Calcaneal fracture reduction effect dynamic evaluation system based on three-dimensional image

Abstract

The invention relates to the technical field of medical image processing, in particular to a calcaneal fracture reduction effect dynamic evaluation system based on a three-dimensional image. According to the system, the broken bone shape features of each bone block are obtained based on Gaussian curvature information of the surfaces of the bone blocks, broken bones corresponding to broken main bones are determined through a classification and direction correlation analysis method, broken bone indexes are quantified, and a bone area is obtained. Dividing the bone region into a to-be-registered region and a damaged region by utilizing the damage characteristic value, determining a stress influence direction of the to-be-registered region by combining information in the damaged region, and obtaining extrusion characteristics and offset characteristics of the bone block in the stress influence direction, and screening out an initial registration point of the bone block in combination with the damage characteristic value, and performing registration and reduction effect evaluation. According to the method, the bone blocks in the three-dimensional model are effectively identified and classified, the accurate initial registration points are screened out for registration by analyzing the fracture influence condition, and the registration effect and the reduction evaluation accuracy are ensured.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, specifically to a dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional images. Background Technology

[0002] Calcaneal fractures are a common type of foot injury in clinical practice, and the quality of surgical reduction directly affects the patient's postoperative functional recovery and quality of life. The calcaneus is an important weight-bearing bone in the foot, with a complex anatomical structure, making it prone to comminuted fractures under high-energy injuries. Calcaneal fractures account for approximately 60% of foot fractures, and in severe cases, can lead to loss of foot function, significantly impacting the patient's daily life and work ability.

[0003] Current technologies for fracture reduction assessment primarily rely on qualitative observation or angle measurement using medical imaging. However, in the dynamic evaluation of reduction effectiveness using 3D images, patient position, scanning direction, and bone tissue condition may differ at different time points in CT scans. Therefore, it is necessary to register the 3D models formed at different time points, unifying all 3D models into a single reference coordinate system to achieve quantitative comparisons of angle changes, bone fragment displacement, and callus formation. Current technologies mainly use the ICP algorithm for model registration. The ICP algorithm uses manual selection of key feature points in the ankle and foot for initial coarse registration to avoid getting trapped in local optima during subsequent iterations. However, in actual registration, different fracture types lead to different skeletal anatomy, making it difficult to determine accurate registration points, resulting in poor registration results and affecting the accuracy of dynamic assessment of reduction effectiveness. Summary of the Invention

[0004] To address the technical problem in existing technologies where inaccurate registration points during the registration of 3D models before and after fracture surgery lead to poor registration results and affect the evaluation of reduction effectiveness, the present invention aims to provide a dynamic evaluation system for calcaneal fracture reduction effectiveness based on 3D images. The specific technical solution adopted is as follows: This invention proposes a dynamic evaluation system for calcaneal fracture reduction based on three-dimensional imaging, the system comprising: The preoperative ankle and foot 3D model construction module is used to obtain a preoperative ankle and foot 3D model containing multiple bone blocks; The skeletal region segmentation module is used to obtain the fragment shape features of each bone block based on the data distribution prominence of the Gaussian curvature of the upper surface of the bone block; to obtain the fractured main bone and fragments based on the fragment shape features; to traverse the fractured main bone as the center in a preset traversal direction, and to obtain the fracture trace features of the fragments in each traversal direction based on the rate of change of the normal direction of the fractured main bone surface in the traversal direction, and to obtain the fragmentation index of each fragment by combining the fragment shape features; and to reclassify the fragments and fractured main bone based on the fragmentation index to obtain multiple skeletal regions. The initial registration point acquisition module is used to obtain damage feature values ​​based on the bone fragmentation index and the number of bone fragments in the skeletal region; classify the skeletal region into a region to be registered and a damaged region using the damage feature values; obtain the force influence direction of the region to be registered based on the traversal direction distribution of bone fragments in the damaged region; obtain compression features and offset features based on the morphological distribution of bone fragments in the region to be registered based on the force influence direction; and obtain the initial registration coefficient for each bone fragment in the region to be registered by combining the damage feature values; and select initial registration points using the initial registration coefficients. The repositioning effect evaluation module is used to register the preoperative 3D model of the ankle and foot with the postoperative model based on the initial registration point, and evaluate the repositioning effect based on the registered model.

[0005] Furthermore, the method for obtaining the shape features of the bone fragments includes: The Gaussian curvature of the surface voxel points is arranged in order of magnitude to obtain a Gaussian curvature sequence; the surface voxel point corresponding to the largest element in the difference sequence of the Gaussian curvature sequence is the surface mutation point; If the surface abrupt change point is a maximum value in the Gaussian curvature sequence, the mean of the Gaussian curvature difference between the surface abrupt change point and the adjacent surface voxel point is normalized to obtain the sharpness. The Gaussian curvature of the surface abrupt change point is increased according to the sharpness to obtain the sharpness weight; otherwise, the Gaussian curvature is used as the sharpness weight of the surface abrupt change point. Obtain the surface mutation vector pointing from the centroid of the bone block to each surface mutation point, and obtain the cosine similarity between each surface mutation vector and the average surface mutation vector; The ratio of sharpness weight to cosine similarity is used as the first significant value of bone fragments at the surface mutation point; the ratio of the mean Gaussian curvature of the surface voxel points to the number of surface voxel points is used as the second significant value of bone fragments; the bone fragment shape characteristics of the bone block are obtained based on the mean of the second and first significant values ​​of bone fragments.

