Three-dimensional reconstruction method and device for tubular part, computer equipment and storage medium

Abstract

The invention relates to a three-dimensional reconstruction method and device for a tubular part, computer equipment and a storage medium. The method comprises the following steps: acquiring original reconstruction information of a tubular part; the original reconstruction information comprises a center line position of the tubular part and an original section radius corresponding to the center line position; the original section radius smaller than the section radius threshold value is expanded to the section radius threshold value, and the expanded section radius is obtained; performing three-dimensional reconstruction on the tubular part according to the expanded section radius and the center line position to obtain an initial three-dimensional reconstruction result of the tubular part; and correcting the initial three-dimensional reconstruction result according to the original reconstruction information to obtain a target three-dimensional reconstruction result of the tubular part. By adopting the method, the accuracy of three-dimensional reconstruction of the tubular part can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of medical image processing, and in particular to a method, device, computer equipment and storage medium for three-dimensional reconstruction of a tubular part. Background Art

[0002] The blood vessels around liver tumors are usually complex in shape, with many branches and small terminal structures. Accurate reconstruction of liver blood vessels, especially small blood vessels, and application of the reconstruction results in clinical practice can not only improve the diagnostic efficiency of liver tumors and avoid missed or misdiagnosis, but also avoid damage to the small blood vessels around the tumor during surgery.

[0003] At present, in the process of three-dimensional reconstruction of liver blood vessels, the acquired radiological image data (computer tomography data or magnetic resonance imaging data, etc.) are usually segmented to obtain parameters for blood vessel reconstruction, and these parameters are reconstructed into three-dimensional surfaces using methods such as isosurface extraction and traditional convolution surface. After obtaining the three-dimensional surface model, surface rendering technology is used to present the three-dimensional reconstruction results of blood vessels based on the three-dimensional surface model. However, due to the presence of a lot of noise in radiological image data, the three-dimensional surface model needs to be denoised. The denoising process can easily cause small blood vessels in the three-dimensional reconstruction results to break or deform, resulting in inaccurate three-dimensional reconstruction results.

[0004] Therefore, the current three-dimensional reconstruction technology has the problem of inaccurate three-dimensional reconstruction results. Summary of the invention

[0005] Based on this, it is necessary to provide a three-dimensional reconstruction method, device, computer equipment, computer-readable storage medium and computer program product for tubular parts that can improve accuracy in order to solve the above technical problems.

[0006] In a first aspect, the present application provides a method for three-dimensional reconstruction of a tubular part. The method comprises:

[0007] Acquire original reconstruction information of the tubular part; the original reconstruction information includes the centerline position of the tubular part and the original cross-sectional radius corresponding to the centerline position;

[0008] Expanding the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain the expanded cross-sectional radius;

[0009] Performing three-dimensional reconstruction on the tubular portion according to the enlarged cross-sectional radius and the centerline position to obtain an initial three-dimensional reconstruction result of the tubular portion;

[0010] The initial three-dimensional reconstruction result is corrected according to the original reconstruction information to obtain a target three-dimensional reconstruction result of the tubular part.

[0011] In a second aspect, the present application also provides a three-dimensional reconstruction device for a tubular part. The device comprises:

[0012] An acquisition module, used to acquire original reconstruction information of the tubular part; the original reconstruction information includes the centerline position of the tubular part and the original cross-sectional radius corresponding to the centerline position;

[0013] An expansion module, used to expand an original cross-sectional radius that is smaller than a cross-sectional radius threshold to the cross-sectional radius threshold to obtain an expanded cross-sectional radius;

[0014] A reconstruction module, used to perform three-dimensional reconstruction of the tubular portion according to the enlarged cross-sectional radius and the centerline position, to obtain an initial three-dimensional reconstruction result of the tubular portion;

[0015] The correction module is used to correct the initial three-dimensional reconstruction result according to the original reconstruction information to obtain the target three-dimensional reconstruction result of the tubular part.

[0016] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0017] Acquire original reconstruction information of the tubular part; the original reconstruction information includes the centerline position of the tubular part and the original cross-sectional radius corresponding to the centerline position;

[0018] Expanding the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain the expanded cross-sectional radius;

[0019] Performing three-dimensional reconstruction on the tubular portion according to the enlarged cross-sectional radius and the centerline position to obtain an initial three-dimensional reconstruction result of the tubular portion;

[0020] The initial three-dimensional reconstruction result is corrected according to the original reconstruction information to obtain a target three-dimensional reconstruction result of the tubular part.

[0021] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0022] Acquire original reconstruction information of the tubular part; the original reconstruction information includes the centerline position of the tubular part and the original cross-sectional radius corresponding to the centerline position;

[0023] Expanding the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain the expanded cross-sectional radius;

[0024] Performing three-dimensional reconstruction on the tubular portion according to the enlarged cross-sectional radius and the centerline position to obtain an initial three-dimensional reconstruction result of the tubular portion;

[0025] The initial three-dimensional reconstruction result is corrected according to the original reconstruction information to obtain a target three-dimensional reconstruction result of the tubular part.

[0026] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0027] Acquire original reconstruction information of the tubular part; the original reconstruction information includes the centerline position of the tubular part and the original cross-sectional radius corresponding to the centerline position;

[0028] Expanding the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain the expanded cross-sectional radius;

[0029] Performing three-dimensional reconstruction on the tubular portion according to the enlarged cross-sectional radius and the centerline position to obtain an initial three-dimensional reconstruction result of the tubular portion;

[0030] The initial three-dimensional reconstruction result is corrected according to the original reconstruction information to obtain a target three-dimensional reconstruction result of the tubular part.

[0031] The above-mentioned three-dimensional reconstruction method, device, computer equipment, storage medium and computer program product of the tubular part obtain the original reconstruction information of the tubular part, expand the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain the expanded cross-sectional radius, and perform three-dimensional reconstruction on the tubular part according to the expanded cross-sectional radius and the centerline position to obtain an initial three-dimensional reconstruction result of the tubular part, and correct the initial three-dimensional reconstruction result according to the original reconstruction information to obtain a target three-dimensional reconstruction result of the tubular part; the initial three-dimensional reconstruction result with a larger radius can be obtained by first expanding the original cross-sectional radius, and then the initial three-dimensional reconstruction result can be corrected to obtain a target three-dimensional reconstruction result that is close to the real fine structure at the end of the tubular part, so as to accurately reconstruct the end of the tubular part and improve the accuracy of the three-dimensional reconstruction of the tubular part. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 is a schematic flow chart of a method for three-dimensional reconstruction of a tubular part in one embodiment;

[0033] Figure 2is a schematic flow chart of the initial blood vessel model correction step in one embodiment;

[0034] Figure 3 is a schematic diagram of a three-dimensional reconstruction result of a blood vessel based on a traditional convolution surface method in one embodiment;

[0035] Figure 4 is a schematic diagram of a three-dimensional reconstruction result of a blood vessel based on a three-dimensional reconstruction method of a tubular part in one embodiment;

[0036] Figure 5 is a schematic diagram of a three-dimensional reconstruction result of a blood vessel based on a traditional convolution surface method in another embodiment;

[0037] Figure 6 is a schematic diagram of a three-dimensional reconstruction result of a blood vessel based on a three-dimensional reconstruction method of a tubular part in another embodiment;

[0038] Figure 7 is a schematic diagram of a three-dimensional reconstruction result of the liver and hepatic blood vessels in one embodiment;

[0039] Figure 8 is a schematic flow chart of a method for three-dimensional reconstruction of a tubular part in another embodiment;

[0040] Fig. 9 is a structural block diagram of a three-dimensional reconstruction device for a tubular part in one embodiment;

[0041] Fig.10 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0043] In one embodiment, Figure 1 As shown, a method for 3D reconstruction of a tubular part is provided. This embodiment uses the method applied to a terminal as an example for illustration. It is understandable that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0044] Step S110, obtaining original reconstruction information of the tubular part; the original reconstruction information includes the centerline position of the tubular part and the original cross-sectional radius corresponding to the centerline position.

