A method, device and equipment for three-dimensional accurate reconstruction of a spherical marker

Abstract

The application provides a spherical marker three-dimensional accurate reconstruction method, device and equipment, relates to the technical field of medical image processing, and comprises the following steps: obtaining a two-dimensional target detection result of each axial slice image of a CT three-dimensional model; wherein each two-dimensional detection result corresponds to a cross-section circle of a spherical marker detected on an axial slice image; generating a three-dimensional candidate point set of the CT three-dimensional model based on the two-dimensional target detection result of each axial slice image; adaptively clustering the three-dimensional candidate point set to obtain a plurality of three-dimensional clustering clusters; wherein each three-dimensional clustering cluster corresponds to a spherical marker; and for any three-dimensional clustering cluster, the center of a sphere is reconstructed to obtain the center coordinates and the physical radius of the corresponding spherical marker. The application can effectively improve the positioning accuracy and reconstruction stability of the spherical marker.

Description

Technical Field

[0001] This application relates to the field of medical image processing technology, and more specifically, to a method, apparatus, and equipment for precise three-dimensional reconstruction of spherical markers. Background Technology

[0002] In the field of spherical landmark detection and localization based on CT 3D models, traditional methods often directly perform independent target detection on a single axial slice image, lacking a reasonable mechanism to effectively aggregate the 2D detection results of multiple slices into 3D space. This easily leads to problems such as the same spherical landmark being split into multiple discrete detection results and serious interference from false targets. At the same time, existing clustering and reconstruction methods often use fixed parameter configurations, which are difficult to adapt to spherical landmarks of different sizes and distribution patterns. This results in low 3D clustering accuracy and poor stability of sphere center reconstruction, failing to meet the practical application requirements of high-precision intraoperative navigation and 3D localization. Summary of the Invention

[0003] The purpose of this application is to provide a method, apparatus, and device for accurate three-dimensional reconstruction of spherical markers, in order to solve the above-mentioned problems existing in the prior art, and to effectively improve the positioning accuracy and reconstruction stability of spherical markers.

[0004] Firstly, a method for accurate three-dimensional reconstruction of spherical markers is provided, which may include: Obtain the two-dimensional target detection results of each axial slice image of the CT three-dimensional model; where each two-dimensional detection result corresponds to the cross-sectional circle of the spherical marker detected on an axial slice image; Based on the two-dimensional target detection results of slice images along each axis, a three-dimensional candidate point set for the CT three-dimensional model is generated; Adaptive spatial clustering is performed on the three-dimensional candidate point set to obtain multiple three-dimensional clusters; wherein each three-dimensional cluster corresponds to a spherical marker; For any three-dimensional cluster, the center of the sphere is reconstructed to obtain the center coordinates and physical radius of the corresponding spherical marker.

[0005] In an optional implementation, the two-dimensional target detection result includes: The physical coordinates and physical radius of the center of the spherical marker in each axial slice image, as well as the layer information of the corresponding axial slice image, the quality assessment result of the two-dimensional target detection result, and the spherical confidence score; the layer information includes: the slice index and the physical z-coordinate of the slice in the axial slice image.

[0006] In an optional implementation, a set of 3D candidate points for the CT 3D model is generated based on the 2D target detection results of the slice images along each axis, including: The physical coordinates of the center of each two-dimensional target detection result are combined with the physical Z coordinates of the slice to construct the three-dimensional spatial points corresponding to each two-dimensional target detection result. The layer information, physical radius, quality assessment results, and spherical confidence score of each two-dimensional target detection result are used as the attribute information of the corresponding three-dimensional spatial points. Based on the three-dimensional spatial points and attribute information corresponding to each two-dimensional target detection result, a three-dimensional candidate point set is generated.

[0007] In an optional implementation, adaptive spatial clustering is performed on the set of 3D candidate points to obtain multiple 3D clusters, including: For any three-dimensional candidate point set, perform statistical analysis on the three-dimensional candidate point set and calculate the overall density of the three-dimensional candidate point set; For any three-dimensional spatial point in the set of three-dimensional candidate points, with the three-dimensional spatial point as the center and the configured initial radius as the neighborhood radius, the number of points in the neighborhood of the three-dimensional spatial point is counted and used as the local density of the three-dimensional spatial point. An adaptive neighborhood radius strategy is adopted to dynamically adjust the neighborhood radius of the three-dimensional spatial point based on the local density and the overall density of the three-dimensional spatial point, thereby generating the target neighborhood radius of the three-dimensional spatial point. Clustering is performed based on the target neighborhood radius and the minimum number of configured points to obtain multiple three-dimensional clusters.

