Multi-source data-based orthopedic surgery execution state planning system

Abstract

The application discloses a kind of orthopedic surgery execution state planning systems of multi-source data, comprising: acquisition module, for collecting the multi-source data of current time in the process of orthopedic surgery;Preprocessing module, for preprocessing multi-source data, obtain preprocessed data set;Data correlation module, for based on the risk degree that data is characterized in preprocessed data set, the data in preprocessed data set is carried out different feature extraction processing, obtain multi-source feature set, and the target data set is obtained by nonlinear processing to multi-source feature set, integrate the data in target data set, obtain the current execution state of orthopedic surgery at current time;State planning module, for by improving the decision network in PPO algorithm model to the current execution state is handled to obtain control signal.This scheme can realize the accurate safety planning of orthopedic surgery execution state, can improve the execution result of orthopedic surgery.

Description

Technical Field

[0001] This application relates to the field of intelligent decision-making technology in orthopedic surgery, and to, but is not limited to, a bone surgery execution state planning system for multi-source surgery. Background Technology

[0002] The process of orthopedic surgery relies on the doctor's clinical experience and relevant medical test results. However, during actual orthopedic surgery, the doctor's, surgical instruments', and the patient's condition are constantly changing, all of which affect the execution of the surgery and thus its outcome. Therefore, there is an urgent need for a technological solution that can precisely plan the execution of orthopedic surgery. Summary of the Invention

[0003] Based on the above technical problems, this application provides a multi-source data orthopedic surgery execution state planning system, which can realize accurate and safe planning of orthopedic surgery execution state and improve the execution results of orthopedic surgery.

[0004] The technical solution provided in this application is as follows: This application provides a multi-source data-based orthopedic surgery execution state planning system, including: The acquisition module is used to acquire multi-source data at the current moment during orthopedic surgery; wherein, the multi-source data includes image data of the lesion area, relative position data between the lesion area and the target tissue, and vibration transmission data between the surgical instruments and the target tissue; the target tissue includes bones, nerves and blood vessels; The preprocessing module is used to preprocess the multi-source data to obtain a preprocessed data set; The data association module is used to perform differential feature extraction processing on the data in the preprocessed data set based on the risk level represented by the data in the preprocessed data set to obtain a multi-source feature set, and to perform nonlinear processing on the multi-source feature set to obtain a target data set. The data in the target data set are then integrated to obtain the current execution status of the orthopedic surgery at the current moment. The state planning module is used to process the current execution state through the decision network in the improved Proximal Policy Optimization (PPO) algorithm model to obtain control signals; wherein, the control signals are used to indicate and regulate the pose and / or movement of the surgical instruments in order to plan the future execution state of the orthopedic surgery at future moments.

[0005] The orthopedic surgery execution state planning system provided in this application embodiment includes an acquisition module for collecting multi-source data at the current moment during orthopedic surgery. This multi-source data includes image data of the lesion area, relative position data between the lesion area and target tissue, and vibration transmission data between surgical instruments and target tissue during orthopedic surgery. Target tissue includes bones, nerves, and blood vessels. This not only improves the comprehensiveness and completeness of the multi-source data but also enables tracking acquisition of multi-source data during orthopedic surgery, thereby improving the real-time performance of the multi-source data. Furthermore, a preprocessing module preprocesses the multi-source data to obtain a preprocessed data set. A data association module performs differentiated feature extraction processing on the data in the preprocessed data set based on the risk level represented by the data, obtaining a multi-source feature set. This enables differentiated and targeted processing of the data in the preprocessed data set, thereby improving the completeness and precision of the features in the multi-source feature set. Based on this, the data association module performs execution processing on the multi-source feature set. Nonlinear processing is performed to obtain the target data set, thus reducing its redundancy. On the other hand, the data association module integrates the data in the target data set to obtain the current execution state of the orthopedic surgery at the current moment. This allows the current execution state to be displayed comprehensively from the dimensions of the lesion area and target tissue, and the surgical instruments and target tissue. Furthermore, the state planning module processes the current execution state through an improved decision network in the PPO algorithm to obtain control signals. These control signals are used to instruct and regulate the pose and / or movement of the surgical instruments to plan the future execution state of the orthopedic surgery. This allows the control signals to be dynamically, in real-time, and flexibly associated with the current surgical state, improving their accuracy and specificity. The control signals can respond to the dynamic changes associated with the lesion area, target tissue, and surgical instruments during the orthopedic surgery, improving the specificity and safety of the future execution state of the orthopedic surgery. This enables precise and safe planning of the orthopedic surgery execution state, thereby improving the execution result of the orthopedic surgery. Attached Figure Description

[0006] Figure 1 This is a schematic diagram of the structure of the orthopedic surgery execution state planning system provided in the embodiments of this application; Figure 2 A schematic diagram of the data processing architecture for training the improved PPO algorithm model provided in the embodiments of this application; Figure 3 This is another structural schematic diagram of the bone surgery execution state planning system provided in the embodiments of this application. Detailed Implementation

[0007] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0008] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0009] The process of a doctor performing orthopedic surgery relies on the doctor's clinical experience, as well as relevant medical test and examination results. During actual orthopedic surgery, the doctor's control over surgical instruments, the direction and force applied, and the patient's body tissues' response to the instruments all dynamically change. These factors influence the actual state of the orthopedic surgery and thus its outcome.

[0010] Based on the above technical problems, this application provides a multi-source data-based orthopedic surgery execution state planning system. Figure 1 This is a schematic diagram of the structure of the orthopedic surgery execution state planning system provided in the embodiments of this application, as shown below. Figure 1 The orthopedic surgery execution state planning system 100 may include: The acquisition module 101 is used to acquire multi-source data at the current moment during orthopedic surgery. Preprocessing module 102 is used to preprocess multi-source data to obtain a preprocessed data set; The data association module 103 is used to perform differential feature extraction processing on the data in the preprocessed data set based on the risk level represented by the data in the preprocessed data set, to obtain a multi-source feature set, and to perform nonlinear processing on the multi-source feature set to obtain a target data set. The data in the target data set are then integrated to obtain the current execution status of the orthopedic surgery at the current moment. The state planning module 104 is used to process the current execution state by improving the decision network in the PPO algorithm model to obtain control signals.

