherapy injection.

[0028] Step S260, a temporary data area is built in, the third feature library is used as the decision target, the image triangular mesh is used as the generation standard for parameter adjustment, and training is performed based on the injection positioning sample to determine the dynamic adjustment branch, wherein the temporary data area is used to store the injection positioning strategy generated by the static positioning branch. Specifically, a dynamic adjustment branch is constructed in the cervical chemotherapy injection system, and a temporary data area with fast reading and writing and efficient storage characteristics is first opened in the core processor to store the injection positioning strategy generated by the static positioning branch, such as the initial target position, injection angle and depth, so that it can be called at any time. Then, training preparation is carried out, and the third feature library storing the relationship between the motion deformation trend and the injection parameter adjustment trend is clearly used as the decision target, and the image triangular mesh presenting the spatial structure of the tumor and surrounding tissue is used as the parameter adjustment generation standard. At the same time, the injection positioning samples are sorted and classified according to the tumor motion deformation factors. Then, training is started, and a machine learning algorithm suitable for dynamic adjustment tasks such as a deep Q network is selected, and hyperparameters such as the learning rate and discount factor are initialized, and the samples, feature library and triangular mesh data are input into the model. In each iteration, the model generates injection parameter adjustment suggestions based on tumor motion deformation information, feature library and triangular mesh, and calculates the loss function by comparing the actual records, and adjusts the parameters accordingly. During training, the validation set is used regularly to evaluate the accuracy of parameter adjustment, the degree of improvement in treatment effect, generalization ability and other indicators. According to the evaluation results, if the accuracy is low, the hyperparameters are adjusted, and if the generalization ability is poor, the sample diversity is increased or regularization technology is used. After multiple optimizations, reliable dynamic adjustment branches are determined to ensure accurate treatment of cervical chemotherapy injections.

[0029] In a possible implementation, the cervical scan image is transmitted back to the processor, and the cervical scan image is subjected to multi-scale triangulated mesh division and target injection positioning decision based on the static positioning branch of the intelligent positioning module to determine the injection positioning strategy, wherein the intelligent positioning module is built into the processor, and includes a static positioning branch and a dynamic adjustment branch with lateral interaction, and the injection method is single-needle injection or multi-needle coordinated injection, and step S200 further includes step S270, and the injection targeting feature is determined according to the physicochemical properties of the injection, wherein the injection targeting feature includes the directional position and non-directional position of the target area. Specifically, when designing an injection treatment plan, it is key to accurately determine the injection targeting feature, which is based on the physicochemical properties of the injection. First, the physicochemical property data of the injection are comprehensively collected, covering chemical composition analysis, clarifying the main ingredients, active substances and excipients, understanding the chemical structure judgment of the active substance and the mode of action of the target cell; physical property determination is performed to grasp the influence of molecular weight, solubility, etc. on the in vivo behavior of the drug; stability research is carried out to clarify the validity period and storage conditions. Then analyze the interaction between drugs and target areas, conduct in-depth research on the mechanism of drug action, and understand its treatment methods for tumor cells; explore the characteristics of tumor cells and master information such as surface markers; based on this, determine the directional position, use immunohistochemistry and other technologies to locate the distribution of receptors on the surface of tumor cells to determine the specific injection position, and for drugs whose mechanism of action does not depend on a specific position, combine the size and shape of the tumor to determine the non-directional position. Finally, integrate the directional and non-directional position information, draw a three-dimensional tumor model map to mark the position, record relevant parameters such as drug concentration and action time, and form a detailed injection targeting feature report, which provides a key basis for the formulation of subsequent injection treatment plans.