[0006] Furthermore, obtaining the fractured main bone and bone fragments based on the shape characteristics of the bone fragments includes: Using the shape features of bone fragments, the bone blocks were clustered based on the K-means clustering algorithm to obtain three types of bone block clusters. The bone block cluster with the largest average value of bone fragment shape features is the set of bone fragments, the second largest is the set of fractured main bones, and the smallest is the set of normal bone blocks.

[0007] Furthermore, the method for obtaining the fracture trace features includes: Obtain the minimum bounding cuboid of the fractured main bone. Using each surface of the minimum bounding cuboid as a unit, traverse from the center point of the fractured main bone to the six surfaces; stop traversing when the traversed space contains normal bone blocks. The connected domain formed by the surface mutation points of the fractured main bone is called the surface mutation connected domain; the variance of the rate of change of the normal direction of all voxel points on the surface mutation connected domain is obtained; If the surface abrupt connected domain has the largest projected area and the closest distance on the surface of the smallest enclosing cuboid, then the surface abrupt connected domain is taken as the reference connected domain of the cuboid surface; the mean of the variance of the normal direction change rate of all reference connected domains is normalized by the product of the number of reference connected domains to obtain the fracture trace characteristics of each cuboid surface. If the bone fragments belong to multiple traversal spaces, then the fracture trace characteristics of the bone fragments are the fracture trace characteristics of the cuboid surface corresponding to the smallest traversal space; otherwise, the fracture trace characteristics of the bone fragments are the fracture trace characteristics of the cuboid surface corresponding to the traversal space.

[0008] Furthermore, the method for obtaining the bone fragmentation index includes: For each bone fragment, the fracture trace features are normalized to obtain an amplification ratio. Based on the amplification ratio, the shape features of the bone fragment are amplified to obtain the bone fragment index.

[0009] Furthermore, the method for dividing the skeletal region includes: For each fractured main bone, the fragment shape characteristics are enlarged according to the maximum enlargement ratio to obtain the fragmentation index of the fractured main bone. Based on the centroid coordinates of the bone fragments and the fragmentation index, the K-means clustering algorithm was used to cluster the fractured main bone and fragments to obtain multiple skeletal regions.

[0010] Furthermore, the classification of the skeletal region into a region to be registered and a damaged region using damage feature values ​​includes: The product of the average bone fragmentation index and the number of bone fragments in each skeletal region is used as the damage feature value. Threshold segmentation is performed using the damage feature value to obtain two types of skeletal regions. The type with the smallest damage feature value is used as the region to be registered, and the other type is the damaged region.

[0011] Furthermore, the method for obtaining the direction of force influence includes: Select any region to be registered as the target region to be registered; If the target registration area and the damaged area are adjacent, the sum of the traversal directions of the traversal space corresponding to all bone fragments in the damaged area during the fracture trace feature assignment process is taken as the force influence direction of the target registration area. If the target registration area is not adjacent to the damaged area, the force influence direction of the adjacent registration area is taken as the reference direction, and the sum of the reference directions passing through the target registration area is selected as the force influence direction of the target registration area.

[0012] Furthermore, the method for obtaining the extrusion feature and the offset feature includes: Obtain the contact surface between adjacent bone blocks in the registration area; on the contact surface, perpendicular to the direction of force influence, traverse the surface voxel points on the contact surface column by column to obtain the average distance between the surface voxel points in each column and the voxel points of the adjacent bone blocks; obtain the average distance sequence according to the traversal order; normalize the absolute value of the sum of the elements of the difference sequence of the average distance sequence to obtain the extrusion feature; Obtain the first angle between the principal axis of each bone block and the three-dimensional XY plane; obtain the second angle between the line connecting the center point of the region to be registered and the center of the calcaneus and the XY plane; obtain the offset feature based on the difference between the first angle and the second angle.

[0013] Furthermore, the method for obtaining the initial registration point includes: The initial registration coefficients are thresholded to obtain two types of initial registration coefficients. The bone block corresponding to the largest type of initial registration coefficient is selected as the registration bone block, and the registration points on the registration bone block obtained by the ICP registration algorithm are used as the initial registration points.

[0014] The present invention has the following beneficial effects: After obtaining a 3D model of the ankle and foot, this invention considers the presence of various bone fragments in the preoperative 3D model due to fracture factors. Model registration requires using bone fragments less affected by the fracture as references. Therefore, this invention first obtains the fragment shape characteristics of each bone fragment based on the Gaussian curvature information of the bone fragment surface. Then, through classification and directional correlation analysis, it identifies the bone fragments corresponding to the fractured main bone and quantifies the fragment index to obtain the skeletal region. Furthermore, using damage feature values, the skeletal region can be divided into a region to be registered and a damaged region; that is, the region to be registered is a set of bone fragments that can be used as registration references. By combining information from the damaged region, the direction of force influence in the region to be registered can be determined, and the compression and displacement characteristics of the bone fragments can be obtained in the direction of force influence. Further, the initial registration coefficient is obtained by combining the damage feature values. In other words, by quantifying the initial registration coefficients based on the compression and displacement characteristics exhibited by the bone fragments in the region to be registered, the degree of impact of the fracture on the bone fragments can be effectively characterized. This allows for the selection of initial registration points for bone fragments for registration and evaluation of the reduction effect. This invention effectively identifies and classifies bone fragments in a three-dimensional model, and selects accurate initial registration points for registration by analyzing the impact of fractures, thus ensuring the registration effect and the accuracy of repositioning assessment. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a block diagram of a dynamic evaluation system for calcaneal fracture reduction based on three-dimensional imaging, provided as an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a dynamic evaluation system for calcaneal fracture reduction based on three-dimensional imaging proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details the specific scheme of a dynamic evaluation system for calcaneal fracture reduction based on three-dimensional imaging provided by the present invention.