[0011] The multi-source data includes imaging data of the lesion area, relative position data between the lesion area and the target tissue, and vibration transmission data between the surgical instruments and the target tissue; the target tissue includes bones, nerves, and blood vessels; the control signals are used to indicate and regulate the position and / or movement of the surgical instruments used in orthopedic surgery in order to plan the future execution state of the orthopedic surgery at future moments.

[0012] In some embodiments, the lesion area may include the area where at least one body part that has suffered a fracture is located.

[0013] In some embodiments, the bones in the target tissue may include the bones contained in the lesion region and a collection of bones whose distance from the lesion region is less than or equal to a first threshold; the nerves in the target tissue may include at least one of nerve bundles, nerve fibers, or nerve trunks, wherein the nerves in the target tissue may be distributed in the lesion region or may be located at a distance from the lesion region less than or equal to a second threshold; the blood vessels in the target tissue may include at least one of arteries, veins, and capillaries, wherein the aforementioned blood vessels may be distributed in the lesion region or may be located at a distance from the lesion region less than or equal to a third threshold.

[0014] In some embodiments, vibration transmission data may include at least one of the following: vibration amplitude, vibration frequency, and vibration direction, as the vibration generated by the surgical instrument during the surgical procedure is transmitted from one end of the surgical instrument to the target tissue.

[0015] In some embodiments, the acquisition module may include a vibration sensor, thereby enabling the acquisition of vibration transmission data using the vibration sensor and a real-time calculation method.

[0016] In some embodiments, relative location data may include the distance and relative orientation between the lesion area and the target tissue, and may also characterize the spatial overlap between the lesion area and the target tissue in at least one dimension.

[0017] In some embodiments, the acquisition module may include a measuring device capable of image-guided measurement and mechanical measurement, thereby acquiring relative position data through the aforementioned measuring device.

[0018] In some embodiments, the image data may include medical images of the lesion area and associated areas prior to orthopedic surgery, and may also include medical images of the lesion area and associated areas during orthopedic surgery; for example, the associated areas may include at least one area in the patient's body where the lesion status of the lesion area has an influence greater than or equal to a fourth threshold; for example, the image data may include computed tomography (CT) images and / or cone-beam computed tomography (CBCT) images.

[0019] In some embodiments, data acquisition may include intraoperative imaging equipment, and image data may be acquired through intraoperative imaging equipment.

[0020] In some embodiments, the preprocessing module can be electrically connected to the acquisition module. After receiving multi-source data, the preprocessing module can perform preprocessing operations on the data in the multi-source data according to the data types in the multi-source data, as shown below: For image data in multi-source datasets, preprocessing can be achieved in the following ways: The image data is subjected to grayscale normalization and spatial resolution alignment to obtain the first processing result. Then, a deep learning denoising network is used to remove speckle noise and motion artifacts in the first processing result to obtain a standardized image.

[0021] For relative location data in multi-source datasets, preprocessing can be performed on it in the following ways: The surface mesh of the lesion area and the target tissue is extracted by three-dimensional reconstruction and segmentation algorithm, and the spatial geometric parameters between the lesion area and the target tissue are calculated. Then, the missing values ​​are filled by spatial interpolation method to obtain the second processing result. Then, the extreme distance or extreme overlap between the target value and the lesion area in the second processing result is verified to eliminate measurement errors, thereby obtaining standardized relative position data.

[0022] For vibration transmission data, preprocessing can be performed in the following ways: The high-frequency interference in vibration transmission data was removed by using sliding window smoothing and wavelet denoising techniques, and the abnormal vibration transmission data was corrected by combining a biomechanical model to obtain the third processing result.

[0023] Accordingly, the preprocessed dataset may include standardized imagery, standardized relative position data, and third-party processing results.

[0024] In some embodiments, the risk level may include the level of risk of the intraoperative state as represented by multi-source data at the current moment; for example, the risk level may include high risk, medium risk and low risk; wherein, high risk may be related to blood vessels, medium risk may be related to nerves and low risk may be related to bones.

[0025] In some embodiments, the multi-source feature set may include a set of multi-source features corresponding to data in the multi-source data; correspondingly, the multi-source feature set may be implemented in the following ways: The first extraction network extracts features from the data corresponding to high risk in the preprocessed dataset, resulting in a first extraction result. The second extraction network extracts features from the data corresponding to medium risk in the preprocessed dataset, resulting in a second extraction result. The third extraction network extracts features from the data corresponding to low risk in the preprocessed dataset, resulting in a third extraction result. Compared to the second extraction network, the first extraction network can extract more complete, comprehensive, and refined features. Compared to the third extraction network, the second extraction network can extract more complete, comprehensive, and refined features. The multi-source feature set can include the first extraction result, the second extraction result, and the third extraction result.

[0026] In some embodiments, the target dataset may include data from a multi-source feature set excluding the target interval; exemplarily, the target interval may be predetermined or adjusted, and may change with changes in the patient's age or physical condition; correspondingly, nonlinear processing may be performed in the following manner: By sequentially performing nonlinear processing on the multi-source features in the multi-source feature set using a nonlinear function, the data in the target interval contained in the multi-source features in the multi-source feature set are removed, thus obtaining the target data set.

[0027] In some embodiments, the current execution state may include a feature representation of the risk state and progress state of the orthopedic surgery as characterized by multi-source data at the current moment; for example, the risk state may include the current pose of the surgical instruments and / or the threat state of the operation performed to the target tissue; for example, the progress state may include the current pose of the surgical instruments and / or the repair progress of the operation performed on the lesion area; accordingly, the current execution state may be implemented in the following ways: The data in the target dataset is concatenated to obtain concatenated features, and these features are then used to determine the current execution state.

[0028] In some embodiments, the position of the surgical instrument may include the current position and / or orientation of the surgical instrument; the action of the surgical instrument may include positional movement and / or orientation change of the surgical instrument.

[0029] In some embodiments, the control signal may indicate at least one of the following: the amplitude, frequency, direction, and duration of the pose and / or movement of the surgical instrument.