[0030] Step S280, determine the injection conditions, wherein the injection conditions are long-time micropump injection or short-time instant injection. Specifically, determining the injection conditions is critical to ensuring the treatment effect, and it is necessary to comprehensively consider various factors to decide whether to use long-time micropump injection or short-time instant injection. First, assess the patient's condition and treatment needs, analyze the tumor type, stage, growth rate and diffusion range, consider the patient's physical conditions such as liver and kidney function, cardiopulmonary function, nutritional status, and clarify the treatment goals such as radical, palliative or preventive. Then study the characteristics of the injection, analyze the metabolic kinetic characteristics of drug absorption, distribution, metabolism, excretion, etc., pay attention to its stability in vivo and in vitro, and judge according to the drug mechanism of action. Then weigh the advantages and disadvantages of the two injection methods. Long-time micropump injection can maintain stable blood drug concentration and realize personalized treatment, but the operation time is long, restricts patient activities, has high cost and has the risk of abnormal infusion; short-time instant injection is simple and fast to operate, can take effect quickly, but may lead to excessive drug concentration and require frequent injection. Finally, a decision is made based on the patient's condition, treatment needs, injection characteristics, and the advantages and disadvantages of the two methods. A detailed injection plan and emergency plan including parameters such as time, dosage, and speed are formulated. Full communication is then carried out with the patient and his or her family to ensure that the patient understands and cooperates, and to ensure smooth treatment.

[0031] Step S290, taking the injection targeting feature and the injection condition as constraints, the target injection positioning decision is made. Specifically, the target injection positioning decision takes the injection targeting feature and the injection condition as key constraints. First, the injection targeting feature is comprehensively sorted out, the key area of ​​the directional position and the range of the non-directional position are clarified, and the injection conditions are confirmed at the same time, such as the flow rate and interval of the long-time micropump injection, the dose and speed of the short-time instant injection, and then the two are associated, such as considering the precise delivery of the drug to the directional position during the long-time micropump injection. Then, according to the integrated information, a suitable method is selected to establish a decision model, determine the input and output, and use a machine learning algorithm to train with a large number of historical cases and optimize with a validation set. Then, through computer simulation evaluation, taking into account factors such as breathing and body movement, a team of clinicians and imaging experts will be formed to evaluate, and the decision will be adjusted according to the evaluation results. If there is a risk of injury, the coordinate angle will be readjusted, and if there is a dose adjustment suggestion, the parameters will be modified. Finally, the decision is converted into a detailed implementation plan, covering the operation steps, equipment and instruments, and personnel division of labor. A monitoring mechanism is also established to ensure the safety and smoothness of the injection with the help of imaging equipment and vital signs monitoring.

[0032] Step S300, establish the connection between the intelligent positioning module and the injection navigation system, the connection between the injection navigation system and the syringe, transmit the injection positioning strategy to the injection navigation system, and guide and control the syringe. Specifically, to achieve precise injection, it is necessary to establish the connection between the intelligent positioning module, the injection navigation system and the syringe and transmit the injection positioning strategy. In terms of hardware connection, select the adapter cable according to the interface type, connect the intelligent positioning module to the injection navigation system, and then connect the injection navigation system to the syringe. After the connection, perform hardware self-test and functional test respectively. At the software level, set a unified communication protocol in the software system of each device, specify parameters such as data format, encoding, and verification, and perform data transmission test and optimization after completion. When transmitting the strategy, the intelligent positioning module generates and organizes the strategy containing information such as target coordinates and injection angle, and transmits it to the injection navigation system through an encrypted link in the protocol format. After receiving, the injection navigation system parses the strategy and converts it into a syringe control instruction, guides in real time during injection, monitors and adjusts the syringe position according to the positioning information, receives feedback on its working status, and also has a safety mechanism to check parameters before injection, monitor the patient's physiological indicators during injection, and stop and deal with emergency immediately in case of abnormality, so as to ensure the safety and accuracy of the entire injection process.

