A method and system for predicting cement dispersion in fractured vertebral bodies

Abstract

The application discloses a method and system for predicting cement dispersion distribution of fractured vertebral body. The application belongs to the technical field of medical image processing. The method comprises the following steps: acquiring a three-dimensional image of a fractured vertebral segment of a patient, and segmenting a fractured vertebral body in the three-dimensional image; calculating a permeability coefficient of each voxel in the fractured vertebral body to construct a flow dispersion model of cement in the fractured vertebral body; calculating the cumulative resistance of each voxel in the fractured vertebral body to cement according to the flow dispersion model to obtain a cost function; and calculating all voxels in the fractured vertebral body that can be dispersed to by cement based on the cost function by using a cost priority traversal algorithm, and then obtaining a prediction result of the cement dispersion distribution. The technical scheme can predict the cement dispersion distribution in the fractured vertebral body, effectively reduce the risk during and after the operation, reduce complications, significantly improve the operation treatment effect, and improve the operation efficiency.

Description

Technical Field

[0001] This application belongs to the field of medical image processing technology, specifically relating to a method and system for predicting the diffusion distribution of bone cement in fractured vertebrae. Background Technology

[0002] With the rapid development of orthopedic surgical techniques, bone cement is increasingly used in the treatment of vertebral body repair, vertebral fractures, and other related diseases. Accurate establishment and injection of bone cement channels can enable rapid treatment and rehabilitation of these conditions. Among bone cement-related surgeries, percutaneous vertebroplasty (PVP) and percutaneous kyphoplasty (PKP) are the preferred treatments for osteoporotic vertebral compression fractures. These procedures are characterized by minimal surgical trauma and bleeding, allowing for rapid relief of symptoms, correction of kyphosis, and rapid postoperative recovery.

[0003] In the aforementioned surgeries, the accuracy of bone cement injection is crucial in spinal vertebroplasty (such as percutaneous kyphoplasty). However, current bone cement injection procedures cannot accurately control the flow and distribution of the cement, nor can they ensure the stability of the injection pressure. Furthermore, the dynamic changes in the viscosity of the bone cement are not adequately considered, leading to leakage into vital structures such as the spinal canal, nerves, or blood vessels, resulting in complications such as nerve damage and bleeding. Simultaneously, leakage or improper injection may cause pressure changes within the vertebral body, vertebral rupture, or further damage to the spinal structure, increasing postoperative pain and rehabilitation difficulty for the patient. Moreover, current bone cement injection methods cannot predict the diffusion distribution of the cement, requiring multiple X-ray examinations during injection to confirm its diffusion status. This not only significantly increases the patient's exposure time to radiation, causing harm, but also fails to ensure adequate filling of vertebral cavities, potentially leading to excessive local compression or insufficient support due to uneven cement distribution. This compromises the stability of the fractured vertebra, affecting the surgical outcome, potentially increasing the risk of postoperative fracture recurrence, and increasing postoperative pain. Finally, each patient's fractured vertebral body morphology, size, and pathological characteristics are different. The current bone cement injection procedure cannot plan the bone cement injection according to the patient's individual vertebral body structure. This may lead to problems such as bone cement leakage or improper injection, as mentioned above, which may cause interference and damage to normal tissues.

[0004] Therefore, it is necessary to provide a method for predicting the diffusion distribution of bone cement in fractured vertebrae by establishing a flow model of bone cement in the porous medium of the patient's vertebra based on the individualized three-dimensional image of the vertebra. Summary of the Invention

[0005] This application provides a method and system for predicting the diffusion distribution of bone cement in fractured vertebral bodies. It can calculate the diffusion distribution of bone cement in the vertebral body, aiming to solve the problem of inaccurate prediction of the diffusion state of bone cement during vertebral fracture surgery. It can reduce bone cement leakage and seepage, improve surgical efficiency, and has a wide range of applications in related medical fields.

[0006] In a first aspect, embodiments of this application provide a method for predicting the diffusion distribution of bone cement in a fractured vertebral body, the method comprising: S1. Obtain three-dimensional images of the fractured vertebrae of the patient and segment the fractured vertebral bodies; S2. Calculate the permeability coefficient of each voxel in the fractured vertebral body to construct a flow and diffusion model of bone cement in the fractured vertebral body; S3. Calculate the cumulative resistance of each voxel in the fractured vertebral body to the bone cement based on the flow diffusion model to obtain the cost function; S4. Based on the cost function, the cost-first traversal algorithm is used to calculate all voxels in the fractured vertebra that can be diffused by bone cement, thereby obtaining the predicted diffusion distribution of bone cement.

[0007] Furthermore, S4 includes: S41. Configure a cost priority queue for storing voxels in ascending order according to the first cost function value, wherein, in the initial state, the cost priority queue contains only injection point voxels. S42. Take the first voxel of the cost priority queue as a specific voxel, traverse all its adjacent voxels, and determine whether its adjacent voxels can be filled. When there is a fillable voxel among its neighboring voxels, calculate the first cost function value of entering the fillable voxel from the injection point voxel and the second cost function value of entering the fillable voxel through a specific voxel. When the second cost function value is not greater than the first cost function value, replace the first cost function value with the second cost function value, and put the corresponding fillable voxel into the cost priority queue. S43. Repeat S42 until the cost priority queue is empty, obtain all the voxels that can be filled and their corresponding cost function values, sort them in ascending order according to the cost function values, and obtain the ascending sort result. S44. Obtain the amount of bone cement injected, and perform voxel filling according to the ascending sorting results to obtain all voxels in the fractured vertebral body that can be diffused by bone cement, thereby obtaining the predicted results of bone cement diffusion distribution.

[0008] Furthermore, in S42, the determination of whether adjacent voxels of a specific voxel can be filled is as follows: The flow dispersion model is used to obtain a metric value indicating whether the adjacent voxels of a specific voxel can be filled. If the metric value is not less than a set voxel filling threshold, it is determined that the adjacent voxels of the specific voxel can be filled.