[0030] In some embodiments, the improved PPO algorithm model may include a policy network; specifically, the policy network may receive and comprehensively analyze the multi-source features carried in the current execution state to determine the discrete actions that the surgical instrument can perform in the current execution state, and determine the discrete actions as control signals.

[0031] In some embodiments, the accuracy of the discrete actions output by the improved PPO algorithm model can be greater than the accuracy threshold.

[0032] In some embodiments, a future moment may include at least one next moment after the current moment; correspondingly, a future execution state may include the pose and / or movement of surgical instruments at a future moment, and may also include the execution progress of orthopedic surgery at a future moment.

[0033] In some embodiments, the state planning module can output control signals for reference by the surgeon performing orthopedic surgery. During the orthopedic surgery, the acquisition module can continuously acquire multi-source data, the preprocessing module can preprocess the newly acquired multi-source data to obtain a new preprocessed data set, and the data association module can process the new preprocessed data set to obtain a new current execution state, which is then used by the state planning module to output new control signals. This process is repeated, enabling the orthopedic surgery execution state planning system to track and determine the current execution state of the orthopedic surgery in real time, thereby improving the real-time performance and accuracy of the control signals.

[0034] As can be seen from the above, the orthopedic surgery execution state planning system with multi-source data provided in this application embodiment has an acquisition module used to acquire multi-source data at the current moment during the orthopedic surgery. This multi-source data includes image data of the lesion area, relative position data between the lesion area and the target tissue, and vibration transmission data between surgical instruments and the target tissue during the orthopedic surgery. The target tissue includes bones, nerves, and blood vessels. This not only improves the comprehensiveness and completeness of the multi-source data but also enables tracking acquisition of multi-source data during the orthopedic surgery execution, thereby improving the real-time performance of the multi-source data. Furthermore, the preprocessing module is used to preprocess the multi-source data to obtain a preprocessed data set. The data association module is used to perform differential feature extraction processing on the data in the preprocessed data set based on the risk level represented by the data in the preprocessed data set, obtaining a multi-source feature set. This enables differentiated and targeted processing of the data in the preprocessed data set, thereby improving the completeness of the features in the multi-source feature set and the precision of the feature differentiation. Based on this... The data association module performs nonlinear processing on the multi-source feature set to obtain the target data set, thus reducing the redundancy of the target data set. On the other hand, the data association module integrates the data in the target data set to obtain the current execution state of the orthopedic surgery at the current moment. This allows the current execution state to be displayed comprehensively from the dimensions of the lesion area and target tissue, and the surgical instruments and target tissue. Furthermore, the state planning module processes the current execution state through an improved decision network in the PPO algorithm to obtain control signals. The control signals are used to instruct and regulate the pose and / or movement of the surgical instruments to plan the future execution state of the orthopedic surgery at future moments. This allows the control signals to be dynamically, in real time, and flexibly associated with the current surgical state in the orthopedic surgery, thereby improving the accuracy and specificity of the control signals. The control signals can respond to the dynamic changes associated with the lesion area, target tissue, and surgical instruments during the execution of the orthopedic surgery, improving the specificity and safety of the future execution state of the orthopedic surgery at future moments, and thus improving the execution result of the orthopedic surgery.

[0035] Based on the above technical issues, in the orthopedic surgery execution status planning system with multi-source data provided in this application embodiment, the acquisition module is also used to acquire the instrument position of the surgical instruments during the orthopedic surgery execution process; the preprocessing module is used to determine the set of positional relationships between the surgical instruments and the target tissue based on the instrument position, and to determine the risk level based on the set of positional relationships and the target threshold.

[0036] In some implementations, the instrument position may include the relative position of the surgical instrument to the lesion area; for example, as the orthopedic surgery progresses, the acquisition module may continuously track and acquire the instrument position; for example, the instrument position may be represented in the form of three-dimensional spatial coordinates.

[0037] In some embodiments, before orthopedic surgery is initiated and during orthopedic surgery, the acquisition module can also acquire the relative position of the target tissue with respect to the lesion area in real time to obtain the tissue position; for example, the tissue position can be represented in the form of three-dimensional spatial coordinates.

[0038] In some embodiments, the set of positional relationships may include the distance and / or orientation between surgical instruments and target tissue; accordingly, the set of positional relationships may be determined in the following ways: Based on the three-dimensional spatial coordinates corresponding to the instrument position, a first positional relationship between the surgical instrument and the reference target is determined. Based on the three-dimensional spatial coordinates corresponding to the tissue position, a second positional relationship between the target tissue and the reference target is determined. Then, based on the first and second positional relationships, a set of positional relationships between the surgical instrument and the target tissue is determined. The reference target may include a pre-selected reference plane or reference direction in the three-dimensional coordinate system.

[0039] In some embodiments, the number of thresholds in the target threshold may be the same as the number of tissues in the target tissue, and the thresholds corresponding to different tissues in the target threshold may be the same or different; for example, the tissues may include bones, nerves, and blood vessels; accordingly, the degree of risk can be determined in the following ways: The risk level is determined based on the relationship between the m-th distance between the surgical instrument and the m-th tissue in the target tissue, and the corresponding threshold in the target threshold. For example, if the m-th distance is less than the fifth threshold, the risk level of the surgical instrument relative to the m-th tissue can be determined as high risk; if the m-th distance is greater than the fifth threshold, the risk level of the surgical instrument relative to the m-th tissue can be determined as low risk; and if the m-th distance is equal to the fifth threshold, the risk level of the surgical instrument relative to the m-th tissue can be determined as medium risk. Here, m is an integer greater than or equal to 1 and less than or equal to 3, and the first to third tissues can include bones, nerves, and blood vessels. The fifth threshold can correspond to the m-th distance.

[0040] In some embodiments, the value of the target threshold can be determined by a professional technician, or it can be adjusted according to the patient's age, gender, and physical condition.