[0033] In a possible implementation, the connection between the intelligent positioning module and the injection navigation system is established, and the connection between the injection navigation system and the injector is established, and the injection positioning strategy is transmitted to the injection navigation system to guide and control the injector. Step S300 further includes step S310, using the image triangular mesh to determine the virtual positioning baseline. Specifically, when determining the virtual positioning baseline, the image triangular mesh is first preprocessed, and the noise, isolated points and wrongly connected meshes are removed by using a data cleaning algorithm. The mesh shape and quality are optimized by subdivision, merging, smoothing and other operations to better fit the tissue morphology. Then, feature points are extracted, and points with large changes in tumor boundary curvature, internal vascular bifurcation points, etc. are selected based on geometric features. Connected area boundary points, hole boundary points, etc. are extracted based on topological features. Then, a virtual positioning baseline is generated based on the feature points. When the feature point distribution is regular, the initial baseline is constructed by connecting the points using linear interpolation. If the distribution is complex, spline curve fitting optimization is used. Finally, the generated baseline is compared with the original image, and the visualization tool is used to check whether it fits the tumor boundary and passes through important feature points. If there is a deviation, the feature point selection and curve fitting parameters are rechecked, adjusted and optimized, and multiple verifications are performed to ensure that the baseline is accurate and reliable.

[0034] Step S320, taking the cervical region based on the cervical scan image as the physical space, the physical space and the virtual positioning baseline are combined to build a virtual reality space. Specifically, to build a virtual reality space that combines the virtual and the real, first use MRI or CT and other equipment to collect cervical region images, and after pre-processing such as noise reduction and image enhancement, use deep learning algorithms such as U-Net to segment the cervical region and complete three-dimensional reconstruction, accurately restore its size and spatial position. Then unify the coordinate system of the physical space (three-dimensional model of the cervical region) and the virtual positioning baseline, compare and adjust according to the key size of the cervix, and complete scale matching. Then, in a computer graphics environment, according to the unified coordinates and matching scale, embed the virtual positioning baseline into the three-dimensional cervical model, set the parameters such as lighting and material for rendering, and add interactive functions such as zooming and marking. Finally, check the fusion and function status from multiple perspectives, invite medical experts to evaluate, optimize the display, rendering and interaction according to the feedback, improve the space quality and application value, and provide strong support for medical operations such as cervical injection.

[0035] Step S330, for the virtual reality space, the injection navigation system executes the injection positioning strategy in response to the injection positioning strategy, and performs the positioning guidance control of the syringe. Specifically, in the virtual reality space, the injection navigation system needs to accurately control the syringe. First, it reads the injection positioning strategy data stored or transmitted, such as information stored in XML format, uses a parser to read key parameters, and then parses the semantics, converts the abstract strategy into instructions that the system can recognize, and matches it to the coordinate system of the virtual reality space. Then, the initial position of the syringe is determined and calibrated, and the A* algorithm and other algorithms are selected according to the target and the current position to plan the path, combined with the operation restrictions and the patient's physiological state optimization, and adjusted in real time according to the patient's state. Then, according to the planned path, motion control instructions such as motor rotation and piston propulsion are generated, transmitted to the syringe drive device, and sensor feedback is received in real time. The syringe state is compared with the preset monitoring state, and any deviation is corrected. The real-time position and parameter auxiliary operation are also displayed visually. Finally, safety mechanisms such as upper and lower limits of injection parameters and limited range of motion are set, and emergency response plans such as equipment failure and patient abnormality are formulated to ensure the safety and accuracy of the injection process in all aspects.