[0008] In an optional implementation, after obtaining multiple three-dimensional clusters, the method further includes: Calculate the statistical characteristics of each three-dimensional cluster; Three-dimensional clusters whose statistical characteristics satisfy the configured filtering rules are identified as target three-dimensional clusters.

[0009] In an optional implementation, for any three-dimensional cluster, the sphere center is reconstructed to obtain the sphere center coordinates and physical radius of the spherical marker, including: For any target 3D cluster, calculate the centroid of the target 3D cluster, and calculate the distance from the centroid to all 3D spatial points within the target 3D cluster; Delete 3D spatial points whose distance to the centroid is greater than the configured distance threshold to obtain the target 3D spatial points of the target 3D cluster; Calculate the standard deviation of the spatial coordinates of all target 3D spatial points within the target 3D cluster; Based on the standard deviation of the spatial coordinates, a target fusion strategy is determined; Based on the target fusion strategy, all three-dimensional spatial points of the target within the target's three-dimensional cluster are fused to obtain the center coordinates and physical radius of the spherical marker.

[0010] Secondly, a device for precise three-dimensional reconstruction of spherical markers is provided, the device including: The acquisition unit is used to acquire the two-dimensional target detection results of each axial slice image of the CT three-dimensional model; wherein, each two-dimensional detection result corresponds to the cross-sectional circle of a spherical marker detected on an axial slice image; The generation unit is used to generate a set of three-dimensional candidate points for the CT three-dimensional model based on the two-dimensional target detection results of the slice images along each axis. Clustering units are used to perform adaptive spatial clustering on the set of three-dimensional candidate points to obtain multiple three-dimensional clusters; wherein each three-dimensional cluster corresponds to a spherical marker; The reconstruction unit is used to reconstruct the center of a sphere for any three-dimensional cluster, thereby obtaining the center coordinates and physical radius of the corresponding spherical marker.

[0011] Thirdly, an electronic device is provided, which includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements any of the steps described in the first aspect above.

[0012] Fourthly, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed by a processor, the computer program implements the steps of any of the methods described in the first aspect above.

[0013] This application constructs a three-dimensional candidate point set from the two-dimensional target detection results of multi-axial slices, and uses adaptive spatial clustering to effectively aggregate the detection points corresponding to the same spherical marker. This can significantly suppress the interference of discrete false targets and improve the accuracy and robustness of three-dimensional target clustering. By independently reconstructing the center of the sphere for each three-dimensional cluster, the coordinates and physical radius of the center of the spherical marker can be accurately obtained, realizing a reliable conversion from two-dimensional slice detection to three-dimensional spatial target positioning, and effectively improving the positioning accuracy and reconstruction stability of spherical markers. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1An architecture diagram of a spherical marker three-dimensional precision reconstruction system provided in this application embodiment; Figure 2 A flowchart illustrating a method for precise three-dimensional reconstruction of a spherical marker provided in this application embodiment; Figure 3 A schematic diagram of the structure of a spherical marker three-dimensional precision reconstruction device provided in this application embodiment; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0016] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Unless otherwise defined, the technical or scientific terms used in this application should have the ordinary meaning understood by those skilled in the art. The words "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are only used to distinguish different components. The words "comprising" or "including," etc., mean that the element or object preceding the word covers the element or object listed after the word and its equivalents, but do not exclude other elements or objects. The words "connected," "coupled," or "connected," etc., are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up," "down," "left," "right," etc., are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0017] The three-dimensional accurate reconstruction method for spherical markers provided in this application embodiment can be applied to... Figure 1 In the system architecture shown, such as Figure 1 As shown, the system may include: a server and a data acquisition component. The server may be a physical server, a server cluster composed of multiple physical servers, or a distributed system. It may also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0018] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0019] Figure 2 This is a flowchart illustrating a method for precise three-dimensional reconstruction of a spherical marker, as provided in an embodiment of this application. Figure 2 As shown, the method may include: Step S210: Obtain the two-dimensional target detection results of slice images of each axis of the CT three-dimensional model.