[0020] Please see Figure 1 The diagram illustrates a block diagram of a dynamic evaluation system for calcaneal fracture reduction based on three-dimensional images, provided by an embodiment of the present invention. The system includes a preoperative three-dimensional model construction module 101 for the ankle and foot, a bone region division module 102, an initial registration point acquisition module 103, and a reduction effect evaluation module 104.

[0021] The preoperative ankle-foot 3D model construction module 101 is used to obtain a preoperative ankle-foot 3D model containing multiple bone blocks. The purpose of this embodiment is to find bone blocks less affected by fractures and obtain registration points in the preoperative ankle-foot 3D model. In the process of constructing the preoperative ankle-foot 3D model, it is first necessary to determine the basic information it contains, namely, the information of each bone block and the center point of the calcaneus. When constructing the ankle-foot 3D model, different bone blocks are usually treated as independent model blocks to help with bone block matching and defect identification. Here, only a brief description of the model construction method in this embodiment is provided: (1) High-resolution ankle and foot CT scan images were acquired at multiple time points before and after surgery (such as 1 month, 3 months and 6 months after surgery) using a multi-slice spiral CT scanner (such as GE Revolution 256-slice spiral CT scanner). Thin-slice scanning (slice thickness ≤ 0.625 mm) was set, and the scanning range covered the lower 1 / 3 of the tibia to the distal end of the foot. (2) Import DICOM format images into 3D Slicer software for preprocessing: (3) Denoising and resampling of the acquired CT images; (4) Using the Segment Editor module, each bone block is treated as an independent segment, and the skeletal part in the CT image is segmented from the cartilage and soft tissue; (5) Generate a 3D model of the ankle and foot (STL format) using the Marching Cubes algorithm; (6) Artificially mark all key feature points in the preoperative three-dimensional model of the ankle and foot (such as the apex of the calcaneal tuberosity, the center of the posterior articular surface, the center of the calcaneus, etc.).

[0022] Each voxel in the 3D model has 3D (X-axis, Y-axis, Z-axis) coordinate attributes, and the CT value of the point is inherited from the original 2D pixel.

[0023] In 3D models of the ankle and foot, bone fragments resulting from fractures may appear. These fragments are treated as independent bone regions during 3D reconstruction. Directly registering these fragments can lead to poor registration results. Therefore, the system proposed in this embodiment of the invention needs to analyze the properties of each bone fragment, assess the degree of impact from the fracture, and then select effective registration points. The bone region segmentation module 102 mainly analyzes the properties of the bone fragments in each fragment, integrates and divides the fragments into multiple bone regions, and reduces the interference of bone fragments in the fracture damage area on the selection of registration points through integrated analysis.

[0024] The skeletal region segmentation module 102 considers that bone fragments and fractured main bones have obvious corner points compared to normal bone fragments. When a fracture occurs, the bone breaks under stress, forming multiple cross-sections. At this time, bone fragments are mostly asymmetrical small pieces with sharp edges and distinct corners. Normal bone fragments are not affected by fractures and are smoother, usually with areas of greater curvature only on the surface of the bone endpoints. Both fractured main bones and bone fragments are affected by the varying sizes of fractures, producing more prominent corner points. Therefore, the shape characteristics of each bone fragment can be obtained by the prominence of the Gaussian curvature data distribution on the bone fragment surface. That is, the more prominent the Gaussian region distribution on the bone fragment surface, the more likely the corresponding bone fragment is to be a fractured bone fragment or a fractured main bone, and the larger the shape characteristics of the bone fragment.

[0025] The specific calculation method for Gaussian curvature is a well-known technique in the field and will not be elaborated here.

[0026] Therefore, the fractured main bone and bone fragments of bone blocks can be screened based on the shape characteristics of the bone fragments. It should be noted that the screening of fractured main bone and bone fragments of bone blocks in the embodiments of the present invention does not mean that there are only these two types of bone blocks in the three-dimensional model. In fact, it also includes normal bone blocks with small bone fragment shape characteristics or even no obvious bone fragment characteristics, which will not be elaborated here.

[0027] Bone fragments can be considered as small pieces formed by the fracture of the main bone. Therefore, it is necessary to match the bone fragments with the fractured main bone as a reference and analyze the fracture trace characteristics under the influence of the fracture. The bone region segmentation module uses the fractured main bone as the center and presets multiple traversal directions. That is, it traverses from the center point of the fractured main bone towards each traversal direction. The bone fragments traversed in the traversal space can be considered as bone fragments generated by the stress on the fractured main bone under this traversal direction. Therefore, further analysis of the normal change rate of the fractured main bone surface in the traversal direction is performed. The greater the fluctuation of the normal change rate of the fractured main bone surface in this traversal direction, the more irregular, rougher, and serrated the surface is. It is more likely that external force caused the bone fragments to be generated, forming an uneven and irregular bone block surface. Therefore, the fracture trace characteristics of bone fragments in each traversal direction can be obtained based on this. Further combining the bone fragment shape characteristics, that is, the larger the fracture trace characteristics and the larger the bone fragment shape characteristics, the greater the influence of the fracture, and the larger the bone fragment index. Based on the fragmentation index, bone fragments and fractured main bones can be reclassified to obtain multiple skeletal regions. Each skeletal region represents a group of bone fragments with a clear affiliation. For example, a skeletal region containing both fractured main bones and bone fragments indicates that the region is affected by the fracture, and the fracture has some impact, but not a significant one. If a skeletal region only contains bone fragments, it indicates a more obvious injury area, belonging to a region with a greater impact from the fracture, and should not be used as a registration reference during 3D model registration.

