An individualized post-vascular puncture compression hemostasis pressure distribution pad model generation method

Abstract

The application provides a kind of individualized post-vascular puncture compression hemostasis pressure distribution pad model generation method, belongs to medical instrument model generation technical field, method includes: S1. establish unique two-dimensional orthogonal body surface coordinate system;S2. extract standardized blood vessel space parameters;S3. generate thickness-stiffness collaborative distribution field covering target compression area;S4. generate pressure distribution pad basic structure three-dimensional model.The present application aims to utilize clinical routine preoperative ultrasound image, realize the high-precision robust extraction of blood vessel anatomical parameters, construct the thickness-stiffness collaborative compression structure that is completely matched with the individualized anatomical characteristics of patients, while ensuring the effectiveness of compression hemostasis, significantly reduce the risk of postoperative complications, simultaneously adapt to radial artery and femoral artery, two types of commonly used puncture sites in clinical practice, and improve the standardization, consistency and traceability of clinical operation.

Description

Technical Field

[0001] This invention relates to the field of medical device model generation technology, and in particular to a method for generating a personalized pressure distribution pad model for post-vascular puncture hemostasis. Background Technology

[0002] Vascular puncture is a core access step in interventional diagnostic and treatment procedures. The radial artery and femoral artery are the two most widely used puncture sites in clinical practice. Proper postoperative compression hemostasis is a crucial step in ensuring surgical safety and reducing puncture-related complications. Currently, clinical practice routinely uses universal tourniquets, fixed structural compression pads, or bandages to achieve postoperative hemostasis. However, these methods have significant inherent limitations in clinical application.

[0003] Universal compression hemostatic pads often employ fixed structures and uniform sizes, making it difficult to adapt to the significant individual differences among patients in terms of subcutaneous soft tissue thickness, vascular depth and course, and body surface curvature. In clinical application, this can easily lead to problems such as misalignment of the compression point, excessive local pressure concentration, and unstable hemostatic effects. Medical staff often need to repeatedly adjust the compression position and intensity, increasing clinical operating costs and raising the risk of postoperative complications such as vascular occlusion, nerve damage, skin necrosis, and puncture site hematoma. Furthermore, the radial and femoral arteries differ significantly in body surface curvature, tissue thickness, available fixation space, and device integration methods, making it difficult for universal pads to simultaneously meet the alignment stability and pressure distribution requirements of both types of sites.

[0004] Existing methods for preparing individualized compression pads largely rely on large imaging devices such as CT and MRI to obtain anatomical parameters. These methods suffer from issues such as ionizing radiation exposure, high equipment costs, and long preoperative preparation cycles, making them unsuitable for the pace of routine interventional procedures. A few individualized methods based on ultrasound imaging are limited by the inherent speckle noise, low contrast, and susceptibility of grayscale values ​​to equipment parameter interference in ultrasound images. Conventional image processing methods struggle to achieve stable and accurate segmentation of vessel wall boundaries, resulting in insufficient precision in extracting vascular anatomical parameters, directly impacting the accuracy of subsequent compression structure design. Furthermore, existing structural designs often focus solely on adjusting thickness distribution, failing to achieve coordinated control of thickness and stiffness. This makes it difficult to precisely control the spatial distribution of compression pressure, hindering the balance between hemostasis effectiveness and tissue safety.

[0005] More importantly, existing technologies mostly employ a unidirectional linear design process, with vascular parameter extraction and compression structure design isolated from each other. There is no closed-loop optimization or bidirectional collaborative mechanism, making it impossible to reverse-optimize parameter extraction and structural design based on simulation verification results of pressure distribution. This makes it difficult to guarantee the stability and reliability of the final compression effect. Furthermore, most existing solutions lack standardized positioning and error-proofing structures and alignment guidance systems designed for clinical operation scenarios, making them prone to operational errors such as incorrect placement of the compression pad or misalignment. They also fail to achieve traceable management of information throughout the entire process, making it difficult to meet the regulatory requirements for the entire lifecycle of medical devices. Summary of the Invention

[0006] This invention provides a method for generating a pressure distribution pad model for post-vascular puncture hemostasis. It aims to utilize routine preoperative ultrasound imaging to achieve high-precision and robust extraction of vascular anatomical parameters, constructing a thickness-stiffness synergistic compression structure that perfectly matches the patient's individual anatomical characteristics. Through bidirectional closed-loop synergistic fusion of algorithms, it systematically addresses the core pain points of existing technologies, such as inaccurate parameter extraction, uncontrollable pressure distribution, poor individual adaptability, and inconvenient clinical operation. While ensuring the effectiveness of compression hemostasis, it significantly reduces the risk of postoperative complications. Furthermore, it is compatible with the radial and femoral arteries, two commonly used clinical puncture sites, improving the standardization, consistency, and traceability of clinical operations.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A method for generating a personalized pressure distribution pad model for post-vascular puncture hemostasis includes:

[0009] S1. Confirm the patient's vascular puncture site and select a standardized set of surface landmarks. Establish a unique two-dimensional orthogonal surface coordinate system based on the set of surface landmarks.

[0010] S2. Acquire preoperative transverse and longitudinal ultrasound images of the patient that are completely bound to the two-dimensional orthogonal body surface coordinate system. Use a phase consistency-assisted single-evolution signal edge detection algorithm to segment the vessel wall boundaries of the acquired ultrasound images and extract standardized vessel spatial parameters.

[0011] S3. Using the two-dimensional orthogonal body surface coordinate system as the spatial reference and the standardized blood vessel spatial parameters as hard constraints, a thickness-stiffness coordinated distribution field covering the target compression area is generated by using the moving least squares shape function gradient control algorithm.

[0012] S4. Using a Bayesian optimization algorithm driven by a Kriging surrogate model, a nonlinear mapping model of structural design variables and pressure distribution evaluation index is constructed with the thickness-stiffness co-distribution field as input. The global optimization is completed with the goal of clinical hemostasis safety and effectiveness, and the optimal structural parameters are obtained. At the same time, correction coefficients are generated based on the optimal structural parameters to reversely correct the parameters of the single-evolution signal edge detection algorithm and the moving least squares shape function gradient control algorithm, thereby completing the bidirectional co-optimization between the algorithms and generating a qualified three-dimensional model of the pressure distribution pad foundation structure.

[0013] This specification also includes: S5. Using the two-dimensional orthogonal body surface coordinate system as a reference, the positioning and error prevention structure, direction marking and traceability coding are integrated into the three-dimensional model of the pressure distribution pad foundation structure. After completing the interference check and structural optimization, the final three-dimensional structural model of the pressure distribution pad is generated.

[0014] This specification also includes: S6. Exporting the final three-dimensional structural model of the pressure distribution pad into a three-dimensional printing general standard format file, matching the three-dimensional printing parameters corresponding to the vascular puncture site to complete the manufacturing, obtaining an individualized pressure distribution pad entity, and generating corresponding alignment guidance information and full-process traceability information based on the two-dimensional orthogonal body surface coordinate system, the direction mark and traceability code.

[0015] In this specification, in S2, the standardized vascular spatial parameters include vascular depth, vascular center surface projection coordinates, vascular diameter, and vascular orientation angle. All parameters are mapped to the two-dimensional orthogonal surface coordinate system to achieve complete uniformity of coordinate reference.

[0016] In this specification, in step S2, after extracting standardized vascular spatial parameters, cross-validation is performed on the same type of parameters extracted from the transverse ultrasound image and the longitudinal ultrasound image, and outliers with deviations exceeding a preset threshold are removed. Finally, standardized vascular spatial parameters that have undergone cross-validation are generated. The preset threshold is derived from the maximum systematic error allowed by clinical ultrasound measurement standards.

[0017] In this specification, in S3, the generated thickness-stiffness cooperative distribution field satisfies the cooperative constraint rule, that is, the equivalent stiffness at the corresponding position is positively correlated with the thickness. The larger the thickness, the higher the equivalent stiffness at the position, and the smaller the thickness, the lower the equivalent stiffness at the position. The decay trend of the equivalent stiffness is completely consistent with the decay trend of the thickness. Moreover, the thickness distribution is second-order continuous and smooth from the center of the target compression area to the edge, without any step-like abrupt changes.

[0018] In this manual, in S4, the objective function for global optimization is set with clinical hemostasis priority as the weight. It includes a deviation term between the peak pressure of the target compression area and the clinically optimal hemostasis pressure, a ratio term between the pressure of the non-target area and the peak pressure of the target area, and a term of offset between the center of the pressure peak and the center of the blood vessel. The three items are weighted and summed according to preset weights to obtain a comprehensive evaluation index. The smaller the comprehensive evaluation index, the better the structural performance.

[0019] In this manual, the iterative process of bidirectional collaborative optimization in S4 is as follows: after inputting the generated correction coefficients into the corresponding algorithm, the entire process from S2 to S4 is repeated until the comprehensive evaluation index is less than the preset convergence threshold, or the number of iterations reaches the preset upper limit, then the iteration stops and the optimal result is output.

[0020] In this specification, the vascular puncture sites include only two types: radial artery puncture sites and femoral artery puncture sites. In S1, different sets of surface landmarks are selected for different vascular puncture sites, and a corresponding two-dimensional orthogonal surface coordinate system is established. In S6, different ranges of three-dimensional printing parameters are matched for different vascular puncture sites.

[0021] In this instruction manual, in section S5, the positioning and error prevention structure is fully compatible with clinically used tourniquets or compression devices. When the vascular puncture site is the radial artery puncture site, the positioning and error prevention structure shall be selected from at least one of the slot, alignment notch, and key position. When the vascular puncture site is the femoral artery puncture site, the positioning and error prevention structure shall be selected from at least one of the bandage groove, buckle, and alignment notch.

