Interventional ultrasound method and system for minimally invasive surgery

Abstract

The invention relates to the technical field of medical instruments, and discloses an interventional ultrasound method and system for minimally invasive surgery, and the method comprises the steps: obtaining ultrasound image data and operation state parameters; a three-dimensional ultrasonic image model is constructed through multi-layer scanning, and an initial tissue feature evaluation model is analyzed and generated; dynamic blood vessel data and operation constraint conditions are loaded, a comprehensive evaluation model is generated, and puncture point dynamic simulation is carried out; constructing a dynamic blood vessel prediction model, training to obtain a puncture point adjustment model, and generating initial adjustment parameters; obtaining dynamic adjustment simulation information based on the parameters and the model; constructing a multi-path optimization model, and carrying out iterative optimization to obtain an optimal path planning parameter; the safety puncture probability is inferred in combination with real-time images, and a real-time puncture strategy is adjusted to achieve dynamic matching. The system comprises a data acquisition module, an image modeling and analysis module and the like. The safety and accuracy of minimally invasive surgery puncture are improved, and the method is suitable for interventional ultrasound guided minimally invasive surgery scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of medical devices, and in particular to an interventional ultrasound method and system for minimally invasive surgery. Background Art

[0002] In modern medicine, minimally invasive surgery is widely used due to its advantages such as minimal trauma and rapid recovery. Interventional ultrasound, as one of the key technologies in minimally invasive surgery, plays an important role in guiding surgical instruments for precise puncture and avoiding vascular damage. However, traditional interventional ultrasound technology faces many challenges in practical application and cannot meet the precision requirements of complex surgical scenarios.

[0003] From the perspective of image modeling and analysis, traditional methods usually only construct two-dimensional or simple three-dimensional models based on a single-level ultrasound scan, which cannot fully capture the dynamic tissue characteristics of the surgical site. For example, human tissue has complex spatial structures and physiological characteristics. The density, vascular distribution and other characteristics of different tissues may change dynamically during surgery with physiological activities such as breathing and heartbeat. Traditional modeling methods lack the support of multi-level and multi-angle ultrasound imaging data, making it difficult to accurately classify tissue safety levels. This leads to inaccurate safety assessments of puncture paths and increases surgical risks.

[0004] When it comes to puncture path planning, existing technologies often fail to fully integrate dynamic vascular distribution data with surgical operation constraints. Blood vessels are structures that must be avoided during surgery, and their position and morphology may change in real time with changes in hemodynamics. Traditional methods often use static vascular models for path planning, which cannot adapt to such dynamic changes. At the same time, surgical instruments have operational constraints such as maximum puncture angle restrictions and minimum turning radius. Traditional path planning does not comprehensively consider these factors, which can easily lead to the planned puncture path being difficult to implement in actual operation, or increase the operation time and patient pain due to excessive operational difficulty.

[0005] Regarding puncture point adjustment and dynamic simulation, traditional systems lack effective dynamic prediction models and real-time adjustment mechanisms. When unexpected situations such as tissue displacement and changes in vascular morphology occur during surgery, traditional methods struggle to quickly generate accurate puncture point adjustment parameters, making it impossible to achieve real-time dynamic optimization of the puncture strategy. Furthermore, when it comes to multi-path optimization, traditional algorithms often employ a single optimization strategy, failing to achieve an effective balance between multiple objectives such as path length, adjustment response speed, and operational safety, resulting in poor overall puncture path performance.

[0006] With the continuous development of medical imaging technology and artificial intelligence algorithms, how to introduce advanced three-dimensional reconstruction technology, dynamic modeling methods and intelligent optimization algorithms into the field of interventional ultrasound to build a system that can dynamically evaluate tissue characteristics in real time, accurately plan puncture paths and achieve adaptive adjustment of puncture strategies has become the key to solving the bottleneck of traditional technologies. It is against this technical background that the present invention addresses the shortcomings of existing interventional ultrasound methods in terms of comprehensive image modeling, dynamic path planning, and real-time puncture adjustment. It proposes an interventional ultrasound method and system based on multi-faceted ultrasound scanning, dynamic tissue feature analysis and intelligent optimization algorithms to improve the safety and accuracy of minimally invasive surgery. Summary of the Invention

[0007] The object of the present invention is to provide an interventional ultrasound method and system for minimally invasive surgery to solve the problems raised in the above background technology.

[0008] To achieve the above objectives, the present invention provides the following technical solution: an interventional ultrasound method for minimally invasive surgery, the method comprising:

[0009] Obtaining ultrasound imaging data of the patient's surgical site and operating status parameters of surgical instruments;

[0010] Perform multi-level ultrasound scanning of the surgical site and construct a three-dimensional ultrasound image model. Combined with the ultrasound image data, perform dynamic tissue feature analysis on the three-dimensional ultrasound image model to generate an initial tissue feature evaluation model.

[0011] Dynamic vascular distribution data and surgical operation constraints are loaded into the initial tissue feature evaluation model to generate a comprehensive tissue feature evaluation model. The puncture point dynamic simulation is performed in combination with the operation state parameters to obtain puncture point control simulation data.

[0012] A dynamic vascular prediction model is constructed and trained using puncture point control simulation data to obtain a puncture point adjustment model, thereby generating initial adjustment parameters, wherein the initial adjustment parameters include at least an initial puncture point position, an initial obstacle avoidance path, and an initial adjustment response parameter;

[0013] Based on the initial adjustment parameters, the comprehensive tissue characteristic evaluation model and the operation state parameters, dynamic adjustment simulation information is obtained;

[0014] By combining dynamic adjustment of simulation information with surgical plan design parameters, a multi-path optimization model is constructed and the design parameters are iteratively optimized to obtain the optimal path planning parameters;

[0015] Based on the real-time ultrasound imaging model, the path planning parameters of the current period are inferred to generate the safe puncture probability of the current period. Combined with the optimal path planning parameters of the current period and the actual path parameters, the real-time puncture strategy of the surgical instrument is adjusted to achieve the dynamic puncture point matching goal.

[0016] Preferably, the puncture point control simulation data at least includes tissue density distribution, blood vessel density distribution, safe puncture area set and dynamically adjusted priority sequence;

[0017] The dynamic adjustment simulation information at least includes a puncture point offset, a path obstacle avoidance success rate, and a stability evaluation index.

[0018] Preferably, performing multi-level ultrasound scanning of the surgical site and constructing a three-dimensional ultrasound image model, performing dynamic tissue feature analysis on the three-dimensional ultrasound image model in combination with ultrasound image data, and generating an initial tissue feature evaluation model comprises the following steps:

[0019] Perform multi-angle scanning of the surgical site with an ultrasound probe to obtain ultrasound image sequences of tissue morphology and vascular distribution;

[0020] Preprocessing the ultrasound image sequence to obtain standardized ultrasound image data, wherein the preprocessing includes one or more of image denoising, coordinate registration, feature extraction, grid subdivision, and data alignment;

[0021] Generate a 3D ultrasound image model based on standardized ultrasound image data and combined with 3D reconstruction technology;

[0022] Dynamic tissue feature analysis is performed on the three-dimensional ultrasound image model to generate an initial tissue feature assessment model, wherein the dynamic tissue feature analysis includes at least grid weight allocation and feature parameter mapping, and tissue safety levels are divided by grid weight allocation, and corresponding ultrasound image data are assigned to each level area.