[0020] Each two-dimensional detection result corresponds to a spherical marker cross-section circle detected on an axial slice image; the CT three-dimensional model can include multiple axial slice images; the two-dimensional target detection results of each axial slice image can include: the physical coordinates and physical radius of the center of the spherical marker in the corresponding axial slice image, as well as the slice information of the corresponding axial slice image, the quality assessment result of the two-dimensional target detection result, and the spherical confidence score; the slice information can include: the slice index and physical z-coordinate of the slice of the axial slice image; the quality assessment result can include roundness (reflecting the degree of fit between the cross-section circle and the ideal circle), edge sharpness (reflecting the gradient strength of the circle boundary), gradient strength (reflecting the overall texture features of the circle region), and the number of edge points (reflecting the amount of information used in the detection); the spherical confidence score is a value between 0 and 1 obtained by normalizing the quality assessment result, used to represent the reliability of the detection result.

[0021] In practice, the physical coordinates of the circle center are obtained by transforming the pixel coordinates of the circle center detected by the target using a spatial affine matrix, and are used to represent the X and Y positions of the cross-sectional circle in physical space; the physical Z coordinates of the slice are obtained by calculating the slice index and layer thickness, and are used to represent the Z-axis height of the slice in physical space; the physical radius is obtained by transforming the pixel radius detected by the target and the pixel spacing of the configured CT 3D model, and is used to represent the physical size of the cross-sectional circle.

[0022] In another embodiment of this application, the method for obtaining the two-dimensional target detection results of slice images along each axis may include: Obtain the initial two-dimensional target detection results of slice images along each axis of the CT three-dimensional model; The initial two-dimensional target detection result with a spherical confidence score not greater than the configured spherical confidence threshold is taken as the first two-dimensional target detection result; wherein, the spherical confidence threshold can be 0.3-0.5; the spherical confidence threshold can be customized according to the user's actual application scenario; The detection result of the first two-dimensional target whose physical radius is within the configured physical radius threshold range is used as the detection result of the second two-dimensional target; wherein, the physical radius threshold range is determined according to the specifications of the intraoperative marker and can be from 1.5mm to 4mm; The second two-dimensional target detection results whose quality assessment results are not greater than the configured quality assessment threshold are used as the two-dimensional target detection results of each axial slice image; the values ​​of each index in the quality assessment results are compared with the corresponding quality assessment thresholds, and the second two-dimensional target detection results that are not greater than the corresponding quality assessment thresholds are filtered to obtain the final two-dimensional target detection results.

[0023] Step S220: Based on the two-dimensional target detection results of the slice images along each axis, generate a three-dimensional candidate point set for the CT three-dimensional model.

[0024] In practice, the physical coordinates of the center of each two-dimensional target detection result are combined with the physical Z-coordinates of the slice to construct the three-dimensional spatial points corresponding to each two-dimensional target detection result; the other data of each two-dimensional target detection result, except for the physical coordinates of the center of the circle and the physical Z-coordinates of the slice, are used as the attribute information of the corresponding three-dimensional spatial points; and a set of three-dimensional candidate points is generated based on the three-dimensional spatial points and attribute information corresponding to each two-dimensional target detection result.

[0025] Step S230: Perform adaptive spatial clustering on the three-dimensional candidate point set to obtain multiple three-dimensional clusters.

[0026] Each 3D cluster can contain at least one set of 3D candidate points; each 3D cluster corresponds to a spherical marker.