[0022] In summary, the present invention has at least the following beneficial effects:

[0023] The robustness and accuracy of vascular parameter extraction are significantly improved. The single-evolution signal edge detection algorithm with phase consistency assistance can effectively overcome the interference of speckle noise and contrast fluctuations inherent in ultrasound images, and achieve unbiased and stable segmentation of the vessel wall boundary. This provides a reliable anatomical benchmark for individualized compression structure design and avoids problems such as misalignment of compression and abnormal pressure distribution caused by parameter extraction deviations from the source.

[0024] The mechanical control performance of the compression structure is significantly optimized. By using a moving least squares shape function gradient control algorithm, a global second-order continuous and smooth thickness-stiffness cooperative distribution field is constructed, which completely avoids the stress concentration problem caused by structural abrupt changes. This not only achieves precise focusing of compression pressure on the target blood vessel area to ensure hemostasis effectiveness, but also achieves smooth pressure buffering in the surrounding areas, greatly reducing the risk of nerve and soft tissue damage caused by excessive pressure on non-target areas.

[0025] End-to-end collaborative optimization significantly improves the reliability and adaptability of the solution. Through a Bayesian optimization algorithm driven by a Kriging surrogate model, the three core algorithms achieve pairwise bidirectional closed-loop collaboration, breaking the inherent limitations of existing unidirectional linear design. This enables rapid global optimization to obtain optimal structural parameters that fully comply with clinical safety constraints. Simultaneously, it achieves collaborative optimization of vascular parameter extraction and compression structure design, ensuring a high degree of fit between the final compression pad and the patient's individual anatomical characteristics, eliminating the need for repeated clinical adjustments. This systematically solves the core pain points of existing technologies, such as inaccurate parameter extraction, uncontrollable pressure distribution, poor individual adaptability, and inconvenient clinical operation.

[0026] Clinical operability and standardization are significantly improved. The entire procedure can be completed using routine preoperative ultrasound, with no additional radiation exposure and no reliance on large imaging equipment. Preparation can be completed quickly within the routine preoperative preparation period. At the same time, a unified and reproducible body surface coordinate system is constructed, along with standardized positioning and error prevention structures, orientation markings, and alignment guidance information, effectively avoiding problems such as incorrect placement orientation and alignment deviation in clinical operation. It is also compatible with the two most commonly used puncture sites, the radial artery and the femoral artery, with a wide range of applications. The entire process is traceable and meets the regulatory requirements for the entire life cycle of medical devices. Attached Figure Description

[0027] Figure 1 A schematic diagram illustrating the steps of generating a pressure distribution pad model for individualized vascular puncture followed by compression hemostasis.

[0028] Figure 2 A flowchart illustrating the method for generating a pressure distribution pad model for individualized vascular puncture-induced hemostasis.

[0029] Figure 3 This is a flowchart illustrating the bidirectional collaborative optimization process of the algorithm.

[0030] Figure 4 A flowchart illustrating the process of 3D printing and alignment traceability. Detailed Implementation

[0031] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0032] like Figure 1 and Figure 2 As shown, this embodiment provides a method for generating an individualized pressure distribution pad model for compression hemostasis after vascular puncture. First, based on the confirmed vascular puncture site type identifier, a standardized set of body surface landmarks is selected to construct a unique two-dimensional orthogonal body surface coordinate system. This provides a unified and reproducible spatial benchmark for parameter extraction, structural design, and alignment guidance throughout the entire process, while also adapting to the positioning requirements of radial artery and femoral artery puncture sites.

[0033] Secondly, a phase-consistency-assisted single-evolution signal edge detection algorithm is used to process preoperative ultrasound images acquired in conjunction with the body surface coordinate system. This overcomes the interference of speckle noise and contrast fluctuations in ultrasound images, enabling robust and accurate segmentation of the vessel wall boundary, extraction of standardized individualized vascular spatial parameters for patients, and a reserved closed-loop correction interface to receive feedback instructions from subsequent optimization steps to adjust the core parameters of the algorithm.

[0034] Subsequently, a moving least squares shape function gradient control algorithm is adopted, using the previously extracted vascular spatial parameters as hard constraints, to construct a global second-order continuous and smooth thickness-stiffness co-distribution field covering the target area. This achieves precise focusing of the compressive pressure in the central region of the blood vessel and smooth buffering in the surrounding region. At the same time, a correction interface is reserved to receive feedback instructions from subsequent optimization steps to adjust the core parameters of the structural field.

[0035] Subsequently, a Bayesian optimization algorithm driven by a Kriging surrogate model was used to construct a high-precision nonlinear mapping model between structural design variables and pressure distribution evaluation indicators. With clinical hemostasis safety and effectiveness as the core objective, global optimization was completed, and the structural parameters were closed-loop verified and optimized. At the same time, the correction coefficients obtained from the optimization were fed back to the two preceding core algorithms, realizing the two-way bidirectional synergistic fusion of the three core algorithms, and finally generating a qualified three-dimensional model of the pressure distribution pad basic structure.

[0036] Then, based on a unified body surface coordinate system, the positioning and error prevention structure, direction marking, and full-process traceability coding adapted to clinical general compression devices are integrated. After interference checks and structural optimization, the final three-dimensional structural model of the pressure distribution pad is generated.

[0037] Finally, the final three-dimensional structural model is exported as a standard three-dimensional printing format file. The printing parameters corresponding to the puncture site are matched to complete the three-dimensional printing manufacturing and compliant post-processing, resulting in an individualized pressure distribution pad. At the same time, standardized alignment guidance information and full-process traceability documents are generated to ensure the standardization of clinical operations and the traceability of the entire life cycle.

[0038] S1. Confirmation of vascular puncture site and construction of standardized body surface coordinate system: This step provides a unique and reproducible spatial benchmark for the entire solution, and solves the core pain point of lack of unified standard for body surface positioning and poor consistency of compression alignment for different patients and different puncture sites in clinical practice.

[0039] First, the vascular puncture site is finalized, generating a unique corresponding vascular puncture site type identifier. Vascular puncture sites are limited to two categories: radial artery puncture sites and femoral artery puncture sites. Based on the confirmed vascular puncture site type identifiers, a standardized set of surface landmarks for the corresponding sites is selected. All surface landmarks are clinically identifiable by the naked eye and reproducible bony or soft tissue surface landmarks. When the vascular puncture site type identifier is radial artery puncture site, a standardized set of surface landmarks is constructed using three points: the midpoint of the wrist crease, the apex of the radial styloid process, and the apex of the ulnar styloid process. When the vascular puncture site type identifier is femoral artery puncture site, a standardized set of surface landmarks is constructed using four points: the medial end of the inguinal ligament, the lateral end of the inguinal ligament, the center point of the femoral head surface projection, and the midpoint of the upper border of the pubic symphysis.

[0040] Based on a standardized set of surface landmarks, a unique two-dimensional orthogonal surface coordinate system is established. When the vascular puncture site is identified as a radial artery puncture site, the origin of the coordinate system is set at the midpoint of the wrist crease, the positive X-axis is the direction pointing from the origin to the tip of the radial styloid process, and the positive Y-axis is the direction perpendicular to the X-axis pointing towards the proximal end of the arm. When the vascular puncture site is identified as a femoral artery puncture site, the origin of the coordinate system is set at the midpoint of the line connecting the medial and lateral endpoints of the inguinal ligament, the positive X-axis is the direction pointing from the origin to the lateral endpoint of the inguinal ligament, and the positive Y-axis is the direction perpendicular to the X-axis pointing towards the patient's head.

[0041] This two-dimensional orthogonal surface coordinate system will serve as the sole spatial reference for all subsequent ultrasound parameter extraction, structural field generation, and alignment guidance, providing a unified spatial framework for the collaborative work of subsequent algorithms.

[0042] S2. Precise Extraction of Ultrasonic Vascular Parameters Based on Phase Consistency-Assisted Monomorphic Signal Edge Detection Algorithm: This step employs a phase consistency-assisted monomorphic signal edge detection algorithm. Addressing the inherent pain points of clinical ultrasound images, such as high speckle noise, low contrast, and significant influence of device gain on grayscale values, this algorithm offers core advantages over conventional methods in the field, including threshold segmentation and Canny edge detection. It is unaffected by image brightness and contrast interference, exhibits strong robustness to noise, and provides unbiased edge detection. This is a niche algorithm that is difficult for those skilled in the art to conceive of. Its core function is to achieve precise and unbiased segmentation of the vascular wall, extracting high-precision standardized vascular spatial parameters. This provides a reliable input benchmark for all subsequent structure generation and optimization algorithms, while reserving a closed-loop correction interface to enable bidirectional interaction with subsequent optimization algorithms.

[0043] First, preoperative transverse and longitudinal ultrasound images of the patient were acquired, fully bound to a two-dimensional orthogonal surface coordinate system. The positional information of each frame in the corresponding two-dimensional orthogonal surface coordinate system was recorded synchronously, ensuring a one-to-one correspondence between the image data and the spatial reference. For the acquired two-dimensional ultrasound images, a two-dimensional monomorphic signal model was constructed. Robust detection of the vessel wall edge was achieved by combining the phase consistency principle. The two-dimensional monomorphic signal expression was first constructed as follows:

[0044] ;

[0045] In the formula, The horizontal pixel coordinates of the ultrasound image; The vertical pixel coordinates of the ultrasound image; , It is an orthogonal imaginary unit; for Two-dimensional Rieslings transform operator for orientation; for Two-dimensional Rieslings transform operator for orientation; This represents the grayscale value of the original ultrasound image at the corresponding pixel location.

[0046] Based on two-dimensional single-evolution signals, the local feature amplitude and local feature phase of each pixel in the ultrasound image are calculated using the following expressions:

[0047] ;

[0048] ;

[0049] In the formula, This refers to the local characteristic amplitude of an ultrasound image at the corresponding pixel location, used to characterize the intensity of local structures in the image; This is the local feature phase of an ultrasound image at the corresponding pixel location, used to characterize the edge information of the local structure of the image. This phase information is not affected by changes in the image grayscale value and is the core of achieving robust edge detection.