[0045] The tubular part may be a tubular part to be reconstructed, including but not limited to blood vessels, trachea, bronchus and urethra of the human body.

[0046] The original reconstruction information may be the centerline position and the original cross-sectional radius of the tubular part, wherein the centerline position may be the coordinates of the centerline of the tubular part, and the original cross-sectional radius may be the original value of the cross-sectional radius of the tubular part.

[0047] In a specific implementation, a tubular portion image containing a tubular portion may be input into a terminal, and the terminal segments the tubular portion image to obtain original reconstruction information of the tubular portion. The original reconstruction information may include the centerline position of the tubular portion and the original cross-sectional radius corresponding to the centerline position.

[0048] In practical applications, taking blood vessels as an example, the blood vessels to be reconstructed can be divided into multiple small segments, each small segment is used as a blood vessel unit, and the CT (Computed Tomography) image data or MRI (Magnetic Resonance Imaging) image data of the blood vessels to be reconstructed is input into the terminal. The terminal segments the blood vessels in the image data to obtain the centerline position and original cross-sectional radius corresponding to each blood vessel unit, and uses the centerline positions and original cross-sectional radius corresponding to all blood vessel units as the original reconstruction information of the blood vessels to be reconstructed. The terminal can also identify each vascular branch of the blood vessel to be reconstructed to obtain a vascular branch index.

[0049] Step S120, expanding the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain an expanded cross-sectional radius.

[0050] The cross-sectional radius threshold may be a preset cross-sectional radius threshold.

[0051] In the specific implementation, the terminal can compare the original section radius with the section radius threshold. If the original section radius is smaller than the section radius threshold, the original section radius is expanded to the section radius threshold to obtain the expanded section radius. Otherwise, if the original section radius is greater than or equal to the section radius threshold, the original section radius can be directly used as the expanded section radius.

[0052] In practical applications, the cross-sectional radius threshold R0 can be set in advance. For each vascular unit of the blood vessel to be reconstructed, if the original cross-sectional radius corresponding to the vascular unit is less than R0, the original cross-sectional radius is expanded to R0; otherwise, if the original cross-sectional radius corresponding to the vascular unit is greater than or equal to R0, the original cross-sectional radius is kept unchanged.

[0053] Step S130 , performing three-dimensional reconstruction on the tubular portion according to the expanded cross-sectional radius and the centerline position, to obtain an initial three-dimensional reconstruction result of the tubular portion.

[0054] The initial three-dimensional reconstruction result may be a three-dimensional reconstruction result with an enlarged cross-sectional radius.

[0055] In a specific implementation, a scalar field can be constructed in advance, and the terminal uses the expanded cross-sectional radius and centerline position as the expanded reconstruction information of the tubular part, performs convolution surface processing on the expanded reconstruction information, and obtains a first contribution value of the expanded reconstruction information to the pre-constructed scalar field. By comparing the first contribution value with a preset isosurface parameter, an isosurface is determined from the scalar field, and an initial three-dimensional reconstruction result of the tubular part is obtained according to the isosurface. Specifically, the first contribution value equal to the isosurface parameter can be selected as the target contribution value, and the position of the target contribution value in the scalar field is determined as the target position. A triangular mesh reconstruction is performed according to the target position to obtain an initial three-dimensional reconstruction result.

[0056] In practical applications, the maximum bounding box of the blood vessel to be reconstructed can be determined in advance based on the image data, and a scalar field that can accommodate the entire blood vessel to be reconstructed can be constructed based on the maximum bounding box. The initial field value of the scalar field can be set to the default value, for example, 0, and the voxel size of the scalar field can be set based on the image data. The section radius threshold should be neither too large nor too small. If it is too large, the 3D reconstruction result will have vascular adhesion, and if it is too small, the 3D reconstruction result will have vascular rupture. The section radius threshold can be set to

[0057]

[0058] Among them, k is the preset coefficient, the value range is 0.55~0.95, x space is the size of the voxel in the x-axis direction of the scalar field, y space is the size of the voxel in the scalar field in the y-axis direction, z space is the size of the voxel of the scalar field in the z-axis direction.

[0059] The enlarged cross-sectional radius and centerline position corresponding to each vascular unit are used as the enlarged reconstruction information, and the enlarged reconstruction information is convolved according to the pre-set convolution kernel to obtain the first contribution value value1 of the enlarged reconstruction information to the scalar field. The isosurface parameter ISO is pre-set, and the value1 equal to ISO is determined as the target contribution value. The position of the target contribution value in the scalar field is determined as the target position. The triangular mesh is reconstructed according to the target position to obtain the initial three-dimensional reconstruction result of the blood vessel to be reconstructed. Among them, the calculation formula of value1 can be

[0060]

[0061] Among them, S is the reconstruction information after enlargement, including the radius of the section and the centerline position after enlargement, represents the convolution operation, h(r) is the convolution kernel formula, r is the distance between the position on the scalar field and the centerline position, i is the vascular branch index, s is the number of vascular branches, dl represents the differential of the vascular centerline, and λ is the preset coefficient.

[0062] Step S140, correcting the initial three-dimensional reconstruction result according to the original reconstruction information to obtain a target three-dimensional reconstruction result of the tubular part.

[0063] The target three-dimensional reconstruction result may be the three-dimensional reconstruction result of the finally determined tubular part.

[0064] In a specific implementation, the original reconstruction information can be used to perform convolution surface processing on the vertices of the triangular mesh in the initial three-dimensional reconstruction result to obtain the second contribution value of the vertex to the scalar field, and the second contribution value is compared with the isosurface parameter. If the difference between the second contribution value and the isosurface parameter does not exceed the preset threshold, there is no need to correct the vertex, and the vertex is directly used as the corrected vertex. Otherwise, if the difference between the second contribution value and the isosurface parameter exceeds the preset threshold, the vertex needs to be corrected according to the difference between the second contribution value and the isosurface parameter, until the difference between the second contribution value and the isosurface parameter does not exceed the preset threshold, and the corrected vertex is output. The above processing is performed on each vertex of the triangular mesh in the initial three-dimensional reconstruction result, and the target three-dimensional reconstruction result of the tubular part can be obtained according to all the corrected vertices finally obtained.