[0041] As can be seen from the above, the orthopedic surgery execution status planning system with multi-source data provided in this application embodiment has an acquisition module used to acquire the instrument positions of surgical instruments during the orthopedic surgery execution process, thus enabling tracking acquisition of the instrument positions; and a preprocessing module used to determine the set of positional relationships between the surgical instruments and the target tissue based on the instrument positions, thereby improving the dynamism and accuracy of the positional relationship set; on this basis, the risk level is determined based on the positional relationship set and the target threshold. As the orthopedic surgery progresses, the positional relationship set can also change dynamically, thereby allowing the risk level to change accordingly, further improving the dynamism and accuracy of the risk level.

[0042] Based on the above technical issues, in the orthopedic surgery execution status planning system for multi-source data provided in this application embodiment, the risk level includes high risk, medium risk, and low risk. Correspondingly, the data association module is used to perform feature extraction processing on the data corresponding to high risk in the preprocessed data set through the first convolutional layer to obtain high-risk features, perform feature extraction processing on the data corresponding to medium risk in the preprocessed data set through the second convolutional layer to obtain medium-risk features, and perform feature extraction processing on the data corresponding to low risk in the preprocessed data set through the third convolutional layer to obtain low-risk features.

[0043] The multi-source feature set includes high-risk features, medium-risk features, and low-risk features; the kernel size of the first convolutional layer is larger than that of the second convolutional layer, and the kernel size of the second convolutional layer is larger than that of the third convolutional layer.

[0044] In some embodiments, the number of convolutional kernels included in the first to third convolutional layers can be at least one. For example, when the number of convolutional kernels included in the first to third convolutional layers is at least two, the kernel sizes of the convolutional kernels included in the first convolutional layer can be the same, the kernel sizes of the convolutional kernels included in the second convolutional layer can be the same, and the kernel sizes of the convolutional kernels included in the third convolutional layer can also be the same. Specifically, the first to third convolutional layers can be arranged in parallel, and the kernel size of the first convolutional layer... The kernel size can be 7×7×7, allowing the first convolutional layer to capture more complete high-risk features from the data corresponding to high risk in the preprocessed dataset. The kernel size of the second convolutional layer can be 5×5×5, thus achieving a balance between feature completeness and feature precision when processing data corresponding to medium risk in the preprocessed dataset. The kernel size of the third convolutional layer can be 3×3×3, thus balancing computational efficiency and feature precision when processing data corresponding to low risk in the preprocessed dataset.

[0045] It should be noted that the number and / or size of the convolutional kernels contained in the first to third convolutional layers can be predetermined or flexibly adjusted, and this application embodiment does not limit this.

[0046] For example, the first convolutional layer can be a first extraction network, the second convolutional layer can be a second extraction network, and the third convolutional layer can be a third extraction network.

[0047] As can be seen from the above, the orthopedic surgery execution state planning system for multi-source data provided in this application embodiment has a data association module that uses a first to a third convolutional layer with different kernel sizes to perform feature extraction processing on data corresponding to different risk levels in the preprocessed data set. This improves the targeting and differentiated processing of data corresponding to different risk levels in the preprocessed data set, and thus can take into account the comprehensiveness, precision and timeliness of features in the multi-source feature set.

[0048] Based on the above technical problems, in the orthopedic surgery execution state planning system with multi-source data provided in this application embodiment, the data association module is used to obtain the k-th distance gradient feature and the k-th azimuth feature from the k-th multi-source feature set, and perform nonlinear processing on the k-th distance gradient feature and the k-th azimuth feature based on the first weight to obtain the k-th target data.

[0049] Wherein, k is an integer greater than or equal to 1 and less than or equal to K; K is an integer greater than 1 and is used to characterize the number of types of multi-source features in the multi-source feature set; the target data set includes the first target data to the Kth target data; the kth distance gradient feature includes the gradient feature of the distance between the lesion region and / or the surgical instrument and the kth tissue in the target tissue; the kth azimuth feature includes the feature of the azimuth angle of the lesion region and / or the surgical instrument relative to the kth tissue.

[0050] In some embodiments, during the preprocessing of multi-source data, the preprocessing module can process each data in the multi-source data into two dimensions: distance gradient and azimuth angle. For example, the distance gradient can include the rate of change of distance between the lesion region and / or the surgical instrument along the extension direction of the bone contained in the lesion region and the surface of the target tissue, and the azimuth angle can characterize the spatial orientation of the lesion region and / or the surgical instrument along the extension direction of the bone contained in the lesion region and the surface of the target tissue.

[0051] In some embodiments, a network module with feature extraction capabilities can be used to extract features from the distance gradient and azimuth angle respectively, thereby obtaining the k-th distance gradient feature and the k-th azimuth angle feature.

[0052] In some embodiments, the k-th target data can be obtained in the following way: A nonlinear activation function is constructed based on the first weight, and nonlinear processing is performed on the k-th distance gradient feature and the k-th azimuth feature through the nonlinear activation function to obtain the k-th target data; for example, the nonlinear activation function can be as shown in equation (1): (1) in, This can correspond to the gradient features from the first distance to the Kth distance. The first weight can correspond to the first azimuth feature to the Kth azimuth feature, and the first weight can include... and , It can be the target data in the target data set obtained by calculating the distance gradient feature and the azimuth feature.

[0053] In some embodiments, the first weight can be preset or flexibly adjusted; specifically, the first weight can be adjusted through supervised training based on sample data corresponding to historical orthopedic surgeries.

[0054] As can be seen from the above, the orthopedic surgery execution state planning system based on multi-source data provided in this application embodiment uses a data association module to perform nonlinear processing on the k-th distance gradient feature and the k-th azimuth feature in the multi-source feature set based on a first weight, to obtain the k-th target data. This allows the target data in the target data set to comprehensively display the spatial positional relationship between the lesion area and / or surgical instruments relative to the tissue in the target tissue from the dimensions of distance gradient and azimuth.

[0055] Based on the above technical issues, the orthopedic surgery execution state planning system for multi-source data provided in this application embodiment includes an initial policy network, a value network, a reward function, a dynamic advantage function, and a PPO pruning and updating module in its PPO algorithm model; the PPO algorithm model is associated with an improved PPO algorithm model.