[0009] Furthermore, in S42, if the adjacent voxels of a specific voxel cannot be filled or the second cost function value of the adjacent voxels that can be filled is greater than the first cost function value, the next adjacent voxel is sequentially judged to see if it can be filled, until all the adjacent voxels of the specific voxel have been traversed. If all adjacent voxels of a particular voxel cannot be filled, or the second cost function value of all adjacent voxels that can be filled is greater than the first cost function value, then there is no corresponding fillable voxel that can be put into the cost priority queue for that particular voxel, and the traversal for that particular voxel ends.

[0010] Furthermore, in S43, S42 is repeated. When none of the voxels in the cost priority queue have adjacent voxels that can be filled, or when the second cost function values ​​of all adjacent voxels that can be filled are greater than the first cost function value, all voxels in the cost priority queue are taken and set to empty.

[0011] Furthermore, S44 includes: Based on the amount of bone cement injected, the voxels in S43 are filled sequentially in ascending order until they are exhausted, and each voxel filled is marked to obtain all the voxels in the fractured vertebral body that can be diffused by bone cement, thus obtaining the diffusion distribution of bone cement.

[0012] Furthermore, S2 includes: Based on the measurement value of each voxel in the fractured vertebral body, the void density is calculated, and then the permeability coefficient of each voxel is obtained. The flow and diffusion model of the constructed bone cement in each voxel of the fractured vertebral body is specifically calculated using the following formula: ; Where S is a measure of whether the voxel can be filled, k is the permeability coefficient of the voxel, μ is the viscosity of the bone cement, and P is the injection pressure. P is the pressure gradient.

[0013] Furthermore, the calculation method for the permeability coefficient of each voxel includes: ; Where k is the permeability coefficient of the voxel, E is the geometric structure coefficient, which is determined based on the skeletal structure, and Φ is the porosity of the voxel, which is calculated based on the energy value of the voxel.

[0014] Furthermore, in S1, the target vertebral segment in the three-dimensional image containing the fractured vertebral segment of the patient is first segmented using the vertebral segmentation model, and then the fractured vertebral body in the fractured vertebral segment of the patient is segmented using the vertebral body segmentation model. The vertebral segmentation model was pre-trained using a 3D-Unet network, and the vertebral body segmentation model was pre-trained using a SwinUnetR network.

[0015] Secondly, embodiments of this application provide a device for predicting the diffusion distribution of bone cement in a fractured vertebral body, the device comprising: The fractured vertebral body segmentation module is used to acquire three-dimensional images of the fractured vertebral segments of the patient and segment out the fractured vertebral bodies; The flow and diffusion model construction module is used to calculate the permeability coefficient of each voxel in the fractured vertebral body in order to construct a flow and diffusion model of bone cement in the fractured vertebral body. The cost function determination module is used to calculate the cumulative resistance of each voxel in the fractured vertebral body to the bone cement based on the flow diffusion model, and obtain the cost function. The distribution prediction module is used to calculate all voxels in the fractured vertebral body that can be diffused by bone cement based on the cost function and using a cost-first traversal algorithm, thereby obtaining the diffusion distribution prediction result of bone cement.

[0016] Thirdly, embodiments of this application provide a bone cement diffusion distribution prediction system for fractured vertebral bodies. The bone cement diffusion distribution prediction system for fractured vertebral bodies includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor. When the program or instructions are executed by the processor, they implement the steps of the method described in the first aspect.

[0017] Fourthly, embodiments of this application provide a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method described in the first aspect.

[0018] Fifthly, embodiments of this application provide a chip, the chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the method as described in the first aspect.

[0019] Beneficial effects: This invention combines cutting-edge imaging technology with clinical surgical needs. By constructing a flow and diffusion model of bone cement in the fractured vertebral body and using a cost-priority traversal algorithm, it can predict the diffusion distribution of bone cement in the fractured vertebral body. This solves the problem of inaccurate prediction of the diffusion state of bone cement in vertebral fracture surgery. It can not only effectively reduce the risks during and after surgery and reduce complications, but also significantly improve the surgical treatment effect and increase surgical efficiency. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the method for predicting the diffusion distribution of bone cement in fractured vertebrae provided in Embodiment 1 of this application. Figure 2 Example image of extracting fractured vertebrae from three-dimensional images provided in this application embodiment; Figure 3 An example image of the brightness and darkness of the voxel Hu value of the fractured vertebral segment provided in the embodiments of this application; Figure 4 The flowchart of the calculation cost Ci value and the diffuse voxel provided for the embodiments of this application; Figure 5 An example diagram illustrating the prediction of bone cement diffusion status in fractured vertebral segments provided in this application embodiment; Figure 6 This is a schematic diagram of the bone cement diffusion distribution prediction device for fractured vertebrae provided in Embodiment 2 of this application; Figure 7 This is a schematic diagram of the structure of the bone cement diffusion distribution prediction system for fractured vertebrae provided in Embodiment 3 of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, specific embodiments of this application will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely for explaining this application and not for limiting it. It should also be noted that, for ease of description, only the parts relevant to this application are shown in the drawings, not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe operations (or steps) as sequential processes, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but may also have additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0022] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0023] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0024] The following detailed description, in conjunction with the accompanying drawings, of the bone cement diffusion distribution prediction method and system for fractured vertebrae provided in this application, through specific embodiments and application scenarios, will be provided in detail.

[0025] Example 1 Figure 1 This is a flowchart illustrating the method for predicting the diffusion distribution of bone cement in fractured vertebrae provided in Embodiment 1 of this application. Figure 1 As shown, the specific steps include the following: S1. Obtain three-dimensional images of the fractured vertebrae of the patient and segment the fractured vertebral bodies.

[0026] A fractured vertebral segment refers to a vertebral segment in the spine that has suffered an osteoporotic compression fracture or a fracture caused by external force. For example, it could be a thoracic vertebra or a lumbar vertebra.

[0027] Three-dimensional imaging refers to medical imaging data that can present three-dimensional spatial information of human tissue structure. For example, it can be three-dimensional computed tomography (CT) images.

[0028] Segmentation is the process of extracting the anatomical structure of a target from an overall image using image segmentation algorithms or deep learning segmentation models.

[0029] This solution can acquire three-dimensional images of the fractured vertebrae of the patient, and through image segmentation processing, separate and extract the fractured vertebrae from the three-dimensional images to achieve precise localization and extraction of the target anatomical region.