[0027] In practice, for any three-dimensional candidate point set, statistical analysis is performed on the three-dimensional candidate point set to calculate the overall density of the three-dimensional candidate point set. For any three-dimensional spatial point in the set of three-dimensional candidate points, take the three-dimensional spatial point as the center and the configured initial radius as the neighborhood radius, count the number of points in the neighborhood of the three-dimensional spatial point, and use it as the local density of the three-dimensional spatial point. An adaptive neighborhood radius strategy is adopted to dynamically adjust the neighborhood radius of the three-dimensional spatial point based on its local and global density, thereby generating the target neighborhood radius of the three-dimensional spatial point. DBSCAN clustering is performed based on the target neighborhood radius and the configured minimum number of points to obtain multiple 3D clusters. Specifically, for each 3D spatial point, the number of points in the neighborhood centered on that 3D spatial point and with the target neighborhood radius as the neighborhood radius is determined. If the number of points in the neighborhood is not less than the configured minimum number of points, then that 3D spatial point is determined as the target 3D spatial point. All 3D spatial points in the neighborhood of the target 3D spatial point are grouped into the same cluster to obtain multiple 3D clusters. Points in the neighborhood of any 3D spatial point are 3D spatial points whose weighted Euclidean distance from that 3D spatial point is not greater than the target neighborhood radius of that 3D spatial point. If the number of points in the neighborhood is not less than the configured minimum number of points, then that 3D spatial point is deleted.

[0028] In another embodiment of this application, after obtaining multiple three-dimensional clusters, the method may further include: The three-dimensional clusters in which the number of all three-dimensional spatial points within a cluster is not less than the minimum number of points configured are designated as the first three-dimensional clusters. For any first three-dimensional cluster, calculate the maximum and minimum values ​​of the z-coordinates of all three-dimensional spatial points within the first three-dimensional cluster to obtain the z-direction span; The first three-dimensional cluster with a span in the z-direction not greater than the configured span threshold is used as the second three-dimensional cluster; wherein, the span threshold is 1.5-2 times the diameter of the spherical marker target; For any second 3D cluster, calculate the slice index of all 3D spatial points within the second 3D cluster, sort them, and calculate the difference between any two adjacent slice indices; if the difference between any two adjacent slice indices is greater than the configured tortuosity threshold, delete the corresponding second 3D cluster; based on the undeleted second 3D clusters, obtain the third 3D cluster; For any third 3D cluster, calculate the standard deviation of the 3D coordinates of all 3D spatial points within the third 3D cluster, and take the third 3D cluster whose 3D coordinate standard deviation is not greater than the configured standard deviation threshold as the fourth 3D cluster. Calculate the mean and standard deviation of the physical radii of all three-dimensional spatial points within each fourth three-dimensional cluster to obtain the radius variation coefficient; the fourth three-dimensional clusters whose radius variation coefficient is not greater than the configured radius variation coefficient threshold are designated as the fifth three-dimensional clusters; Calculate the centroid of all three-dimensional spatial points within each fourth three-dimensional cluster, and perform principal component analysis on all three-dimensional spatial points within each fourth three-dimensional cluster to obtain three eigenvalues ​​and corresponding eigenvectors; calculate the slenderness ratio; and select the fifth three-dimensional cluster whose slenderness ratio is not greater than the configured slenderness ratio threshold as the target three-dimensional cluster.

[0029] Step S240: For any three-dimensional cluster, reconstruct the center of the sphere to obtain the center coordinates and physical radius of the spherical marker.

[0030] In practice, for any target 3D cluster, the centroid of the target 3D cluster is calculated, and the distance from all 3D spatial points within the target 3D cluster to the centroid is calculated; 3D spatial points whose distance to the centroid is greater than the configured distance threshold are deleted, and the target 3D spatial points of the target 3D cluster are obtained; wherein, the distance threshold is determined based on the standard deviation of the distances from all 3D spatial points within the target 3D cluster to the centroid; Calculate the standard deviation of the spatial coordinates of all three-dimensional points of the target within the target's three-dimensional cluster; Based on the standard deviation of spatial coordinates, a target fusion strategy is determined; if the standard deviation of spatial coordinates is less than the configured first standard deviation threshold, mean fusion is used; if the standard deviation of spatial coordinates is greater than the configured second standard deviation threshold, a robust weighted fusion strategy is used; if the standard deviation of spatial coordinates is not less than the first standard deviation threshold and not greater than the second standard deviation threshold, a confidence-weighted fusion strategy is used. Based on the target fusion strategy, all target 3D spatial points within the target 3D cluster are fused to obtain the center coordinates and physical radius of the spherical marker. Specifically, mean fusion is to take the arithmetic mean of all target 3D spatial points within the 3D cluster to obtain the center coordinates of the spherical marker; confidence-weighted fusion is to use the spherical confidence score in the attribute of each target 3D spatial point as a weight to take the weighted average of the coordinates of the corresponding target 3D spatial points to obtain the center coordinates of the spherical marker; robust weighted fusion is to use iterative reweighted least squares, which gradually reduces the influence of outliers through iteration, and can effectively suppress their interference even if outliers are not completely removed.