[0056] The preprocessing module is used to preprocess the data in the sample data to obtain a sample preprocessed data set. The data association module is used to perform differential feature extraction on the data in the sample preprocessed data set based on the risk level represented by the data, to obtain a sample multi-source feature set, and to perform nonlinear processing on the sample multi-source feature set to obtain a sample target data set. The data in the sample target data set are integrated to obtain the surgical state at time t; where t is an integer greater than or equal to 1. The sample data is used to describe the path planning process of historical orthopedic surgeries.

[0057] It should be noted that the preprocessing module preprocesses the sample data in the same way as the preprocessing module preprocesses the multi-source data in the previous embodiment. Furthermore, the data association module processes the data in the sample preprocessed data set in the same way as the data association module processes the preprocessed data set in the previous embodiment, and will not be repeated here.

[0058] In some embodiments, the surgical state at time t may correspond to the progress of the orthopedic surgery at time t; time t may include historical times covered by the sample data.

[0059] Figure 2 A schematic diagram illustrating the data processing architecture for training the improved PPO algorithm model provided in this application embodiment, as shown below. Figure 2 As shown, the improved PPO algorithm model 200 may include: An initial policy network is used to process the t-th surgical state to obtain the (t+1)-th control signal; A value network is used to determine the state score corresponding to the surgical state t. The reward function is used to process the (t+1)th control signal and sample data based on the second weight to obtain the t-th reward score; The dynamic advantage function is used to process the t-th state score, the t-th instrument displacement vector corresponding to the t-th pose data of the surgical instrument, the t-th vibration vector corresponding to the t-th vibration amplitude of the surgical instrument, and the t-th damage vector corresponding to the t-th damage volume based on the third weight, so as to obtain the t-th advantage score.

[0060] Among them, the t-th pose data, the t-th vibration amplitude, and the t-th damage volume correspond to the t-th surgical state; the t-th damage volume includes the volume of damage caused by the surgical instrument to at least one site.

[0061] In some embodiments, the t+1 control signal can be used to control the pose and / or movement of surgical instruments at time t+1.

[0062] In some embodiments, the t-th state score can be obtained by comprehensively evaluating the t-th surgical state and the t-th advantage score output by the dynamic advantage function through a value network; for example, the t-th state score can characterize the expected cumulative reward obtained from the t-th surgical state under the constraint of the t-th advantage score.

[0063] In some embodiments, the t-th reward score can be obtained by processing the t-th surgical state and the t+1-th control signal corresponding to the sample data through a reward function; for example, the t-th reward score can be used to quantify the quality of the discrete action represented by the t+1-th control signal.

[0064] In some embodiments, the t-th advantage score may include the value representation of the surgical state after performing the discrete action represented by the t+1-th control signal in the case of the t-th surgical state.

[0065] In this embodiment, the dynamic dominance function has been improved; specifically, the dynamic dominance function is shown in equation (2): (2) in, Characterizing the t-th advantage score, The t-th instrument displacement vector is represented by the t-th pose data. The t-th vibration vector represents the vibration amplitude corresponding to the t-th vibration. The damage vector representing the damage volume at time t is described. The score for state t is a combination of information on dissection risk and target progress. This is the weight vector.

[0066] In some embodiments, the t-th damage volume can be obtained by analyzing the image data corresponding to the t-th surgical state in the sample data. It can characterize the size of the damage caused by the surgical instrument to at least one site during the orthopedic surgery up to the t-th surgical state. The t-th instrument displacement vector can characterize the displacement path of the surgical instrument determined by the change in the pose data of the surgical instrument corresponding to the t-th surgical state. The t-th vibration vector can characterize the vibration amplitude, vibration direction, and vibration frequency of the surgical instrument corresponding to the t-th surgical state.

[0067] In some embodiments, the t-th pose data, t-th vibration amplitude, and t-th damage volume can be processed by a multi-objective motion value decomposition algorithm to obtain the t-th instrument displacement vector, t-th vibration vector, and t-th damage vector respectively.

[0068] It should be noted that the third weight can be predetermined or adjusted and updated during the training of the PPO algorithm model.

[0069] Figure 3 Another structural schematic diagram of the bone surgery execution state planning system provided in the embodiments of this application is shown below. Figure 3 As shown, the orthopedic surgery execution state planning system may also include an algorithm improvement module 105, which is used to obtain an improved PPO algorithm model by jointly regulating the parameters of the initial policy network and the value network through the PPO pruning and updating module based on the t-th advantage score, the t+1-th control signal, the t-th state score and the t-th reward score.

[0070] For example, the process of adjusting the parameters of the policy network and the value network described above can be a supervised process. That is, the sample data can include surgical state t and surgical state t+1, and the state transition between surgical state t and surgical state t+1 can be achieved through surgical operations between surgical state t and surgical state t+1 in the sample data. These surgical operations can be label data of discrete actions corresponding to control signal t+1. In this way, the initial policy network and value network can be continuously adjusted by the degree of matching between the label data and control signal t+1, and the degree of contribution of control signal t+1 to the surgical results contained in the sample data. For example, when the degree of matching between label data and control signal t+1 is greater than or equal to a degree threshold, and the degree of contribution of control signal t+1 to the surgical results in the sample data is greater than or equal to a contribution threshold, the policy network can be obtained, and the policy network is determined as the updated PPO algorithm model.

[0071] As can be seen from the above, the orthopedic surgery execution state planning system based on multi-source data provided in this application, after processing the sample data through the preprocessing module and the data association module to obtain the t-th surgical state represented by the sample data, trains the PPO algorithm model through the t-th surgical state to obtain an improved PPO algorithm model. Compared with related technologies, the dynamic advantage function in the PPO algorithm model is used to process the t-th state score, the t-th instrument displacement vector corresponding to the t-th pose data of the surgical instrument, the t-th vibration vector corresponding to the t-th vibration amplitude of the surgical instrument, and the t-th damage vector corresponding to the t-th damage volume based on the third weight to obtain the t-th advantage score, which improves the comprehensiveness and dynamism of the t-th advantage score. On this basis, the algorithm improvement module, based on the t-th advantage score, the t+1-th control signal, the t-th state score, and the t-th reward score, jointly regulates the parameters of the initial policy network and the value network through the PPO algorithm model to obtain an improved PPO algorithm model, which can improve the pertinence of the control signal output by the policy network of the improved PPO algorithm model.