[0028] In 3D reconstruction and registration analysis, it is necessary to analyze the spatial relationships and structural features of the bones surrounding the calcaneus to achieve more accurate model segmentation, registration, and postoperative dynamic change assessment. The calcaneus forms joints with multiple bones, typically exhibiting stable relative positional relationships and morphological characteristics. To further optimize the selection of initial registration points, it is necessary to analyze the relative positional changes and degree of bone damage after fractures in different bone segments. In the 3D model of the ankle and foot, the calcaneus is the core bone of the hindfoot. Calcaneal fractures cause cascading spatial changes in surrounding structures. To ensure the accuracy of model registration, key feature points on unaffected or weakly affected bone segments should be selected as initial registration points. Within the same skeletal region, the degree of damage to bone segments is relatively similar. After selecting a bone region with a lower degree of damage, the key location points in the current skeletal region can be assessed by analyzing the degree and direction of damage in adjacent bone regions. Therefore, after region segmentation by the bone region segmentation module 102, the initial registration point module 103 can further analyze the damage characteristics in the skeletal region and select the initial registration points.

[0029] For a given skeletal region, the more bone fragments it contains and the higher the fragmentation index, the more likely it is to be a region of significant injury, exhibiting more pronounced injury characteristics. Therefore, after obtaining the injury characteristics, the initial registration module 103 classifies the skeletal region into a region to be registered and an injured region. Because the injured region is a significantly injured area clearly affected by fractures, it should not be used as a reference for registration. However, since the injured region is heavily affected by fractures, the direction of force influence on the region to be registered can be obtained based on the distribution of bone fragments within the injured region.

[0030] When the calcaneus is fractured, the calcaneal angle changes, and the previously well-matched articular surfaces become irregular or misaligned, leading to local stress concentration and displacement of the joint space. This displacement is a factor influencing the instantaneous force during fracture. Normally, the contact surfaces between bone fragments are relatively smooth. Therefore, the distribution of gaps between bones can be used as one of the parameters influencing the selection of initial registration points within the registration area. Additionally, the bone arrangement in the ankle and foot is also affected: taking a standing posture as an example, the calcaneus is located in the lowest posterior region of the ankle and foot. After a calcaneal fracture, other ankle and foot bones may experience a collapse-like displacement, meaning the bone arrangement angle concaves towards the calcaneus. This concavity is a factor influencing the gravitational structure after fracture. Normally, vertically supporting bones (such as the talus) and laterally supporting bones (such as the cuboid bone) maintain their corresponding support structures. Therefore, the orientation of the bone fragments can be used as another parameter for selecting registration point locations.

[0031] Therefore, analyzing the bone fragment morphology distribution under the direction of force in the registration region allows us to determine the compression and displacement characteristics of the bone fragments after being subjected to fracture forces. A larger compression characteristic indicates a more pronounced fracture impact, with the bone fragment exhibiting a significant downward displacement under the force. A larger displacement characteristic indicates a greater deviation of the bone fragment from the calcaneal center after being subjected to external forces, signifying a greater impact from the fracture. Therefore, combining the damage characteristic values ​​in the registration region yields the initial registration coefficients for each bone fragment within that region. Based on these initial registration coefficients, bone fragments with higher reference value can be selected from the registration region, and the registration points in these fragments can then be used as the initial registration points for initial registration.

[0032] The reduction effect evaluation module 104 can register the preoperative three-dimensional model of the ankle and foot with the postoperative ankle and foot based on the initial registration point, and then evaluate the reduction effect based on the registered model.

[0033] It should be noted that the registration process in this embodiment of the invention is implemented using the ICP registration algorithm. The purpose of the initial coarse registration is to initially align the two point clouds, reduce spatial differences, and avoid subsequent iterations from getting trapped in local optima. Registration is achieved through steps such as nearest point matching and fine registration iteration, which are technical means well known to those skilled in the art and will not be elaborated here. After obtaining the registered model, the degree of recovery is evaluated by extracting changes in anatomical angles (such as the calcaneal tuberosity angle, the angle formed by the posterior articular surface of the calcaneus and the line connecting the calcaneal groove to the anterior process) and bone block alignment (such as the Dice coefficient) of the preoperative and postoperative models. In this embodiment of the invention, only the evaluation process in one embodiment is briefly described; the specific evaluation process is a technical means well known to those skilled in the art and will not be elaborated here.

[0034] The specific evaluation process is as follows: 1. The difference between the anatomical angle and the healthy angle; a) The Gissane angle (the angle formed by the posterior articular surface of the calcaneus and the line connecting the calcaneal groove to the anterior process) was (123.3±8.5)° in the male group and (122.7±8.9)° in the female group. b) The Böhler (calcaneal tuberosity angle) was (35.98±4.34)° in the male group and (38.31±4.43)° in the female group. The smaller the difference between the postoperative anatomical angle and the healthy angle compared to the difference between the preoperative anatomical angle and the healthy angle, the better the repositioning effect.