[0023] Preferably, the method of loading the dynamic vascular distribution data and surgical operation constraints into the initial tissue feature evaluation model to generate a comprehensive tissue feature evaluation model, performing dynamic simulation of the puncture point in combination with the operation state parameters, and obtaining puncture point control simulation data comprises the following steps:

[0024] loading dynamic vascular distribution data into the initial tissue characteristic evaluation model to simulate real-time vascular changes, thereby generating a first tissue characteristic evaluation model;

[0025] Loading surgical operation constraints on the first tissue feature evaluation model to generate a comprehensive tissue feature evaluation model, wherein the surgical operation constraints include a maximum puncture angle limit, a minimum turning radius, and instrument operation redundancy parameters;

[0026] Based on the comprehensive tissue feature evaluation model and operation status parameters, dynamic simulation of the puncture point is performed. The specific process includes:

[0027] Constructing a multi-constraint path planning equation, which includes at least an obstacle avoidance distance equation, a trajectory smoothness equation, and an operation difficulty equation. Combining this with a comprehensive tissue feature assessment model, the multi-constraint path planning equation is numerically solved using a piecewise linearization method to obtain tissue density distribution, vascular density distribution, a set of safe puncture areas, and a dynamically adjusted priority sequence.

[0028] Obtaining a safe puncture area set includes the following steps:

[0029] Based on the vascular density distribution, the vascular coverage index of each grid cell is calculated;

[0030] Identify the areas where the vascular coverage index is not greater than the preset threshold in the comprehensive tissue feature evaluation model and generate a set of safe puncture areas, which are expressed as follows:

[0031] The safe puncture area set is a spatial distribution mapping result of a blood vessel coverage index and a preset threshold.

[0032] Preferably, the construction of the dynamic blood vessel prediction model and training through puncture point control simulation data to obtain a puncture point adjustment model, and then generating initial adjustment parameters, includes the following steps:

[0033] Constructing a dynamic vascular prediction model based on three-dimensional ultrasound image features;

[0034] The dynamic vascular prediction model is trained and verified through puncture point control simulation data to obtain a puncture point adjustment model;

[0035] Input the real-time operation status parameters into the puncture point adjustment model to predict the set of safe puncture areas;

[0036] Based on the predicted safe puncture area set, initial adjustment parameters are generated, wherein the initial adjustment parameters at least include an initial puncture point position, an initial obstacle avoidance path, and an initial adjustment response parameter.

[0037] Preferably, the method further includes reconstructing the puncture point control simulation data, specifically:

[0038] Based on tissue density distribution, blood vessel density distribution and safe puncture area set, an initial multi-dimensional path input tensor is constructed;

[0039] Normalize and perform feature layering on the initial multi-dimensional path input tensor to generate the final multi-dimensional path input tensor;

[0040] Based on the dynamic adjustment priority sequence, the puncture point adjustment label tensor is constructed;

[0041] Combine the final multi-dimensional path input tensor and the puncture point adjustment label tensor to form a training sample set.

[0042] Preferably, generating the initial adjustment parameters based on the predicted safe puncture area set comprises the following steps:

[0043] Extract the grid cell with the lowest blood vessel density in the predicted safe puncture area set to generate the initial puncture point position;

[0044] According to the spatial connectivity of the predicted safe puncture area set, the feasible topological structure of the initial obstacle avoidance path is fitted to generate the initial adjustment response parameters;

[0045] The tissue density gradient direction of the predicted safe puncture area set is calculated and normalized into a path reference vector, which is the initial obstacle avoidance path direction.

[0046] Preferably, obtaining dynamic adjustment simulation information based on the initial adjustment parameters, the comprehensive tissue characteristic evaluation model and the operation state parameters comprises the following steps:

[0047] The initial puncture point position is mapped to the comprehensive tissue feature evaluation model, the initial obstacle avoidance path is matched with the initial adjustment response parameters, and the path nodes in the adjustment area are encrypted. The comprehensive tissue feature evaluation model is updated, and the dynamic adjustment trigger conditions, path update rules, and response parameter increments are defined to generate a dynamic adjustment simulation model.

[0048] Based on the dynamic adjustment simulation model, the multi-constraint path planning equations are iteratively solved by the piecewise linearization method to obtain dynamic adjustment simulation information, including:

[0049] When the dynamic adjustment trigger conditions are met, the puncture point offset, path obstacle avoidance success rate, and stability evaluation index are updated, and the multi-constraint path planning equation is re-solved until the simulation termination conditions are met;

[0050] The dynamic adjustment trigger condition includes triggering the response parameter update when the current path obstacle avoidance success rate is no greater than a preset success rate threshold; the path update rule includes adjusting the initial obstacle avoidance path based on the tissue density gradient direction; the response parameter increment is in a piecewise linear relationship with the current stability evaluation index.

[0051] Preferably, the construction of the multi-path optimization model and iterative optimization of the design parameters to obtain the optimal path planning parameters includes the following steps:

[0052] A multi-path optimization model was constructed, where the optimization variables included path node density, turning angle threshold, and operation force distribution ratio. The optimization objectives included minimizing path length and maximizing adjustment response speed. The constraints included the upper limit of surgical instrument operation and the ultrasound imaging accuracy threshold.

[0053] The multi-path optimization model is preliminarily solved using the A* algorithm to generate an initial optimized path population;

[0054] Based on the initial optimized path population, the Dijkstra algorithm is used for global optimization to generate the optimal path planning parameters.

[0055] Preferably, the present invention further includes an interventional ultrasound system for minimally invasive surgery, the system comprising:

[0056] A data acquisition module is used to obtain ultrasonic image data of the patient's surgical site and operating status parameters of surgical instruments;

[0057] The image modeling and analysis module is used to perform multi-level ultrasound scanning of the surgical site and construct a three-dimensional ultrasound image model. The three-dimensional ultrasound image model is then dynamically analyzed for tissue characteristics based on the ultrasound image data to generate an initial tissue characteristic assessment model.

[0058] The comprehensive evaluation and simulation module is used to load dynamic vascular distribution data and surgical operation constraints into the initial tissue feature evaluation model to generate a comprehensive tissue feature evaluation model. It then combines the operation status parameters to perform dynamic simulation of the puncture point and obtain puncture point control simulation data.

[0059] A parameter generation module is used to construct a dynamic vascular prediction model and train it through puncture point control simulation data to obtain a puncture point adjustment model, thereby generating initial adjustment parameters, wherein the initial adjustment parameters at least include an initial puncture point position, an initial obstacle avoidance path, and an initial adjustment response parameter;

[0060] A dynamic adjustment simulation module is used to obtain dynamic adjustment simulation information based on initial adjustment parameters, a comprehensive tissue feature evaluation model, and operation status parameters;

[0061] The path optimization module is used to combine dynamic adjustment simulation information with surgical plan design parameters, build a multi-path optimization model, and iteratively optimize the design parameters to obtain the optimal path planning parameters;

[0062] The strategy adjustment module is used to infer the path planning parameters of the current period based on the real-time ultrasound imaging model, generate the safe puncture probability of the current period, combine the optimal path planning parameters of the current period with the actual path parameters, and adjust the real-time puncture strategy of the surgical instrument to achieve the dynamic puncture point matching goal.

[0063] Compared with the prior art, the present invention has the following beneficial effects:

[0064] In terms of image modeling and analysis, ultrasound probes are used to scan the surgical site from multiple angles to acquire ultrasound image sequences of tissue morphology and vascular distribution. Preprocessing steps, such as image denoising and coordinate registration, are then performed, and three-dimensional reconstruction techniques are used to generate a three-dimensional ultrasound image model. Dynamic tissue feature analysis, performed through grid weight assignment and feature parameter mapping, enables comprehensive and precise classification of tissue safety levels, providing more realistic and detailed tissue feature information for subsequent puncture path planning. This effectively addresses the limitations of traditional single-layer scanning modeling and improves the accuracy of understanding the anatomical structure of the surgical site.

[0065] In the comprehensive evaluation and dynamic simulation of puncture points, the initial tissue feature evaluation model is loaded with dynamic vascular distribution data to simulate real-time vascular changes, and then loaded with surgical operation constraints such as the maximum puncture angle limit to generate a comprehensive tissue feature evaluation model. The multi-constraint path planning equation is constructed in combination with the operation state parameters, and numerically solved using the piecewise linearization method. It can simultaneously consider multiple factors such as vascular distribution, tissue density, and instrument operation constraints, and accurately calculate puncture point control simulation data such as tissue density distribution, vascular density distribution, and safe puncture area collection. This process realizes the dynamic simulation of the puncture path and accurate evaluation under multiple constraints, effectively circumventing the problem that traditional static models cannot adapt to dynamic vascular changes and instrument operation limitations, and greatly improves the safety and feasibility of the puncture path.