[0036] Step S400, assisting CT scanning monitoring, when there is target area motion deformation, the target area scanning image is transmitted back, combined with the dynamic adjustment branch of the intelligent positioning module, the injection strategy adjustment decision is made, and the injection adjustment strategy is determined. Specifically, in order to ensure the accuracy of injection treatment, the CT scanning equipment is reasonably deployed near the treatment area during treatment, and the scanning layer thickness, tube voltage and other parameters are accurately set according to the target area. During treatment, real-time scanning is performed at fixed intervals, and the image is transmitted to the data processing center using high-speed transmission technology. After the image is transmitted back, it is pre-processed first, and the quality is improved by image enhancement and noise reduction. Then, the image registration technology is used to compare with the reference image, and the displacement, rotation and other parameters are calculated to detect the target area motion deformation. Then the deformation information is transmitted to the dynamic adjustment branch of the intelligent positioning module, which integrates the third feature library, injection positioning samples and other data, uses the algorithm model to generate multiple injection strategy adjustment schemes, predicts the feasibility of the treatment effect evaluation, and comprehensively selects the optimal scheme as the injection adjustment strategy based on factors such as the treatment effect and patient impact. Finally, the strategy is transmitted to the injection device control system to adjust the injection parameters. After the injection, continuous CT scanning is performed to collect feedback on the treatment effect. If the effect is not good or there are new changes in the target area, feedback is sent to the dynamic adjustment branch to optimize the strategy again, forming a closed-loop control.

[0037] In a possible implementation, CT scanning monitoring is assisted. When there is target area motion deformation, the target area scan image is transmitted back, and the injection strategy adjustment decision is made in combination with the dynamic adjustment branch of the intelligent positioning module. The injection adjustment strategy is determined. Step S400 further includes step S410, setting the preset degrees of freedom of target area motion deformation. Specifically, to set the preset degrees of freedom of target area motion deformation, relevant information should be collected comprehensively. First, the patient's medical records are consulted to understand the past medical history, disease diagnosis, and especially the characteristics of the target area disease, such as the type and stage of the tumor; then, the past CT and MRI images are analyzed in depth, and the images of different periods are compared by image registration technology to understand the location, size and morphological changes of the target area; at the same time, clinical experience and medical research results are referred to to obtain the general law of target area movement. Then, the dimension and direction are determined, and the spatial dimension is clarified according to the target area location and treatment needs, such as front and back, left and right, up and down. The rotational freedom is also considered for special parts, and the target area movement direction is analyzed for each dimension, such as the chest target area mainly moves up and down. Finally, the values ​​are quantified and a large amount of case data is analyzed statistically to preliminarily estimate the range of values. Doctors, physicists and other experts then make adjustments based on the patient's condition assessment, simulate the target area movement before treatment or monitor it in real time in the early stages, and dynamically verify and optimize the preset degrees of freedom values.

[0038] Step S420, based on the preset degrees of freedom, set the image return instruction of the injection feedback. Specifically, in order to obtain images in time when the target area moves and deforms, the image return instruction of the injection feedback is set based on the preset degrees of freedom. First, clarify the trigger logic, take the preset degrees of freedom as the benchmark, and stipulate that the target area is triggered when the movement deformation in the three-dimensional space and rotation direction reaches or exceeds the corresponding degrees of freedom, while taking into account factors such as movement duration and change rate to avoid invalid triggering. Then design the instruction content structure, including key information such as trigger time, cause, motion parameters, and auxiliary information such as patient identification and treatment stage, to facilitate subsequent analysis and processing. Then, according to the hospital network architecture, select wired or wireless network transmission instructions, use TCP / IP protocol and optimize, to ensure data integrity and efficient transmission. Finally, build a simulation test environment, simulate various target area movement conditions to test instructions, optimize the trigger logic, content encoding, transmission method, etc. according to the test results, so that the instructions can play an accurate and timely role and help precision treatment.

[0039] Step S430, assisting CT scanning monitoring, performing target area motion deformation judgment, if the preset degree of freedom is met, generating the image return instruction, controlling the CT device to return the target area scan image to the dynamic adjustment branch. Specifically, when performing auxiliary CT scanning monitoring, first fully calibrate and maintain the CT scanning equipment to ensure the normal operation of each component, and use the calibration module to calibrate the image reconstruction algorithm. Then, according to the specific conditions of the target area and the clinical monitoring accuracy requirements, customize the scanning layer thickness, interval, tube voltage, tube current and other parameters. Then set the device to continuous scanning mode, scan uninterruptedly at the set time interval, and store the image in DICOM format locally after each scan and transmit it to the data processing center via a high-speed network. After the image is obtained, it is first pre-processed using image enhancement and noise reduction algorithms, and then the motion parameters of the target area in three-dimensional space are calculated using image registration and motion analysis algorithms, and compared with the preset degree of freedom. Once the preset degree of freedom conditions are met, an image return instruction containing key information is generated in a format, transmitted to the CT device control system through the network, and the device returns the image after receiving the command, and the verification and error correction technology is used to ensure the accuracy and completeness of the data. Finally, the dynamic adjustment branch receives the images and stores them in the database, which is combined with clinical information and treatment history analysis to adjust the injection treatment strategy accordingly to improve the accuracy and effectiveness of the treatment.