[0031] In another embodiment of this application, the method may further include: constructing constraints based on the geometric properties of the configured spherical marker to be identified; constructing an optimization objective function with constraints; and fusing all target three-dimensional spatial points within the three-dimensional cluster based on a target fusion strategy to obtain the center coordinates and physical radius of the spherical marker.

[0032] In one embodiment of this application, after obtaining the center coordinates and physical radius of each spherical marker, the method may further include: For any target 3D cluster, calculate the characteristic factors of the target 3D cluster. These characteristic factors may include: observation quantity factor, spatial consistency factor, radius consistency factor, two-dimensional confidence mean factor, slice continuity factor, and roundness mean factor. Specifically, the observation quantity factor is the ratio of the number of all target 3D spatial points within the target 3D cluster to the maximum number of points obtained by configuration, reflecting how many slices observed the spherical marker. The spatial consistency factor can be calculated by: calculating the Euclidean distance from all target 3D spatial points within the target 3D cluster to the center of the fusion sphere, and calculating the standard deviation of the Euclidean distance to obtain the distance standard deviation; using 1 / (1 + distance standard deviation) as the spatial consistency factor to reflect the discreteness of each target 3D spatial point within the target 3D cluster relative to the center of the fusion sphere. The calculation method for the radius consistency factor may include: calculating the absolute deviation between the radius and the fusion radius of all target 3D spatial points within the target 3D cluster, and calculating the radius variation coefficient; using 1 / (1+radius variation coefficient) as the radius consistency factor; the two-dimensional confidence mean factor is the arithmetic mean of the spherical confidence scores of all target 3D spatial points within the target 3D cluster; the calculation method for the slice continuity factor may include: extracting the slice index of all target 3D spatial points within the target 3D cluster, sorting it to obtain a sequence; calculating the coverage rate of the slice index, i.e., the slice continuity factor, based on the sequence; the coverage rate of the slice index is the proportion of the number of actually observed slices to the total span in the Z direction; the roundness mean factor is the arithmetic mean of the roundness values ​​of all target 3D spatial points within the target 3D cluster; Based on the characteristic factors of the target three-dimensional cluster, calculate the comprehensive confidence level of the target three-dimensional cluster; Calculate the Euclidean distance between the center coordinates of any two target 3D clusters, and use it as the center distance; classify two target 3D clusters whose center distance is less than the configured center distance threshold as two conflicting 3D clusters; where the center distance threshold is the sum of the physical radii of the two target 3D clusters and the configured safety distance coefficient; If the difference in the overall confidence of two conflicting 3D clusters is greater than the configured difference threshold, then the conflicting 3D cluster with the higher overall confidence is retained, and the conflicting 3D cluster with the lower overall confidence is deleted. If the difference in the overall confidence of two conflicting 3D clusters is not greater than the difference threshold, then the number of all target 3D spatial points within the two conflicting 3D clusters is used as the observation count for the two conflicting 3D clusters; the difference in the number of observations is calculated; if the difference in the number of observations is not less than the configured difference threshold, then the conflicting 3D cluster with more observations is retained; the conflicting 3D cluster with fewer observations is deleted; if the difference in the number of observations is less than the difference threshold, then the two conflicting 3D clusters are merged into a new target 3D cluster; return to step S240; if the confidence and number of observations of the conflicting 3D clusters are similar, but the spatial distance is indeed less than the sum of the radii, and the overall confidence after merging is significantly reduced, then it is determined that the two real targets are too close, and they remain independent. Perform secondary clustering on the target 3D spatial points within each target 3D cluster, using the configured minimum neighborhood radius for DBSCAN clustering; if the secondary clustering produces two or more sub-clusters, and the centroid distance between each sub-cluster is greater than the sum of the sub-cluster radii, then determine that the corresponding target 3D cluster contains multiple spherical markers; separate each sub-cluster into an independent cluster, and execute step S240 respectively.