[0065] In practical applications, for the vertex P(x, y, z) of the triangular mesh in the initial 3D reconstruction result, the center line position and the original cross-sectional radius corresponding to P(x, y, z) can be obtained, and the center line position and the original cross-sectional radius can be convolved with the preset convolution kernel to obtain the second contribution value value2. Value2 is compared with the preset isosurface parameter ISO. If |value2-ISO| does not exceed the preset threshold, it means that the two are close, and there is no need to correct the vertex P(x, y, z). The vertex P(x, y, z) can be directly used as the corrected vertex. Otherwise, if |value2-ISO| exceeds the preset threshold, it means that the two are far away. At this time, the gradient step Δ needs to be determined according to |value2-ISO|. The specific formula is:

[0066] Δ=θ(e |value2-ISO| -1)

[0067] Among them, θ is the control factor of the gradient step and e is a natural constant.

[0068] Calculate the gradient of vertex P(x, y, z) according to Δ The specific formula is

[0069]

[0070] Among them, V x , V y , V z It is the third contribution value obtained by performing convolution surface processing on the points P(x+Δ, y, z), P(x, y+Δ, z), and P(x, y, z+Δ) according to the center line position and the original cross-sectional radius.

[0071] According to the gradient The vertex P(x, y, z) is corrected, and a new second contribution value value2 is calculated according to the corrected vertex P′(x, y, z), until the new |value2-ISO| does not exceed the preset threshold, wherein the correction formula can adopt the Newton iteration formula, specifically:

[0072]

[0073] Among them, δ is the correction coefficient, and P′(x, y, z) is the corrected vertex.

[0074] The above method is used to correct each vertex in the triangular mesh of the initial 3D reconstruction result, and the target 3D reconstruction result is obtained based on all the corrected vertices.

[0075] The above-mentioned three-dimensional reconstruction method of the tubular part obtains the original reconstruction information of the tubular part, expands the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain the expanded cross-sectional radius, and performs three-dimensional reconstruction on the tubular part according to the expanded cross-sectional radius and the centerline position to obtain an initial three-dimensional reconstruction result of the tubular part, and corrects the initial three-dimensional reconstruction result according to the original reconstruction information to obtain a target three-dimensional reconstruction result of the tubular part; the initial three-dimensional reconstruction result with a larger radius can be obtained by first expanding the original cross-sectional radius, and then the initial three-dimensional reconstruction result can be corrected to obtain a target three-dimensional reconstruction result that is close to the real fine structure at the end of the tubular part, so as to accurately reconstruct the end of the tubular part and improve the accuracy of the three-dimensional reconstruction of the tubular part.

[0076] In one embodiment, the above step S130 may specifically include: obtaining the expanded reconstruction information of the tubular part based on the expanded cross-sectional radius and the centerline position; performing convolution surface processing on the expanded reconstruction information to obtain a first contribution value of the tubular part to a pre-constructed scalar field; and determining the initial three-dimensional reconstruction result of the tubular part based on the first contribution value and pre-set isosurface parameters.

[0077] The first contribution value may be a contribution value of the surface of the tubular portion to the scalar field calculated based on the centerline position and the radius of the expanded cross section.

[0078] The convolution surface processing may be a convolution operation.

[0079] In a specific implementation, a scalar field can be constructed in advance, and the terminal uses the expanded cross-sectional radius and centerline position as the expanded reconstruction information of the tubular part, performs convolution surface processing on the expanded reconstruction information, and obtains a first contribution value of the tubular part to the scalar field, pre-sets isosurface parameters, extracts a first contribution value equal to the isosurface parameters from the scalar field, forms an isosurface, and obtains an initial three-dimensional reconstruction result of the tubular part according to the isosurface. Specifically, the first contribution value equal to the isosurface parameters can be selected as the target contribution value, and the position of the target contribution value in the scalar field is determined as the target position. A triangular mesh reconstruction is performed according to the target position to obtain an initial three-dimensional reconstruction result of the tubular part.

[0080] In practical applications, the contribution factor h(r) of the blood vessel centerline to the point at a distance r from the blood vessel centerline can be used as the convolution kernel. The setting of h(r) needs to satisfy monotonicity, that is, the contribution factor h(r) of the blood vessel centerline to the point at a distance r is smaller. The calculation formula of the first contribution value1 can be

[0081]

[0082] Among them, S is the reconstruction information after enlargement, including the radius of the section and the centerline position after enlargement, represents the convolution operation, r is the distance between the position on the scalar field and the centerline position, i is the vascular branch index, s is the number of vascular branches, dl represents the differential of the vascular centerline, and λ is the preset coefficient.

[0083] Among them, λ should satisfy that when the specified blood vessel unit approaches infinite length, the contribution of the blood vessel centerline to the expanded cross-section radius R is equal to the isosurface parameter ISO, that is,

[0084]

[0085] It should be noted that the field value of each voxel in the scalar field can be calculated in parallel to increase the speed of field value calculation.

[0086] In this embodiment, the expanded reconstruction information of the tubular part is obtained according to the expanded cross-sectional radius and the centerline position; convolution surface processing is performed on the expanded reconstruction information to obtain a first contribution value of the tubular part to a pre-constructed scalar field; based on the first contribution value and pre-set isosurface parameters, an initial three-dimensional reconstruction result of the tubular part is determined, and the tubular part can be roughly reconstructed in three dimensions according to the expanded cross-sectional radius, thereby improving the efficiency of three-dimensional reconstruction.

[0087] In one embodiment, the above step S140 may specifically include: determining an initial vertex from the initial three-dimensional reconstruction result, determining a second contribution value of the initial vertex to the scalar field according to the original reconstruction information, and determining the difference between the second contribution value and the isosurface parameter as a contribution value deviation; when the contribution value deviation exceeds a first preset threshold, correcting the initial vertex to obtain a corrected vertex, and returning to the step of determining the initial vertex from the initial three-dimensional reconstruction result, determining the second contribution value of the initial vertex to the scalar field according to the original reconstruction information, and determining the difference between the second contribution value and the isosurface parameter as a contribution value deviation, until the preset stop condition is met; according to the corrected vertex, obtaining the target three-dimensional reconstruction result of the tubular part.

[0088] The initial vertex may be a vertex of a triangular mesh in the initial three-dimensional reconstruction result, for example, a point in the first contribution value value1 that is equal to the isosurface parameter ISO.

[0089] The second contribution value may be a contribution value of the initial vertex to the scalar field calculated according to the center line position and the original cross-sectional radius.

[0090] The contribution value deviation may be a deviation between the second contribution value and the isosurface parameter.

[0091] The preset stop condition may be that there are no uncorrected initial vertices in the initial three-dimensional reconstruction result.