[0066] In terms of parameter generation and dynamic adjustment simulation, a dynamic vascular prediction model is constructed based on three-dimensional ultrasound image features, and is trained and verified using puncture point control simulation data to obtain a puncture point adjustment model. This model can predict the set of safe puncture areas based on real-time operation status parameters, and then generate initial adjustment parameters including the initial puncture point position and initial obstacle avoidance path. At the same time, by constructing a dynamic adjustment simulation model, dynamic adjustment simulation information such as the puncture point offset is updated in real time when the dynamic adjustment trigger conditions are met, thereby achieving dynamic optimization of the puncture path. This mechanism based on dynamic prediction and real-time adjustment enables the system to quickly respond to changes in tissue and blood vessels during surgery, significantly improving the adaptability and accuracy of puncture operations.

[0067] In terms of multipath optimization and strategy adjustment, a multipath optimization model was constructed using path node density as optimization variables, with minimizing path length and maximizing adjustment response speed as optimization objectives. Iterative optimization was performed using the A* and Dijkstra algorithms to obtain the optimal path planning parameters. This optimization process achieved a balance between multiple objectives, ensuring optimal overall performance of the puncture path. Furthermore, a real-time ultrasound imaging model was used to generate the probability of safe puncture during the current time period. The real-time puncture strategy was adjusted by combining the optimal path planning parameters with the actual path parameters, enabling dynamic puncture point matching. This further improved the safety and efficiency of surgical procedures, reducing operative time and patient trauma. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is a working principle diagram of the interventional ultrasound method and system for minimally invasive surgery according to the present invention;

[0069] Figure 2 Design diagram generated from simulation data for puncture point control;

[0070] Figure 3 Design graph generated for initial tuning parameters;

[0071] Figure 4 Design drawings generated for dynamic adjustment of simulation information;

[0072] Figure 5 Design diagram constructed for the multi-path optimization model. DETAILED DESCRIPTION

[0073] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0074] See also Figure 1-Figure 5 The present invention relates to an interventional ultrasound method and system for minimally invasive surgery, and the specific implementation steps are as follows:

[0075] Data acquisition steps: Ultrasonic image data of the patient's surgical site is obtained through ultrasonic imaging equipment, and at the same time, operating status parameters are collected in real time through sensors carried by surgical instruments. The operating status parameters include at least instrument position, movement speed and puncture angle.

[0076] Three-dimensional modeling and initial assessment steps: Perform multi-level ultrasound scanning of the surgical site and use three-dimensional reconstruction technology to construct a three-dimensional ultrasound image model; combine ultrasound image data, divide tissue safety levels through grid weight distribution, and assign corresponding ultrasound image data to each level area to complete dynamic tissue feature analysis and generate an initial tissue feature assessment model.

[0077] Comprehensive evaluation and dynamic simulation steps: load dynamic vascular distribution data into the initial tissue feature evaluation model to simulate real-time vascular changes and generate a first tissue feature evaluation model; further load surgical operation constraints (including maximum puncture angle limit, minimum turning radius and instrument operation redundancy parameters) to generate a comprehensive tissue feature evaluation model; based on this model and operation state parameters, construct a multi-constraint path planning equation including obstacle avoidance distance equation, trajectory smoothness equation and operation difficulty equation, and numerically solve it through piecewise linearization method to obtain puncture point control simulation data.

[0078] Parameter generation steps: A dynamic vascular prediction model is constructed based on three-dimensional ultrasound image features, and it is trained and verified using puncture point control simulation data to obtain a puncture point adjustment model; real-time operation status parameters are input into the model to predict the set of safe puncture areas and generate initial adjustment parameters including the initial puncture point position, initial obstacle avoidance path, and initial adjustment response parameters.

[0079] Dynamic adjustment simulation steps: map the initial adjustment parameters to the comprehensive tissue feature evaluation model, encrypt and update the model through path nodes, define the dynamic adjustment trigger conditions, path update rules and response parameter increments, and build a dynamic adjustment simulation model; iteratively solve the multi-constraint path planning equation based on the model to obtain dynamic adjustment simulation information including puncture point offset, path obstacle avoidance success rate and stability evaluation index.

[0080] Path optimization steps: Construct a multi-path optimization model with path node density, turning angle threshold, and operation force distribution ratio as optimization variables, and minimize path length and maximize adjustment response speed as optimization goals. Combined with the surgical plan design parameters, the initial optimized path population is generated through the A* algorithm, and then the Dijkstra algorithm is used for global optimization to obtain the optimal path planning parameters.

[0081] Strategy adjustment steps: Based on the real-time ultrasound imaging model, the path planning parameters of the current period are inferred to generate the probability of safe puncture; combining the optimal path planning parameters with the actual path parameters, the surgical instrument puncture strategy is dynamically adjusted to achieve the puncture point matching goal.

[0082] The present invention will be further described below in conjunction with Examples 1 to 5:

[0083] Example 1:

[0084] In the three-dimensional modeling and initial assessment steps, the specific implementation method is as follows: First, the surgical site is scanned at multiple angles using an ultrasound probe, with the scanning angle range covering 0° to 180°, and the interval between adjacent scanning angles not exceeding 15°, to ensure that complete tissue morphology and vascular distribution information is obtained. The ultrasound probe can be a convex array probe or a linear array probe, and the appropriate type is selected according to the depth and range of the surgical site. For example, a convex array probe is used for abdominal surgery, and a linear array probe is used for superficial tissue surgery. During the scanning process, the probe remains in contact with the skin surface, and the probe movement is controlled manually or by a robotic arm to obtain a continuous ultrasound image sequence, which contains multiple frames of two-dimensional ultrasound images, with each frame having a resolution of not less than 512×512 pixels and a frame rate of not less than 30 frames per second.

[0085] After acquiring an ultrasound image sequence, it must be preprocessed to obtain standardized ultrasound image data. This preprocessing process sequentially includes image denoising, coordinate registration, feature extraction, grid subdivision, and data alignment. Image denoising utilizes a non-local means filtering algorithm, which calculates the similarity between image blocks and performs a weighted average, effectively suppressing speckle noise while preserving edge details. Coordinate registration unifies ultrasound images acquired from different scanning angles into a common coordinate system. A registration method based on feature point matching is employed. The scale-invariant feature transform (SIFT) algorithm is first used to extract feature points from each frame. Mismatched points are then removed using the random sampling consensus (RANSAC) algorithm. Finally, coordinate transformation is achieved using an affine transformation matrix. In the feature extraction phase, a Gaussian pyramid combined with the Laplacian operator is used to extract edge, texture, and corner features from the image and generate a feature image. Grid subdivision divides the ultrasound image into regular grid cells. The grid size is determined by the ultrasound imaging resolution. For example, when the resolution is 0.5 mm, the grid side length is set to 1 mm to ensure that each grid cell contains sufficient tissue feature information. Data alignment is to spatially correspond the feature-extracted image to the grid cells so that each grid cell corresponds to a specific area in the image.

[0086] Based on standardized ultrasound image data, a 3D ultrasound image model is generated using 3D reconstruction technology. This 3D reconstruction utilizes volume rendering technology. The specific steps are: first, the standardized 2D ultrasound image sequence is interpolated to fill the gaps between adjacent slices. Cubic spline interpolation is used to improve the smoothness of the reconstructed model. The interpolated image sequence is then converted into volume data, with the grayscale value of each voxel corresponding to the echo intensity of the ultrasound image. Finally, the volume data is rendered using a ray casting algorithm to generate a realistic 3D ultrasound image model that can intuitively display the tissue layers and vascular distribution of the surgical site.