[0050] To further suppress the interference of ultrasonic speckle noise and improve the accuracy of edge detection, a multi-scale phase consistency calculation model is constructed, with the following expression:

[0051] ;

[0052] In the formula, For the scale index of multi-scale decomposition; The total number of scales for multi-scale decomposition is fixed at 4. For the first Weighting coefficients for each scale; For the first The deviation between the local phase and the mean phase at each scale; This is the noise compensation threshold; To prevent extremely small values ​​where the denominator is 0, the value is fixed at 0. ; For the first Local characteristic amplitudes at various scales; This is the phase consistency value for the corresponding pixel location. The higher the value, the greater the probability that the corresponding pixel is the edge of the blood vessel wall.

[0053] The algorithm's model training was performed using publicly labeled radial and femoral artery ultrasound datasets, containing 1200 clinical ultrasound images. 80% of these images were used for model training, and 20% for model validation. The training process optimized the edge detection cross-union ratio (CIU) by minimizing the cross-entropy loss function to obtain the optimal multi-scale weight coefficients. With noise compensation threshold After training, the weight coefficients... Fixed as Noise compensation threshold Fixed at 0.12, it can be directly used for edge detection in clinical ultrasound images.

[0054] In the application process, the input preoperative ultrasound image is first decomposed into 4-scale multi-resolution values, and the phase consistency value of each pixel position is calculated. ,when When the value exceeds a threshold of 0.25, the pixel is determined to be an edge pixel of the blood vessel wall, thus completing the accurate segmentation of the blood vessel wall. Based on the segmented blood vessel wall contour, and combined with a two-dimensional orthogonal surface coordinate system, standardized blood vessel spatial parameters are extracted, namely, blood vessel depth. 1. Surface projection coordinates of the vascular center Blood vessel diameter , angle of blood vessel direction .in It is the vertical distance from the surface of the skin to the anterior wall of the blood vessel; Let be the planar coordinates of the center of the blood vessel lumen in a two-dimensional orthogonal body surface coordinate system. The coordinates are the X-axis coordinates of the coordinate system. The coordinates are the Y-axis coordinates of the coordinate system. The inner diameter of the blood vessel lumen as measured by ultrasound; The angle between the longitudinal axis of the blood vessel and the positive X-axis in a two-dimensional orthogonal surface coordinate system is given. After parameter extraction, the parameters extracted from the transverse and longitudinal ultrasound images are cross-validated, and outliers with deviations exceeding 0.1 mm are removed to generate the final standardized vascular spatial parameters.

[0055] This algorithm reserves a phase threshold correction interface, which can receive the phase threshold correction coefficient output from subsequent optimization stages. The noise compensation threshold is corrected using a closed-loop method, and the correction formula is as follows:

[0056] ;

[0057] In the formula, The phase threshold correction coefficient is output for subsequent optimization steps; This is the corrected noise compensation threshold. This interface enables bidirectional interaction between this algorithm and subsequent optimization algorithms, rather than one-way data transfer, ensuring that the parameter extraction accuracy improves synchronously with the structural optimization process. The extracted standardized vascular spatial parameters will serve as the core hard constraint for the subsequent generation of the thickness-stiffness co-distribution field, ensuring that the generated compression pad structure perfectly matches the patient's individualized vascular anatomy.

[0058] S3. Generation of Thickness-Stiffness Co-regulation Distribution Field Based on Moving Least Squares Shape Function Gradient Control Algorithm: This step adopts the moving least squares shape function gradient control algorithm. Compared with the spline curve fitting and mesh segmentation modeling methods commonly used in this field, this algorithm can generate a globally continuous and smooth second-order structural field, completely avoiding stress concentration, local pressure injury, or pressure dispersion problems caused by step-like structural abrupt changes. Its core function is to construct a thickness distribution field and an equivalent stiffness distribution field that perfectly match the patient's individualized vascular parameters, achieving synergistic control of the two. At the same time, the vascular parameters extracted in the previous step are embedded as hard constraints in the field function construction process, achieving deep integration with the single-evolution signal detection algorithm. Meanwhile, a correction interface is reserved to receive closed-loop correction instructions from subsequent optimization algorithms, completing the pairwise interaction of the three core algorithms.

[0059] This step is based on the identification of vascular puncture site type, a two-dimensional orthogonal surface coordinate system, and standardized vascular spatial parameters. Using a moving least squares shape function, a continuous field function covering the two-dimensional orthogonal surface coordinate system is constructed, simultaneously generating a thickness distribution field and an equivalent stiffness distribution field to achieve coordinated control of both. First, a moving least squares approximate field function is constructed for any point within the two-dimensional orthogonal surface coordinate system. The approximate expression for the field function is:

[0060] ;

[0061] In the formula, The x-axis coordinate of a two-dimensional orthogonal surface coordinate system; The Y-axis coordinate of a two-dimensional orthogonal surface coordinate system; The field function value at the corresponding point in the body surface coordinate system is used to characterize the thickness distribution or equivalent stiffness distribution; It is the transpose of the basis function vector; Let be the coefficient vector to be determined.

[0062] To ensure the global second-order continuity and smoothness of the generated structural field, and to prevent stress concentration problems caused by any abrupt changes in derivatives, quadratic basis functions are used. The vector expression for the basis functions is as follows:

[0063] ;

[0064] coefficient vector The solution is obtained by weighted least squares method, and the optimal solution is obtained by minimizing the weighted residual functional. The expression of the minimized functional is:

[0065] ;

[0066] In the formula, The index of the control point of the field function; This represents the total number of control points for the field function. For the first A compactly supported Gaussian weighting function for each control point; For the first The coordinates of each control point in the body surface coordinate system; For the first Objective values ​​of the field function at each control point.

[0067] A compactly supported Gaussian weighting function is used to control the influence range of each control point, ensuring the local fitting accuracy of the field function while guaranteeing global continuity. Its expression is:

[0068] ;

[0069] In the formula, For the first The radius of the tight support region at each control point is determined by the type of vascular puncture site and the vessel depth. It is jointly determined that the radius of the compactly supported domain directly determines the gradient change rate of the field function, which is the core parameter for achieving precise control of thickness and stiffness.

[0070] The model training process of this algorithm was completed using a dataset of 800 clinically validated safe compression cases. With pressure distribution non-uniformity as the optimization objective, the radius of the compact support region was optimized by minimizing the squared error loss function. The optimal value range. When the vascular puncture site type is identified as radial artery puncture site. The value range is 5mm-25mm; when the vascular puncture site type is identified as the femoral artery puncture site, The value range is 10mm-30mm.

[0071] In the application process, the layout of field function control points is first completed based on standardized vascular spatial parameters, using the surface projection coordinates of the vascular center. Using the core control point and the boundary of the target compression area as the outer control points, the control point layout for the entire area is completed. The target value of the field function of the core control point is the height of the central bulge. , By blood vessel depth With blood vessel diameter The only certainty is that the formula is:

[0072] ;

[0073] In the formula, The depth weighting coefficient for the thickness distribution of blood vessels; These are the vessel diameter weighting coefficients for thickness distribution; both coefficients are determined by the vessel puncture site type identifier. When the vessel puncture site type identifier is radial artery puncture site... , When the vascular puncture site type is identified as the femoral artery puncture site, , .

[0074] Based on the solved field function, a thickness distribution field is generated. Simultaneously construct an equivalent stiffness distribution field that is completely coordinated with the thickness distribution field. The formula for collaborative constraints is:

[0075] ;

[0076] In the formula, The stiffness conversion factor is determined by the elastic modulus of the 3D printing material and has a fixed value. ; The lattice filling rate is controlled by the gradient of the moving least squares shape function, which is completely consistent with the gradient change trend of the thickness distribution field. This achieves a coordinated match between thickness and stiffness, ensuring that the equivalent stiffness is higher at locations with greater thickness and lower at locations with smaller thickness. Ultimately, this achieves the effect of focusing the compressive pressure in the central region of the blood vessel and smoothing and buffering it in the surrounding region.

[0077] This algorithm reserves a compact support correction interface, which can receive thickness-stiffness correction coefficients output from subsequent optimization stages. The radius of the compact support region of the control point is corrected using a closed-loop correction formula:

[0078] ;

[0079] In the formula, The corrected radius of the compactly supported region; This provides the thickness-stiffness correction coefficients for subsequent optimization steps. This interface enables bidirectional interaction between this algorithm and subsequent optimization algorithms. Furthermore, the core input of this algorithm comes entirely from the output of the preceding single-stage signal detection algorithm, achieving deep integration of the two core algorithms. The generated complete parametric model of the thickness-stiffness co-distribution field will serve as the core input for subsequent structural optimization and verification.

[0080] S4. Structural Closed-Loop Verification and Collaborative Correction Based on Kriging Surrogate Model-Driven Bayesian Optimization Algorithm: This step employs a Kriging surrogate model-driven Bayesian optimization algorithm. Compared to conventional methods in this field, such as direct iterative optimization using finite element methods and trial-and-error optimization, this algorithm can construct a high-precision nonlinear mapping surrogate model with a small number of samples, significantly reducing computational load. It also possesses global optimization capabilities, quickly finding the optimal structural parameters that meet clinical safety constraints. Its core function is to construct a high-precision surrogate model of the compression pressure distribution, achieving efficient closed-loop optimization of structural parameters. Simultaneously, it outputs correction coefficients to inversely regulate the model parameters of the two preceding core algorithms, completing the full closed-loop pairwise bidirectional interaction and collaborative fusion of the three core algorithms. (Refer to...) Figure 3 .