[0092] In a specific implementation, the terminal can select an initial vertex from the triangular mesh of the initial three-dimensional reconstruction result, perform convolution surface processing on the centerline position and the original cross-sectional radius corresponding to the initial vertex, obtain a second contribution value value2, calculate the contribution value deviation |value2-ISO| between the second contribution value value2 and the isosurface parameter ISO, if |value2-ISO| does not exceed the first preset threshold, then there is no need to correct the initial vertex, and the initial vertex can be directly used as the corrected vertex, otherwise, if |value2-ISO| exceeds the first preset threshold, then the initial vertex needs to be corrected according to |value2-ISO| until the contribution value deviation of the corrected vertex does not exceed the first preset threshold. The above processing is performed on each initial vertex in the triangular mesh of the initial three-dimensional reconstruction result, and all the corrected vertices are surface drawn to obtain the target three-dimensional reconstruction result of the tubular part.

[0093] It should be noted that the correction of each initial vertex in the initial three-dimensional reconstruction result can be performed by parallel calculation to increase the correction speed.

[0094] In this embodiment, an initial vertex is determined from an initial three-dimensional reconstruction result, a second contribution value of the initial vertex to a scalar field is determined according to the original reconstruction information, and the difference between the second contribution value and the isosurface parameter is determined as a contribution value deviation; when the contribution value deviation exceeds a first preset threshold value, the initial vertex is corrected to obtain a corrected vertex, and the process returns to the step of determining the initial vertex from the initial three-dimensional reconstruction result, determining the second contribution value of the initial vertex to a scalar field according to the original reconstruction information, and determining the difference between the second contribution value and the isosurface parameter as a contribution value deviation, until a preset stop condition is met; based on the corrected vertex, a target three-dimensional reconstruction result of the tubular part is obtained, and the initial three-dimensional reconstruction result can be continuously corrected according to the real reconstruction information of each vertex in the initial three-dimensional reconstruction result, gradually restoring the fine structure at the end of the tubular part, and improving the accuracy of the three-dimensional reconstruction of the tubular part.

[0095] In one embodiment, the step of correcting the initial vertex to obtain a corrected vertex may specifically include: determining the gradient of the initial vertex; inputting the gradient and contribution value deviation into a preset Newton iteration model to obtain a corrected vertex.

[0096] In a specific implementation, if the contribution value deviation |value2-ISO| exceeds the first preset threshold, the terminal can determine the gradient step Δ according to |value2-ISO|. The specific formula is:

[0097] Δ=θ(e |value2-ISO| -1)

[0098] Among them, θ is the control factor of the gradient step, e is a natural constant, and the first preset threshold is usually not greater than 1% of ISO. This formula can convert the closeness between value2 and ISO into the control step of gradient calculation. The larger |value2-ISO| is, the longer the control step is required, thereby simulating the nonlinear change of the scalar field. The control step should not be too large. By adjusting θ and ISO, the range of Δ can be guaranteed to be within the range of (0.01 to 0.1) unit lengths.

[0099] Then the gradient of the initial vertex P(x, y, z) can be calculated based on the gradient step Δ The specific formula is

[0100]

[0101] Among them, V x , V y , V z It is the third contribution value obtained by performing convolution surface processing on the points P(x+Δ, y, z), P(x, y+Δ, z), and P(x, y, z+Δ) according to the center line position and the original cross-sectional radius.

[0102] The gradient The contribution value deviation |value2-ISO| is input into the Newton iteration model to calculate the corrected vertex, and the above process is repeated until the contribution value deviation |value2-ISO| corresponding to the corrected vertex does not exceed the preset threshold. The Newton iteration model can be

[0103]

[0104] Among them, P′(x, y, z) is the corrected vertex, δ is the correction coefficient, and the correction coefficient δ affects both the number of iterations and the iterative convergence. The reference range is usually 0.3 to 0.95.

[0105] In this embodiment, by determining the gradient of the initial vertex and inputting the gradient and contribution value deviation into a pre-set Newton iterative model to obtain a corrected vertex, the fine structure at the end of the tubular part in the initial three-dimensional reconstruction result can be corrected to improve the accuracy of the three-dimensional reconstruction.

[0106] In one embodiment, the above-mentioned step of determining the gradient of the initial vertex may specifically include: determining the gradient step of the initial vertex based on the contribution value deviation; determining the neighboring partial differential points of the initial vertex based on the gradient step; determining the third contribution value of the neighboring partial differential points based on the original reconstruction information; and obtaining the gradient of the initial vertex based on the second contribution value, the third contribution value and the gradient step.

[0107] The third contribution value may be a contribution value of a neighboring partial differential point to the scalar field calculated according to the center line position and the original cross-sectional radius.

[0108] In the specific implementation, the terminal can determine the gradient step Δ of the initial vertex P(x, y, z) according to the contribution value deviation |value2-ISO|. The specific formula is:

[0109] Δ=θ(e |value2-ISO| -1)

[0110] Among them, θ is the control factor of the gradient step and e is a natural constant.

[0111] Then, the neighboring partial differential points P(x+Δ, y, z), P(x, y+Δ, z), and P(x, y, z+Δ) of the initial vertex P(x, y, z) can be determined according to the gradient step Δ. According to the center line position and the original cross-sectional radius, convolution surface processing is performed on P(x+Δ, y, z), P(x, y+Δ, z), and P(x, y, z+Δ) to obtain the third contribution value V x , V y , V z According to the third contribution value V x , V y , Vz , the second contribution value value2, and the gradient step Δ, the gradient of the initial vertex P(x, y, z) can be calculated. The specific formula is

[0112]

[0113] In this embodiment, the gradient step of the initial vertex is determined based on the contribution value deviation; the neighboring partial differential points of the initial vertex are determined based on the gradient step; the third contribution value of the neighboring partial differential points is determined based on the original reconstruction information; the gradient of the initial vertex is obtained based on the second contribution value, the third contribution value and the gradient step, and the gradient of the vertex in the initial three-dimensional reconstruction result can be calculated, so as to facilitate the correction of the vertices in the initial three-dimensional reconstruction result according to the gradient and improve the accuracy of three-dimensional reconstruction.

[0114] In one embodiment, before the above-mentioned step of performing convolution surface processing on the expanded reconstructed information to obtain the first contribution value of the tubular part to the pre-constructed scalar field, it can also specifically include: determining the maximum bounding box of the tubular part based on the original reconstruction information; determining the pre-constructed scalar field based on the maximum bounding box and the pre-set voxel size of the tubular part.

[0115] In a specific implementation, the terminal can determine a maximum bounding box that can accommodate the entire tubular part based on the original reconstruction information of the tubular part, and can also pre-set the voxel size of the tubular part. The terminal pre-constructs a scalar field based on the maximum bounding box and the pre-set voxel size.

[0116] In practical applications, the maximum bounding box that can accommodate the entire vessel to be reconstructed can be obtained by traversing all the original reconstruction information of the vessel to be reconstructed, and the voxel size (x space ,y space , z space ), based on the maximum bounding box and the predetermined voxel size, the three-dimensional size of the scalar field (x dim ,y dim , z dim ), where x space ,y space , z space Represents the size of the voxel in the x, y, and z directions of the scalar field, respectively. dim ,y dim , z dim Represent the magnitude of the scalar field in the x, y, and z directions respectively.