[0035] 2. Assessment of bone block alignment: The overlap rate between the postoperative bone block and the preoperative model: Dice coefficient and Hausdorff distance: Calculate the overlap and maximum surface distance of the registered model; the larger the Dice coefficient and Hausdorff distance between the postoperative model and the preoperative model, the better the reduction effect.

[0036] It should be noted that, in this embodiment of the invention, the patient's calcaneus can also undergo regular CT scans at 1 month, 3 months, and 6 months post-surgery to repeat the modeling and registration process, obtaining three-dimensional models at consecutive time points to observe the dynamic trend of bone healing, bone block stability, and remodeling. Other evaluation cycles can also be set in other embodiments of the invention, which are not limited or elaborated here.

[0037] In summary, after obtaining a three-dimensional model of the ankle and foot, this invention obtains the fragment shape characteristics of each bone fragment based on the Gaussian curvature information of the bone fragment surface. Then, through classification and directional correlation analysis, it identifies the fragments corresponding to the fractured main bone and quantifies the fragment index to obtain the skeletal region. Using damage feature values, the skeletal region can be divided into a registration area and a damaged area. Combining information from the damaged area, the direction of force influence on the registration area can be determined, and the compression and displacement characteristics of the bone fragments can be obtained in the direction of force influence. Furthermore, the initial registration coefficient is obtained by combining the damage feature values. Initial registration points for bone fragments are selected for registration, and the reduction effect is evaluated. This invention, through effective identification and classification of bone fragments in the three-dimensional model and analysis of the impact of fractures to select accurate initial registration points for registration, ensures the registration effect and the accuracy of the reduction evaluation.

[0038] Preferably, in one embodiment of the present invention, the method for obtaining the shape features of bone fragments includes: The Gaussian curvature of surface voxel points is arranged in ascending order to obtain a Gaussian curvature sequence. The surface voxel point corresponding to the largest element in the difference sequence of the Gaussian curvature sequence is the surface abrupt change point. That is, the surface abrupt change point can be considered as the corner point formed by the serrated texture of the bone block surface. Because the surface abrupt change point is the surface voxel point corresponding to the largest element in the difference sequence of the Gaussian curvature sequence, it indicates that this point has a significantly prominent Gaussian curvature compared to other surface voxel points. It should be noted that in this embodiment of the invention, the Gaussian curvature is arranged in ascending order, and the difference sequence is the value of each element minus the previous element in the Gaussian curvature sequence. That is, each element in the difference sequence corresponds to two surface voxel points. After filtering out the largest element in the difference sequence, the surface voxel point with the larger Gaussian curvature among the corresponding two surface voxel points is selected as the surface abrupt change point. Furthermore, there may be multiple largest elements in the difference sequence, meaning that there may be multiple surface abrupt change points.

[0039] If a surface abrupt change point is a maximum value in the Gaussian curvature sequence, it indicates that the abrupt change point is a significant bone corner point. The sharpness at this location can be assessed by the difference in Gaussian curvature between the abrupt change point and adjacent surface voxel points. Therefore, the absolute value of the difference in Gaussian curvature between the surface abrupt change point and each adjacent surface voxel point is first calculated. The mean of all the calculated absolute values ​​of difference is then normalized to obtain the sharpness. The Gaussian curvature of the surface abrupt change point is increased according to the sharpness to obtain the sharpness weight; otherwise, if the surface abrupt change point is not a maximum value in the Gaussian curvature sequence, the Gaussian curvature is used as the sharpness weight for the surface abrupt change point.

[0040] It should be noted that the normalization algorithm in this embodiment of the invention can choose range standardization. By statistically analyzing the range of sharpness values ​​before normalization, sharpness is obtained by normalizing based on the maximum and minimum values. Since the range of sharpness values ​​after normalization is between 0 and 1, the positive integer 1 can be added to the sharpness to obtain the amplification factor. That is, the range of the amplification factor is between 1 and 2. The sharpness weight can be obtained by directly multiplying the amplification factor by the Gaussian curvature.

[0041] It should be noted that, since the Gaussian curvature sequence is arranged in ascending order, the maximum value is the maximum value in the sequence.

[0042] Obtain the surface mutation vector pointing from the centroid of the bone block to each surface mutation point, and then obtain the cosine similarity between each surface mutation vector and the average surface mutation vector. The smaller the cosine similarity, the more the surface mutation vector of the corresponding surface mutation point deviates from the overall vector.

[0043] Because a higher sharpness weight and lower cosine similarity indicate that the surface abrupt change point is more likely to be a prominent, irregular, or jagged corner point in the bone fragment, the ratio of sharpness weight to cosine similarity is used as the first significant value of the surface abrupt change point. Furthermore, a higher Gaussian curvature and a smaller number of surface voxel points indicate a smaller and more irregular bone fragment, suggesting it is more likely to be a fracture fragment. Therefore, the ratio of the mean Gaussian curvature of the surface voxel points to the number of surface voxel points is used as the second significant value of the bone fragment. The shape characteristics of the bone fragments are obtained based on the mean of the second and first significant values.

[0044] In this embodiment of the invention, after calculating the mean value of the first bone fragment significance value, multiplying it by the second bone fragment significance value yields the bone fragment shape characteristics.