[0030] S2. Calculate the permeability coefficient of each voxel in the fractured vertebral body to construct a flow and diffusion model of bone cement in the fractured vertebral body.

[0031] Among them, a voxel is the smallest volume unit that represents a three-dimensional spatial position in a three-dimensional medical image, and it is the basic unit that constitutes a three-dimensional vertebral model.

[0032] The permeability coefficient is a physical parameter that characterizes the ability of a fluid to permeate within a porous medium. It is used to reflect the ease with which bone cement flows through the trabecular gaps in the vertebral body.

[0033] The flow and diffusion model is a mathematical model based on fluid mechanics and porous media theory, used to describe the flow, diffusion and filling of bone cement inside a fractured vertebral body.

[0034] Specifically, by traversing all voxels contained in the fractured vertebral body, calculating the permeability coefficient corresponding to each voxel, and using the permeability coefficient as the core parameter, a flow and diffusion model of bone cement in the fractured vertebral body can be constructed.

[0035] S3. Calculate the cumulative resistance of each voxel in the fractured vertebral body to the bone cement based on the flow diffusion model to obtain the cost function.

[0036] Cumulative resistance refers to the total resistance value of bone cement as it flows from the injection point to the target voxel along the entire path.

[0037] The cost function is a function used to quantify the flow resistance required for bone cement to reach each voxel. The smaller the cost function value, the easier it is for the bone cement to reach that voxel.

[0038] This approach can use the established flow dispersion model to calculate the cumulative resistance to bone cement flow on a voxel-by-voxel basis, and then convert the cumulative resistance into a cost function to form a correspondence between voxels and costs.

[0039] S4. Based on the cost function, the cost-first traversal algorithm is used to calculate all voxels in the fractured vertebra that can be diffused by bone cement, thereby obtaining the predicted diffusion distribution of bone cement.

[0040] The cost-first traversal algorithm is a path search algorithm that visits voxels in order of priority based on their cost function values ​​from smallest to largest.

[0041] The diffusion distribution prediction result refers to the three-dimensional prediction data of the range, location and shape of bone cement that can be filled in the fractured vertebral body under the set injection volume and injection pressure.

[0042] The cost function can be used as the traversal basis to run the cost-first traversal algorithm, which filters out all voxel sets that can be diffusely filled by bone cement, and finally outputs the predicted results of the diffusion distribution of bone cement in the vertebral body.

[0043] The technical solution provided in this embodiment achieves individualized and accurate prediction of bone cement dispersion distribution through three-dimensional image segmentation, voxel permeability coefficient calculation, cumulative resistance cost function construction, and cost-first traversal. This solves the problem of unpredictable bone cement distribution in vertebral fracture surgery, reduces leakage risk, and improves surgical efficiency.

[0044] In one embodiment, optionally, S4 includes the following steps.

[0045] S41. Configure a cost priority queue for storing voxels in ascending order according to the first cost function value, wherein, in the initial state, the cost priority queue contains only injection point voxels. S42. Take the first voxel of the cost priority queue as a specific voxel, traverse all its adjacent voxels, and determine whether its adjacent voxels can be filled. When there is a fillable voxel among its neighboring voxels, calculate the first cost function value of entering the fillable voxel from the injection point voxel and the second cost function value of entering the fillable voxel through a specific voxel. When the second cost function value is not greater than the first cost function value, replace the first cost function value with the second cost function value, and put the corresponding fillable voxel into the cost priority queue. S43. Repeat S42 until the cost priority queue is empty, obtain all the voxels that can be filled and their corresponding cost function values, sort them in ascending order according to the cost function values, and obtain the ascending sort result. S44. Obtain the amount of bone cement injected, and perform voxel filling according to the ascending sorting results to obtain all voxels in the fractured vertebral body that can be diffused by bone cement, thereby obtaining the predicted results of bone cement diffusion distribution.

[0046] The cost priority queue is a data structure that can automatically sort voxels in ascending order based on their cost function values, and can retrieve the voxel with the lowest current cost each time.

[0047] The first cost function value refers to the initial value of the cost function set initially, which is the cost function that directly reaches the target voxel from the injection point voxel.

[0048] Injection point voxels refer to the voxels at the starting position where bone cement begins to be injected into the vertebral body during surgery.

[0049] This solution can initialize the traversal algorithm by configuring a cost-priority queue and adding the injection point voxel as the initial element to the queue.

[0050] A specific voxel refers to the central voxel currently taken from the cost priority queue for neighborhood search.

[0051] Adjacent voxels refer to the set of voxels that are directly adjacent to a specific voxel in three-dimensional space.

[0052] The first element is taken from the cost priority queue as a specific voxel. All adjacent voxels of the specific voxel are visited in turn, and a fillability check is performed on each adjacent voxel.

[0053] The second cost function value refers to the updated cost function value of reaching the target adjacent voxel after passing through the current specific voxel.

[0054] Bone cement injection volume refers to the total volume of bone cement to be injected into the fractured vertebral body according to the surgical plan.

[0055] Specifically, two path costs are calculated for adjacent voxels that can be filled. When the path cost through a specific voxel is better, the cost function value of that voxel is updated, and the voxel is added to a cost priority queue to await subsequent traversal. The voxel retrieval, neighborhood traversal, cost update, and queue enqueue operations in S42 are executed repeatedly until there are no voxels to be processed in the queue. Then, all fillable voxels are sorted in ascending order of cost. The preset bone cement injection volume is then obtained, and voxels are filled sequentially according to the ascending sorting results until the injection volume is exhausted, finally determining the diffusion distribution of the bone cement.

[0056] This technical solution achieves the optimal search for the flow path of bone cement by comparing the cost priority queue with the path cost, ensuring that the diffusion prediction is more in line with the actual physical flow law.

[0057] In one embodiment, optionally, in S42, the determination of whether adjacent voxels of a specific voxel can be filled is as follows: The flow dispersion model is used to obtain a metric value indicating whether the adjacent voxels of a specific voxel can be filled. If the metric value is not less than a set voxel filling threshold, it is determined that the adjacent voxels of the specific voxel can be filled.