[0072] Based on the above technical problems, in the orthopedic surgery execution state planning system with multi-source data provided in this application embodiment, the algorithm improvement module is also used to determine the t-th urgency level of the distance dimension based on the t-th distance, determine the t-th urgency level of the velocity dimension based on the t-th displacement velocity, and update the second weight based on the t-th urgency level of the distance dimension and the t-th urgency level of the velocity dimension.

[0073] Wherein, the t-th distance includes the distance between the surgical instrument and the blood vessel corresponding to the t-th surgical state, and the t-th displacement velocity includes the velocity of the surgical instrument relative to the blood vessel corresponding to the t-th surgical state.

[0074] In some embodiments, blood vessels may include arteries and / or veins distributed within the lesion area or at a distance from the surgical instruments less than or equal to a sixth threshold.

[0075] In some embodiments, the t-th distance may include the straight-line distance between the surgical instrument and the blood vessel corresponding to the t-th surgical state; correspondingly, the t-th urgency of the distance dimension can characterize the degree of threat that the current operation of the surgical instrument may pose to the blood vessel when the surgical instrument and the blood vessel are at the aforementioned straight-line distance; correspondingly, the t-th urgency of the distance dimension can be determined in the following manner: The t-th urgency level of the distance dimension is determined based on the relationship between the t-th distance and the seventh threshold; for example, if the t-th distance is less than the seventh threshold, the t-th urgency level of the distance dimension can be determined as high urgency.

[0076] In some embodiments, the t-th displacement velocity may include the velocity of the surgical instrument relative to the blood vessel during the transition from the previous state to the t-th surgical state; correspondingly, the t-th urgency of the velocity dimension can characterize the degree of threat that the current operation of the surgical instrument may pose to the blood vessel under the t-th displacement velocity; correspondingly, the t-th urgency of the velocity dimension can be determined in the following manner: The urgency level of the velocity dimension is determined based on the relationship between the t-th displacement velocity and the eighth threshold. For example, if the t-th displacement velocity is greater than the eighth threshold, the t-th urgency level of the velocity dimension can be determined as high urgency.

[0077] It should be noted that the first and second thresholds can be preset or flexibly adjusted during the training of the PPO algorithm model.

[0078] In some embodiments, the second weight can be updated in any of the following ways: We perform weighted statistics on the t-th urgency of the distance dimension and the t-th urgency of the velocity dimension to obtain the first index, and update the weight indicated by the first index in the first set to the second weight.

[0079] The urgency levels corresponding to the two dimensions are quantified in advance to obtain a first quantization result and a second quantization result. The larger value between the first quantization result and the second quantization result is determined as the second index. Then, the weight indicated by the second index in the first set is updated to the second weight. For example, the urgency levels of the two dimensions can increase as the urgency quantization result increases.

[0080] For example, a first correspondence can be pre-established between the indices in the first index set and each second weight in the first set. Thus, after determining the first index or the second index, the second weight can be determined from the first set based on the first correspondence and the first index or the second index. The indices in the first index set may include the identifiers or storage locations of each weight in the first set, and the indices in the first index set may include the first index and the second index.

[0081] As can be seen from the above, the orthopedic surgery execution state planning system based on multi-source data provided in this application embodiment further includes an algorithm improvement module for determining the t-th urgency level based on the t-th distance and the t-th urgency level based on the t-th displacement velocity. The t-th distance includes the distance between the surgical instrument and the blood vessel corresponding to the t-th surgical state, and the t-th displacement velocity includes the velocity of the surgical instrument relative to the blood vessel corresponding to the t-th surgical state. Thus, by using the urgency levels in two dimensions, the threat state of the surgical instrument to the blood vessel can be dynamically characterized from the dimensions of the distance and displacement velocity of the surgical instrument relative to the blood vessel. Furthermore, updating the second weight based on the t-th urgency level in the distance and velocity dimensions enables targeted updates to the second weight, thereby improving the accuracy of the second weight update.

[0082] Based on the above technical problems, in the orthopedic surgery execution state planning system with multi-source data provided in this application embodiment, the algorithm improvement module is also used to determine the t-th sub-reward set, and to determine the penalty data based on the safety distance threshold and the t-th displacement distance of the surgical instrument; the reward function is used to process the t-th sub-reward set and the penalty data based on the second weight to obtain the t-th reward score.

[0083] The t-th sub-reward set includes the t-th path reward corresponding to the t-th displacement path, the t-th damage reward corresponding to the t-th damage volume, and the t-th vibration reward corresponding to the t-th vibration amplitude; the t-th displacement path includes the path length of the surgical instrument relative to the target tissue, corresponding to the t-th surgical state.

[0084] In some embodiments, the t-th displacement path may include the displacement distance of the surgical instrument relative to the blood vessel in the t-th surgical state.

[0085] In some embodiments, the reward for the t-th path can characterize the degree of influence of the t-th displacement path on the surgical outcome contained in the sample data; for example, the t-th displacement path may include the length of the path of displacement of the surgical instruments between the start of the orthopedic surgery and the time corresponding to the t-th surgical state contained in the sample data.

[0086] Accordingly, the reward for path t can be calculated as follows: The reward for the t-th path is determined based on the relationship between the length of the t-th displacement path and the path threshold.

[0087] In some embodiments, the t-th vibration reward can characterize the t-th vibration amplitude of the surgical instrument and its influence on the surgical outcomes contained in the sample data; accordingly, the t-th vibration reward can be calculated as follows: The reward for vibration t is determined based on the relationship between the amplitude of vibration t and the amplitude threshold.

[0088] In some embodiments, the damage reward of the tth injury can characterize the degree of influence of the tth injury volume on the surgical outcomes contained in the sample data; accordingly, the damage reward of the tth injury can be calculated as follows: The reward for damage t is determined based on the relationship between the damage volume at t and the volume threshold.