[0045] Preferably, in this embodiment of the invention, obtaining the fractured main bone and bone fragments based on the shape characteristics of the bone fragments includes: Using the shape features of bone fragments, the bone fragments were clustered based on the K-means clustering algorithm, resulting in three clusters. The cluster with the largest average value of the bone fragment shape features represents the set of bone fragments, the second largest represents the set of fractured main bones, and the smallest represents the set of normal bone fragments. Specifically, K was set to 3 for clustering to obtain the three clusters. The specific clustering algorithm is a conventional technique for those skilled in the art and will not be elaborated upon here.

[0046] Preferably, in this embodiment of the invention, the method for obtaining fracture trace features includes: The smallest bounding cuboid of the fractured main bone is obtained. Using each surface of this cuboid as a unit, the traversal proceeds from the center point of the fractured main bone towards all six surfaces. The traversal stops when a normal bone fragment is found within the traversed space. In other words, this embodiment of the invention sets up traversal in six directions. During the traversal, in three-dimensional space, each surface of the cuboid is used as a scanning plane, and the traversal is performed step-by-step with a step size of 1.

[0047] The connected regions formed by the surface abrupt change points of the fractured main bone are called surface abrupt change connected regions. It should be noted that the connected regions obtained by directly performing connected region detection on the surface abrupt change points may not be complete. Therefore, in this embodiment of the invention, morphological closing operations are further used to fill the pores inside the initially obtained connected regions to obtain surface abrupt change connected regions.

[0048] The variance of the rate of change of the normal direction of all voxel points on the surface abrupt connected domain is obtained. In this embodiment of the invention, for a voxel point, its normal is obtained, and the average of the angles between the found voxel point and the normals of its adjacent voxel points is taken as the rate of change of the normal. The larger the variance of the rate of change, the more non-uniform the current surface connected domain is, and the larger the fracture trace on the fracture surface that serves as the main fracture bone is.

[0049] If a surface abrupt change in a connected domain results in the largest projected area and the shortest distance on the surface that least encloses the cuboid, then this surface abrupt change is considered a reference connected domain for the cuboid's surface. In other words, these reference connected domains represent the fracture surface of the cuboid in that direction. Therefore, the mean of the variance of the rate of change of the normal direction for all reference connected domains corresponding to the cuboid surface is calculated, and the product of this variance and the number of reference connected domains is normalized to obtain the fracture trace characteristics for each cuboid surface. That is, the larger the variance of the rate of change of the normal direction and the larger the number of reference connected domains, the more significant the fracture trace in that direction on the cuboid surface, and thus the stronger the fracture trace characteristics.

[0050] If the bone fragment belongs to multiple traversal spaces, then the fracture trace characteristics of the bone fragment are the fracture trace characteristics of the cuboid surface corresponding to the smallest traversal space. This is because the smaller the traversal space, the closer the bone fragment is to the fractured main bone, and the more likely it belongs to the bone fragment generated in that direction. Otherwise, if the bone fragment belongs to only one traversal space, then the fracture trace characteristics of the bone fragment are the fracture trace characteristics of the cuboid surface corresponding to that traversal space.

[0051] Preferably, in this embodiment of the invention, for bone fragments, the greater the sharpness of the fracture and the larger the shape of the bone fragment, the more likely the bone fragment is a small bone fragment caused by a fracture. Therefore, for each bone fragment, the fracture trace characteristics are normalized to obtain an amplification ratio. Based on the amplification ratio, the shape characteristics of the bone fragment are amplified to obtain the bone fragment index. It should be noted that the method for increasing the amplification ratio here is the same as the method for increasing the amplification ratio of the sharpness weight described above, and will not be repeated here.

[0052] Preferably, in this embodiment of the invention, the method for dividing the skeletal region includes: For bone fragments with a high fragmentation index, since they are characteristic fragments of the fractured main bone, meaning they are both significantly affected by the fracture, they should be grouped into the same skeletal region as the fractured main bone. Bone fragments with a low fragmentation index should be grouped into the same skeletal region. This reduces interference from fracture-affected fragments on the registration points of intact bone fragments.

[0053] Therefore, for each fractured main bone, the shape features of the bone fragments are enlarged according to the maximum enlargement ratio to obtain the fragmentation index of the fractured main bone. The specific enlargement method will not be elaborated further. After enlargement, the fractured main bone has a larger fragmentation index, so the fractured main bone and bone fragments can be directly clustered using the K-means clustering algorithm based on the centroid coordinates of the bone fragments and the fragmentation index to obtain multiple skeletal regions. The K value of the clustering algorithm can be obtained using the elbow method, ultimately resulting in multiple skeletal regions. The specific clustering algorithm will not be elaborated further.

[0054] Preferably, in this embodiment of the invention, classifying the bone region into a region to be registered and a damaged region using damage feature values ​​includes: The product of the average fragmentation index and the number of bone fragments in each skeletal region is used as the damage feature value. That is, the higher the fragmentation index and the more bone fragments in a skeletal region, the greater the fracture impact and the more bone fragments produced, resulting in a higher damage feature value. Thresholding is performed using the damage feature value to obtain two types of skeletal regions. The region with the lowest damage feature value is designated as the region to be registered, and the other is the damaged region. It should be noted that thresholding can use Otsu's thresholding method, or it can be done by sorting by size and then selecting the point with the greatest difference between adjacent elements as the segmentation point to obtain the two types of regions. Further details are omitted.