[0058] The metric can be a value calculated by a flow dispersion model to characterize the ability of voxels to be filled by bone cement.

[0059] Voxel filling threshold refers to a pre-set critical value used to determine whether a voxel meets the conditions for bone cement filling.

[0060] This scheme can calculate the metric values ​​of adjacent voxels based on the flow dispersion model, and compare the metric values ​​with the voxel filling threshold. If the conditions are met, it is determined that the voxel can be filled.

[0061] This technical solution achieves standardized judgment of voxel fillability by comparing quantitative metric values ​​with thresholds, thereby improving the accuracy and consistency of prediction.

[0062] In one embodiment, optionally, in S42, if the adjacent voxels of a specific voxel cannot be filled or the second cost function value of the adjacent voxels that can be filled is greater than the first cost function value, the next adjacent voxel is sequentially determined to be filled until all the adjacent voxels of the specific voxel have been traversed. If all adjacent voxels of a particular voxel cannot be filled, or the second cost function value of all adjacent voxels that can be filled is greater than the first cost function value, then there is no corresponding fillable voxel that can be put into the cost priority queue for that particular voxel, and the traversal for that particular voxel ends.

[0063] The process involves sequentially checking adjacent voxels and skipping over any that do not meet the criteria. When all adjacent voxels fail to meet the enqueue criteria, the traversal of the current voxel is terminated.

[0064] This technical solution clarifies the termination rules for neighborhood traversal, avoids invalid calculations, and improves the algorithm's execution efficiency and logical rigor.

[0065] In one embodiment, optionally, in S43, S42 is repeated, and when none of the voxels in the cost priority queue have adjacent voxels that can be filled, or when the second cost function values ​​of all adjacent voxels that can be filled are greater than the first cost function value, all voxels in the cost priority queue are taken and set to empty.

[0066] In the continuous loop traversal process, when no new fillable neighboring voxels are generated in all voxels in the queue, the queue is automatically cleared and the algorithm moves to the next step.

[0067] This technical solution clearly defines the conditions for clearing the cost priority queue, ensuring that the entire traversal process has complete termination logic and guaranteeing the stable operation of the algorithm.

[0068] In one embodiment, optionally, S44 includes: Based on the amount of bone cement injected, the voxels in S43 are filled sequentially in ascending order until they are exhausted, and each voxel filled is marked to obtain all the voxels in the fractured vertebral body that can be diffused by bone cement, thus obtaining the diffusion distribution of bone cement.

[0069] The labeling refers to assigning a recognizable status identifier to a voxel that has been filled with bone cement, which is used to distinguish between filled and unfilled areas.

[0070] This scheme can allocate bone cement volume sequentially in order of increasing cost until the total injection volume is reached, and mark the filler voxels to form the final three-dimensional diffuse distribution.

[0071] This technical solution achieves diffusion prediction that matches the clinical injection volume by prioritizing filling and voxel labeling, resulting in more clinically instructive outcomes.

[0072] In one embodiment, optionally, S2 includes: Based on the measurement value of each voxel in the fractured vertebral body, the void density is calculated, and then the permeability coefficient of each voxel is obtained.

[0073] Among them, porosity refers to the proportion of the volume occupied by the interstices of bone within a voxel, which is used to reflect the porosity of vertebral osteoporosis and fractures.

[0074] The flow and diffusion model of the constructed bone cement in each voxel of the fractured vertebral body is specifically calculated using the following formula: ; Where S is a measure of whether the voxel can be filled, k is the permeability coefficient of the voxel, μ is the viscosity of the bone cement, and P is the injection pressure. P is the pressure gradient.

[0075] Bone cement viscosity refers to the degree of viscosity of the bone cement itself, which directly affects its flowability inside the vertebral body.

[0076] This approach first calculates the void density from the voxel metric, then obtains the permeability coefficient from the void density, and finally substitutes it into the formula to construct a flow dispersion model.

[0077] This technical solution constructs a model using explicit physical formulas, providing fluid dynamics theory support for the prediction of bone cement dispersion, making the prediction more scientific and reliable.

[0078] In one embodiment, optionally, the permeability coefficient of each voxel is calculated in the following ways: ; Where k is the permeability coefficient of the voxel, E is the geometric structure coefficient, which is determined based on the skeletal structure, and Φ is the porosity of the voxel, which is calculated based on the energy value of the voxel.

[0079] Among them, the geometric structure coefficient refers to the correction coefficient used to reflect the geometric morphology, pore connectivity and pore shape of the vertebral trabeculae.

[0080] Energy value refers to the CT value corresponding to a voxel in a 3D CT image, also known as the HU value (Hounsfield Unit), which is used to reflect tissue density.

[0081] Specifically, the porosity is calculated based on the energy value of the voxel, and then the permeability coefficient is calculated using the formula in combination with the geometric structure coefficient.

[0082] This technical solution calculates the permeability coefficient using a formula related to pore structure and density, making the parameters more closely match the individualized vertebral structure of the patient and improving prediction accuracy.

[0083] In one embodiment, optionally, in S1, the target vertebral segment in the three-dimensional image containing the fractured vertebral segment of the patient is first segmented using a vertebral segmentation model as the fractured vertebral segment, and then the fractured vertebral body in the fractured vertebral segment of the patient is segmented using a vertebral body segmentation model.

[0084] The vertebral segmentation model was pre-trained using a 3D-Unet network, and the vertebral body segmentation model was pre-trained using a SwinUnetR network.

[0085] A vertebral segmentation model is a deep learning model used to segment individual vertebral regions from three-dimensional images of the spine.

[0086] Vertebral segmentation model refers to a deep learning model used to further and more accurately segment the vertebral bone region from a single vertebral segment.

[0087] The 3D-Unet network is a fully convolutional deep learning network suitable for 3D medical image segmentation.

[0088] The SwinUnetR network is a 3D medical image segmentation network based on the Transformer architecture.

[0089] This approach first uses a segmental segmentation model to locate the fractured vertebral segment, and then uses a vertebral body segmentation model to extract the fractured vertebral body, achieving two levels of precise segmentation.