[0040] Step S500, the injection adjustment strategy is fed back to the injection navigation system to perform positioning feedback control on the syringe. Specifically, after receiving the injection adjustment strategy, the injection navigation system will deeply analyze the key information such as the injection target coordinates, angles, depths, speeds, etc., and convert them into executable control instructions, such as movement, rotation, propulsion, and speed instructions. Then, the user's posture is monitored using an accelerometer and a gyroscope, and the movement of the cervical target tissue is tracked with the help of ultrasound and MRI, and a feedback loop is established from the monitoring end to the injection navigation system and then to the syringe. The syringe adjusts the position, posture, injection speed, depth and other parameters in real time according to the instructions of the navigation system to cope with changes in the user's posture and target tissue. During the control process, the actual state information of the syringe is collected in real time through the sensor, compared with the instructions, and the control effect is evaluated. If there is a deviation, the causes such as sensor error and actuator accuracy are analyzed and optimized to ensure accurate injection.

[0041] The embodiment of the present application uses a CT device to scan and acquire cervical images, and transmits the images back to the processor, and combines the static positioning branch of the intelligent positioning module to make injection positioning decisions based on multi-scale triangular mesh division and target area, and determine the injection strategy. The intelligent positioning module is built into the processor, and includes static positioning and dynamic adjustment branches. The injection method is single-needle or multi-needle collaborative injection. The intelligent positioning module is connected to the injection navigation system, and the positioning strategy is transmitted to the injection navigation system to guide and control the syringe. Under the auxiliary CT scanning monitoring, when the target area undergoes motion deformation, the new scanning image is transmitted back and the strategy is adjusted in combination with the dynamic adjustment branch. Finally, the adjusted injection strategy is fed back to the injection navigation system for positioning feedback regulation. By combining CT scanning and dynamic adjustment mechanism, the technical effect of accurately locating the cervical target area and adapting to the deformation of the target area in real time is achieved.

[0042] In the above, refer to Figure 1 The intelligent positioning method for chemotherapy injection for cervix according to an embodiment of the present invention is described in detail. Figure 2 An intelligent positioning system for chemotherapy injection for the cervix according to an embodiment of the present invention is described.

[0043] The intelligent positioning system for chemotherapy injection for cervix according to the embodiment of the present invention is used to solve the technical problems of injection deviation caused by inaccurate positioning and target area deformation during cervical chemotherapy injection in the prior art, and achieves the technical effect of accurately positioning the cervical target area and adapting to the deformation of the target area in real time by combining CT scanning and dynamic adjustment mechanism. The intelligent positioning system for chemotherapy injection for cervix includes: a cervical scanning image acquisition module 10, an injection positioning strategy determination module 20, a guidance control module 30, an injection adjustment strategy determination module 40, and a feedback control module 50.

[0044] The cervical scan image acquisition module 10 is used to acquire the cervical scan image of the target user through scanning with a CT device.

[0045] The injection positioning strategy determination module 20 is used to transmit the cervical scanning image back to the processor, and in combination with the static positioning branch of the intelligent positioning module, perform multi-scale triangulated mesh division and target injection positioning decision on the cervical scanning image to determine the injection positioning strategy, wherein the intelligent positioning module is built into the processor, including a static positioning branch and a dynamic adjustment branch with lateral interaction, and the injection method is single-needle injection or multi-needle coordinated injection.