[0033] In one embodiment of this application, after obtaining the center coordinates and physical radius of each spherical marker, the method may further include: Each CT 3D model corresponds to a specific point in time. Calculate the coordinate deviation between the center coordinates of the spherical marker corresponding to the CT 3D model at the current time point and the center coordinates of the spherical marker corresponding to the CT 3D model at the previous time point; Calculate the radius deviation between the physical radius of the spherical marker corresponding to the CT 3D model at the current time point and the physical radius of the spherical marker corresponding to the CT 3D model at the previous time point; If the coordinate deviation is less than the configured spatial distance threshold and the radius deviation is less than the configured radius change rate threshold, then the spherical marker corresponding to the CT 3D model at the current time point and the spherical marker corresponding to the CT 3D model at the previous time point are determined to be the same spherical marker. Statistical analysis of the center coordinates, physical radius, and duration of existence of the same spherical marker in CT 3D models at different time points; Remove spherical markers detected in the CT 3D model at only one time point; and adjust the overall confidence level according to the duration of their presence.

[0034] Corresponding to the above method, this application also provides a device for precise three-dimensional reconstruction of spherical markers, such as... Figure 3 As shown, the device includes: The acquisition unit 310 is used to acquire the two-dimensional target detection results of each axial slice image of the CT three-dimensional model; wherein, each two-dimensional detection result corresponds to the cross-sectional circle of a spherical marker detected on an axial slice image; The generation unit 320 is used to generate a set of three-dimensional candidate points for the CT three-dimensional model based on the two-dimensional target detection results of the slice images along each axis. Clustering unit 330 is used to perform adaptive spatial clustering on the three-dimensional candidate point set to obtain multiple three-dimensional clusters; wherein, each three-dimensional cluster corresponds to a spherical marker; The reconstruction unit 340 is used to reconstruct the center of a 3D cluster for any given 3D cluster, thereby obtaining the center coordinates and physical radius of the corresponding spherical marker.

[0035] The functions of each functional unit in the spherical marker three-dimensional accurate reconstruction device provided in the above embodiments of this application can be realized through the above methods and steps. Therefore, the specific working process and beneficial effects of each unit in the spherical marker three-dimensional accurate reconstruction device provided in the embodiments of this application will not be repeated here.

[0036] This application also provides an electronic device, such as... Figure 4 As shown, it includes a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440.

[0037] Memory 430 is used to store computer programs; When the processor 410 executes the program stored in the memory 430, it performs the following steps: Obtain the two-dimensional target detection results of each axial slice image of the CT three-dimensional model; where each two-dimensional detection result corresponds to the cross-sectional circle of the spherical marker detected on an axial slice image; Based on the two-dimensional target detection results of slice images along each axis, a three-dimensional candidate point set for the CT three-dimensional model is generated; Adaptive spatial clustering is performed on the three-dimensional candidate point set to obtain multiple three-dimensional clusters; each three-dimensional cluster corresponds to a spherical marker. For any three-dimensional cluster, the center of the cluster is reconstructed to obtain the center coordinates and physical radius of the corresponding spherical marker.

[0038] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0039] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0040] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0041] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0042] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments for solving the problem can be found in [reference needed]. Figure 2 The steps in the illustrated embodiments are used to implement the electronic device. Therefore, the specific working process and beneficial effects of the electronic device provided in this application will not be repeated here.

[0043] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform any of the three-dimensional accurate reconstruction methods for spherical markers described in the above embodiments.

[0044] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the three-dimensional accurate reconstruction methods for spherical markers described in the above embodiments.

[0045] Those skilled in the art will understand that the embodiments in this application can be provided as methods, systems, or computer program products. Therefore, the embodiments in this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments in this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0046] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0047] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0048] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0049] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of this application.

[0050] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of this application and its equivalents, then these modifications and variations are also intended to be included in the embodiments of this application.

Claims

1. A method for accurate three-dimensional reconstruction of spherical markers, characterized in that, The method includes: Obtain the two-dimensional target detection results of each axial slice image of the CT three-dimensional model; where each two-dimensional detection result corresponds to the cross-sectional circle of the spherical marker detected on an axial slice image; Based on the two-dimensional target detection results of slice images along each axis, a three-dimensional candidate point set for the CT three-dimensional model is generated; Adaptive spatial clustering is performed on the three-dimensional candidate point set to obtain multiple three-dimensional clusters; wherein each three-dimensional cluster corresponds to a spherical marker; For any three-dimensional cluster, the center of the sphere is reconstructed to obtain the center coordinates and physical radius of the corresponding spherical marker.