[0055] Preferably, in this embodiment of the invention, the method for obtaining the direction of force influence includes: Select any region to be registered as the target region to be registered; If the target registration region and the damaged region are adjacent, then because they are adjacent, the damaged region is a region with significant fracture influence, and therefore the traversal direction in the damaged region can be used as a reference. The sum of the traversal directions of the traversal space corresponding to all bone fragments in the damaged region during the fracture trace feature assignment process is taken as the force influence direction of the target registration region. If the target registration area is not adjacent to the damaged area, the adjacent registration area whose force influence direction has been determined can be used as a reference. The force influence direction of the adjacent registration area can be used as the reference direction. The sum of the reference directions passing through the target registration area can be selected as the force influence direction of the target registration area.

[0056] Preferably, in this embodiment of the invention, the method for obtaining the extrusion feature and the offset feature includes: The contact surfaces between adjacent bone blocks in the region to be registered are obtained. In this embodiment of the invention, the contact surfaces between adjacent bone blocks can be directly determined using nearest neighbor and normal consistency, that is, the closest surface and the surface with the most consistent normal direction in three-dimensional space are selected as the contact surfaces.

[0057] On the contact surface, perpendicular to the direction of force, traverse the surface voxel points column by column to obtain the average distance between each voxel point and its adjacent bone block. Obtain the average distance sequence according to the traversal order; normalize the absolute value of the sum of the elements of the difference sequence of the average distance sequence to obtain the compression characteristic. That is, if all elements of the difference sequence have the same sign, i.e., all are positive or negative values, it indicates that voxel points at different positions have a consistent distance change trend as traversed, such as all tending to move closer to the adjacent bone block or all tending to move away from the adjacent bone block. Therefore, the larger the absolute value of the sum of the elements, the greater the compression characteristic under the influence of external force. The hyperbolic tangent function mapping method can be used here for normalization.

[0058] It should be noted that the method described above for obtaining the average distance between surface voxel points and adjacent bone blocks in each column is to obtain the distance between each surface voxel point in a column and its adjacent bone blocks, and then average the distances of all surface pixels in that column to obtain the average distance.

[0059] The first angle between the principal axis of each bone segment and the XY plane in three-dimensional space is obtained; the second angle between the line connecting the center point of the region to be registered and the center of the calcaneus and the XY plane is obtained. Since both angles are obtained with reference to the XY plane, the smaller the difference between the two angles, the closer the direction of the bone's principal axis is to the direction between the region and the center of the calcaneus, meaning the greater the probability of bone collapse and displacement, and thus the greater the displacement feature. Therefore, the displacement feature can be obtained based on the difference between the first and second angles. In this embodiment of the invention, the reciprocal of the difference between the two angles is used as the displacement feature.

[0060] It should be noted that, because the compression characteristics, offset characteristics, and damage characteristics of the corresponding region to be registered for a bone block are all negatively correlated with the registration reference degree, in this embodiment of the invention, the reciprocal of the product of the three is used as the initial registration coefficient. That is, the larger the initial registration coefficient, the more likely the registration point on the bone block should be used as the initial registration point.

[0061] Preferably, in this embodiment of the invention, the method for obtaining the initial registration point includes: Thresholding is applied to the initial registration coefficients to obtain two classes of initial registration coefficients. The bone block corresponding to the largest initial registration coefficient is selected as the registration bone block, and the registration points on the registration bone block obtained using the ICP registration algorithm are used as the initial registration points. It should be noted that the thresholding method used here is the same as that used for thresholding the skeletal region described above, and will not be repeated here.

[0062] It should be noted that, in order to ensure the accuracy of the initial registration, at least three initial registration points should be selected, that is, at least three registration bone blocks should be selected. If the number of selected registration bone blocks is less than three, bone blocks with larger initial registration coefficients can be selected based on the size of the initial registration coefficients until the quantity requirement is met.

[0063] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0064] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional imaging, characterized in that, The system includes: The preoperative ankle and foot 3D model construction module is used to obtain a preoperative ankle and foot 3D model containing multiple bone blocks; The skeletal region segmentation module is used to obtain the fragment shape features of each bone block based on the data distribution prominence of the Gaussian curvature of the upper surface of the bone block; to obtain the fractured main bone and fragments based on the fragment shape features; to traverse the fractured main bone as the center in a preset traversal direction, and to obtain the fracture trace features of the fragments in each traversal direction based on the rate of change of the normal direction of the fractured main bone surface in the traversal direction, and to obtain the fragmentation index of each fragment by combining the fragment shape features; and to reclassify the fragments and fractured main bone based on the fragmentation index to obtain multiple skeletal regions. The initial registration point acquisition module is used to obtain damage feature values ​​based on the bone fragmentation index and the number of bone fragments in the skeletal region; classify the skeletal region into a region to be registered and a damaged region using the damage feature values; obtain the force influence direction of the region to be registered based on the traversal direction distribution of bone fragments in the damaged region; obtain compression features and offset features based on the morphological distribution of bone fragments in the region to be registered based on the force influence direction; and obtain the initial registration coefficient for each bone fragment in the region to be registered by combining the damage feature values; and select initial registration points using the initial registration coefficients. The repositioning effect evaluation module is used to register the preoperative 3D model of the ankle and foot with the postoperative model based on the initial registration point, and evaluate the repositioning effect based on the registered model.