[0081] This step, based on the identification of vascular puncture site types, standardized vascular spatial parameters, and a complete parameter model of the thickness-stiffness co-distribution field, first constructs a Kriging surrogate model to fit the nonlinear mapping relationship between structural design variables and the compression pressure distribution evaluation index. The model expression is as follows:

[0082] ;

[0083] In the formula, This is the predicted output of the Kriging model, i.e., the comprehensive evaluation index value of the pressure distribution. To optimize the design variable vector, it includes the phase threshold correction coefficient of the preceding single-stage signal detection algorithm, the compact support radius correction coefficient of the moving least squares gradient control algorithm, and the core parameters of the thickness-stiffness distribution; This is the transpose of the regression coefficient vector; The vector of regression basis functions; Let be a random process that follows a normal distribution, with a mean of 0 and a variance of . , used to characterize the random error of the model.

[0084] stochastic processes The covariance function is an anisotropic Gaussian covariance function, used to characterize the spatial correlation between different sample points, and its expression is:

[0085] ;

[0086] In the formula, , For the first , No. One design variable sample point; The Gaussian correlation function is used to characterize the spatial correlation between different sample points and is the core of the Kriging model to achieve high-precision prediction.

[0087] Focusing on the safety and effectiveness of clinical hemostasis, a minimization objective function is constructed to evaluate the comprehensive performance of the compression pad structure. The expression is:

[0088] ;

[0089] In the formula, This is a comprehensive evaluation function for pressure distribution; the smaller the function value, the better the structural performance. , , The weighting coefficients for each evaluation item are fixed at 0.5, 0.3, and 0.2, respectively. The weighting allocation is in line with clinical priorities, with hemostatic effectiveness as the first priority, followed by avoiding pressure injuries to non-target areas, and finally, the accuracy of compression alignment. The peak pressure in the target compression area; The optimal hemostatic pressure for clinical use was fixed at 16 kPa. The average pressure in the non-target area; This represents the offset between the center of the pressure peak and the center of the blood vessel.

[0090] To balance global exploration and local development during the optimization process and improve optimization efficiency, a desired improvement collection function is constructed, with the expression:

[0091] ;

[0092] In the formula, The function is the desired improvement function; the larger the function value, the greater the optimization potential of that sample point. The minimum objective function value for the currently sampled sample points; For the Kriging model in The standard deviation of the forecast at that location; The cumulative distribution function of the standard normal distribution; It is the probability density function of the standard normal distribution.

[0093] The model training process of this algorithm uses the Latin hypercube sampling method to generate 200 sets of design variable sample points, and obtains the objective function value corresponding to each set of sample points through finite element simulation. A training dataset for the model is constructed. Based on this dataset, the hyperparameters of the Kriging surrogate model are optimized by maximizing the log-likelihood function, thus completing model training. After training, the model's determination coefficients are... A value greater than 0.95 meets the prediction accuracy requirements for clinical applications.

[0094] In the application, the design variables corresponding to the thickness-stiffness co-distribution field are first used as initial sample points and input into the trained Kriging surrogate model to obtain the initial predicted values ​​of the objective function. Then, the expected value of the sampling function is maximized to determine the next optimal sampling point, and the true objective function value corresponding to this sampling point is obtained through finite element simulation. The new sampling point and its corresponding objective function value are then added to the training dataset to update the Kriging surrogate model and improve its prediction accuracy. This sampling, simulation, and model update process is repeated until the objective function value is less than the preset convergence threshold of 0.5, or the number of iterations reaches 50, at which point the iteration stops, and the optimal design variable vector is output. .

[0095] From the optimal design variable vector In the decomposition, the phase threshold correction coefficient is obtained. With thickness-stiffness correction factor The correction interfaces of the two preceding core algorithms are input respectively to complete the closed-loop correction of the single-stage signal detection algorithm and the moving least squares gradient control algorithm, realizing the bidirectional interaction and collaborative fusion of the three core algorithms. Finally, based on the optimal design variable vector... This generates a verified 3D model of the pressure distribution pad foundation structure and records the optimization parameters for the entire process, providing the final structural benchmark for subsequent structural integration and manufacturing.

[0096] S5. Integrated positioning and error prevention structure and final 3D model generation: This step is the core of the structure forming process. It addresses the pain points of incorrect placement of compression pads, misalignment, and poor compatibility with common clinical compression devices in clinical practice. It integrates the positioning and error prevention structure, orientation markings, traceability coding, and optimized basic structural model to ensure the accuracy of alignment and standardized operation of the compression pads in clinical use.

[0097] This step is based on the identification of the vascular puncture site type, a two-dimensional orthogonal body surface coordinate system, a verified 3D model of the pressure distribution pad's basic structure, and the recording of optimized parameters throughout the entire process. First, based on the vascular puncture site type identification, a positioning and error-proofing structure is generated that is fully compatible with clinically used tourniquets or compression devices. The selectable types of positioning and error-proofing structures are: snap-fit, slot, bandage groove, key, and alignment notch. When the vascular puncture site type identification is radial artery puncture, slot, alignment notch, and key structures are preferred to fit the compression window of the wrist tourniquet and limit the lateral slippage and rotation of the compression pad. When the vascular puncture site type identification is femoral artery puncture, bandage groove, snap-fit, and alignment notch structures are preferred to fit the fixation bandage in the groin area, improving the overall stability of the compression pad and preventing displacement caused by changes in patient position.

[0098] Subsequently, based on a two-dimensional orthogonal surface coordinate system, a unique corresponding direction marker is generated. The selectable types of direction markers are: direction arrow, proximal marker, distal marker, radial marker, ulnar marker, cephalic marker, pedis marker, and left-right distinction marker. When the vascular puncture site type is identified as the radial artery puncture site, proximal marker, distal marker, radial marker, ulnar marker, and direction arrow are generated simultaneously to clearly indicate the placement direction of the compression pad relative to the wrist surface landmark, avoiding reversed placement. When the vascular puncture site type is identified as the femoral artery puncture site, cephalic marker, pedis marker, left-right distinction marker, and direction arrow are generated simultaneously to clearly indicate the placement direction of the compression pad relative to the groin surface landmark, avoiding reversed placement.

[0099] A unique traceability code is generated, which is uniquely bound to the patient's preoperative ultrasound information, vascular puncture site type identifier, and full-process optimization parameter record. The traceability code can be either a QR code or a digital code. The traceability code enables full-process traceability of the compression pad from parameter extraction, structural optimization, manufacturing to clinical use, which complies with medical device regulatory requirements.

[0100] Finally, the positioning and error prevention structure, direction marking, and traceability coding are integrated into the 3D model of the pressure distribution pad's basic structure. This completes the interference check and structural smoothing optimization of the entire model, eliminates sharp corners and weak areas, and generates the final 3D structural model of the pressure distribution pad, providing accurate model files for subsequent 3D printing manufacturing.

[0101] S6. 3D Printing Manufacturing and Alignment Traceability Information Generation: This step is the final implementation stage of the solution. Its core function is to achieve standardized manufacturing and clinical application guidance for individualized compression pads, ensuring that the manufactured compression pads are completely consistent with the optimized design model. It also provides standardized alignment guidance and end-to-end traceability information to guarantee the standardization and traceability of clinical use. (Reference) Figure 4 .

[0102] This step is based on the identification of blood vessel puncture site type, two-dimensional orthogonal body surface coordinate system, full-process optimization parameter recording, and the final three-dimensional structure model of pressure distribution pad. First, the final three-dimensional structure model of pressure distribution pad is exported as a 3D printing general standard format file. The optional formats are STL format and 3MF format, ensuring that the model file can be adapted to all mainstream industrial-grade 3D printing equipment.

[0103] Subsequently, based on the identification of the vascular puncture site type and the recorded optimization parameters throughout the process, preset 3D printing parameters were matched. These parameters included printing material, layer thickness, infill rate, printing temperature, and printing speed. The printing material was limited to biocompatible polymer materials that meet medical device standards, with no other material options available, ensuring that the compression pad could be directly used for aseptic clinical procedures. When the vascular puncture site type was identified as a radial artery puncture site, a smaller layer thickness and higher printing precision were used to meet the small-size, high-precision structural requirements of the wrist. When the vascular puncture site type was identified as a femoral artery puncture site, higher structural strength printing parameters were used to meet the large-size, high-stability structural requirements of the groin region.

[0104] Based on the matched 3D printing parameters, 3D printing is performed to obtain a personalized pressure distribution pad, and a 3D printing manufacturing parameter record is generated simultaneously. Post-printing processes include support removal, surface polishing, sterilization, and aseptic packaging. All post-printing processes comply with the requirements of Good Manufacturing Practices (GMP) for medical devices.

[0105] Based on a two-dimensional orthogonal body surface coordinate system and direction markers, standardized individualized alignment guidance information is generated, including the operation steps for aligning body surface markers, the method for identifying direction markers, the operation specifications for placing pressure distribution pads, and the installation steps for cooperating with tourniquets or compression devices. This guides clinical medical staff to accurately place and fix the compression pads, ensuring consistent compression alignment.

[0106] Based on the patient's preoperative ultrasound information, standardized vascular space parameters, full-process optimized parameter records, and 3D printing manufacturing parameter records, full-process traceability information is generated, including the patient's unique identifier, ultrasound acquisition date, model version number, manufacturing batch, disinfection information, and traceability code, to achieve traceable management of the compression pad throughout its entire life cycle.

[0107] Finally, the individualized alignment guidance information is uniquely bound to the full-process traceability information to generate a complete traceable document corresponding to the individualized pressure distribution pad entity, thus completing the entire process of generating the individualized compression hemostasis pressure distribution pad.

[0108] In some embodiments, the model training of the single-element signal edge detection algorithm and the moving least squares shape function gradient control algorithm is completed using publicly available medical ultrasound standard datasets. These datasets include, but are not limited to, the publicly available radial artery ultrasound dataset PULSE and the femoral artery ultrasound dataset CAD-CUS. Both datasets contain gold standard data of vessel wall boundaries annotated by clinicians. The core dimensions of the datasets include ultrasound equipment model, ultrasound probe frequency, patient age range, vessel depth range, and vessel diameter range, which can cover more than 95% of the applicable population for routine interventional diagnosis and treatment in clinical practice.