[0046] The guidance control module 30 is used to establish a connection between the intelligent positioning module and the injection navigation system, and between the injection navigation system and the syringe, transmit the injection positioning strategy to the injection navigation system, and guide and control the syringe.

[0047] The injection adjustment strategy determination module 40 is used to assist CT scanning monitoring. When there is motion deformation of the target area, the target area scanning image is transmitted back, and combined with the dynamic adjustment branch of the intelligent positioning module, an injection strategy adjustment decision is made to determine the injection adjustment strategy.

[0048] The feedback control module 50 is used to feed back the injection adjustment strategy to the injection navigation system to perform positioning feedback control on the syringe.

[0049] The specific configuration of the injection positioning strategy determination module 20 will be described in detail below. As described above, the cervical scan image is transmitted back to the processor, and combined with the static positioning branch of the intelligent positioning module, the cervical scan image is subjected to multi-scale triangulated mesh division and target injection positioning decision-making to determine the injection positioning strategy, wherein the intelligent positioning module is built into the processor, and includes a static positioning branch and a dynamic adjustment branch with lateral interaction, and the injection method is single-needle injection or multi-needle collaborative injection. The injection positioning strategy determination module 20 further includes: a first feature library construction unit, the first feature library construction unit is used to call injection positioning samples, mine tumor visual features, construct a first feature library, mine tumor geometric features-injection parameters, and construct a second feature library, wherein the tumor visual features include the injection parameters. The geometric features of the tumor; an injection parameter composition unit, wherein the injection parameters at least include the three-dimensional position of the target, the three-dimensional angle of the target, and the burial depth. If it is a multi-needle collaborative injection, the injection parameters also include the target spacing; a positioning sample training unit, wherein the positioning sample training unit is used to use the first feature library as the recognition target and the multi-scale triangular mesh based on the tumor target area as the initial position as the division target, and perform one-step training based on the injection positioning sample; a static positioning branch determination unit, wherein the static positioning branch determination unit is used to use the second feature library as the decision target and the image triangular mesh as the parameter generation standard, and perform two-step training based on the injection positioning sample to determine the static positioning branch.

[0050] Among them, taking the first feature library as the recognition target, taking the multi-scale triangular mesh based on the tumor target area as the initial position as the division target, and performing one-step training based on the injection positioning sample, the positioning sample training unit further includes: a grid size setting subunit, the grid size setting subunit is used to set the first grid size and the second grid size, wherein the accuracy of the first grid size is higher than that of the second grid size; an image triangular mesh determination subunit, the image triangular mesh determination subunit is used to take any boundary position of the tumor target area as the initial position, divide the tumor target area based on the first grid size, and perform three-dimensional division of the non-tumor target area with the second grid size to determine the image triangular mesh; an injection parameter conversion generation subunit, the injection parameter conversion generation subunit is used to convert and generate injection parameters with the image triangular mesh as the reference baseline.

[0051] Among them, the injection positioning strategy determination module 20 further includes: a third feature library mining unit, the third feature library mining unit is used to mine a third feature library based on motion deformation trend-injection parameter adjustment trend according to the injection positioning sample; a dynamic adjustment branch determination unit, the dynamic adjustment branch determination unit is used to have a built-in temporary data area, take the third feature library as the decision target, take the image triangular mesh as the generation standard for parameter adjustment, train based on the injection positioning sample, and determine the dynamic adjustment branch, wherein the temporary data area is used to store the injection positioning strategy generated by the static positioning branch.

[0052] Among them, the injection positioning strategy determination module 20 further includes: an injection targeting feature determination unit, the injection targeting feature determination unit is used to determine the injection targeting feature according to the physical and chemical properties of the injection, wherein the injection targeting feature includes the directional position and the non-directional position of the target area; an injection condition determination unit, the injection condition determination unit is used to determine the injection condition, wherein the injection condition is a long-time micropump injection, or a short-time instant injection; a target area injection positioning decision unit, the target area injection positioning decision unit is used to make a target area injection positioning decision based on the injection targeting feature and the injection condition as constraints.