[0089] In some embodiments, the t-th reward score can be calculated in the following way: The penalty data is determined based on the difference between the length of the t-th displacement path and the safety distance threshold. Then, each reward in the t-th sub-reward set is weighted based on the second weight to obtain the reward weighting result. The difference between the reward weighting result and the penalty data is then determined as the t-th reward score, as shown in Equation (3). (3) in, For safe distance threshold, Let be the length of the path for the t-th displacement. To punish data, For the t-th reward score, and As the second weight, ( ) are the path reward for the tth step, the vibration reward for the tth step, and the damage reward for the tth step, respectively.

[0090] As can be seen from the above, in the orthopedic surgery execution state planning system with multi-source data provided in this application embodiment, the algorithm improvement module is used to determine the t-th sub-reward set, including the t-th path reward corresponding to the t-th displacement path, the t-th damage reward corresponding to the t-th damage volume, and the t-th vibration reward corresponding to the t-th vibration amplitude. Thus, the t-th sub-reward set can characterize the influence of the t-th surgical state on the surgical outcome from multiple dimensions. Furthermore, by determining the penalty data based on the safety distance threshold and the t-th displacement path, the negative impact of the displacement length of the surgical instrument in the orthopedic surgery described by the sample data on the surgical outcome can be accurately quantified. On this basis, the reward function processes the t-th sub-reward set and the penalty data based on the second weight to obtain the t-th reward score, which can improve the accuracy of the t-th reward score.

[0091] Based on the above technical problems, in the orthopedic surgery execution state planning system with multi-source data provided in this application embodiment, the algorithm improvement module is used to process the displacement path, damage volume and vibration amplitude corresponding to the t-th surgical state based on the threshold set to obtain the sub-reward set of t.

[0092] In some embodiments, the reward for the t-th path can be calculated using equation (4): (4) in, The reward for path t is... For path threshold, Let t be the length of the displacement path.

[0093] In some embodiments, the vibration reward at time t can be calculated using equation (5): (5) in, For the t-th vibration reward, For amplitude threshold, Let t be the amplitude of the vibration.

[0094] In some embodiments, the damage reward at time t can be calculated using equation (6): (6) in, For the t-th damage reward, Let t be the damage volume. This is the volume threshold.

[0095] For example, the threshold set may include path thresholds, amplitude thresholds, and volume thresholds.

[0096] As can be seen from the above, in the orthopedic surgery execution state planning system based on multi-source data provided in this application, the algorithm improvement module is used to process the t-th displacement path, t-th damage volume, and t-th vibration amplitude corresponding to the t-th surgical state based on a threshold set, thereby obtaining the t-th sub-reward set. In this way, tracking processing of the t-th displacement path, t-th damage volume, and t-th vibration amplitude is achieved, thereby improving the comprehensiveness of the rewards in the t-th sub-reward set.

[0097] Based on the above technical problems, in the orthopedic surgery execution state planning system with multi-source data provided in this application embodiment, the algorithm improvement module is also used to determine the t-th urgency level of the distance dimension based on the t-th distance, determine the urgency level of the velocity dimension based on the t-th displacement velocity, and update the third weight based on the t-th urgency level of the distance dimension and the t-th urgency level of the velocity dimension.

[0098] Wherein, the t-th distance includes the distance between the surgical instrument and the blood vessel corresponding to the t-th surgical state, and the t-th displacement velocity includes the velocity of the surgical instrument relative to the blood vessel corresponding to the t-th surgical state.

[0099] In some embodiments, the third weight can be updated in any of the following ways: We perform weighted statistics on the t-th urgency of the distance dimension and the t-th urgency of the velocity dimension to obtain the third index, and update the weight indicated by the third index in the second set to the third weight.

[0100] The urgency of the two dimensions is pre-quantified to obtain a third quantization result and a fourth quantization result. The larger value between the third quantization result and the fourth quantization result is determined as the fourth index. The weight indicated by the fourth index in the second set is updated to the third weight. For example, the urgency of the two dimensions can increase as the second quantization result increases.

[0101] For example, a second correspondence can be pre-established between the indices in the second index set and each third weight in the second set. Thus, after determining the third index or the fourth index, the third weight can be determined from the second set based on the second correspondence and the third index or the fourth index. The indices in the second index set may include the identifiers or storage locations of each weight in the second set, and the indices in the second index set may include the third index and the fourth index.

[0102] As can be seen from the above, the orthopedic surgery execution state planning system based on multi-source data provided in this application embodiment further includes an algorithm improvement module for determining the t-th urgency level based on the t-th distance and the t-th urgency level based on the t-th displacement velocity. The t-th distance includes the distance between the surgical instrument and the blood vessel corresponding to the t-th surgical state, and the t-th displacement velocity includes the velocity of the surgical instrument relative to the blood vessel corresponding to the t-th surgical state. Thus, by using the urgency levels in two dimensions, the threat state of the surgical instrument to the blood vessel can be dynamically characterized from the dimensions of the distance and displacement velocity of the surgical instrument relative to the blood vessel. Furthermore, updating the third weight based on the t-th urgency level in the distance and velocity dimensions enables targeted updates to the third weight, thereby improving the accuracy of the third weight update.

[0103] The description of the various embodiments above tends to emphasize the differences between the various embodiments. The similarities or similarities between them can be referred to, and for the sake of brevity, they will not be repeated here.

[0104] The methods disclosed in the various method embodiments provided in this application can be arbitrarily combined to obtain new method embodiments without conflict.

[0105] The features disclosed in the various product embodiments provided in this application can be arbitrarily combined without conflict to obtain new product embodiments.

[0106] The features disclosed in the various method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method or device embodiments.

[0107] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0108] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0109] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware nodes. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0110] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0111] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0112] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0113] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A multi-source data-based orthopedic surgical execution state planning system, characterized in that, include: The acquisition module is used to acquire multi-source data at the current moment during orthopedic surgery; wherein, the multi-source data includes image data of the lesion area, relative position data between the lesion area and the target tissue, and vibration transmission data between the surgical instruments and the target tissue; the target tissue includes bones, nerves and blood vessels; The preprocessing module is used to preprocess the multi-source data to obtain a preprocessed data set; The data association module is used to perform differential feature extraction processing on the data in the preprocessed data set based on the risk level represented by the data in the preprocessed data set to obtain a multi-source feature set, and to perform nonlinear processing on the multi-source feature set to obtain a target data set. The data in the target data set are then integrated to obtain the current execution status of the orthopedic surgery at the current moment. The state planning module is used to process the current execution state through an improved decision network in the PPO algorithm model to obtain a control signal; wherein the control signal is used to indicate and regulate the pose and / or movement of the surgical instruments in order to plan the future execution state of the orthopedic surgery at future moments.