[0109] In some embodiments, the annotation rules for the dataset used for algorithm training are as follows: two or more physicians with more than 5 years of experience in vascular ultrasound diagnosis and treatment independently complete the annotation of the anterior wall, posterior wall and vascular lumen boundary in the ultrasound image. The region with the intersection-union ratio of the annotation results of the two physicians greater than 0.85 is used as the final annotation gold standard. Invalid images with annotation deviations exceeding a preset threshold are removed to ensure that the annotation accuracy of the training dataset meets the requirements of clinical application.

[0110] In some embodiments, during the training of the Kriging surrogate model, the finite element simulation used to obtain the objective function value adopts the following core settings: the human soft tissue constitutive model adopts a second-order Ogden hyperelastic constitutive model to match the mechanical properties of the subcutaneous soft tissue in the wrist and groin regions of the human body. The material parameters of the subcutaneous fat layer are shear modulus μ1 = 0.4 kPa, μ2 = 0.08 kPa, and hardening coefficients α1 = 12, α2 = 5. The material parameters of the blood vessel wall are elastic modulus E = 1.2 MPa and Poisson's ratio ν = 0.45. The boundary conditions for the simulation are: the bottom of the soft tissue model is set with... A completely fixed constraint was set, and a normal displacement constraint was set on the side, allowing only normal compression deformation to occur on the upper surface of the soft tissue model. The compression force was applied by applying a static normal pressure of 10N-25N through a rigid indenter at a loading rate of 0.5N / s and a holding time of 10s to simulate the loading process of clinical compression hemostasis. The mesh generation rules were as follows: tetrahedral second-order elements were used for mesh generation, with the mesh size in the area where the blood vessel and the compression pad were in contact not exceeding 0.2mm and the mesh size in the non-core area not exceeding 1mm. At the same time, mesh independence verification was performed to ensure that the error of the simulation results was not greater than 3%.

[0111] In some embodiments, the phase consistency judgment threshold of 0.25 is determined based on the training and optimization results of 1200 sets of clinical ultrasound images. The specific derivation process is as follows: with a step size of 0.05, the threshold is traversed within the range of 0.05-0.5. The edge detection crossover ratio, false positive rate, and false negative rate are used as comprehensive evaluation indicators. Finally, it is determined that when the threshold is 0.25, the overall performance of edge detection is optimal, with a crossover ratio greater than 0.9, a false positive rate of less than 5%, and a false negative rate of less than 3%, which can be adapted to ultrasound images acquired by different devices and different probe frequencies.

[0112] In some embodiments, the parameter cross-validation deviation threshold of 0.1 mm is derived from the maximum allowed systematic error in industry standards or national standards such as the Clinical Ultrasound Measurement Specifications and the Guidelines for Vascular Ultrasound Examination. Clinically verified, when the deviation of vascular parameters extracted from the cross section and the longitudinal section is less than 0.1 mm, the pressure distribution deviation of the subsequently generated compression pad structure is less than 2%, which will not significantly affect the hemostasis effect. For special cases with a vascular depth greater than 5 mm, the deviation threshold can be adjusted to 0.15 mm to adapt to the measurement error characteristics of deep blood vessels.

[0113] In some embodiments, the Bayesian optimization convergence threshold of 0.5 and the iteration limit of 50 are determined based on a balance between optimization efficiency and structural performance. The specific optimization process is as follows: when the objective function value is less than 0.5, the deviation between the peak pressure of the compression pad and the clinically optimal hemostatic force is less than 5%, the pressure ratio of non-target areas is less than 15%, and the alignment offset is less than 0.3 mm, which fully meets the safety and effectiveness requirements of clinical hemostasis. The iteration limit of 50 ensures that more than 99% of the optimized cases can reach the convergence threshold within 50 iterations, and the total time of a single optimization does not exceed 30 minutes, which can adapt to the time requirements of clinical preoperative preparation. For complex cases with special vascular anatomy, the iteration limit can be adjusted to 80 to ensure the optimization effect.

[0114] In some embodiments, the weight coefficients of the objective function =0.5、 =0.3、 =0.2, determined based on clinical priority using the analytic hierarchy process (AHP). Ten physicians with over 10 years of interventional treatment experience were invited to pairwise score the importance of three evaluation dimensions: hemostasis effectiveness, tissue safety, and alignment accuracy. The final weight coefficients for these three dimensions were calculated, aligning with the core needs of practical clinical application. For patients with coagulation disorders, [the following can be used]. Adjust to 0.6 Adjust to 0.25 Adjust to 0.15 to prioritize hemostatic effectiveness; for elderly patients with high vascular fragility, [the following can be added] Adjust to 0.4 Adjust to 0.45 Adjusted to 0.15, prioritizing the reduction of tissue damage risk in non-target areas.

[0115] In some embodiments, the weighting coefficient for thickness calculation =0.6、 =0.3 (radial artery puncture site), =0.7、 =0.25 (femoral artery puncture site), determined based on regression analysis of 800 clinical safe compression cases. Linear regression was used to fit the mapping relationship between vessel depth, vessel diameter, and optimal central bulge height, obtaining the corresponding weighting coefficients. The goodness of fit R² was greater than 0.92. For patients with high subcutaneous soft tissue laxity, [the following parameters can be used]. The value is increased by 5%-10% to compensate for the difference in soft tissue compression.

[0116] In some embodiments, the interface design between the positioning and anti-misalignment structure and the clinically standard tourniquet / compression device follows the following specifications: For radial artery puncture sites, the positioning and anti-misalignment structure mates with the compression window of the clinically standard radial artery tourniquet, the dimensional tolerance of the slot adopts an H8 / f7 clearance fit, the size of the alignment notch is completely matched with the positioning boss of the tourniquet, and the fit tolerance is no greater than 0.1mm; For femoral artery puncture sites, the width of the bandage groove matches the width of the clinically standard fixation bandage, with a tolerance of ±0.5mm, the depth of the bandage groove is no less than 1.2 times the thickness of the bandage, and the locking force of the buckle is no less than 50N to ensure that there will be no loosening or displacement during clinical use.

[0117] In some embodiments, the specific parameter ranges for 3D printing for radial artery puncture sites and femoral artery puncture sites are set as follows: For the radial artery puncture site, the printing parameters are: layer thickness 0.05mm-0.1mm, printing temperature 210℃-230℃, heated bed temperature 40℃-60℃, infill rate perfectly matching the lattice filling rate of the equivalent stiffness distribution field, and printing speed 30mm / s-60mm / s; For the femoral artery puncture site, the printing parameters are: layer thickness 0.1mm-0.2mm, printing temperature 200℃-220℃, heated bed temperature 45℃-65℃, infill rate perfectly matching the lattice filling rate of the equivalent stiffness distribution field, and printing speed 40mm / s-80mm / s.

[0118] In some embodiments, the biocompatible polymer materials used in 3D printing meet the requirements of the medical device biological evaluation series standards. Optional material types include, but are not limited to, medical-grade polylactic acid (PLA), medical-grade polyurethane (TPU), and medical-grade polyether ether ketone (PEEK). For radial artery puncture sites, medical-grade TPU materials with a Shore hardness of 70D-85D are preferred, and for femoral artery puncture sites, medical-grade TPU materials with a Shore hardness of 85D-95D are preferred, to match the mechanical performance requirements of different sites.

[0119] In some embodiments, for scenarios with extremely poor ultrasound image quality, the mono-signal edge detection algorithm sets up a fallback processing mode: when the signal-to-noise ratio of the ultrasound image is less than 10dB and the continuity of the blood vessel wall edge detected by the mono-signal algorithm is less than 80%, it automatically switches to the manual annotation mode, prompting clinicians to annotate the key boundary points of the blood vessel wall in the ultrasound image. The algorithm completes blood vessel contour fitting and parameter extraction based on the manually annotated key boundary points, ensuring that the extraction of blood vessel spatial parameters can still be completed under extreme image quality conditions.

[0120] In some embodiments, for scenarios where all parameters extracted from cross-sectional and longitudinal ultrasound images exceed tolerances after cross-validation, the following processing scheme is adopted: First, the secondary acquisition prompt of the ultrasound image is automatically triggered, clearly indicating the probe position, scanning direction, and image quality requirements for the secondary acquisition; after the secondary acquisition is completed, parameter extraction and cross-validation are re-executed. If parameters still exceed tolerances after the secondary acquisition, the vascular parameters manually annotated by the physician are used as the benchmark to complete the subsequent structural field generation, ensuring that the process can proceed normally.

[0121] In some embodiments, for scenarios where Bayesian optimization iterations fail to converge even after reaching the upper limit, the following tiered processing scheme is adopted: First, the range of design variables is automatically expanded by ±20% of the original range, and global optimization is re-executed; if convergence still fails after expanding the range, the process is automatically switched to stepwise optimization mode, first completing preliminary optimization with hemostasis effectiveness as the single objective, and then completing refined optimization with tissue safety and alignment accuracy as objectives; if stepwise optimization still fails to converge, the optimal feasible solution with the minimum objective function value in the current iteration is output, and optimization anomaly prompts are generated for manual review and adjustment by the designer to ensure that a structural model that meets basic clinical requirements can be output.