[0053] The specific configuration of the guidance control module 30 will be described in detail below. As described above, the connection between the intelligent positioning module and the injection navigation system, the connection between the injection navigation system and the syringe are established, the injection positioning strategy is transmitted to the injection navigation system, and the syringe is guided and controlled. The guidance control module 30 further includes: a virtual positioning baseline determination unit, the virtual positioning baseline determination unit is used to determine the virtual positioning baseline with the image triangular grid; a virtual reality space construction unit, the virtual reality space construction unit is used to use the cervical area based on the cervical scan image as the physical space, and combine the physical space with the virtual positioning baseline to build a virtual reality space; a positioning guidance control unit, the positioning guidance control unit is used for the injection navigation system to execute the injection positioning strategy in response to the virtual reality space, and perform positioning guidance control of the syringe.

[0054] The specific configuration of the injection adjustment strategy determination module 40 will be described in detail below. As described above, in order to assist CT scanning monitoring, when there is target area motion deformation, the target area scanning image is transmitted back, and the injection strategy adjustment decision is made in combination with the dynamic adjustment branch of the intelligent positioning module to determine the injection adjustment strategy. The injection adjustment strategy determination module 40 further includes: a preset degree of freedom setting unit, which is used to set the preset degree of freedom of target area motion deformation; an image transmission instruction setting unit, which is used to set the image transmission instruction of injection feedback based on the preset degree of freedom; a target area motion deformation determination unit, which is used to assist CT scanning monitoring, perform target area motion deformation determination, and if the preset degree of freedom is met, generate the image transmission instruction, and control the CT device to transmit the target area scanning image back to the dynamic adjustment branch.

[0055] The intelligent positioning system for cervical chemotherapy injection provided by the embodiment of the present invention can execute the intelligent positioning method for cervical chemotherapy injection provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0056] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0057] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0058] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. An intelligent positioning method for chemotherapy injection in the cervix, characterized in that: The method comprises: Scanning with CT equipment to obtain the target user's cervical scan image; The cervical scan image is transmitted back to the processor, and combined with the static positioning branch of the intelligent positioning module, the cervical scan image is subjected to multi-scale triangulated mesh division and target injection positioning decision, and the injection positioning strategy is determined, wherein the intelligent positioning module is built into the processor, and includes a static positioning branch and a dynamic adjustment branch with lateral interaction, and the injection method is single-needle injection or multi-needle coordinated injection; Establishing a connection between the intelligent positioning module and the injection navigation system, and between the injection navigation system and the syringe, transmitting the injection positioning strategy to the injection navigation system, and guiding and controlling the syringe; Assisted CT scanning monitoring, when there is target area motion deformation, the target area scanning image is transmitted back, combined with the dynamic adjustment branch of the intelligent positioning module, the injection strategy adjustment decision is made to determine the injection adjustment strategy; The injection adjustment strategy is fed back to the injection navigation system to perform positioning feedback regulation on the syringe.

2. The intelligent positioning method for cervical chemotherapy injection according to claim 1, characterized in that: Contains static positioning branches, including: Calling injection positioning samples, mining tumor visual features, constructing a first feature library, mining tumor geometric features-injection parameters, and constructing a second feature library, wherein the tumor visual features include the tumor geometric features; The injection parameters at least include the three-dimensional position of the target, the three-dimensional angle of the target, and the embedding depth. If a multi-needle coordinated injection is used, the injection parameters also include the target spacing; Taking the first feature library as the recognition target, taking the multi-scale triangular mesh based on the tumor target area as the initial position as the segmentation target, and performing one-step training based on the injection positioning sample; The second feature library is used as a decision target, the image triangular mesh is used as a parameter generation standard, and two-step training is performed based on the injection positioning sample to determine the static positioning branch.