[0087] Dynamic tissue feature analysis is performed on the three-dimensional ultrasound image model to generate an initial tissue feature assessment model. Dynamic tissue feature analysis includes at least grid weight allocation and feature parameter mapping. During the grid weight allocation process, a safety level weight is assigned to each grid unit based on the physiological characteristics of the tissue and the degree of surgical risk. First, the vascular density parameter in the grid unit is calculated. The vascular density is determined by the proportion of vascular pixels in the grid unit. The calculation formula is: vascular density = number of vascular pixels / total number of grid unit pixels. Then, the grid units are divided into three safety levels: high, medium, and low according to the vascular density. For example, when the vascular density is greater than 0.4, it is classified as a high risk level, between 0.2 and 0.4, it is classified as a medium risk level, and less than 0.2, it is classified as a low risk level. Different weight coefficients are assigned to grid units of different safety levels. The weight coefficient of the high risk level is 1.5, the medium risk level is 1.0, and the low risk level is 0.5.

[0088] Feature parameter mapping involves associating feature parameters in ultrasound image data with grid cells. Feature parameters include grayscale value, texture features (such as entropy, energy, and correlation), vascular diameter, and blood flow velocity. Grayscale value directly reflects the echogenicity of tissue. Hyperechoic regions correspond to calcification or bone tissue, while hypoechoic regions correspond to fluid or cystic tissue. Texture features are calculated using the gray-level co-occurrence matrix (GLCM). Entropy reflects the complexity of image texture, energy reflects texture uniformity, and correlation reflects the correlation between texture elements. Vascular diameter is calculated by extracting vascular boundaries using edge detection algorithms (such as the Canny operator), and blood flow velocity is obtained from color Doppler ultrasound data. These feature parameters are normalized to the [0, 1] range and then mapped to the corresponding grid cells, forming a feature vector for each grid cell.

[0089] An initial tissue feature assessment model is generated through grid weight assignment and feature parameter mapping. This model, based on a three-dimensional grid structure, contains safety level weights and feature vector information for each grid cell, comprehensively reflecting the tissue characteristics and risk distribution of the surgical site. During subsequent puncture point planning and path optimization, this initial tissue feature assessment model provides fundamental data support for comprehensive evaluation and dynamic simulation, enabling dynamic adjustment of puncture strategies based on tissue characteristics, improving surgical safety and accuracy.

[0090] Throughout the 3D modeling and initial assessment process, each step is closely linked, and the data processing process is strictly logical and standardized. Multi-angle scanning of ultrasound images ensures data integrity, preprocessing steps improve data quality and consistency, 3D reconstruction technology enables the conversion from 2D images to 3D models, and dynamic tissue feature analysis gives the model the evaluation capabilities required for clinical application. The parameter settings and algorithm selections for each link are based on ultrasound imaging principles and clinical practice requirements, ensuring that the generated initial tissue feature assessment model can accurately reflect the actual conditions of the surgical site and provide a reliable basis for subsequent interventional ultrasound surgery.

[0091] Example 2:

[0092] The comprehensive assessment and dynamic simulation steps are implemented as follows: First, dynamic vascular distribution data is loaded into the initial tissue feature assessment model to simulate real-time vascular changes, generating a first tissue feature assessment model. The dynamic vascular distribution data is derived from a preoperative vascular dynamic sequence acquired via four-dimensional ultrasound. This sequence contains information on vascular position, caliber, and blood flow velocity changes during the cardiac cycle, with a sampling frequency of at least 50 Hz and a temporal resolution of milliseconds. During the loading process, a temporal interpolation algorithm is used to spatially and temporally match the dynamic vascular distribution data with the three-dimensional mesh of the initial tissue feature assessment model. Specifically, the spatial displacement vectors of the vascular mesh cells in the initial model are calculated based on the changes in vascular position in the time series. Cubic spline interpolation is used to smooth the displacement vectors to ensure the continuity of vascular dynamic changes. Simultaneously, the attribute parameters of the vascular mesh cells are adjusted based on the blood flow velocity data. For example, a higher energy loss coefficient is set for areas with high blood flow velocity to facilitate resistance calculation in subsequent path planning.

[0093] Subsequently, surgical operation constraints are applied to the first tissue feature assessment model to generate a comprehensive tissue feature assessment model. These constraints include a maximum puncture angle limit, a minimum turning radius, and instrument operation redundancy parameters. The maximum puncture angle limit is determined based on the mechanical properties of the surgical instrument. For example, the maximum puncture angle for flexible needles is set at 60°, while that for rigid needles is set at 45°. This angle is the angle between the needle axis and the tissue surface normal. The minimum turning radius is based on the bending performance parameters of the instrument. For surgical instruments with active bending capabilities, the minimum turning radius is set at twice the instrument diameter, and for passive bending instruments, it is set at three times. The instrument operation redundancy parameter is defined as the ratio of the maximum output power of the instrument power system to the actual power used. It is typically set at 1.2-1.5 to ensure sufficient power reserve when the instrument encounters resistance. When applying the constraints, each parameter is embedded into the grid cell attributes of the comprehensive tissue feature assessment model through a constraint matrix. Each grid cell corresponds to a set of constraint vectors, including the maximum allowable puncture angle, the minimum turning radius threshold, and the power redundancy coefficient.

[0094] Based on the comprehensive tissue feature evaluation model and operation state parameters, dynamic simulation of the puncture point is performed. First, a multi-constraint path planning equation is constructed, which includes at least the obstacle avoidance distance equation, the trajectory smoothness equation, and the operation difficulty equation. The obstacle avoidance distance equation is used to ensure that the puncture path maintains a safe distance from dangerous structures such as blood vessels. The expression is: d obstacle ≥d safe , where d obstacle is the distance between the puncture path and the dangerous structure, d safe The preset safety distance is 2-5 mm, which is adjusted according to the tissue risk level. The trajectory smoothness equation is measured by the integral of the path curvature, and the expression is: where κ(s) is the curvature of the path at position s, C smooth is the smoothness threshold, with a value range of 0.5-1.0rad / mm. The operation difficulty equation takes into account factors such as puncture force and angle change rate, and the expression is: F force ≤F max and Among them F force is the puncture force, F max is the maximum permissible force of the device (usually 5-10N), is the puncture angle change rate, is the maximum allowable angle change rate (taken as 10° / s).

[0095] The multi-constraint path planning equation is numerically solved using a piecewise linearization method. First, the puncture path is discretized into N nodes, and the node spacing is determined according to the ultrasound imaging resolution, usually 1-3 mm. The path segments between adjacent nodes are regarded as straight line segments, and the node positions are adjusted to meet the various constraints. For each node, the parameters such as the vascular density and tissue hardness of the grid unit where it is located are calculated and substituted into the multi-constraint path planning equation for solution. During the solution process, the obstacle avoidance distance constraint is first processed to ensure that the distance between the node position and the dangerous structure is greater than the safe distance; then the trajectory smoothness is optimized, and the path curvature integral is minimized by adjusting the node coordinates; finally, the operation difficulty constraint is verified. If the puncture force or angle change rate exceeds the threshold, the node position is adjusted and recalculated.

[0096] During the solution process, the tissue density distribution, vascular density distribution, safe puncture area set, and dynamically adjusted priority sequence are obtained. The safe puncture area set is generated as follows: Based on the vascular density distribution, the vascular coverage index of each grid cell is calculated. The vascular coverage index is defined as the ratio of the vascular pixel area within the grid cell to the total area of ​​the grid cell. The calculation formula is: Among them A v is the pixel area of ​​blood vessels, A g =The total area of ​​the grid cells. A threshold segmentation algorithm was used to identify areas within the comprehensive tissue feature assessment model where the vascular coverage index (VCI) was no greater than a preset threshold. The threshold was determined based on the surgical risk level: 0.3 for low-risk surgery, 0.2 for medium-risk surgery, and 0.1 for high-risk surgery. Grid cells that met these criteria constituted a safe puncture area set. This set was the spatial distribution mapping of the VCI and the preset threshold, stored as a three-dimensional grid index. Each safe grid cell recorded its coordinates, VCI, and tissue density parameters.