[0117] In this embodiment, the maximum bounding box of the tubular part is determined according to the original reconstruction information; the pre-constructed scalar field is determined according to the maximum bounding box and the pre-set voxel size of the tubular part, so that the pre-constructed scalar field can accommodate the entire tubular part to be reconstructed, and three-dimensional reconstruction is performed based on the scalar field, which can improve the efficiency of three-dimensional reconstruction.

[0118] In one embodiment, the above-mentioned step of performing convolution surface processing on the expanded reconstructed information to obtain the first contribution value of the tubular part to the pre-constructed scalar field may specifically include: when the distance between the target voxel of the tubular part and the target position of the scalar field does not exceed a second preset threshold, determining the target contribution value of the target voxel to the target position; when the distance between the target voxel and the target position exceeds the second preset threshold, determining the target contribution value of the target voxel to the target position as a preset contribution value; and obtaining the first contribution value based on the target contribution values ​​of each target voxel of the tubular part.

[0119] The target position may be a coordinate position in a scalar field.

[0120] The target voxel may be a small section of a tubular part.

[0121] The target contribution value may be a contribution value of the target voxel to the target position.

[0122] Among them, the preset contribution value can be 0.

[0123] In a specific implementation, after constructing a scalar field according to the maximum bounding box of the tubular part, the terminal can determine any position in the scalar field as the target position, and can also determine each small segment as a target voxel for multiple voxels in the tubular part. If the distance between the target voxel and the target position does not exceed the second preset threshold, the target contribution value of the target voxel to the target position can be determined by convolution surface processing. Otherwise, if the distance between the target voxel and the target position exceeds the second preset value, the target contribution value of the target voxel to the target position can be determined as 0. The target contribution values ​​of each target voxel in the tubular part are summed to obtain the first contribution value of the tubular part to the target position. The above processing is performed on each position in the scalar field to obtain the first contribution value of the tubular part to the entire scalar field.

[0124] In practical applications, in the process of determining the first contribution value of the blood vessel to be reconstructed to the scalar field through convolution surface processing, the contribution value of points far away from the center line of the blood vessel to the scalar field can be determined to be 0 based on the principle of elimination outside the cylindrical range, so as to reduce the computational complexity of convolution surface processing.

[0125] In this embodiment, the target contribution value of the target voxel to the target position is determined when the distance between the target voxel in the tubular part and the target position of the scalar field does not exceed the second preset threshold; when the distance between the target voxel and the target position exceeds the second preset threshold, the target contribution value of the target voxel to the target position is determined as the preset contribution value; and the first contribution value is obtained based on the target contribution value of each target voxel in the tubular part, which can reduce the calculation amount of three-dimensional reconstruction and improve the speed of three-dimensional reconstruction.

[0126] In order to facilitate those skilled in the art to have a deeper understanding of the embodiments of the present application, a specific example will be described below.

[0127] This application proposes a method for reconstructing a tubular structure geometric model, which can not only effectively reconstruct a small blood vessel model, but also does not require excessive subdivision of voxels, can control memory consumption at a normal level, and supports parallel computing to ensure the execution efficiency of the reconstruction algorithm. The method mainly includes the following steps:

[0128] Step 10: Use the traditional convolution surface method to generate a coarser initial blood vessel model, that is, the initial three-dimensional reconstruction result.

[0129] In the specific implementation, a scalar field is first constructed. The requirements for the scalar field setting are: it is large enough to accommodate the entire vascular model, can characterize the vascular characteristics, and has an appropriate memory capacity. The range of the scalar field can be determined by traversing the centerline position of the vessel to be reconstructed and the radius of the original cross section to obtain the maximum bounding box of the vessel to be reconstructed. The voxel size (x space ,y space , z space ) can be set according to the original image data of the blood vessel to be reconstructed. When the layer thickness of the original image data is large, the voxel size can also be customized. According to the size of the entire scalar field and the voxel size of the scalar field, the three-dimensional size of the scalar field (x dim ,y dim , z dim ), where x space ,y space , z space Represents the size of the voxel in the x, y, and z directions of the scalar field, respectively. dim ,y dim , z dim Represent the magnitude of the scalar field in the x, y, and z directions respectively.

[0130] Then set the section radius threshold R0 according to the following formula:

[0131]

[0132] Where k is a preset coefficient. The original cross-sectional radius of the blood vessel is enlarged by setting the original cross-sectional radius that is smaller than the cross-sectional radius threshold as the cross-sectional radius threshold, and keeping the original cross-sectional radius that is greater than or equal to the cross-sectional radius threshold unchanged; at the same time, the original cross-sectional radius is recorded for use in the next step. The setting of the cross-sectional radius threshold should not be too large or too small. If it is too large, the reconstructed blood vessel model will have vascular adhesion, and if it is too small, it will cause vascular rupture. Therefore, k is usually set in the range of 0.55 to 0.95.

[0133] Using the enlarged cross-sectional radius of the blood vessel, the field value of each voxel in the scalar field can be calculated. The process is a convolution calculation process. The convolution kernel h(r) needs to satisfy monotonicity, that is, the contribution factor h(r) of the blood vessel centerline to the point with a greater distance r is smaller. The convolution calculation formula is as follows:

[0134]

[0135] Calculate the first contribution value value1 of the scalar field, use value1 to annotate the scalar field, and obtain a scalar field with annotated field values. According to value1 and isosurface parameter ISO, an initial blood vessel model with a relatively thick diameter can be generated. Specifically, the triangular mesh can be reconstructed according to the scalar field position corresponding to value1 equal to ISO to obtain the initial blood vessel model. Among them, S is the reconstruction information after expansion, including the cross-sectional radius and centerline position after expansion. represents the convolution operation, h(r) is the convolution kernel formula, r is the distance between the position on the scalar field and the centerline position, i is the vascular branch index, s is the number of vascular branches, dl represents the differential of the vascular centerline, and λ is the preset coefficient.

[0136] Among them, the coefficient λ should satisfy that when a specific blood vessel segment l approaches infinite length, the contribution value of the point at the distance R from the center line of the blood vessel segment is equal to the isosurface parameter, where R is the radius of the blood vessel after expansion. The specific formula is as follows:

[0137]

[0138] Step 11, using the convolution surface method to correct the coarser initial blood vessel model, that is, using the original cross-sectional radius of the blood vessel to correct the vertices in the initial blood vessel model, so as to gradually restore the small blood vessels at the end to the real state.

[0139] Figure 2 A schematic flow chart of the initial vascular model modification steps is provided. Figure 2, we can first restore the enlarged cross-sectional radius in step 10, use the original cross-sectional radius to perform convolution surface calculation on the vertex P (x, y, z) of the initial blood vessel model, and solve the second contribution value value2 of the vertex to the scalar field.