[0090] This technical solution improves the accuracy and robustness of fractured vertebral body extraction through a two-level deep learning segmentation model, providing reliable three-dimensional structural data for subsequent calculations.

[0091] To enable those skilled in the art to better understand this solution, this application also provides a preferred embodiment.

[0092] S1. Obtain a three-dimensional image containing the fractured vertebral segment of the patient, and segment the fractured vertebral body within it; S2. Based on the energy value of each voxel in the fractured vertebral body obtained from S1 segmentation, calculate its pore density, and then calculate its permeability coefficient, thereby constructing a flow and diffusion model of bone cement in the fractured vertebral body. S3. Based on the flow and diffusion model of bone cement in the fractured vertebral body constructed in S2, calculate the cost function of each voxel in the fractured vertebral body. S4. Combining S2 and S3, a cost-priority traversal algorithm is used to calculate all voxels in the fractured vertebral body that can be diffused by bone cement, thereby obtaining the diffusion distribution of bone cement.

[0093] In this invention, the target vertebral segment in the three-dimensional image containing the fractured vertebral segment of the patient can first be segmented using a vertebral segmentation model, i.e., the fractured vertebral segment of the patient. Then, the fractured vertebral body in the fractured vertebral segment of the patient can be segmented using a vertebral body segmentation model, such as... Figure 2 As shown, the left image is the obtained three-dimensional image containing the fractured vertebral segment of the patient, the middle image is the segmented three-dimensional image of the fractured vertebral segment of the patient, and the right image is the segmented three-dimensional image of the fractured vertebral body.

[0094] In this invention, both the vertebral segmentation model and the vertebral body segmentation model can be obtained through pre-training.

[0095] In this embodiment, the vertebral segmentation model can be pre-trained using a 3D-Unet network, and the vertebral body segmentation model can be pre-trained using a SwinUnetR network.

[0096] In this invention, the three-dimensional image can be a three-dimensional CT image, which generally includes multiple spinal vertebrae, including fractured vertebrae.

[0097] In this invention, image analysis of three-dimensional images can yield three-dimensional voxel information for each vertebral segment, including the position and energy value of each voxel, such as... Figure 3 As shown, its position is generally represented by three-dimensional coordinates (x, y, z) or one-dimensional index, and its energy value is represented by the standard HU value. This standard defines the HU value of water as 0, air as -1000, and dense skeleton as approximately +1000.

[0098] In this invention, the porosity of a voxel in the fractured vertebral body obtained by S1 segmentation is calculated as follows: ; Where Φ is the void density of the voxel, Hu is the energy value of the voxel, and ρbone is the bone tissue density constant; a and b are the linear mapping coefficients from the known energy value of the voxel to the bone tissue density, which can be obtained through pre-calibration or by using empirical parameters known in relevant literature in the prior art.

[0099] Based on the foregoing, in this invention, the permeability coefficient k of a certain voxel is calculated as follows: ; Here, E is the geometric structure coefficient, used to reflect the geometry of porous media, particularly the connectivity, shape, and size of pores. For typical trabecular bone structures, E is 5-15; for dense bone structures, E may be smaller (e.g., less than or equal to 5); for cancellous bone or vertebral bodies in patients with osteoporosis, due to the sparseness of trabeculae and increased porosity, E is 10-15.

[0100] Therefore, the flow and diffusion model of bone cement in a certain voxel of the fractured vertebral body constructed in this invention is as follows: ; Where S is a measure of whether the voxel can be filled, μ is the viscosity of the bone cement, and P is the injection pressure. P is the pressure gradient.

[0101] The flow and diffusion model of bone cement in a certain voxel of a fractured vertebral body constructed in this invention can reflect the ability of each voxel of the vertebral body to be filled when bone cement is injected.

[0102] In this invention, a voxel filling threshold St can be set. When the S value corresponding to a certain voxel in the fractured vertebral body is greater than or equal to the voxel filling threshold St, it indicates that it can be filled with bone cement, as follows: .

[0103] In this invention, the cost function represents the cumulative resistance of a voxel to bone cement. It can be obtained by summing the reciprocals of the measures of whether all voxels j passing through the bone cement injection point to the voxel can be filled, calculated according to S2. Therefore, the smaller the cost function value of the flow from the injection point to the voxel, the smaller its cumulative resistance to bone cement, the easier it is for bone cement to flow to the voxel, and the easier it is for the voxel to be filled by bone cement.

[0104] In this embodiment, the cost function Ci of a voxel in the fractured vertebral body is calculated as follows: ; Where path represents the set of voxels that the path from the bone cement injection point to voxel i passes through; Sj represents the measure of whether voxel j in path can be filled.

[0105] The purpose of this invention is to predict the diffusion distribution of bone cement after injection into the fractured vertebral body. In essence, it involves calculating whether each voxel of the fractured vertebral body can diffuse after bone cement injection. Therefore:

[0106] Therefore, a cost-priority traversal algorithm is used to calculate all voxels that can be diffused by bone cement in order to predict the diffusion distribution of bone cement.

[0107] In this invention, reference is made to Figure 4 A cost-priority traversal algorithm is used to calculate all voxels in the fractured vertebral body that can be diffused by bone cement, specifically including: S41. Configure a cost priority queue Q for storing voxels. All voxels in the cost priority queue Q are sorted in ascending order according to their first cost function values. Initially, the cost priority queue contains only injection point voxels s.

[0108] In this invention, the cost function of each voxel in the fractured vertebra is the first cost function value.

[0109] In this invention, the cost function Cs corresponding to the injection point voxel s is initialized to 0. At the same time, the cost function value of all voxels in the fractured vertebral body is set to infinity, i.e., C=+∞. This means that the cost function, i.e. the first cost function value, from the bone cement injection point voxel s to all other voxels in the fractured vertebral body is initially infinity, which means that the bone cement does not reach any other voxels at the beginning.

[0110] S42. Take the first voxel in the cost priority queue as a specific voxel, traverse all its adjacent voxels, and determine whether its adjacent voxels can be filled. If there is a fillable voxel among its adjacent voxels, calculate the first cost function value of entering the fillable voxel from the bone cement injection point voxel and the second cost function value of entering the fillable voxel through the specific voxel. If the second cost function value is not greater than the first cost function value, replace the first cost function value with the second cost function value, and put the corresponding fillable voxel into the cost priority queue.