[0097] The generation of the dynamically adjusted priority sequence is based on the comprehensive assessment of tissue density and vascular density. First, the risk index of each grid cell is calculated as follows: R = w1·D t +w2·D v , where D t is tissue density, D v is the vascular density, w1 and w2 are weight coefficients (typically w1 = 0.6 and w2 = 0.4). Regions with higher risk indices have higher dynamic adjustment priorities. All grid cells are sorted from high to low risk indices to generate a dynamic adjustment priority sequence. This sequence is used to guide the order of puncture path adjustments, ensuring that path planning for high-risk areas is addressed first.

[0098] Throughout the comprehensive evaluation and dynamic simulation process, data interaction and algorithm execution are strictly time-sequential. The loading of dynamic vascular distribution data ensures that the model can reflect vascular movement in real time. The embedding of surgical operation constraints makes the simulation results consistent with the physical characteristics of the instrument. The construction of multi-constraint path planning equations comprehensively considers safety, smoothness, and operational feasibility. The piecewise linearization solution method transforms complex continuous optimization problems into computable numerical problems through discretization processing. The generation of safe puncture area sets and dynamically adjusted priority sequences provides key data support for subsequent puncture point adjustments and path optimization. The parameter settings and algorithm selections of each link are based on clinical surgical requirements and instrument performance parameters, ensuring that the simulation results can be directly applied to the formulation of puncture strategies in actual surgery, providing reliable technical support for the precise implementation of minimally invasive surgery.

[0099] Example 3:

[0100] In the parameter generation step, the dynamic vascular prediction model is constructed and trained as follows: First, a dynamic vascular prediction model is constructed based on 3D ultrasound image features. 3D ultrasound image features encompass both spatial geometric and dynamic characteristics of blood vessels. Spatial geometric features include parameters such as vessel diameter, branching angles, and tortuosity. These are calculated by extracting vessel boundaries using an edge detection algorithm. For example, the Canny operator is used to detect vessel cross-sections in 3D ultrasound images and ellipse fitting is used to obtain diameter parameters. Dynamic features include blood flow velocity and vessel wall motion amplitude. These are calculated using color Doppler ultrasound data and an optical flow algorithm. The optical flow algorithm tracks the displacement of vessel wall pixels across consecutive frames to determine motion velocity. The dynamic vascular prediction model utilizes a convolutional neural network (CNN) architecture, consisting of an input layer, convolutional layers, pooling layers, a fully connected layer, and an output layer. The input layer receives 3D ultrasound image volume data. The convolutional layer extracts multi-scale vascular features using convolutional kernels of varying scales. The pooling layer reduces feature dimensionality and improves translation invariance. The fully connected layer integrates global features. The output layer predicts the location, morphology, and blood flow parameters of blood vessels in future time periods.

[0101] Next, the dynamic vascular prediction model is trained and verified using puncture point control simulation data. Before training, the puncture point control simulation data needs to be reconstructed. The specific steps are as follows: First, based on the tissue density distribution, blood vessel density distribution, and safe puncture area set, an initial multi-dimensional path input tensor is constructed. The tissue density distribution is obtained by grayscale value statistics of ultrasound images. The grayscale value is positively correlated with tissue density. The calculation formula is: ρ t =k·mean(G)+b, where ρ tis the tissue density, G is the grayscale value matrix within the grid cell, k and b are calibration coefficients (determined by in vitro tissue sample measurement); the vascular density distribution is the proportion of vascular pixels within the grid cell; the safe puncture area set is represented by a three-dimensional binary mask, 1 represents a safe area, and 0 represents a dangerous area. These three types of data are spliced ​​according to the channel dimension to form the initial multi-dimensional path input tensor Where H, W, and D are the height, width, and depth of the three-dimensional grid, respectively.

[0102] The initial multi-dimensional path input tensor is normalized and feature-stratified to generate the final multi-dimensional path input tensor. The standardization process uses the Z-score standardization method to calculate the mean μ and standard deviation σ of the data of each channel, and then calculate the final value of ... Normalization is performed so that the data of each channel obeys the standard normal distribution. Feature hierarchical processing maps tissue density, vascular density, and safe area to different feature levels. For example, tissue density is divided into three levels: low, medium, and high (the corresponding normalized value range is [-1, -0.3), [-0.3, 0.3], and (0.3, 1]). Vascular density is divided into three levels: sparse, medium, and dense. The safe area maintains binary attributes. The semantic distinction of the features is enhanced through hierarchical mapping. The final generated tensor

[0103] Based on the dynamic adjustment priority sequence, the puncture point adjustment label tensor is constructed. The dynamic adjustment priority sequence is an index list sorted according to the risk index of the grid cell (obtained by the weighted sum of tissue density and vascular density, the higher the risk index, the higher the adjustment priority). It is constructed in the following way: for each grid cell, if its index in the dynamically adjusted priority sequence is i, then the label value is (N is the total number of grid cells), so that the label values ​​are distributed in the interval [0, 1]. Larger values ​​indicate higher adjustment priorities. This label tensor is used to supervise the model to learn the priority mapping relationship of puncture point adjustment.

[0104] Combine the final multi-dimensional path input tensor and the puncture point adjustment label tensor according to the sample dimension to form a training sample set Where M is the total number of samples. The dynamic vascular prediction model was trained using a supervised learning method. The loss function was the mean square error (MSE). The Adam optimizer was selected as the optimizer, and the initial learning rate was set to 10. -4 During training, early stopping was used to prevent overfitting. Training was terminated when the validation set loss did not decrease for five consecutive epochs. After training, the model was validated using test samples containing different tissue types and vascular distributions to ensure that the model could accurately predict the set of safe puncture areas.

[0105] Real-time operating state parameters are input into the puncture point adjustment model to predict the set of safe puncture areas. These parameters, including the current position coordinates (x, y, z) of the surgical instrument, puncture velocity v, and puncture angle θ, are collected in real time by the six-dimensional force sensor and inertial measurement unit (IMU) at the end of the instrument at a sampling frequency of 100 Hz. Based on the dynamic vascular prediction model and combined with real-time operating parameters, the puncture point adjustment model outputs the probability distribution of the set of safe puncture areas through forward propagation calculation. Grid cells with probability values ​​greater than a preset threshold (e.g., 0.5) are identified as safe areas.

[0106] Based on the predicted safe puncture area set, the initial adjustment parameters are generated. First, the grid cell with the lowest vascular density in the predicted safe puncture area set is extracted. The vascular density is determined by the second channel value of this grid cell in the input tensor (the normalized vascular density parameter). The smaller the value, the lower the vascular density. The center coordinates of the grid cell with the smallest value are selected as the initial puncture point location (x0, y0, z0).

[0107] Based on the spatial connectivity of the predicted set of safe puncture areas, a feasible topological structure for the initial obstacle avoidance path is fitted. Spatial connectivity analysis is implemented using a three-dimensional connected domain detection algorithm. Adjacent safe grid cells (six-neighborhood connectivity) are grouped into the same connected domain. The volume and center of mass of each connected domain are calculated, and the connected domain with the largest volume is selected as the primary path channel. The feasible topological structure of the initial obstacle avoidance path is based on the centroid line of the connected domain. A smooth path is formed by Bezier curve fitting. The control points of the curve are determined by the coordinates of the boundary grid cells of the connected domain to ensure that the path is completely within the safe puncture area.

[0108] The tissue density gradient direction of the predicted safe puncture area set is calculated and normalized to the path reference vector. The tissue density gradient is calculated by numerically differentiating the three-dimensional tissue density field, and the gradient vector at each grid cell is The central difference method is used to approximate the solution. The average gradient direction is obtained by taking the average value of the gradient vectors of all grid cells in the safe puncture area set. After normalization, we get the path reference vector This vector is the direction of the initial obstacle avoidance path, which is used to guide the puncture path to extend along the direction where the tissue density changes most gently, so as to reduce the puncture resistance.