[0140] Then use value2 and the expected ISO value (isosurface parameter ISO) to calculate the gradient step Δ. The specific formula is:

[0141] Δ=θ(e |value2-ISO| -1)

[0142] Among them, |value2-ISO| refers to the absolute value of the difference between value2 and the expected ISO value, which can represent the degree of proximity between the two. The smaller the value, the closer the two values ​​are. When the value is less than a certain threshold, it means that the point has been corrected and no subsequent steps are required. The threshold is usually not greater than 1% of the ISO value; θ is the control factor of the gradient step, which can convert the degree of proximity between value2 and the expected ISO value into the control step length of the gradient calculation. The larger the error |value2-ISO|, the longer the gradient calculation step length is required to simulate the nonlinear change of the scalar field. The step length should not be too large. Finally, by setting both θ and ISO, the range of Δ can be guaranteed to be within the range of (0.01~0.1) unit lengths.

[0143] After getting the gradient step, you need to calculate the gradient of the current point P(x, y, z) The calculation process can first calculate the third contribution value V of P(x+Δ, y, z), P(x, y+Δ, z), and P(x, y, z+Δ) respectively according to the center line position and the original cross-sectional radius through the convolution surface method. x , V y , V z , with V x For example, the calculation formula is

[0144]

[0145] Where S” represents the original reconstruction information corresponding to the point P(x+Δ, y, z). The gradient The calculation formula can be

[0146]

[0147] Finally, the gradient and the second contribution value value2 are used in combination with the correction coefficient δ to correct the vertices of the initial vascular model. The correction coefficient δ affects both the number of iterations and the convergence of the iterations. The reference range is usually 0.3 to 0.95. The correction formula can use the Newton iteration formula, which is as follows:

[0148]

[0149] After the calculation, the above steps are followed to iterate other vertices in the initial blood vessel model until the second contribution value value2 is close to the expected ISO value, that is, |value2-ISO| does not exceed a certain threshold.

[0150] The second contribution value involved in this process is Value2, V x , V y , V z , is also obtained according to the convolution calculation formula in step 10, but the reconstruction information of the blood vessels substituted in is different.

[0151] It should be noted that the convolution calculations in the above tubular structure geometric model reconstruction method can all be accelerated in parallel, and the performance of the calculation can be optimized based on the principle of eliminating points outside the cylinder range. Parallel acceleration means that each voxel of the scalar field in step 10 is an independent calculation unit, so parallel calculations can be used between different voxels; each vertex of the initial blood vessel model in step 11 is an independent calculation unit, so parallel calculations can be used between different vertices. Acceleration by eliminating points outside the cylinder range means that by using the principle of convolution surface processing, the farther the point is from the center of the blood vessel branch, the smaller the contribution of the branch to the point. Therefore, the contribution value of the point that is greater than the specified distance from the center of the blood vessel branch can be set to 0 during convolution surface processing.

[0152] The parallel computing process may include but is not limited to the parallel computing logic commonly used on a GPU or CPU.

[0153] In one embodiment, the above tubular structure geometric model reconstruction method may include the following steps:

[0154] Step 20, acquiring CT image data, applying a segmentation algorithm to the CT image data to obtain information such as the centerline position of the blood vessel, the blood vessel radius, and the branch index, and using the result as input of this embodiment;

[0155] Step 21, traverse the centerline position and the blood vessel radius, calculate the blood vessel radius threshold with k = 0.65, and appropriately expand the radius of small blood vessels; then, based on the expanded blood vessel radius, blood vessel centerline position, branch index and other information, obtain the maximum bounding box of the blood vessel to be reconstructed. In the bounding box statistics process, a quadrangular prism with a side length of 2.0×R is used for statistics, where R is the expanded blood vessel radius, to ensure that it is the minimum bounding box that can accommodate the entire blood vessel to be reconstructed. A scalar field is constructed using the bounding box and the spatial information of the original image data;

[0156] Step 22, use the isosurface ISO=0.5 and the convolution kernel formula

[0157]

[0158] Where s is the filter width factor, s=0.85, and r is the distance between the position on the scalar field and the centerline position. The field value of the scalar field is calculated in parallel on the computer's GPU (graphics processing unit, a core processor that can perform efficient parallel computing). During the parallel computing process, a radius of 2.0×R is used to perform the exclusion logic outside the cylinder range, that is, when the distance from the current point to the centerline of the vascular unit exceeds this value, the contribution value of the vascular unit to the point is considered to be 0, and the calculation can be directly excluded. After the scalar field calculation is completed, a three-dimensional mesh model result of a small blood vessel that is enlarged can be initially obtained based on the isosurface ISO and the scalar field value, that is, the initial vascular model.

[0159] It should be noted that the convolution kernel needs to meet the following characteristics:

[0160] Rotational symmetry, that is, the filter is equally smooth in all directions and is not biased towards any one direction;

[0161] Monotonicity: the farther away a point is from r, the smaller its contribution factor h(r) to any point P is;

[0162] Filter width factor, such as the parameter s in the above convolution kernel formula.

[0163] It should also be noted that the convolution kernel formula can be but is not limited to It can also be a Gaussian function, a polynomial function, etc.

[0164] Step 23, restore the blood vessel radius, and use the real blood vessel radius to correct the vertices of the initial blood vessel model. The vertex correction process can gradually restore the real state of small blood vessels. In the correction process, the same convolution calculation formula as step 22 is used to calculate the real contribution value value2 of the vertex P (x, y, z), that is, the second contribution value. The same isosurface parameters, convolution kernel, and exclusion logic outside the cylinder range can be used in the calculation process.

[0165] Step 24, determine the proximity between the true contribution value value2 and the expected ISO value |value2-ISO|, and set the threshold to 0.005. If the proximity is less than the threshold, it means that the vertex P(x, y, z) does not need to be corrected, and exit the iterative process; otherwise, continue with the following steps to correct the position of the vertex P(x, y, z).

[0166] Step 25, according to the proximity and gradient step Δ, the same convolution calculation formula as step 22 is used to calculate the true contribution value V at points P(x+Δ, y, z), P(x, y+Δ, z), and P(x, y, z+Δ) respectively. x , V y , V z, which is the third contribution value;

[0167] Step 26, calculate the gradient vector of vertex P(x, y, z) The specific formula is

[0168]

[0169] Step 27, substitute the gradient vector and use the correction coefficient δ = 0.8 to correct the position of the vertex P (x, y, z). The specific formula is:

[0170]

[0171] Step 28, using the new vertex position to continue step 23 for iterative correction until |value2-ISO| is less than a threshold value of 0.005, and then the iterative process can be exited.

[0172] In step 29, the correction process of all vertices on the initial blood vessel model can adopt parallel acceleration logic, and the target blood vessel model obtained after all vertices are corrected can well show the characteristics and structure of the small parts of the blood vessel. The target blood vessel model can directly use surface rendering technology to display the blood vessel structure.

[0173] It should be noted that the execution efficiency of the above method is related to the size of the specific reconstruction information parameters. Under the premise of using parallel acceleration, the model reconstruction process can be completed within 180 to 700 ms for common data.