2. The dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional imaging according to claim 1, characterized in that, The method for obtaining the shape features of the bone fragments includes: The Gaussian curvature of the surface voxel points is arranged in order of magnitude to obtain a Gaussian curvature sequence; the surface voxel point corresponding to the largest element in the difference sequence of the Gaussian curvature sequence is the surface mutation point; If the surface abrupt change point is a maximum value in the Gaussian curvature sequence, the mean of the Gaussian curvature difference between the surface abrupt change point and the adjacent surface voxel point is normalized to obtain the sharpness. The Gaussian curvature of the surface abrupt change point is increased according to the sharpness to obtain the sharpness weight; otherwise, the Gaussian curvature is used as the sharpness weight of the surface abrupt change point. Obtain the surface mutation vector pointing from the centroid of the bone block to each surface mutation point, and obtain the cosine similarity between each surface mutation vector and the average surface mutation vector; The ratio of sharpness weight to cosine similarity is used as the first significant value of bone fragments at the surface mutation point; the ratio of the mean Gaussian curvature of the surface voxel points to the number of surface voxel points is used as the second significant value of bone fragments; the bone fragment shape characteristics of the bone block are obtained based on the mean of the second and first significant values ​​of bone fragments.

3. The dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional imaging according to claim 2, characterized in that, The process of obtaining the fractured main bone and bone fragments based on the shape characteristics of the bone fragments includes: Using the shape features of bone fragments, the bone blocks were clustered based on the K-means clustering algorithm to obtain three types of bone block clusters. The bone block cluster with the largest average value of bone fragment shape features is the set of bone fragments, the second largest is the set of fractured main bones, and the smallest is the set of normal bone blocks.

4. The dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional imaging according to claim 3, characterized in that, The method for obtaining the fracture trace features includes: Obtain the minimum bounding cuboid of the fractured main bone. Using each surface of the minimum bounding cuboid as a unit, traverse from the center point of the fractured main bone to the six surfaces; stop traversing when the traversed space contains normal bone blocks. The connected domain formed by the surface mutation points of the fractured main bone is called the surface mutation connected domain; the variance of the rate of change of the normal direction of all voxel points on the surface mutation connected domain is obtained; If the surface abrupt connected domain has the largest projected area and the closest distance on the surface of the smallest enclosing cuboid, then the surface abrupt connected domain is taken as the reference connected domain of the cuboid surface; the mean of the variance of the normal direction change rate of all reference connected domains is normalized by the product of the number of reference connected domains to obtain the fracture trace characteristics of each cuboid surface. If the bone fragments belong to multiple traversal spaces, then the fracture trace characteristics of the bone fragments are the fracture trace characteristics of the cuboid surface corresponding to the smallest traversal space; otherwise, the fracture trace characteristics of the bone fragments are the fracture trace characteristics of the cuboid surface corresponding to the traversal space.

5. The dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional imaging according to claim 1, characterized in that, The method for obtaining the bone fragmentation index includes: For each bone fragment, the fracture trace features are normalized to obtain an amplification ratio. Based on the amplification ratio, the shape features of the bone fragment are amplified to obtain the bone fragment index.

6. The dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional imaging according to claim 5, characterized in that, The method for dividing the skeletal region includes: For each fractured main bone, the fragment shape characteristics are enlarged according to the maximum enlargement ratio to obtain the fragmentation index of the fractured main bone. Based on the centroid coordinates of the bone fragments and the fragmentation index, the K-means clustering algorithm was used to cluster the fractured main bone and fragments to obtain multiple skeletal regions.

7. The dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional imaging according to claim 1, characterized in that, The method of classifying the skeletal region into regions to be registered and damaged regions using damage feature values ​​includes: The product of the average bone fragmentation index and the number of bone fragments in each skeletal region is used as the damage feature value. Threshold segmentation is performed using the damage feature value to obtain two types of skeletal regions. The type with the smallest damage feature value is used as the region to be registered, and the other type is the damaged region.

8. A dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional imaging according to claim 4, characterized in that, The method for obtaining the direction of force influence includes: Select any region to be registered as the target region to be registered; If the target registration area and the damaged area are adjacent, the sum of the traversal directions of the traversal space corresponding to all bone fragments in the damaged area during the fracture trace feature assignment process is taken as the force influence direction of the target registration area. If the target registration area is not adjacent to the damaged area, the force influence direction of the adjacent registration area is taken as the reference direction, and the sum of the reference directions passing through the target registration area is selected as the force influence direction of the target registration area.

9. A dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional imaging according to claim 1, characterized in that, The methods for obtaining the extrusion features and offset features include: Obtain the contact surface between adjacent bone blocks in the registration area; on the contact surface, perpendicular to the direction of force influence, traverse the surface voxel points on the contact surface column by column to obtain the average distance between the surface voxel points in each column and the voxel points of the adjacent bone blocks; obtain the average distance sequence according to the traversal order; normalize the absolute value of the sum of the elements of the difference sequence of the average distance sequence to obtain the extrusion feature; Obtain the first angle between the principal axis of each bone block and the three-dimensional XY plane; obtain the second angle between the line connecting the center point of the region to be registered and the center of the calcaneus and the XY plane; obtain the offset feature based on the difference between the first angle and the second angle.

10. A dynamic evaluation system for calcaneal fracture reduction effect based on three-dimensional imaging according to claim 1, characterized in that, The method for obtaining the initial registration point includes: The initial registration coefficients are thresholded to obtain two types of initial registration coefficients. The bone block corresponding to the largest type of initial registration coefficient is selected as the registration bone block, and the registration points on the registration bone block obtained by the ICP registration algorithm are used as the initial registration points.

Images