[0109] Initial adjustment response parameters, including the path adjustment step size and angle increment, are determined based on the spatial dimensions of the safe puncture zone and the instrument's operational accuracy. For example, if the average width of the safe zone is 5 mm, the path adjustment step size is set to 1 mm, and the angle increment is set to 5°. This ensures that each adjustment effectively avoids hazardous structures without causing excessive path tortuosity.

[0110] Throughout the entire parameter generation process, from data reconstruction to model training and parameter generation, each link strictly adheres to the basic principles of signal processing and machine learning. The multi-dimensional extraction of three-dimensional ultrasound image features ensures the richness of the model input, the standardization and feature layering processing in the data reconstruction process improves the learnability of the data, and the construction of dynamically adjusted priority labels provides a clear supervisory signal for the model. Based on the prediction of the safe area of ​​real-time operating parameters and the generation of initial parameters of geometric features, the system can dynamically generate an adaptive puncture adjustment strategy according to the surgical process, realizing the intelligent and precise planning of puncture points in interventional ultrasound surgery. The algorithm selection and parameter setting of each step are based on actual clinical needs and instrument performance parameters, ensuring that the generated initial adjustment parameters have practical guiding significance and can be directly applied to the motion control of surgical instruments.

[0111] Example 4:

[0112] In the dynamic adjustment simulation step, the specific implementation method is as follows: First, the initial puncture point position is mapped to the comprehensive tissue feature evaluation model, which contains information such as the tissue density of the surgical site, vascular distribution, and surgical operation constraints. For example, assuming that the initial puncture point position corresponds to the grid unit with coordinates (x0, y0, z0) in the model, the unit is evaluated as a low-risk area, the vascular coverage index is 0.1, and the tissue density parameter is at a medium level. Next, the initial obstacle avoidance path is matched with the initial adjustment response parameters. The initial obstacle avoidance path is generated based on the spatial connectivity of the set of safe puncture areas. For example, starting from the initial puncture point, it extends in the direction of the lowest vascular density, passes through multiple continuous low-risk grid units, and forms a preliminarily planned straight path; the initial adjustment response parameters are set according to the instrument operation accuracy, such as the path adjustment step size is 1.5mm, and the angle increment is 8°.

[0113] Subsequently, the path nodes of the adjusted area are encrypted to improve the simulation accuracy. During the encryption process, new nodes are inserted into the initial path at a fixed spacing (such as 0.8mm), so that the total number of nodes is increased to 3 times the original, thereby capturing the changes in tissue characteristics along the path in more detail. For example, if the initial path contains 10 nodes, it is expanded to 30 nodes after encryption. Each node corresponds to a grid unit in the model, recording parameters such as vascular density, tissue hardness, and operational constraints at that location. After the node encryption is completed, the comprehensive tissue feature evaluation model is updated, and the attribute parameters of the new nodes are incorporated into the model calculation to ensure that the model can reflect more detailed path details.

[0114] Dynamic adjustment trigger conditions, path update rules, and response parameter increments are defined to generate a dynamic adjustment simulation model. The dynamic adjustment trigger condition is set to the current path's obstacle avoidance success rate being no greater than a preset success rate threshold, such as 85%. When the calculated obstacle avoidance success rate during the simulation falls below this threshold, the path adjustment mechanism is triggered. The path update rule adjusts the initial obstacle avoidance path based on the direction of the tissue density gradient. For example, if a large tissue density gradient is detected in an area ahead of the path (indicating the presence of hard tissue or potential obstacles), the path is adjusted in the opposite direction of the gradient to avoid areas of high resistance. The response parameter increment exhibits a piecewise linear relationship with the current stability assessment index, which is calculated from parameters such as path curvature and puncture force fluctuation. For example, when the stability assessment index is greater than 0.6 (indicating poor path stability), the response parameter increment is set to a larger value (e.g., a step size increment of 0.5 mm); when the index is less than or equal to 0.6, the increment is set to a smaller value (e.g., a step size increment of 0.2 mm) to balance the adjustment amplitude with path smoothness.

[0115] Based on the dynamic adjustment simulation model, the multi-constraint path planning equations are iteratively solved by the piecewise linearization method to obtain dynamic adjustment simulation information. The piecewise linearization method divides the entire puncture path into multiple segments, each segment corresponds to a linear optimization problem, and the optimal path is approached by solving and connecting the segments. In the iterative solution process, the various parameters of the initial path are first calculated, such as obstacle avoidance distance, trajectory smoothness, and operation difficulty. For example, the obstacle avoidance distance of the initial path is 3.2mm (meeting the preset safety distance requirement of 2mm), the integral value of the trajectory curvature is 0.8rad / mm (within the smoothness threshold range of 0.5-1.0rad / mm), and the puncture force is 4.5N (lower than the maximum allowable force of the instrument 5N). At this time, the obstacle avoidance success rate is calculated to be 90%, which is higher than the preset threshold and the path does not need to be adjusted.

[0116] When a certain iterative step in the simulation process detects that the dynamic displacement of the blood vessels causes the local area blood vessel coverage index to increase, the obstacle avoidance distance of the original path is shortened to 1.8mm (lower than the safety distance), and the obstacle avoidance success rate drops to 80%, triggering the dynamic adjustment trigger condition. At this time, according to the path update rule, the path direction is adjusted along the direction of the tissue density gradient (assuming that the gradient direction points to the right), and the puncture angle of the current node is increased by 5°. At the same time, according to the stability evaluation index (calculated as 0.7 at this time, which is a high fluctuation), the response parameter increment is set to 0.5mm, that is, the path adjustment step size is increased from 1.5mm to 2.0mm. After recalculating the adjusted path parameters, the obstacle avoidance distance is restored to 2.5mm, the obstacle avoidance success rate is increased to 88%, and the stability evaluation index is reduced to 0.55, meeting the conditions for continuing the simulation.

[0117] Repeat the above iterative process until the simulation termination condition is reached. The simulation termination condition can be set as the number of iterations reaching a preset value (such as 50 times) or the obstacle avoidance success rate being stable at more than 95%. In each iteration, the puncture point offset, path obstacle avoidance success rate and stability evaluation index are updated in real time. For example, after 10 iterations, the puncture point offset reaches 4.8mm, the path obstacle avoidance success rate is stable at 92%, and the fluctuation range of the stability evaluation index is reduced to 0.4-0.6, indicating that the path adjustment is stable.

[0118] During the dynamic adjustment simulation process, it is necessary to associate the comprehensive tissue feature evaluation model with the operating state parameters in real time. For example, when the actual puncture speed of the surgical instrument changes (such as from the preset 2mm / s to 1.5mm / s), the speed value in the operating state parameter is updated to the model, causing the puncture force calculation value to change accordingly, thereby affecting the solution of the operation difficulty equation and possibly triggering a new path adjustment. In addition, the model must also take into account the tissue displacement caused by physiological factors such as the patient's respiratory movement, and continuously update the vascular position and risk area through the real-time loading of dynamic vascular distribution data to ensure that the simulation process is synchronized with the actual surgical environment.

[0119] The entire dynamic adjustment simulation step achieves dynamic optimization of the puncture path through refined node encryption, clear trigger conditions and adjustment rules. Its core lies in converting complex tissue characteristics and instrument operation constraints into computable simulation parameters, and gradually approaching a safe and feasible puncture path through iterative solution. This process not only takes into account the static tissue anatomical structure, but also incorporates dynamic factors such as vascular movement and instrument operation status, so that the generated dynamic adjustment simulation information can truly reflect the actual situation during the operation, and provide a reliable basis for subsequent path optimization and puncture strategy adjustment. In clinical application scenarios, such as liver tumor puncture surgery, this simulation process can help doctors predict path risks in advance, dynamically avoid intrahepatic blood vessels, improve puncture accuracy, and reduce the probability of surgical complications.