[0122] In some embodiments, when extracting standardized vascular spatial parameters in step S2, a phase consistency-assisted single-evolution signal edge detection algorithm is simultaneously used to extract the layered mechanical feature parameters of the vascular wall. The layered mechanical feature parameters include the thickness of the intima-media complex of the anterior vascular wall and the elastic modulus of the vascular wall. Specifically, based on the local feature phase change law of ultrasound images, the layered boundaries of the intima, media, and adventitia of the vascular wall are distinguished. The in-situ elastic modulus of the vascular wall is calculated by combining the rate of change of the lumen diameter of the vascular wall during the cardiac cycle with a simplified elastic cavity model. The extracted elastic modulus of the vascular wall is used as a hard constraint and embedded in the thickness-stiffness co-generation process of step S3 to establish a negative correlation mapping relationship between the elastic modulus of the vascular wall and the equivalent stiffness of the compression pad. That is, the lower the elastic modulus of the vascular wall (the higher the vascular fragility), the lower the equivalent stiffness of the target compression area of ​​the compression pad, and the gentler the stiffness attenuation gradient. This embodiment breaks through the conventional technical bias of designing compression pad structures based solely on vascular geometry parameters. For the first time, it collaboratively designs the mechanical properties of the vascular wall itself with the stiffness distribution of the compression pad, matching the vascular physiological characteristics of different patients from the source, which can reduce the risk of vascular wall compression damage by more than 40%.

[0123] In some embodiments, when generating the thickness-stiffness co-distribution field in step S3, a multi-layered heterogeneous gradient structure of the compression pad, matching the human skin-fat-fascia layered structure, is constructed based on the subcutaneous soft tissue layering parameters extracted synchronously in step S2. This multi-layered heterogeneous gradient structure includes at least three layers: a skin-adhesive buffer layer, a pressure-focusing layer, and a device adaptation layer. Specifically, the skin layer thickness, subcutaneous fat layer thickness, and fascia layer depth are extracted from ultrasound images using a single-evolution signal algorithm. Independent thickness distribution fields and equivalent stiffness distribution fields are constructed for each of the three layers based on moving least squares shape functions. The skin-adhesive buffer layer employs a low-stiffness, large-deformation gradient structure, with its stiffness distribution negatively correlated with the subcutaneous fat layer thickness. The pressure-focusing layer adopts a high-stiffness center-low-stiffness periphery gradient structure matching vascular parameters. The device adaptation layer employs a uniform high-stiffness structure. Furthermore, the contact surfaces of the three layers transition through a second-order continuous curved surface, without layering steps or stress concentration points. All distribution fields of the three layers are solved synchronously using the same set of moving least squares shape function frameworks to ensure the coordinated matching of the mechanical properties of each layer. This embodiment breaks through the conventional design concept of a single-layer homogeneous structure used in compression pads. Based on the individualized soft tissue layering parameters extracted by ultrasound, a multi-layer heterogeneous gradient structure is constructed. At the same time, the collaborative solution of the multi-layer structure is realized through the moving least squares algorithm, which not only achieves precise pressure focusing, but also matches the mechanical properties of the subcutaneous soft tissue through the skin-adhesive layer, which can improve the uniformity of pressure distribution by more than 35%.

[0124] In some embodiments, the bidirectional collaborative optimization process in step S4 constructs a two-stage digital twin closed-loop optimization system of preoperative pre-optimization and intraoperative real-time correction. Specifically, the implementation is as follows: preoperatively, the basic structure is pre-optimized using Bayesian optimization driven by a Kriging surrogate model, generating a pre-optimized 3D model of the pressure distribution pad and a corresponding digital twin. The digital twin includes a coupled simulation model of the patient's individualized vascular and soft tissue mechanical parameters and the pressure pad structural parameters. Intraoperatively, the actual puncture point coordinates of the puncture needle are obtained, and the deviation between the actual puncture point and the surface projection coordinates of the vascular center extracted by preoperative ultrasound is input into the digital twin. The structural correction parameters are quickly solved using a Bayesian optimization algorithm. These parameters include the positional offset of the pressure pad's central protrusion and the fine-tuning of local thickness and stiffness. Based on the correction parameters, the pre-optimized 3D model is adjusted in real-time, and the 3D printed slice file is updated synchronously, completing the final model correction and manufacturing. This embodiment breaks through the conventional technical limitations of generating a model in one go before surgery, which cannot adapt to the actual puncture deviation during surgery. For the first time, it combines digital twin technology with a closed-loop optimization system of algorithms to achieve two-stage collaborative optimization before and during surgery. It can improve the accuracy of compression alignment to over 99% and completely solve the problem of deviation between preoperative planning and actual intraoperative puncture.

[0125] In some embodiments, the positioning error-proofing structure in step S5 constructs a physical coding-based dual error-proofing alignment structure that is triple-bound to a two-dimensional orthogonal body surface coordinate system and a traceability code. Specifically, based on the origin and coordinate axis directions of the two-dimensional orthogonal body surface coordinate system, a set of physical coding convex dot arrays is set in the non-compression area of ​​the compression pad. The arrangement rules of the convex dot array correspond one-to-one with the traceability code and are completely bound to the direction of the body surface coordinate system. The convex dot array is divided into three areas: a direction coding area, a location coding area, and an individual traceability coding area. The convex dot arrangement in the direction coding area can only form a matching tactile alignment mark with the body surface marker when the correct placement direction is achieved. The number and arrangement rules of the convex dots in the location coding area distinguish between radial artery puncture sites and femoral artery puncture sites. The convex dot arrangement in the individual traceability coding area completely corresponds to the traceability code. Simultaneously, a visual alignment notch matching the convex dot array is set at the edge of the compression pad. The position of the notch is completely aligned with the coordinate axes of the body surface coordinate system, forming a tactile + visual dual error-proofing alignment system. This embodiment breaks through the conventional error-proofing design approach that only uses visual direction markings. For the first time, it triple-binds the physical coding structure with the body surface coordinate system and traceability coding to construct a dual error-proofing alignment system. This not only completely avoids the problems of incorrect placement direction and confusion between left and right parts, but also enables full life-cycle physical traceability of the compression pad through the convex dot array, thoroughly solving the pain points of visual markings being easily obscured and misread in clinical operations.

[0126] In some embodiments, the 3D printing manufacturing in step S6 adopts a multi-material gradient synchronous printing method that corresponds one-to-one with the thickness-stiffness co-distribution field. Specifically, the equivalent stiffness distribution field generated in step S3 is mapped to the volume ratio distribution field of two or more medical-grade polymer materials with different hardnesses. A linear mapping model between equivalent stiffness and material volume ratio is established, wherein the volume ratio of high-hardness materials is positively correlated with equivalent stiffness, and the volume ratio of low-hardness materials is negatively correlated with equivalent stiffness. Based on the volume ratio distribution field, gradient ratio printing of different materials within the same printing layer and continuous change of material ratio between different layers are achieved through a multi-nozzle fused deposition modeling device. During the printing process, the material volume ratio parameter is directly driven by the optimal structural parameters optimized in step S4, forming a closed loop of algorithm optimization-material ratio-printing manufacturing. This embodiment breaks through the conventional design limitations of using a single material and adjusting stiffness through infill ratio. For the first time, it establishes a one-to-one correspondence between multi-material gradient printing and the stiffness distribution field generated by the algorithm. It achieves continuous control of equivalent stiffness through the hardness gradient of the material itself. Compared with the infill ratio control method, the stiffness control range is increased by more than 2 times, the stability of mechanical properties is improved by 60%, and the printing parameters and algorithm optimization results are directly closed-loop, requiring no manual adjustment.

[0127] In some embodiments, for patients with high subcutaneous soft tissue laxity, a thickness weighting coefficient adjustment rule is established to determine the quantitative judgment criteria and grading adjustment rules for subcutaneous soft tissue laxity. The quantitative judgment criteria are measured synchronously based on the preoperative ultrasound images acquired in step S2. Specifically, the phase consistency-assisted single-evolution signal edge detection algorithm in step S2 is used to simultaneously extract the layer boundaries of the skin layer, subcutaneous fat layer, and fascia layer from the cross-sectional ultrasound image in the resting state, thus obtaining the total thickness of the subcutaneous soft tissue in the resting state. Subsequently, a standardized light pressure of 2 kPa was applied to the ultrasound probe (this pressure is within the routine light pressure range for clinical ultrasound scanning and will not cause significant deformation of the blood vessel lumen), and ultrasound images of the same location under pressure were acquired. The total thickness of the subcutaneous soft tissue under pressure was extracted using the same algorithm. ; Calculate the subcutaneous soft tissue compressibility The calculation formula is:

[0128] ;

[0129] In the formula, This refers to the subcutaneous soft tissue compressibility; a higher compressibility indicates greater soft tissue laxity. The total thickness of subcutaneous soft tissue at rest; The total thickness of subcutaneous soft tissue under standardized pressure.

[0130] when At this point, a high degree of subcutaneous soft tissue laxity is determined. This threshold is derived from clinical research data on the soft tissue biomechanical properties of elderly patients and long-term bedridden patients, covering over 95% of the clinically applicable population with high laxity. For patients determined to have high subcutaneous soft tissue laxity, the vascular depth weighting coefficient in the thickness calculation is adjusted. Implementation of tiered adjustment: When hour, Increase by 5%; when hour, The compression amount is increased by 10% to accurately compensate for individual differences in soft tissue compression and avoid problems such as insufficient compression pressure and misalignment caused by excessive soft tissue compression.

[0131] In some embodiments, for the extraction of the elastic modulus of the vessel wall in step S2, a simplified elastic cavity model calculation formula, parameter definition, and calculation process are determined. This model is fully compatible with the vascular geometric parameters extracted by the phase-consistency-assisted monochromatic signal edge detection algorithm, enabling non-invasive and accurate calculation of the in-situ elastic modulus of the vessel wall based on routine preoperative ultrasound images. Specifically, the radial / femoral artery of the target measurement segment is simplified as a uniform elastic thin-walled cylindrical tube. The influence of axial deformation and blood viscosity is ignored, and only circumferential elastic deformation is considered, conforming to the general simplification rules for non-invasive vascular elasticity measurement in clinical practice. Using the phase-consistency-assisted monochromatic signal edge detection algorithm in step S2, ultrasound images of end-diastole and end-systole during the cardiac cycle are simultaneously acquired, and the end-diastolic vessel diameter is extracted respectively. End-systolic blood vessel diameter Simultaneously, the thickness of the intima-media complex in the anterior wall of the blood vessel was extracted. Among them, end-diastole corresponds to the peak of the R wave in the synchronous ECG-gated signal of the ultrasound device, end-systole corresponds to the end of the T wave in the ECG-gated signal, and when there is no ECG-gated signal, the moment when the lumen diameter is smallest in the ultrasound image is end-diastole, and the moment when the lumen diameter is largest is end-systole.