2. The method as described in claim 1, characterized in that, The two-dimensional target detection results include: The physical coordinates and physical radius of the center of the spherical marker in each axial slice image, as well as the layer information of the corresponding axial slice image, the quality assessment result of the two-dimensional target detection result, and the spherical confidence score; the layer information includes: the slice index and the physical z-coordinate of the slice in the axial slice image.

3. The method as described in claim 2, characterized in that, Based on the two-dimensional target detection results of the slice images along each axis, a three-dimensional candidate point set for the CT three-dimensional model is generated, including: The physical coordinates of the center of each two-dimensional target detection result are combined with the physical Z coordinates of the slice to construct the three-dimensional spatial points corresponding to each two-dimensional target detection result. The layer information, physical radius, quality assessment results, and spherical confidence score of each two-dimensional target detection result are used as the attribute information of the corresponding three-dimensional spatial points. Based on the three-dimensional spatial points and attribute information corresponding to each two-dimensional target detection result, a three-dimensional candidate point set is generated.

4. The method as described in claim 3, characterized in that, Adaptive spatial clustering is performed on the three-dimensional candidate point set to obtain multiple three-dimensional clusters, including: For any three-dimensional candidate point set, perform statistical analysis on the three-dimensional candidate point set and calculate the overall density of the three-dimensional candidate point set; For any three-dimensional spatial point in the set of three-dimensional candidate points, with the three-dimensional spatial point as the center and the configured initial radius as the neighborhood radius, the number of points in the neighborhood of the three-dimensional spatial point is counted and used as the local density of the three-dimensional spatial point. An adaptive neighborhood radius strategy is adopted to dynamically adjust the neighborhood radius of the three-dimensional spatial point based on the local density and the overall density of the three-dimensional spatial point, thereby generating the target neighborhood radius of the three-dimensional spatial point. Clustering is performed based on the target neighborhood radius and the minimum number of configured points to obtain multiple three-dimensional clusters.

5. The method as described in claim 4, characterized in that, After obtaining multiple three-dimensional clusters, the method further includes: Calculate the statistical characteristics of each three-dimensional cluster; Three-dimensional clusters whose statistical characteristics satisfy the configured filtering rules are identified as target three-dimensional clusters.

6. The method as described in claim 5, characterized in that, For any three-dimensional cluster, the sphere center is reconstructed to obtain the coordinates of the sphere center and the physical radius of the spherical marker, including: For any target 3D cluster, calculate the centroid of the target 3D cluster, and calculate the distance from the centroid to all 3D spatial points within the target 3D cluster; Delete 3D spatial points whose distance to the centroid is greater than the configured distance threshold to obtain the target 3D spatial points of the target 3D cluster; Calculate the standard deviation of the spatial coordinates of all target 3D spatial points within the target 3D cluster; Based on the standard deviation of the spatial coordinates, a target fusion strategy is determined; Based on the target fusion strategy, all three-dimensional spatial points of the target within the target's three-dimensional cluster are fused to obtain the center coordinates and physical radius of the spherical marker.

7. A device for precise three-dimensional reconstruction of a spherical marker, characterized in that, The device includes: The acquisition unit is used to acquire the two-dimensional target detection results of each axial slice image of the CT three-dimensional model; wherein, each two-dimensional detection result corresponds to the cross-sectional circle of a spherical marker detected on an axial slice image; The generation unit is used to generate a set of three-dimensional candidate points for the CT three-dimensional model based on the two-dimensional target detection results of the slice images along each axis. Clustering units are used to perform adaptive spatial clustering on the set of three-dimensional candidate points to obtain multiple three-dimensional clusters; wherein each three-dimensional cluster corresponds to a spherical marker; The reconstruction unit is used to reconstruct the center of a sphere for any three-dimensional cluster, thereby obtaining the center coordinates and physical radius of the corresponding spherical marker.

8. An electronic device, characterized in that, The electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method of any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-6.

Images