3. The intelligent positioning method for chemotherapy injection for cervix according to claim 2, characterized in that: The multi-scale triangular mesh based on the tumor target area as the initial position is used as the segmentation target, including: Setting a first grid size and a second grid size, wherein the accuracy of the first grid size is higher than that of the second grid size; Taking any boundary position of the tumor target area as the initial position, dividing the tumor target area based on the first grid size, and performing three-dimensional division of the non-tumor target area with the second grid size to determine the image triangular grid; The injection parameters are converted and generated using the image triangular grid as a reference baseline.

4. The intelligent positioning method for chemotherapy injection for cervix according to claim 2, characterized in that: Contains dynamic adjustment branches, including: According to the injection positioning samples, mining a third feature library based on motion deformation trend-injection parameter adjustment trend; A built-in temporary data area is used, which takes the third feature library as the decision target, the image triangular mesh as the generation standard for parameter adjustment, and is trained based on the injection positioning sample to determine the dynamic adjustment branch, wherein the temporary data area is used to store the injection positioning strategy generated by the static positioning branch.

5. The intelligent positioning method for chemotherapy injection for cervix according to claim 3, characterized in that: The injection positioning strategy is transmitted to the injection navigation system to guide and control the injector, including: Determine a virtual positioning baseline using the image triangulated grid; Taking the cervical area based on the cervical scan image as the physical space, the physical space and the virtual positioning baseline are combined to build a virtual reality space; With respect to the virtual reality space, the injection navigation system executes and responds to the injection positioning strategy to perform positioning guidance control of the syringe.

6. The intelligent positioning method for chemotherapy injection for cervix according to claim 1, characterized in that: Before making a target injection positioning decision, the method further includes: Determine the injection targeting feature according to the physical and chemical properties of the injection, wherein the injection targeting feature includes the directional position and the non-directional position of the target area; Determining injection conditions, wherein the injection conditions are long-time micropump injection or short-time instant injection; The injection targeting feature and the injection condition are used as constraints to make a target injection positioning decision.

7. The intelligent positioning method for chemotherapy injection for cervix according to claim 1, characterized in that: Assisted CT scanning monitoring, when there is target area motion deformation, return target area scanning images, including: Set the preset degrees of freedom for target area motion deformation; Based on the preset degrees of freedom, setting an image return instruction for injection feedback; Assisting CT scanning monitoring, determining the target area movement deformation, generating the image return instruction if the preset degree of freedom is met, and controlling the CT device to return the target area scanning image to the dynamic adjustment branch.

8. An intelligent positioning system for chemotherapy injection in the cervix, characterized in that: The system is used to implement the intelligent positioning method for chemotherapy injection for cervix according to any one of claims 1 to 7, and the system comprises: A cervical scan image acquisition module, which is used to obtain a cervical scan image of a target user by scanning with a CT device; An injection positioning strategy determination module, which is used to transmit the cervical scan image back to the processor, and in combination with the static positioning branch of the intelligent positioning module, perform multi-scale triangulated meshing and target injection positioning decision on the cervical scan image to determine the injection positioning strategy, wherein the intelligent positioning module is built into the processor, and includes a static positioning branch and a dynamic adjustment branch with lateral interaction, and the injection method is single-needle injection or multi-needle coordinated injection; A guidance control module, which is used to establish a connection between the intelligent positioning module and the injection navigation system, and between the injection navigation system and the syringe, transmit the injection positioning strategy to the injection navigation system, and guide and control the syringe; An injection adjustment strategy determination module is used to assist CT scanning monitoring. When there is target area motion deformation, the target area scan image is transmitted back, and the injection strategy adjustment decision is made in combination with the dynamic adjustment branch of the intelligent positioning module to determine the injection adjustment strategy; A feedback control module is used to feed back the injection adjustment strategy to the injection navigation system to perform positioning feedback control on the syringe.

Images