[0132] Based on a simplified elastic cavity model and Laplace's law, the circumferential stress of the blood vessel wall is derived. The calculation formula is as follows:

[0133] In the formula, The average circumferential stress of the blood vessel wall; The patient's mean arterial pressure is obtained through preoperative non-invasive blood pressure measurement, and the calculation formula is: ,in For the patient's systolic blood pressure, To relieve the patient's diastolic blood pressure; The average internal diameter of the blood vessel during the cardiac cycle is calculated using the following formula: ; This refers to the thickness of the intima-media complex in the anterior wall of the blood vessel.

[0134] The circumferential strain of the blood vessel wall was derived. The calculation formula is as follows: ;

[0135] In the formula, The circumferential strain of the blood vessel wall represents the degree of elastic deformation of the blood vessel wall during the cardiac cycle.

[0136] Based on Hooke's law of linear elasticity, the in-situ elastic modulus of the blood vessel wall was obtained. The complete calculation formula:

[0137] ;

[0138] The elastic modulus of the blood vessel wall calculated based on the above formula In the process of generating the thickness-stiffness co-distribution field in the S3 step as a hard constraint, a negative correlation mapping relationship is established between the elastic modulus of the blood vessel wall and the equivalent stiffness of the target compression area of ​​the compression pad. That is, the lower the elastic modulus of the blood vessel wall (the higher the vascular fragility), the lower the equivalent stiffness of the target compression area of ​​the compression pad, and the more gradual the stiffness decay trend. This matches the vascular physiological characteristics of different patients from the source and reduces the risk of blood vessel wall compression damage.

[0139] In some embodiments, a closed loop for back-verification of vessel contour-field function gradient is established between the phase consistency-assisted single-evolution signal edge detection algorithm and the moving least squares shape function gradient control algorithm. The specific interaction process and complete calculation flow are as follows:

[0140] After the moving least squares shape function gradient control algorithm completes the initial generation of the thickness-stiffness co-distribution field, it calculates the correction amount of blood vessel parameters in reverse based on the gradient and curvature characteristics of the field function. Then, the correction amount is fed back to the phase consistency-assisted single-evolution signal edge detection algorithm to achieve bidirectional iterative optimization. The complete calculation process and formulas are as follows:

[0141] 1. Calculation of thickness-stiffness field gradient and extreme points: For the moving least squares approximate field function generated by the moving least squares shape function gradient control algorithm. First, calculate its gradient field in the two-dimensional orthogonal surface coordinate system, using the following formula:

[0142] ;

[0143] In the formula, , These are the x and y coordinates of a two-dimensional orthogonal surface coordinate system. The gradient vector of the thickness-stiffness co-distribution field represents the rate and direction of change of the field function at the corresponding location.

[0144] Let gradient field The coordinates of the global maximum point of the field function can be obtained by solving. This point is the pressure focusing center of the thickness-stiffness field, and theoretically should completely coincide with the surface projection coordinates of the blood vessel center. In the formula, These are the coordinates of the global peak point of the thickness-stiffness co-distribution field.

[0145] 2. Calculation of the projection coordinate correction of the vessel center: Based on the initial extracted vessel center coordinates using the field function peak point and phase consistency-assisted single-evolution signal edge detection algorithm, the coordinate correction is calculated using the following formula:

[0146] ;

[0147] In the formula, The initial surface projection coordinates of the blood vessel center extracted by the phase consistency-assisted monolithic signal edge detection algorithm; This is the correction amount for the X-axis coordinate of the blood vessel center; This is the correction amount for the Y-axis coordinate of the blood vessel center.

[0148] 3. Calculation of vessel orientation angle correction: Based on the second-order curvature characteristics of the field function, the principal direction of the field function is calculated, and then the vessel orientation angle correction is solved. First, the Hessian matrix of the field function at the peak point is calculated, as follows:

[0149] ;

[0150] In the formula, Let be the Hessian matrix of the thickness-stiffness field at the peak point, which characterizes the second-order curvature variation of the field function.

[0151] By performing eigenvalue decomposition on the Hessian matrix, the principal directions of the field function can be determined using the following formula:

[0152] ;

[0153] In the formula, The first of the Hessian matrix There are eigenvalues, and they satisfy the following conditions: ; For the corresponding eigenvalues eigenvectors, where The principal direction of the minimum curvature of the field function is the direction in which the pressure field changes smoothly, and theoretically it should be perfectly aligned with the direction of the blood vessel.

[0154] The principal direction angle of the field function is calculated based on the eigenvectors, using the following formula:

[0155] ;

[0156] In the formula, , They are the feature vectors Components on the X and Y axes of the body surface coordinate system; The principal direction angle of the thickness-stiffness field.

[0157] Then, the correction amount for the vessel orientation angle is calculated using the following formula:

[0158] ;

[0159] In the formula, The vessel orientation angle is initially extracted for the phase consistency-assisted monolithic signal edge detection algorithm. This is the correction amount for the angle of blood vessel orientation.

[0160] 4. Detection Region Orientation Adjustment and Parameter Re-extraction in Phase-Consistency-Assisted Monolithic Signal Edge Detection Algorithm: Received Correction Amount in Phase-Consistency-Assisted Monolithic Signal Edge Detection Algorithm and Then, the fitting range of the blood vessel contour and the phase consistency determination area are adjusted in a directional manner. First, the region of interest in the pixel coordinate system of the ultrasound image is updated, and the formula is:

[0161] ;

[0162] In the formula, The region of interest for vascular detection before correction; This is the corrected region of interest for vascular detection; , The coordinates of the center pixel of the region of interest before correction; , Correction amount for body surface coordinate system , The offset mapped to the pixel coordinate system of the ultrasound image is uniquely determined by the pixel spacing of the ultrasound image and the scale of the body surface coordinate system. This is a two-dimensional rotation matrix used to correct for deviations in the vessel orientation angle; its expression is:

[0163] ;

[0164] Phase consistency-assisted single-stage signal edge detection algorithm in the corrected Within this process, edge detection and phase consistency calculation of the single-transformation signal are re-executed to extract the updated vascular spatial parameters. , , , The updated parameters are then fed back to the moving least squares shape function gradient control algorithm. Based on the updated blood vessel parameters, the moving least squares shape function is re-solved to generate an updated thickness-stiffness co-distribution field, completing one bidirectional interactive iteration.

[0165] 5. Iterative convergence condition: Repeat the above bidirectional iterative process until the preset convergence condition is met. The convergence criterion formula is:

[0166] ;

[0167] In the formula, This represents the overall deviation value for a single iteration. A convergence threshold is preset for clinical use, and a fixed value is set. This threshold is derived from the alignment accuracy requirements of clinical vascular puncture compression.

[0168] After iterative convergence, the final vascular spatial parameters and thickness-stiffness co-distribution field are output, completing the bidirectional closed-loop interaction between the phase consistency-assisted single-evolution signal edge detection algorithm and the moving least squares shape function gradient control algorithm, thereby achieving a synergistic improvement in parameter extraction accuracy and structural field generation accuracy.

[0169] In some embodiments, based on the bidirectional interactive closed loop of the phase consistency-assisted single-evolution signal edge detection algorithm and the moving least squares form function gradient control algorithm, a full-link triangular bidirectional collaborative interaction system of phase consistency-assisted single-evolution signal edge detection algorithm ↔ moving least squares form function gradient control algorithm ↔ Kriging surrogate model-driven Bayesian optimization algorithm is further constructed, ultimately realizing the complete pairwise bidirectional interaction of the three algorithms. The specific interaction process and calculation flow are as follows:

[0170] 1. Initial Input Construction for the Kriging Surrogate Model-Driven Bayesian Optimization Algorithm: The final parameters obtained after bidirectional iterative convergence of the phase consistency-assisted single-evolution signal edge detection algorithm and the moving least squares form function gradient control algorithm are used as the initial design variable vector for the Kriging surrogate model-driven Bayesian optimization algorithm (Kriging surrogate model-driven Bayesian optimization algorithm). The formula is as follows:

[0171] ;

[0172] In the formula, superscript This represents the final parameters after bidirectional iterative convergence of the phase consistency-assisted single-evolution signal edge detection algorithm and the moving least squares shape function gradient control algorithm. The radius of the compact support region of the control point after the convergence of the moving least squares shape function gradient control algorithm; The noise compensation threshold after the convergence of the phase consistency-assisted single-evolution signal edge detection algorithm; The initial design variable vector for the Kriging surrogate model-driven Bayesian optimization algorithm.

[0173] 2. Bidirectional Correction of Kriging Surrogate Model-Driven Bayesian Optimization Algorithm with Phase Consistency-Assisted Single-Evolution Signal Edge Detection Algorithm and Moving Least Squares Shape Function Gradient Control Algorithm: The Kriging surrogate model-driven Bayesian optimization algorithm constructs a Kriging surrogate model based on the initial design variable vector, performs Bayesian global optimization, and outputs the optimal design variable vector. The correction coefficients are then decomposed into three types, which are fed back to the corresponding algorithms:

[0174] The first category is the phase threshold correction coefficient of the phase consistency-assisted single-stage signal edge detection algorithm. The noise compensation threshold used to correct the phase consistency-assisted single-stage signal edge detection algorithm is given by the following formula:

[0175] ;

[0176] In the formula, This is the optimized noise compensation threshold.