[0111] In this invention, all voxels in the cost priority queue Q are sorted in ascending order according to their first cost function values. Therefore, the first voxel in the cost priority queue is the voxel with the smallest first cost function value.

[0112] In this invention, whether adjacent voxels of a specific voxel can be filled can be determined based on the flow dispersion model obtained in S2.

[0113] In this invention, it is sequentially determined whether each adjacent voxel of a specific voxel can be filled. Specifically, the first voxel in the cost priority queue Q, i.e., the specific voxel, is defined as u, and its adjacent voxels are defined as v. When the adjacent voxel v is a fillable voxel, the first cost function value and the second cost function value of the adjacent voxel v are calculated. If the adjacent voxel v cannot be filled, the next adjacent voxel v+1 is sequentially determined whether it can be filled, until all adjacent voxels of the specific voxel u have been traversed. In particular, if all adjacent voxels of the specific voxel u cannot be filled, then the specific voxel u does not have a corresponding fillable voxel that can be put into the cost priority queue Q, and the traversal of the specific voxel u ends.

[0114] In this invention, a first cost function value Cv is defined as the cost function value of voxel entering a fillable voxel from the bone cement injection point, and a second cost function value Cnew = Cu + cv, where Cu is the cost function value of voxel entering a specific voxel from the bone cement injection point, and cv is the cost function value of voxel entering a fillable voxel from a specific voxel u.

[0115] In this invention, the first cost function value and the second cost function value of each fillable adjacent voxel of a specific voxel u are calculated and judged sequentially. Specifically, if the second cost function value of the currently fillable adjacent voxel is greater than the first cost function value, the next adjacent voxel is sequentially judged to see if it can be filled. Based on its fillability, the first cost function value and the second cost function value of the fillable adjacent voxel are calculated and the above judgment is performed until all adjacent voxels of the specific voxel u have been traversed. In particular, if the second cost function value of all fillable adjacent voxels of a specific voxel u is greater than the first cost function value, then there is no corresponding fillable adjacent voxel for the specific voxel u that can be put into the cost priority queue Q, and the traversal of the specific voxel u ends.

[0116] S43. Repeat S42 until the cost priority queue Q is empty. At this point, all voxels that can be filled and their corresponding first cost function values ​​are obtained. Sort them in ascending order according to the first cost function values.

[0117] In this invention, in step S42, after the first voxel of the cost priority queue Q is taken out, that voxel is removed from the cost priority queue Q and a new voxel that can be filled is placed in it. All voxels in the cost priority queue Q will be sorted in ascending order according to their first cost function values. There will be a new first voxel. By repeating step S42, the first voxel can be taken out continuously, and it is determined whether the adjacent voxels of the first voxel can be filled. When there are no adjacent voxels that can be filled in the cost priority queue Q, or when the second cost function values ​​of all adjacent voxels that can be filled are greater than the first cost function value, all voxels in the cost priority queue Q are taken out, that is, it is empty.

[0118] In this invention, in order to facilitate the acquisition of all fillable voxels and their corresponding first cost function values, after the fillable voxels are acquired, they can be placed in a first sequence, their first cost function values ​​can be updated, and they can be sorted in ascending order according to the first cost function values. When the cost priority queue Q is empty, all fillable voxels and their corresponding cost function values ​​are obtained in this sequence, and they are sorted in ascending order according to the first cost function values.

[0119] S44. Obtain the amount of bone cement injected, and fill the voxels according to the sorting sequence obtained in S43 to obtain all the voxels in the fractured vertebral body that can be diffused by bone cement, thereby obtaining the diffusion distribution of bone cement.

[0120] In this invention, based on the bone cement injection volume Vin, the voxels in S43 are sequentially filled according to the ascending order until exhausted. Each filled voxel is marked as Di=1, thus:

[0121] Wherein, Vvi represents the amount of bone cement injected into the filled voxel i, which can be calculated from the physical size parameters of the three-dimensional image of the voxel, and is generally a fixed value.

[0122] For reference Figure 5 The figure shows an example of the prediction effect of the bone cement diffusion distribution prediction method of the present invention on the bone cement diffusion distribution on the fractured vertebral segment. The red area in the figure is the area of ​​bone cement diffusion distribution, and the voxels are marked as Di=1.

[0123] The present invention also provides a bone cement diffusion distribution prediction system for fractured vertebrae based on the aforementioned method for predicting bone cement diffusion distribution, comprising: The image processing unit is used to acquire three-dimensional images containing the fractured vertebrae of the patient and segment the fractured vertebrae. The diffusion model construction unit is used to calculate the pore density of each voxel in the fractured vertebral body based on the energy value obtained by the image processing unit segmentation, and then obtain its permeability coefficient, thereby constructing a flow and diffusion model of bone cement in the fractured vertebral body. The diffusion distribution prediction unit is used to calculate the cost function of each voxel in the fractured vertebra based on the flow diffusion model constructed by the diffusion model construction unit, and to calculate all voxels in the fractured vertebra that can be diffused by bone cement using a cost-first traversal algorithm, thereby obtaining the diffusion distribution of bone cement.

[0124] This invention can also provide a bone cement injection simulation system for fractured vertebrae combined with virtual reality technology, comprising: The aforementioned bone cement diffusion distribution prediction system for fractured vertebral bodies; The planning unit is used to plan the bone cement injection points, dosage, channels and injection methods based on the bone cement diffusion distribution predicted by the aforementioned bone cement diffusion distribution prediction system for fractured vertebrae. The simulation unit is used to simulate bone cement injection according to the planning unit, and to simulate and display the diffusion distribution of bone cement in the fractured vertebral body in real time, so as to verify and optimize the bone cement diffusion distribution prediction system of the fractured vertebral body.