2. The system according to claim 1, characterized in that, The acquisition module is also used to acquire the instrument position of the surgical instruments during the orthopedic surgery. The preprocessing module is used to determine a set of positional relationships between the surgical instruments and the target tissue based on the instrument position, and to determine the risk level based on the set of positional relationships and a target threshold.

3. The system according to claim 1, characterized in that, The risk levels include high risk, medium risk, and low risk. The data association module is used to perform feature extraction processing on the data corresponding to the high risk in the preprocessed data set through a first convolutional layer to obtain high-risk features, perform feature extraction processing on the data corresponding to the medium risk in the preprocessed data set through a second convolutional layer to obtain medium-risk features, and perform feature extraction processing on the data corresponding to the low risk in the preprocessed data set through a third convolutional layer to obtain low-risk features. The multi-source feature set includes the high-risk features, the low-risk features, and the medium-risk features. The kernel size of the first convolutional layer is larger than the kernel size of the second convolutional layer, and the kernel size of the second convolutional layer is larger than the kernel size of the third convolutional layer.

4. The system according to claim 1, characterized in that, The data association module is used to obtain the k-th distance gradient feature and the k-th azimuth feature from the k-th multi-source feature set, and to perform nonlinear processing on the k-th distance gradient feature and the k-th azimuth feature based on a first weight to obtain the k-th target data; wherein, k is an integer greater than or equal to 1 and less than or equal to K, and K is an integer greater than 1 and is used to characterize the number of types of multi-source features in the multi-source feature set; the target data set includes the first target data to the k-th target data; the k-th distance gradient feature includes the gradient feature of the distance between the lesion region and / or the surgical instrument and the k-th tissue in the target tissue; the k-th azimuth feature includes the feature of the azimuth angle of the lesion region and / or the surgical instrument relative to the k-th tissue.

5. The system according to claim 1, characterized in that, The PPO algorithm model includes an initial policy network, a value network, a reward function, a dynamic dominance function, and a PPO pruning and update module; the PPO algorithm model is associated with the improved PPO algorithm model. The preprocessing module is used to preprocess the data in the sample data to obtain a sample preprocessed data set; wherein, the sample data is used to describe the path planning process of historical orthopedic surgeries. The data association module is used to perform differential feature extraction processing on the data in the sample preprocessing data set based on the risk level represented by the data in the sample preprocessing data set, to obtain a sample multi-source feature set, and to perform nonlinear processing on the sample multi-source feature set to obtain a sample target data set, and to integrate the data in the sample target data set to obtain the surgical state at time t; where t is an integer greater than or equal to 1; The initial strategy network is used to process the t-th surgical state to obtain the (t+1)-th control signal; The value network is used to determine the t-th state score corresponding to the t-th surgical state; The reward function is used to process the (t+1)th control signal and the sample data based on the second weight to obtain the t-th reward score; The dynamic advantage function is used to process the t-th state score, the t-th instrument displacement vector corresponding to the t-th pose data of the surgical instrument, the t-th vibration vector corresponding to the t-th vibration amplitude of the surgical instrument, and the t-th damage vector corresponding to the t-th damage volume based on a third weight, to obtain the t-th advantage score; the t-th pose data, the t-th vibration amplitude, and the t-th damage volume correspond to the t-th surgical state; the t-th damage volume includes the volume of damage caused by the surgical instrument to at least one site; The orthopedic surgery execution state planning system also includes an algorithm improvement module, which is used to obtain the improved PPO algorithm model by jointly regulating the parameters of the initial policy network and the value network through the PPO pruning and updating module based on the t-th advantage score, the t+1-th control signal, the t-th state score and the t-th reward score.

6. The system according to claim 5, characterized in that, The algorithm improvement module is further configured to determine the t-th urgency level in the distance dimension based on the t-th distance, determine the t-th urgency level in the velocity dimension based on the t-th displacement velocity, and update the second weight based on the t-th urgency level in the distance dimension and the t-th urgency level in the velocity dimension; wherein, the t-th distance includes the distance between the surgical instrument and the blood vessel corresponding to the t-th surgical state; and the t-th displacement velocity includes the velocity of the surgical instrument relative to the blood vessel corresponding to the t-th surgical state.

7. The system according to claim 5, characterized in that, The algorithm improvement module is further configured to determine the t-th sub-reward set and, based on a safety distance threshold and the t-th displacement path of the surgical instrument, determine penalty data; wherein, the t-th sub-reward set includes the t-th path reward corresponding to the t-th displacement path, the t-th damage reward corresponding to the t-th damage volume, and the t-th vibration reward corresponding to the t-th vibration amplitude; the t-th displacement path includes the path length of the surgical instrument relative to the target tissue, corresponding to the t-th surgical state; The reward function is used to process the t-th sub-reward set and the penalty data based on the second weight to obtain the t-th reward score.

8. The system according to claim 7, characterized in that, The algorithm improvement module is used to process the t-th displacement path, the t-th damage volume, and the t-th vibration amplitude corresponding to the t-th surgical state based on a threshold set, so as to obtain the t-th sub-reward set.

9. The system according to claim 5 or 8, characterized in that, The algorithm improvement module is further configured to determine the t-th urgency level in the distance dimension based on the t-th distance, determine the t-th urgency level in the velocity dimension based on the t-th displacement velocity, and update the third weight based on the t-th urgency level in the distance dimension and the t-th urgency level in the velocity dimension; wherein, the t-th distance includes the distance between the surgical instrument and the blood vessel corresponding to the t-th surgical state; and the t-th displacement velocity includes the velocity of the surgical instrument relative to the blood vessel corresponding to the t-th surgical state.

Images