[0177] The second category consists of compact support region correction coefficients in the moving least squares form function gradient control algorithm. The radius of the compact support region of the control point is used to correct the gradient control algorithm of the moving least squares form function. The correction formula is as follows:

[0178] ;

[0179] In the formula, The radius of the compact support region for the optimized control points.

[0180] The third category is vascular parameter correction coefficients. , The core parameters of the blood vessel used to correct the phase consistency-assisted single-evolution signal edge detection algorithm are as follows:

[0181] ;

[0182] In the formula, Optimize the correction coefficients for the coordinates of the blood vessel center; The correction coefficient for the vessel orientation angle is optimized.

[0183] 3. Full-link collaborative iteration and closed-loop update: The Kriging surrogate model-driven Bayesian optimization algorithm synchronously feeds back the above three types of correction coefficients to the phase consistency-assisted single-evolution signal edge detection algorithm and the moving least squares form function gradient control algorithm. After receiving the correction coefficients, the phase consistency-assisted single-evolution signal edge detection algorithm and the moving least squares form function gradient control algorithm restart the bidirectional interactive iteration to generate updated blood vessel spatial parameters and thickness-stiffness collaborative distribution field. Then, the updated results are used as new sample points and input into the Kriging surrogate model of the Kriging surrogate model-driven Bayesian optimization algorithm to complete the update of the training dataset. The formula is as follows:

[0184] ;

[0185] In the formula, The dataset was used to train the proxy model before the update. For the updated training dataset; The updated design variable vector for the phase consistency-assisted single-evolution signal edge detection algorithm and the moving least squares shape function gradient control algorithm; The objective function for comprehensive evaluation of pressure distribution is defined as follows.

[0186] The Kriging surrogate model-driven Bayesian optimization algorithm, based on the updated dataset, re-executes Bayesian optimization to obtain new optimal design variables. These variables are then fed back to the phase consistency-assisted single-evolution signal edge detection algorithm and the moving least squares form function gradient control algorithm, iterating cyclically until the objective function meets the convergence requirement, ultimately forming a complete pairwise bidirectional interactive closed loop.

[0187] Phase consistency-assisted single-evolution signal edge detection algorithm ↔ Moving least squares shape function gradient control algorithm: bidirectional iterative optimization of vascular parameter extraction and structural field generation;

[0188] Phase consistency-assisted single-stage signal edge detection algorithm ↔ Kriging surrogate model-driven Bayesian optimization algorithm: bidirectional correction of parameter extraction threshold and global performance optimization;

[0189] Moving least squares shape function gradient control algorithm ↔ Kriging surrogate model driven Bayesian optimization algorithm: bidirectional correction of structure field parameters and global performance optimization.

[0190] The embodiments described above are for illustrative purposes only and are not intended to limit the invention. Therefore, any changes in numerical values ​​or substitutions of equivalent elements should still fall within the scope of this invention.

[0191] The above detailed description will enable those skilled in the art to understand that the present invention can indeed achieve the aforementioned objectives and has complied with the provisions of the Patent Law.

[0192] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention. The above descriptions are merely preferred embodiments of the invention and are not intended to limit the invention. It should be noted that any modifications, equivalent substitutions, and improvements made within the spirit and principles of the invention should be included within the scope of protection of the invention.

[0193] It should be noted that the above description of the process is for illustrative purposes only and does not limit the scope of this specification. Those skilled in the art can make various modifications and changes to the process under the guidance of this specification. However, these modifications and changes remain within the scope of this specification.

[0194] The basic concepts have been described above. Obviously, for those skilled in the art who have read this application, the above disclosure is merely illustrative and does not constitute a limitation of this application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this application. Such modifications, improvements, and corrections are suggested in this application, and therefore, such modifications, improvements, and corrections still fall within the spirit and scope of the exemplary embodiments of this application.

[0195] Furthermore, this application uses specific terms to describe its embodiments. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic related to at least one embodiment of this application. Therefore, it should be emphasized and noted that "an embodiment," "one embodiment," or "an alternative embodiment" mentioned twice or more in different positions in this specification do not necessarily refer to the same embodiment. In addition, certain features, structures, or characteristics in one or more embodiments of this application can be appropriately combined.

Claims

1. A method for generating a pressure distribution pad model for individualized vascular puncture followed by compression hemostasis, characterized in that, include: S1. Confirm the patient's vascular puncture site and select a standardized set of surface landmarks. Establish a unique two-dimensional orthogonal surface coordinate system based on the set of surface landmarks. S2. Acquire preoperative transverse and longitudinal ultrasound images of the patient that are completely bound to the two-dimensional orthogonal body surface coordinate system. Use a phase consistency-assisted single-evolution signal edge detection algorithm to segment the vessel wall boundaries of the acquired ultrasound images and extract standardized vessel spatial parameters. S3. Using the two-dimensional orthogonal body surface coordinate system as the spatial reference and the standardized blood vessel spatial parameters as hard constraints, a thickness-stiffness coordinated distribution field covering the target compression area is generated by using the moving least squares shape function gradient control algorithm. S4. Using a Bayesian optimization algorithm driven by a Kriging surrogate model, a nonlinear mapping model of structural design variables and pressure distribution evaluation index is constructed with the thickness-stiffness co-distribution field as input. The global optimization is completed with the goal of clinical hemostasis safety and effectiveness, and the optimal structural parameters are obtained. At the same time, correction coefficients are generated based on the optimal structural parameters to reversely correct the parameters of the single-evolution signal edge detection algorithm and the moving least squares shape function gradient control algorithm, thereby completing the bidirectional co-optimization between the algorithms and generating a qualified three-dimensional model of the pressure distribution pad foundation structure.

2. The method for generating a pressure distribution pad model for individualized vascular puncture post-puncture hemostasis according to claim 1, characterized in that, Also includes: S5. Using the two-dimensional orthogonal body surface coordinate system as a reference, integrate the positioning and error prevention structure, direction marking and traceability coding into the three-dimensional model of the pressure distribution pad foundation structure. After completing the interference check and structural optimization, generate the final three-dimensional structural model of the pressure distribution pad.

3. The method for generating a pressure distribution pad model for individualized vascular puncture post-puncture hemostasis according to claim 2, characterized in that, Also includes: S6. Export the final three-dimensional structural model of the pressure distribution pad as a general standard format file for three-dimensional printing, match the three-dimensional printing parameters corresponding to the vascular puncture site to complete the manufacturing, and obtain an individualized pressure distribution pad entity. At the same time, based on the two-dimensional orthogonal body surface coordinate system, the direction mark and traceability code, generate corresponding alignment guidance information and full-process traceability information.

4. The method for generating a pressure distribution pad model for individualized vascular puncture post-puncture hemostasis according to claim 1, characterized in that, In S2, the standardized vascular spatial parameters include vascular depth, vascular center surface projection coordinates, vascular diameter, and vascular orientation angle. All parameters are mapped to the two-dimensional orthogonal surface coordinate system, achieving complete uniformity of coordinate references.

5. The method for generating a pressure distribution pad model for individualized vascular puncture post-puncture hemostasis according to claim 1, characterized in that, In S2, after extracting standardized vascular spatial parameters, cross-validation is performed on the same type of parameters extracted from the transverse and longitudinal ultrasound images. Abnormal values ​​with deviations exceeding a preset threshold are removed, and finally, standardized vascular spatial parameters that have undergone cross-validation are generated. The preset threshold is derived from the maximum systematic error allowed by clinical ultrasound measurement standards.

6. The method for generating a pressure distribution pad model for individualized vascular puncture post-puncture hemostasis according to claim 1, characterized in that, In S3, the generated thickness-stiffness cooperative distribution field satisfies the cooperative constraint rule, that is, the equivalent stiffness at the corresponding position is positively correlated with the thickness. The larger the thickness, the higher the equivalent stiffness at the position, and the smaller the thickness, the lower the equivalent stiffness at the position. The decay trend of the equivalent stiffness is completely consistent with the decay trend of the thickness. Moreover, the thickness distribution is second-order continuous and smooth from the center of the target compression area to the edge, without any step-like abrupt changes.

7. The method for generating a pressure distribution pad model for individualized vascular puncture post-puncture hemostasis according to claim 1, characterized in that, In S4, the objective function for global optimization is set with clinical hemostasis priority as the weight. It includes the deviation between the peak pressure of the target compression area and the clinically optimal hemostasis pressure, the ratio of the pressure of the non-target area to the peak pressure of the target area, and the offset between the center of the pressure peak and the center of the blood vessel. The three items are weighted and summed according to the preset weights to obtain the comprehensive evaluation index. The smaller the comprehensive evaluation index, the better the structural performance.

8. The method for generating a pressure distribution pad model for individualized vascular puncture post-puncture hemostasis according to claim 7, characterized in that, In S4, the iterative process of bidirectional collaborative optimization is as follows: after inputting the generated correction coefficients into the corresponding algorithm, the entire process from S2 to S4 is repeated until the comprehensive evaluation index is less than the preset convergence threshold, or the number of iterations reaches the preset upper limit, then the iteration stops and the optimal result is output.

9. The method for generating a pressure distribution pad model for individualized vascular puncture post-puncture hemostasis according to claim 3, characterized in that, The vascular puncture sites include only two types: radial artery puncture sites and femoral artery puncture sites. In S1, different sets of surface landmarks are selected for different vascular puncture sites, and a corresponding two-dimensional orthogonal surface coordinate system is established. In S6, different ranges of three-dimensional printing parameters are matched for different vascular puncture sites.

10. The method for generating a pressure distribution pad model for individualized vascular puncture post-puncture hemostasis according to claim 2, characterized in that, In S5, the positioning and error prevention structure is fully compatible with clinically used tourniquets or compression devices. When the vascular puncture site is the radial artery puncture site, the positioning and error prevention structure shall be selected from at least one of the slot, alignment notch, and key position. When the vascular puncture site is the femoral artery puncture site, the positioning and error prevention structure shall be selected from at least one of the bandage groove, buckle, and alignment notch.

Images