[0120] Example 5:

[0121] In the path optimization step, the specific implementation of building a multi-path optimization model and iteratively optimizing the design parameters to obtain the optimal path planning parameters is as follows:

[0122] Taking liver tumor puncture surgery as an example, suppose a 3D ultrasound imaging model of the surgical site shows the tumor is located in the right lobe of the liver, surrounded by the right hepatic vein and its branches, with regional differences in tissue density (higher density in the tumor area, lower density in the tissue surrounding the vessels). The surgical instrument is an active bending needle with a maximum puncture force of 8N, a minimum turning radius of 2mm, and an ultrasound imaging accuracy of 0.3mm.

[0123] First, a multi-path optimization model is constructed. The optimization variables include path node density, turning angle threshold, and operation force distribution ratio. Path node density refers to the number of nodes on a path of unit length, and the value range is set to 5-20 / mm. It is used to control the fineness of the path. For example, when the node density is 10 / mm, it means that each millimeter of path contains 10 planning nodes, which can capture more subtle changes in tissue characteristics. The turning angle threshold is defined as the maximum angle between two adjacent segments in the puncture path, and the value range is 30°-80°. A larger threshold allows for sharper turns in the path, which is suitable for avoiding dense blood vessels. A smaller threshold tends to generate a smoother path. The operation force distribution ratio refers to the proportion of power output of the puncture needle in the directions of the three coordinate axes of X, Y, and Z. For example, when the distribution ratio is 4:3:3, it means that 40% of the power is distributed in the X-axis direction, and 30% is distributed in the Y-axis and Z-axis respectively, to adapt to the difference in puncture resistance in different directions.

[0124] Optimization goals include minimizing path length and maximizing adjustment response speed. Minimizing path length aims to reduce puncture time and tissue damage, such as selecting the shortest feasible path from the skin surface to the center of the tumor while safely avoiding blood vessels. Maximizing adjustment response speed requires that path planning can quickly adapt to intraoperative tissue displacement or instrument operation deviations. For example, when a change in blood vessel position is detected, the path can be replanned in a short time. Constraints include the upper limit of surgical instrument operation (maximum puncture force 8N, minimum turning radius 2mm) and ultrasound imaging accuracy threshold (path node spacing does not exceed 0.3mm to ensure that each node position can be clearly imaged by ultrasound).

[0125] The multi-path optimization model was initially solved using the A* algorithm to generate an initial population of optimized paths. The A* algorithm, combined with a heuristic search function, searches for feasible paths within the safe puncture region of the integrated tissue feature assessment model. Using the tumor center as the target point and the skin surface puncture entrance as the starting point, the algorithm expands the node with the lowest current cost until the target point is reached. During the search, constraints are verified in real time. For example, if the calculated turning angle of a path segment is 90° (exceeding the maximum allowable threshold of 80°), the path segment is eliminated. Using the A* algorithm, a population of 20 initial optimized paths was generated, each of which met the safe puncture requirements. For example, path 1 had a length of 45 mm, a node density of 12 nodes / mm, and a turning angle threshold of 50°; path 2 had a length of 48 mm, a node density of 8 nodes / mm, and a turning angle threshold of 65°. The paths varied in parameters such as path length and node density, resulting in a diverse initial solution.

[0126] Based on the initial optimized path population, the Dijkstra algorithm is used for global optimization to generate the optimal path planning parameters. The Dijkstra algorithm constructs a path graph model based on the initial path population. Each node represents an initial path, and the weight of the edge represents the difference between the paths (such as path length difference, node density difference, etc.). The algorithm gradually updates the global optimal solution of the path through relaxation operations. For example, by comparing the adjustment response speed of path 1 and path 2, it is found that path 1 has a higher node density and can respond to tissue displacement more quickly. Therefore, it is given a higher priority in the global optimization. At the same time, it is verified whether the operating force distribution ratio of each path meets the requirements of the instrument power system. For example, the X-axis force of a certain path accounts for 60%, resulting in insufficient power on the Y-axis and Z-axis, which cannot meet the minimum turning radius constraint. The path is excluded.

[0127] During global optimization, the interplay of various optimization variables must be comprehensively considered. For example, increasing the node density of the path improves the path's adaptability to tissue characteristics, but this increases the computational complexity and reduces the adjustment response speed. Reducing the turning angle threshold improves path smoothness but may force the path to detour further. Through repeated iterative adjustments, the optimal path planning parameters are ultimately generated. For example, the optimal path has a node density of 15 nodes / mm, a turning angle threshold of 55°, an operating force distribution ratio of 5:3:2 (X:Y:Z), and a path length of 42 mm. This path meets both the shortest path requirements and a fast adjustment response speed, and all operating parameters are within the instrument's constraints.

[0128] Once the optimal path planning parameters are generated, they can directly drive the surgical instrument to perform the puncture operation. During the puncture process, ultrasound image data is monitored in real time. If a deviation between the actual tissue characteristics and the model prediction is found (such as tumor position displacement due to respiratory movement), the path optimization model's online update mechanism is triggered, and the optimal path parameters are regenerated based on the latest ultrasound image data, ensuring that the puncture path always adapts to intraoperative changes.

[0129] The entire path optimization process is closely integrated with clinical scenario requirements, and through multivariable optimization and global search algorithms, a balance is achieved between safety, accuracy, and operational feasibility. Taking liver puncture as an example, this method can effectively avoid intrahepatic blood vessels and select the shortest and smoothest puncture path. At the same time, it dynamically adjusts power distribution according to instrument performance to reduce puncture resistance and tissue damage risk. From the generation of the initial path population to the search for the global optimal solution, each link is based on actual surgical parameters and tissue feature data to ensure that the optimization results are clinically operational and provide key technical support for the precise implementation of minimally invasive surgery.

[0130] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0131] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An interventional ultrasound method for minimally invasive surgery, characterized in that: The following steps are involved: Obtaining ultrasound imaging data of the patient's surgical site and operating status parameters of surgical instruments; Perform multi-level ultrasound scanning of the surgical site and construct a three-dimensional ultrasound image model. Combined with the ultrasound image data, perform dynamic tissue feature analysis on the three-dimensional ultrasound image model to generate an initial tissue feature evaluation model. Dynamic vascular distribution data and surgical operation constraints are loaded into the initial tissue feature evaluation model to generate a comprehensive tissue feature evaluation model. The puncture point dynamic simulation is performed in combination with the operation state parameters to obtain puncture point control simulation data. A dynamic vascular prediction model is constructed and trained using puncture point control simulation data to obtain a puncture point adjustment model, thereby generating initial adjustment parameters, wherein the initial adjustment parameters include at least an initial puncture point position, an initial obstacle avoidance path, and an initial adjustment response parameter; Based on the initial adjustment parameters, the comprehensive tissue characteristic evaluation model and the operation state parameters, dynamic adjustment simulation information is obtained; By combining dynamic adjustment of simulation information with surgical plan design parameters, a multi-path optimization model is constructed and the design parameters are iteratively optimized to obtain the optimal path planning parameters; Based on the real-time ultrasound imaging model, the path planning parameters of the current period are inferred to generate the safe puncture probability of the current period. Combined with the optimal path planning parameters of the current period and the actual path parameters, the real-time puncture strategy of the surgical instrument is adjusted to achieve the dynamic puncture point matching goal.

2. The interventional ultrasound method for minimally invasive surgery according to claim 1, characterized in that: The puncture point control simulation data at least includes tissue density distribution, blood vessel density distribution, safe puncture area set and dynamic adjustment priority sequence; The dynamic adjustment simulation information at least includes a puncture point offset, a path obstacle avoidance success rate, and a stability evaluation index.