[0125] This invention combines multiple fields such as medical image processing, porous media fluid mechanics, computational geometry and path planning, and integrates cutting-edge imaging technology with clinical surgical needs. By constructing a flow and diffusion model of bone cement in the fractured vertebral body, it can reflect the ability of each voxel in the vertebral body to be filled during bone cement injection. Based on the amount of bone cement injected, a cost-first traversal algorithm is used to calculate all voxels in the fractured vertebral body that can be diffused by bone cement, which can predict the diffusion distribution of bone cement in the fractured vertebral body. This solves the problem of inaccurate prediction of the diffusion state of bone cement in vertebral fracture surgery. By predicting the diffusion distribution of bone cement, the flow path of bone cement can be accurately grasped, reducing the risk of bone cement leakage and seepage, thereby reducing the occurrence of complications such as nerve damage and bleeding caused by leakage.

[0126] This invention enables effective preoperative planning of injection protocols, helping doctors optimize injection sites, dosages, channels, and injection methods. This ensures bone cement better fills vertebral cavities, enhances the stability of fractured vertebrae, helps alleviate postoperative pain, and promotes early recovery of daily activities. It not only effectively reduces risks during and after surgery and minimizes complications but also significantly improves surgical outcomes and efficiency. It has significant clinical value in ensuring surgical safety and enhancing treatment effectiveness, and has wide applications in related medical fields. This invention can also be used for real-time intraoperative simulation as a reference for actual bone cement injection. Furthermore, it can be combined with virtual reality technology to continuously optimize and validate preoperative bone cement injection planning, ensuring surgical safety and treatment efficacy.

[0127] Understandably, the bone cement diffusion distribution prediction method for fractured vertebrae proposed in this solution differs fundamentally from the "structural reinforcement grouting / adhesive injection" field in building engineering not only in the difference between macroscopic and microscopic scales, but also in the increased bioactivity of the physical medium, the complexity of fluid time-varying mechanisms, and the stringent "zero tolerance" requirements for safety boundaries in clinical medicine.

[0128] Compared to buildings, the model used in this scheme deals with highly heterogeneous and anisotropic living biological tissue. The cancellous trabeculae inside the vertebral body exhibit three-dimensional random fractures and localized cavitation. In contrast, the building structure reinforcement model deals with relatively homogeneous artificial materials (such as concrete and masonry). Its defects are mostly manifested as macroscopic geometric cracks, gaps, or regular cavities. The matrix parameters input during model training are mostly macroscopic and homogeneous physical indicators, without the need to analyze microscopic and discrete pathological structures.

[0129] The ultimate value of the model used in this approach lies in establishing a closed loop for predicting the postoperative diffusion morphology of bone cement for specific fracture types. Before surgery, surgeons can adjust bone cement parameters in the model (such as viscosity gradient, injection pressure, and total injection volume) to perform digital reverse screening from a series of alternative bone cements, intuitively assessing their filling uniformity and mechanical support, thereby accurately selecting the optimal bone cement material and injection plan for the patient's current injury. In contrast, building reinforcement models focus on predicting long-term mechanical performance after construction, rather than optimizing high-time-sensitivity, high-flowability materials before surgery (pre-construction).

[0130] Furthermore, the vertebral body is adjacent to high-risk anatomical structures such as the spinal cord, central nervous system, and vertebral venous plexus. Even a small amount of bone cement leakage can directly lead to paraplegia or fatal pulmonary embolism. Therefore, the core optimization and threshold setting of this model is to treat absolute leakage prevention as a hard penalty, essentially making it a high-precision safety control model that navigates within extremely high-risk anatomical boundaries. In contrast, building reinforcement models have a higher fault tolerance.

[0131] The bone cement diffusion prediction model in this scheme is essentially a precise deterministic platform for controlling fluid leakage and mechanical reconstruction within a microscopic, highly heterogeneous, and high-risk biological barrier. It perfectly decouples and reintegrates the patient's unique fracture spatial topology with the time-varying physical properties of specific bone cement, enabling physicians to move from qualitative, experience-based injection to quantitative, reverse-engineering optimal solutions. In contrast, the building reinforcement model is an engineering application model aimed at improving the overall structural ultimate bearing capacity within macroscopic, relatively homogeneous solid defects. There is a significant disciplinary gap between the two in terms of their physical core, computational accuracy, timeliness, and sensitivity to safety boundaries.

[0132] Example 2 Figure 6 This is a schematic diagram of the bone cement diffusion distribution prediction device for fractured vertebrae provided in Embodiment 2 of this application. Figure 6 As shown, the device includes: The fractured vertebral body segmentation module 601 is used to acquire three-dimensional images of the fractured vertebral segment of the patient and segment the fractured vertebral body therein. The flow diffusion model construction module 602 is used to calculate the permeability coefficient of each voxel in the fractured vertebral body in order to construct a flow diffusion model of bone cement in the fractured vertebral body. The cost function determination module 603 is used to calculate the cumulative resistance of each voxel in the fractured vertebral body to the bone cement based on the flow diffusion model, and obtain the cost function. The distribution prediction module 604 is used to calculate all voxels in the fractured vertebral body that can be diffused by bone cement based on the cost function and using a cost-first traversal algorithm, thereby obtaining the diffusion distribution prediction result of bone cement.

[0133] The bone cement diffusion distribution prediction device for fractured vertebrae in this application embodiment can be a system, or a component, integrated circuit, or chip in a terminal.

[0134] The bone cement diffusion distribution prediction device for fractured vertebrae in this embodiment can be integrated into a device with an operating system. This operating system can be Android, iOS, or other possible operating systems; this embodiment does not impose specific limitations.

[0135] The bone cement diffusion distribution prediction device for fractured vertebrae provided in this application embodiment can realize the various processes of the above embodiments, and will not be described again here to avoid repetition.

[0136] Example 3 like Figure 7 As shown, this application embodiment also provides a bone cement diffusion distribution prediction system 700 for fractured vertebral bodies, including a processor 701, a memory 702, and a program or instruction stored in the memory 702 and executable on the processor 701. When the program or instruction is executed by the processor 701, it implements the various processes of the above-described bone cement diffusion distribution prediction method embodiment for fractured vertebral bodies and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0137] It should be noted that the bone cement diffusion distribution prediction system for fractured vertebrae in this application embodiment includes the bone cement diffusion distribution prediction system for mobile fractured vertebrae and the bone cement diffusion distribution prediction system for non-mobile fractured vertebrae as described above.