[0174] Using the same technical parameters, Figure 3 A schematic diagram of the results of 3D reconstruction of blood vessels based on the traditional convolution surface method is provided. Figure 4 A schematic diagram of the 3D reconstruction results of blood vessels based on the 3D reconstruction method of tubular parts is provided. The voxel size of the scalar field adopts the voxel size of the original image data (0.798828, 0.798828, 1.0) mm, and the radius of the terminal small blood vessels is as small as 0.27 mm.

[0175] Using the same technical parameters, Figure 5 Another schematic diagram of the 3D reconstruction of blood vessels based on the traditional convolution surface method is provided. Figure 6 Another schematic diagram of the 3D reconstruction result of blood vessels based on the 3D reconstruction method of tubular parts is provided. Among them, the voxel size of the scalar field still uses the voxel size of the original image data (0.78125, 0.78125, 1.0) mm, but the radius of the terminal small blood vessels is as small as 0.262 mm.

[0176] Figure 7A schematic diagram of the three-dimensional reconstruction results of the liver and hepatic blood vessels is provided, wherein, except for the liver tissue, the remaining vascular tissues are reconstructed using the three-dimensional reconstruction method of the tubular part of the present application, and the liver model is used as a reference for the display effect of the hepatic blood vessels.

[0177] Depend on Figure 3-7 It can be found that when the blood vessel radius is too small, the traditional convolution surface method is prone to cause blood vessel breakage in the three-dimensional reconstruction results. To solve this problem, the scalar field needs to be subdivided, that is, the size of the voxel is reduced, which will cause the memory for storing the scalar field to increase by a multiple of the cube. However, using the tubular part three-dimensional reconstruction method of the present application, a real blood vessel model can be accurately reconstructed under the same conditions, including small blood vessels with a small blood vessel radius. The reconstructed blood vessel model has a smooth surface and avoids the phenomenon of terminal breakage; using liver tissue as a reference object, the reconstructed blood vessel model can better show the distribution of the overall blood vessels and small blood vessels in the liver tissue.

[0178] In one embodiment, Figure 8 As shown, a method for three-dimensional reconstruction of a tubular part is provided, and the method is applied to a terminal as an example for explanation, and includes the following steps:

[0179] Step S201, obtaining original reconstruction information of the tubular part; the original reconstruction information includes the centerline position of the tubular part and the original cross-sectional radius corresponding to the centerline position;

[0180] Step S202, expanding the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain the expanded cross-sectional radius;

[0181] Step S203, performing three-dimensional reconstruction on the tubular part according to the expanded cross-sectional radius and the centerline position, to obtain an initial three-dimensional reconstruction result of the tubular part;

[0182] Step S204, determining an initial vertex from the initial three-dimensional reconstruction result, determining a second contribution value of the initial vertex to the scalar field according to the original reconstruction information, and determining a difference between the second contribution value and the isosurface parameter as a contribution value deviation;

[0183] Step S205, when the contribution value deviation exceeds the first preset threshold, determine the gradient step of the initial vertex according to the contribution value deviation, determine the neighboring partial differential point of the initial vertex according to the gradient step, and determine the third contribution value of the neighboring partial differential point according to the original reconstruction information;

[0184] Step S206, obtaining the gradient of the initial vertex according to the second contribution value, the third contribution value and the gradient step;

[0185] Step S207, inputting the gradient and contribution value deviation into a preset Newton iteration model to obtain a corrected vertex, and returning to step S204 until a preset stop condition is met;

[0186] Step S208, obtaining a target three-dimensional reconstruction result of the tubular part according to the corrected vertices.

[0187] In a specific implementation, the terminal can obtain the original reconstruction information of the tubular part, including the centerline position and the original cross-sectional radius, expand the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold, and perform three-dimensional reconstruction based on the expanded cross-sectional radius and the centerline position to obtain an initial three-dimensional reconstruction result of the tubular part. For the initial vertex in the initial three-dimensional reconstruction result, the second contribution value of the initial vertex to the pre-constructed scalar field is determined based on the original reconstruction information, and the contribution value deviation between the second contribution value and the isosurface parameter is calculated. The gradient of the initial vertex is determined based on the contribution value deviation, and the initial vertex is corrected based on the determined gradient to obtain a corrected vertex. Based on the corrected vertices corresponding to all the initial vertices in the initial three-dimensional reconstruction result, the target three-dimensional reconstruction result of the tubular part can be obtained.

[0188] The above-mentioned three-dimensional reconstruction method of the tubular part obtains the original reconstruction information of the tubular part, expands the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain the expanded cross-sectional radius, performs three-dimensional reconstruction on the tubular part according to the expanded cross-sectional radius and the centerline position, obtains the initial three-dimensional reconstruction result of the tubular part, determines the initial vertex from the initial three-dimensional reconstruction result, determines the second contribution value of the initial vertex to the scalar field according to the original reconstruction information, and determines the difference between the second contribution value and the isosurface parameter as the contribution value deviation, and when the contribution value deviation exceeds the first preset threshold, determines the gradient step of the initial vertex according to the contribution value deviation, determines the neighboring partial differential points of the initial vertex according to the gradient step, and determines the neighboring partial differential points according to the original reconstruction information. The third contribution value of the partial differential point is used to obtain the gradient of the initial vertex according to the second contribution value, the third contribution value and the gradient step. The gradient and the contribution value deviation are input into a preset Newton iterative model to obtain a corrected vertex, and then return to the step of determining the initial vertex from the initial 3D reconstruction result, and determining the second contribution value of the initial vertex to the scalar field according to the original reconstruction information until the preset stop condition is met. According to the corrected vertex, the target 3D reconstruction result of the tubular part is obtained; the initial 3D reconstruction result with a larger radius can be obtained by first expanding the radius of the original cross-section, and then the initial 3D reconstruction result is corrected to obtain a target 3D reconstruction result close to the real fine structure at the end of the tubular part, and the end of the tubular part is accurately reconstructed, thereby improving the accuracy of the 3D reconstruction of the tubular part.

[0189] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0190] Based on the same inventive concept, the embodiment of the present application also provides a 3D reconstruction device for a tubular part for implementing the 3D reconstruction method for a tubular part involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of the 3D reconstruction device for one or more tubular parts provided below can refer to the limitations of the 3D reconstruction method for a tubular part above, and will not be repeated here.

[0191] In one embodiment, Fig. 9 As shown, a three-dimensional reconstruction device for a tubular part is provided, comprising: an acquisition module 310, an expansion module 320, a reconstruction module 330 and a correction module 340, wherein:

[0192] The acquisition module 310 is used to acquire original reconstruction information of the tubular part; the original reconstruction information includes the centerline position of the tubular part and the original cross-sectional radius corresponding to the centerline position;

[0193] An enlargement module 320 is used to enlarge the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain an enlarged cross-sectional radius;

[0194] A reconstruction module 330, configured to perform three-dimensional reconstruction of the tubular portion according to the enlarged cross-sectional radius and the centerline position, to obtain an initial three-dimensional reconstruction result of the tubular portion;

[0195] The correction module 340 is used to correct the initial three-dimensional reconstruction result according to the original reconstruction information to obtain a target three-dimensional reconstruction result of the tubular part.