3. The interventional ultrasound method for minimally invasive surgery according to claim 1, characterized in that: The method of performing multi-level ultrasound scanning on the surgical site and constructing a three-dimensional ultrasound image model, performing dynamic tissue feature analysis on the three-dimensional ultrasound image model in combination with ultrasound image data, and generating an initial tissue feature evaluation model comprises the following steps: Perform multi-angle scanning of the surgical site with an ultrasound probe to obtain ultrasound image sequences of tissue morphology and vascular distribution; Preprocessing the ultrasound image sequence to obtain standardized ultrasound image data, wherein the preprocessing includes one or more of image denoising, coordinate registration, feature extraction, grid subdivision, and data alignment; Generate a 3D ultrasound image model based on standardized ultrasound image data and combined with 3D reconstruction technology; Dynamic tissue feature analysis is performed on the three-dimensional ultrasound image model to generate an initial tissue feature assessment model, wherein the dynamic tissue feature analysis includes at least grid weight allocation and feature parameter mapping, and tissue safety levels are divided by grid weight allocation, and corresponding ultrasound image data are assigned to each level area.

4. The interventional ultrasound method for minimally invasive surgery according to claim 1, characterized in that: The method of loading the dynamic vascular distribution data and surgical operation constraints on the initial tissue feature evaluation model to generate a comprehensive tissue feature evaluation model, performing dynamic simulation of the puncture point in combination with the operation state parameters, and obtaining puncture point control simulation data includes the following steps: loading dynamic vascular distribution data into the initial tissue characteristic evaluation model to simulate real-time vascular changes, thereby generating a first tissue characteristic evaluation model; Loading surgical operation constraints on the first tissue feature evaluation model to generate a comprehensive tissue feature evaluation model, wherein the surgical operation constraints include a maximum puncture angle limit, a minimum turning radius, and instrument operation redundancy parameters; Based on the comprehensive tissue feature evaluation model and operation status parameters, dynamic simulation of the puncture point is performed. The specific process includes: Constructing a multi-constraint path planning equation, which includes at least an obstacle avoidance distance equation, a trajectory smoothness equation, and an operation difficulty equation. Combining this with a comprehensive tissue feature assessment model, the multi-constraint path planning equation is numerically solved using a piecewise linearization method to obtain tissue density distribution, vascular density distribution, a set of safe puncture areas, and a dynamically adjusted priority sequence. Obtaining a safe puncture area set includes the following steps: Based on the vascular density distribution, the vascular coverage index of each grid cell is calculated; Identify the areas where the vascular coverage index is not greater than the preset threshold in the comprehensive tissue feature evaluation model and generate a set of safe puncture areas, which are expressed as follows: The safe puncture area set is a spatial distribution mapping result of a blood vessel coverage index and a preset threshold.

5. The interventional ultrasound method for minimally invasive surgery according to claim 1, characterized in that: The dynamic blood vessel prediction model is constructed and trained using puncture point control simulation data to obtain a puncture point adjustment model, thereby generating initial adjustment parameters, including the following steps: Constructing a dynamic vascular prediction model based on three-dimensional ultrasound image features; The dynamic vascular prediction model is trained and verified through puncture point control simulation data to obtain a puncture point adjustment model; Input the real-time operation status parameters into the puncture point adjustment model to predict the set of safe puncture areas; Based on the predicted safe puncture area set, initial adjustment parameters are generated, wherein the initial adjustment parameters at least include an initial puncture point position, an initial obstacle avoidance path, and an initial adjustment response parameter.

6. The interventional ultrasound method for minimally invasive surgery according to claim 5, characterized in that: It also includes data reconstruction of the puncture point control simulation data, specifically: Based on tissue density distribution, blood vessel density distribution and safe puncture area set, an initial multi-dimensional path input tensor is constructed; Normalize and perform feature layering on the initial multi-dimensional path input tensor to generate the final multi-dimensional path input tensor; Based on the dynamic adjustment priority sequence, the puncture point adjustment label tensor is constructed; Combine the final multi-dimensional path input tensor and the puncture point adjustment label tensor to form a training sample set.

7. The interventional ultrasound method for minimally invasive surgery according to claim 6, characterized in that: The method of generating initial adjustment parameters based on the predicted safe puncture area set includes the following steps: Extract the grid cell with the lowest blood vessel density in the predicted safe puncture area set to generate the initial puncture point position; According to the spatial connectivity of the predicted safe puncture area set, the feasible topological structure of the initial obstacle avoidance path is fitted to generate the initial adjustment response parameters; The tissue density gradient direction of the predicted safe puncture area set is calculated and normalized into a path reference vector, which is the initial obstacle avoidance path direction.

8. The interventional ultrasound method for minimally invasive surgery according to claim 1, characterized in that: The method of obtaining dynamic adjustment simulation information based on the initial adjustment parameters, the comprehensive tissue characteristic evaluation model and the operation state parameters includes the following steps: The initial puncture point position is mapped to the comprehensive tissue feature evaluation model, the initial obstacle avoidance path is matched with the initial adjustment response parameters, and the path nodes in the adjustment area are encrypted. The comprehensive tissue feature evaluation model is updated, and the dynamic adjustment trigger conditions, path update rules, and response parameter increments are defined to generate a dynamic adjustment simulation model. Based on the dynamic adjustment simulation model, the multi-constraint path planning equations are iteratively solved by the piecewise linearization method to obtain dynamic adjustment simulation information, including: When the dynamic adjustment trigger conditions are met, the puncture point offset, path obstacle avoidance success rate, and stability evaluation index are updated, and the multi-constraint path planning equation is re-solved until the simulation termination conditions are met; The dynamic adjustment trigger condition includes triggering the response parameter update when the current path obstacle avoidance success rate is no greater than a preset success rate threshold; the path update rule includes adjusting the initial obstacle avoidance path based on the tissue density gradient direction; the response parameter increment is in a piecewise linear relationship with the current stability evaluation index.

9. The interventional ultrasound method for minimally invasive surgery according to claim 1, characterized in that: The multi-path optimization model is constructed and the design parameters are iteratively optimized to obtain the optimal path planning parameters, including the following steps: A multi-path optimization model was constructed, where the optimization variables included path node density, turning angle threshold, and operation force distribution ratio. The optimization objectives included minimizing path length and maximizing adjustment response speed. The constraints included the upper limit of surgical instrument operation and the ultrasound imaging accuracy threshold. The multi-path optimization model is preliminarily solved using the A* algorithm to generate an initial optimized path population; Based on the initial optimized path population, the Dijkstra algorithm is used for global optimization to generate the optimal path planning parameters.

10. An interventional ultrasound system for minimally invasive surgery, characterized in that: include: A data acquisition module is used to obtain ultrasonic image data of the patient's surgical site and operating status parameters of surgical instruments; The image modeling and analysis module is used to perform multi-level ultrasound scanning of the surgical site and construct a three-dimensional ultrasound image model. The three-dimensional ultrasound image model is then dynamically analyzed for tissue characteristics based on the ultrasound image data to generate an initial tissue characteristic assessment model. The comprehensive evaluation and simulation module is used to load dynamic vascular distribution data and surgical operation constraints into the initial tissue feature evaluation model to generate a comprehensive tissue feature evaluation model. It then combines the operation status parameters to perform dynamic simulation of the puncture point and obtain puncture point control simulation data. A parameter generation module is used to construct a dynamic vascular prediction model and train it through puncture point control simulation data to obtain a puncture point adjustment model, thereby generating initial adjustment parameters, wherein the initial adjustment parameters at least include an initial puncture point position, an initial obstacle avoidance path, and an initial adjustment response parameter; A dynamic adjustment simulation module is used to obtain dynamic adjustment simulation information based on initial adjustment parameters, a comprehensive tissue feature evaluation model, and operation status parameters; The path optimization module is used to combine dynamic adjustment simulation information with surgical plan design parameters, build a multi-path optimization model, and iteratively optimize the design parameters to obtain the optimal path planning parameters; The strategy adjustment module is used to infer the path planning parameters of the current period based on the real-time ultrasound imaging model, generate the safe puncture probability of the current period, combine the optimal path planning parameters of the current period with the actual path parameters, and adjust the real-time puncture strategy of the surgical instrument to achieve the dynamic puncture point matching goal.

Images