[0138] Example 4 This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described method for predicting the diffusion distribution of bone cement in fractured vertebrae and achieve the same technical effect. To avoid repetition, these will not be described again here.

[0139] The processor is the processor in the bone cement diffusion distribution prediction system for fractured vertebrae described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0140] Example 5 This application also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-described method for predicting the diffusion distribution of bone cement in fractured vertebrae, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0141] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0142] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element. Furthermore, it should be noted that the scope of the methods and systems in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0143] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0144] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above, which are merely illustrative and not restrictive. Those skilled in the art, under the guidance of this application, can make many modifications without departing from the spirit and scope of the claims, all of which fall within the protection scope of this application.

[0145] The above description is merely a preferred embodiment and the technical principles employed in this application. This application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions that can be made by those skilled in the art will not depart from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of this application, the scope of which is determined by the scope of the claims.

Claims

1. A method for predicting the diffusion distribution of bone cement in a fractured vertebral body, characterized in that, The method includes: S1. Obtain three-dimensional images of the fractured vertebrae of the patient and segment the fractured vertebral bodies; S2. Calculate the permeability coefficient of each voxel in the fractured vertebral body to construct a flow and diffusion model of bone cement in the fractured vertebral body; S3. Calculate the cumulative resistance of each voxel in the fractured vertebral body to the bone cement based on the flow diffusion model to obtain the cost function; S4. Based on the cost function, the cost-first traversal algorithm is used to calculate all voxels in the fractured vertebra that can be diffused by bone cement, thereby obtaining the predicted diffusion distribution of bone cement.

2. The method for predicting the diffusion distribution of bone cement in fractured vertebral bodies according to claim 1, characterized in that, S4 include: S41. Configure a cost priority queue for storing voxels in ascending order according to the first cost function value, wherein, in the initial state, the cost priority queue contains only injection point voxels. S42. Take the first voxel of the cost priority queue as a specific voxel, traverse all its adjacent voxels, and determine whether its adjacent voxels can be filled. When there is a fillable voxel among its neighboring voxels, calculate the first cost function value of entering the fillable voxel from the injection point voxel and the second cost function value of entering the fillable voxel through a specific voxel. When the second cost function value is not greater than the first cost function value, replace the first cost function value with the second cost function value, and put the corresponding fillable voxel into the cost priority queue. S43. Repeat S42 until the cost priority queue is empty, obtain all the voxels that can be filled and their corresponding cost function values, sort them in ascending order according to the cost function values, and obtain the ascending sort result. S44. Obtain the amount of bone cement injected, and perform voxel filling according to the ascending sorting results to obtain all voxels in the fractured vertebral body that can be diffused by bone cement, thereby obtaining the predicted results of bone cement diffusion distribution.

3. The method for predicting the diffusion distribution of bone cement in fractured vertebrae according to claim 2, characterized in that, In S42, the determination of whether adjacent voxels of a specific voxel can be filled is as follows: The flow dispersion model is used to obtain a metric value indicating whether the adjacent voxels of a specific voxel can be filled. If the metric value is not less than a set voxel filling threshold, it is determined that the adjacent voxels of the specific voxel can be filled.

4. The method for predicting the diffusion distribution of bone cement in fractured vertebrae according to claim 2, characterized in that, In S42, if the adjacent voxels of a specific voxel cannot be filled or the second cost function value of the adjacent voxels that can be filled is greater than the first cost function value, the next adjacent voxel is sequentially judged to see if it can be filled, until all the adjacent voxels of the specific voxel have been traversed. If all adjacent voxels of a particular voxel cannot be filled, or the second cost function value of all adjacent voxels that can be filled is greater than the first cost function value, then there is no corresponding fillable voxel that can be put into the cost priority queue for that particular voxel, and the traversal for that particular voxel ends.

5. The method for predicting the diffusion distribution of bone cement in fractured vertebral bodies according to claim 2, characterized in that, In S43, S42 is repeated. When none of the voxels in the cost priority queue have adjacent voxels that can be filled, or when the second cost function values ​​of all adjacent voxels that can be filled are greater than the first cost function value, all voxels in the cost priority queue are taken and set to empty.

6. The method for predicting the diffusion distribution of bone cement in fractured vertebral bodies according to claim 2, characterized in that, S44 includes: Based on the amount of bone cement injected, the voxels in S43 are filled sequentially in ascending order until they are exhausted, and each voxel filled is marked to obtain all the voxels in the fractured vertebral body that can be diffused by bone cement, thus obtaining the diffusion distribution of bone cement.

7. The method for predicting the diffusion distribution of bone cement in fractured vertebral bodies according to claim 1, characterized in that, S2 include: Based on the measurement value of each voxel in the fractured vertebral body, the void density is calculated, and then the permeability coefficient of each voxel is obtained. The flow and diffusion model of the constructed bone cement in each voxel of the fractured vertebral body is specifically calculated using the following formula: ； Where S is a measure of whether the voxel can be filled, k is the permeability coefficient of the voxel, μ is the viscosity of the bone cement, and P is the injection pressure. P is the pressure gradient.

8. The method for predicting the diffusion distribution of bone cement in fractured vertebrae according to claim 1 or 7, characterized in that, The calculation methods for the permeability coefficient of each voxel include: ； Where k is the permeability coefficient of the voxel, E is the geometric structure coefficient, which is determined based on the skeletal structure, and Φ is the porosity of the voxel, which is calculated based on the energy value of the voxel.

9. The method for predicting the diffusion distribution of bone cement in fractured vertebral bodies according to claim 1, characterized in that, In S1, the target vertebral segment in the three-dimensional image containing the fractured vertebral segment of the patient is first segmented using the vertebral segmentation model, and then the fractured vertebral body in the fractured vertebral segment of the patient is segmented using the vertebral body segmentation model. The vertebral segmentation model was pre-trained using a 3D-Unet network, and the vertebral body segmentation model was pre-trained using a SwinUnetR network.

10. A system for predicting the diffusion distribution of bone cement in fractured vertebrae, characterized in that, The method includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the method for predicting the diffusion distribution of bone cement in fractured vertebrae as described in any one of claims 1-9.

Images