[0196] In one embodiment, the reconstruction module 330 is further used to obtain the expanded reconstruction information of the tubular portion based on the expanded cross-sectional radius and the centerline position; perform convolution surface processing on the expanded reconstruction information to obtain a first contribution value of the tubular portion to a pre-constructed scalar field; and determine the initial three-dimensional reconstruction result of the tubular portion based on the first contribution value and pre-set isosurface parameters.

[0197] In one embodiment, the correction module 340 is also used to determine an initial vertex from the initial three-dimensional reconstruction result, determine a second contribution value of the initial vertex to the scalar field according to the original reconstruction information, and determine the difference between the second contribution value and the isosurface parameter as a contribution value deviation; when the contribution value deviation exceeds a first preset threshold, the initial vertex is corrected to obtain a corrected vertex, and return to the step of determining the initial vertex from the initial three-dimensional reconstruction result, determining the second contribution value of the initial vertex to the scalar field according to the original reconstruction information, and determining the difference between the second contribution value and the isosurface parameter as a contribution value deviation, until a preset stop condition is met; based on the corrected vertex, the target three-dimensional reconstruction result of the tubular part is obtained.

[0198] In one embodiment, the correction module 340 is further used to determine the gradient of the initial vertex; input the gradient and the contribution value deviation into a preset Newton iteration model to obtain the corrected vertex.

[0199] In one embodiment, the correction module 340 is also used to determine the gradient step of the initial vertex according to the contribution value deviation; determine the neighboring partial differential points of the initial vertex according to the gradient step; determine the third contribution value of the neighboring partial differential points to the scalar field according to the original reconstruction information; and obtain the gradient of the initial vertex according to the second contribution value, the third contribution value and the gradient step.

[0200] In one embodiment, the reconstruction module 330 is further used to determine the maximum bounding box of the tubular part according to the original reconstruction information; and determine the pre-constructed scalar field according to the maximum bounding box and the preset voxel size of the tubular part.

[0201] In one embodiment, the above-mentioned reconstruction module 330 is also used to determine the target contribution value of the target voxel to the target position when the distance between the target voxel of the tubular part and the target position of the scalar field does not exceed a second preset threshold; when the distance between the target voxel and the target position exceeds the second preset threshold, determine the target contribution value of the target voxel to the target position as a preset contribution value; and obtain the first contribution value based on the target contribution value of each target voxel in the tubular part.

[0202] Each module in the above-mentioned three-dimensional reconstruction device of a tubular part can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0203] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Fig.10 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a three-dimensional reconstruction method of a tubular part is realized. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0204] Those skilled in the art will understand that Fig.10The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0205] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0206] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0207] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0208] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0209] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0210] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0211] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A three-dimensional reconstruction method of a tubular part, characterized in that: The method comprises: Acquire original reconstruction information of the tubular part; the original reconstruction information includes the centerline position of the tubular part and the original cross-sectional radius corresponding to the centerline position; Expanding the original cross-sectional radius that is smaller than the cross-sectional radius threshold to the cross-sectional radius threshold to obtain the expanded cross-sectional radius; Performing three-dimensional reconstruction on the tubular portion according to the enlarged cross-sectional radius and the centerline position to obtain an initial three-dimensional reconstruction result of the tubular portion; The initial three-dimensional reconstruction result is corrected according to the original reconstruction information to obtain a target three-dimensional reconstruction result of the tubular part.

2. The method according to claim 1, characterized in that: The three-dimensional reconstruction of the tubular part is performed according to the expanded cross-sectional radius and the centerline position to obtain an initial three-dimensional reconstruction result of the tubular part, including: Obtaining the enlarged reconstruction information of the tubular part according to the enlarged cross-sectional radius and the centerline position; Performing convolution surface processing on the enlarged reconstructed information to obtain a first contribution value of the tubular portion to a pre-constructed scalar field; The initial three-dimensional reconstruction result of the tubular part is determined according to the first contribution value and preset isosurface parameters.

3. The method according to claim 2, characterized in that The correcting process of the initial three-dimensional reconstruction result according to the original reconstruction information to obtain the target three-dimensional reconstruction result of the tubular part includes: Determine an initial vertex from the initial three-dimensional reconstruction result, determine a second contribution value of the initial vertex to the scalar field according to the original reconstruction information, and determine the difference between the second contribution value and the isosurface parameter as a contribution value deviation; In the case where the contribution value deviation exceeds a first preset threshold, the initial vertex is corrected to obtain a corrected vertex, and the process returns to the step of determining the initial vertex from the initial three-dimensional reconstruction result, determining a second contribution value of the initial vertex to the scalar field according to the original reconstruction information, and determining the difference between the second contribution value and the isosurface parameter as the contribution value deviation, until a preset stop condition is met; The target three-dimensional reconstruction result of the tubular part is obtained according to the corrected vertices.

4. The method according to claim 3, characterized in that The step of correcting the initial vertex to obtain a corrected vertex includes: Determining the gradient of the initial vertex; The gradient and the contribution value deviation are input into a preset Newton iteration model to obtain the modified vertex.

5. The method according to claim 4, characterized in that Determining the gradient of the initial vertex includes: Determining the gradient step of the initial vertex according to the contribution value deviation; Determining a partial differential point adjacent to the initial vertex according to the gradient step; Determining a third contribution value of the neighboring partial differential point to the scalar field according to the original reconstruction information; The gradient of the initial vertex is obtained according to the second contribution value, the third contribution value and the gradient step.

6. The method according to claim 2, characterized in that Before performing convolution surface processing on the enlarged reconstructed information to obtain a first contribution value of the tubular portion to the pre-constructed scalar field, the method further includes: Determining a maximum bounding box of the tubular part according to the original reconstruction information; The pre-constructed scalar field is determined according to the maximum bounding box and the preset voxel size of the tubular portion.

7. The method according to claim 6, characterized in that The step of performing convolution surface processing on the enlarged reconstructed information to obtain a first contribution value of the tubular portion to a pre-constructed scalar field includes: In a case where the distance between the target voxel of the tubular portion and the target position of the scalar field does not exceed a second preset threshold, determining a target contribution value of the target voxel to the target position; When the distance between the target voxel and the target position exceeds the second preset threshold, determining the target contribution value of the target voxel to the target position as a preset contribution value; The first contribution value is obtained according to the target contribution value of each target voxel of the tubular part.

8. A three-dimensional reconstruction device for a tubular part, characterized in that: The device comprises: An acquisition module, used to acquire original reconstruction information of the tubular part; the original reconstruction information includes the centerline position of the tubular part and the original cross-sectional radius corresponding to the centerline position; An expansion module, used to expand an original cross-sectional radius that is smaller than a cross-sectional radius threshold to the cross-sectional radius threshold to obtain an expanded cross-sectional radius; A reconstruction module, used to perform three-dimensional reconstruction of the tubular portion according to the enlarged cross-sectional radius and the centerline position, to obtain an initial three-dimensional reconstruction result of the tubular portion; The correction module is used to correct the initial three-dimensional reconstruction result according to the original reconstruction information to obtain the target three-dimensional reconstruction result of the tubular part.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images