Cone beam X-ray three-dimensional reconstruction method and device based on 3D Gaussian sputtering, medium and product

Abstract

The invention discloses a cone beam X-ray three-dimensional reconstruction method and device based on 3D Gaussian sputtering, a medium and a product, and relates to the technical field of three-dimensional reconstruction, and the cone beam X-ray three-dimensional reconstruction method based on 3D Gaussian sputtering comprises the steps: initializing the attributes of a plurality of 3D Gaussian ellipsoids corresponding to a to-be-reconstructed object, and obtaining the initial attribute of each 3D Gaussian ellipsoid; the attributes comprise a center position, a covariance matrix and density; using the rotation matrix and integration, and based on the initial attribute of each 3D Gaussian ellipsoid, carrying out multiple times of iterative updating on the attribute of each 3D Gaussian ellipsoid to obtain the optimized attribute of each 3D Gaussian ellipsoid; and determining each 3D Gaussian ellipsoid with the optimization attribute as a three-dimensional reconstruction result of the object to be reconstructed. According to the invention, the precision of cone-beam X-ray three-dimensional reconstruction is improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of three-dimensional reconstruction, in particular to a cone-beam X-ray three-dimensional reconstruction method and device based on 3D Gaussian sputtering, a medium and a product. BACKGROUND

[0002] In the field of medical imaging, three-dimensional reconstruction technology plays a crucial role. Traditional computed tomography (CT) three-dimensional reconstruction requires the acquisition of a large number of X-ray images, which poses a potential radiation hazard to the human body. Although magnetic resonance imaging (MRI) avoids the problem of radiation, the reconstruction process is time-consuming and the equipment and examination costs are high. Therefore, developing an efficient and low-radiation three-dimensional reconstruction algorithm has important clinical significance. Especially in cone-beam CT (CBCT) imaging, due to its geometric characteristics, higher accuracy and efficiency of the reconstruction algorithm are required.

[0003] Traditional three-dimensional reconstruction algorithms, such as voxel-based or surface-based methods, often have low accuracy and cannot meet the needs of medical diagnosis. SUMMARY

[0004] The application aims to provide a cone-beam X-ray three-dimensional reconstruction method and device based on 3D Gaussian sputtering, a medium and a product to solve the problem of low accuracy of cone-beam X-ray three-dimensional reconstruction.

[0005] To achieve the above-mentioned purpose, the application provides the following solutions:

[0006] In a first aspect, the application provides a cone-beam X-ray three-dimensional reconstruction method based on 3D Gaussian sputtering, comprising:

[0007] initializing the properties of a plurality of 3D Gaussian ellipsoids corresponding to the object to be reconstructed, to obtain the initial properties of each 3D Gaussian ellipsoid; the properties include: center position, covariance matrix and density;

[0008] using a rotation matrix and integration, based on the initial properties of each 3D Gaussian ellipsoid, iteratively updating the properties of each 3D Gaussian ellipsoid multiple times to obtain the optimized properties of each 3D Gaussian ellipsoid;

[0009] determining each 3D Gaussian ellipsoid with optimized properties as the three-dimensional reconstruction result of the object to be reconstructed.

[0010] In an embodiment, in the process of updating the properties of each 3D Gaussian ellipsoid multiple times based on the initial properties of each 3D Gaussian ellipsoid by using the rotation matrix and the integral, to obtain the optimized properties of each 3D Gaussian ellipsoid, the determination process of the properties of each 3D Gaussian ellipsoid at any current iteration number includes:

[0011] determining any 3D Gaussian ellipsoid as a target ellipsoid, determining any light source as a target light source, determining the coordinate system of the target light source as a target light source coordinate system, and determining the camera at the position of the target light source as a target camera;

[0012] determining the updated properties of the target ellipsoid at the last iteration number as the initial properties of the target ellipsoid at the current iteration number;

[0013] determining the rotation matrix of the target ellipsoid in the target light source coordinate system at the current iteration number based on the coordinates of the center position of the target ellipsoid in the target light source coordinate system at the current iteration number;

[0014] rotating the target ellipsoid by using the rotation matrix of the target ellipsoid in the target light source coordinate system at the current iteration number to obtain the rotated target ellipsoid in the target light source coordinate system at the current iteration number, and determining the covariance matrix of the rotated target ellipsoid in the target light source coordinate system at the current iteration number as the rotated covariance matrix of the target ellipsoid in the target light source coordinate system at the current iteration number;

[0015] obtaining the real X-ray image captured by the target camera;

[0016] obtaining the reconstructed X-ray image under the target light source at the current iteration number by using the integral and the rendering operation based on the real X-ray image captured by the target camera, the densities of all 3D Gaussian ellipsoids at the current iteration number, and the rotated covariance matrices of all 3D Gaussian ellipsoids in the target light source coordinate system at the current iteration number;

[0017] determining the loss value at the current iteration number based on the reconstructed X-ray image under each light source at the current iteration number and the corresponding real X-ray image;

[0018] determining whether a stop condition is reached; the stop condition is that a preset iteration number is reached or the loss value at the current iteration number is less than a loss threshold;

[0019] if yes, determining the initial properties of each 3D Gaussian ellipsoid at the current iteration number as the optimized properties of each 3D Gaussian ellipsoid;

[0020] If no, the initial attribute of each 3D Gaussian ellipsoid under the current iteration number is updated to obtain the updated attribute of each 3D Gaussian ellipsoid under the current iteration number, and the current iteration number is updated to the next iteration number for the next iteration until the stopping condition is reached to obtain the optimized attribute of each 3D Gaussian ellipsoid.

[0021] In an embodiment, the rotation matrix of the target ellipsoid in the target light source coordinate system under the current iteration number is determined based on the coordinates of the center position of the target ellipsoid in the target light source coordinate system under the current iteration number, comprising:

[0022] The plurality of normalization factors of the target ellipsoid in the target light source coordinate system under the current iteration number are determined based on the x-axis component, the y-axis component and the z-axis component of the coordinates of the target ellipsoid in the target light source coordinate system under the current iteration number;

[0023] The rotation matrix of the target ellipsoid in the target light source coordinate system under the current iteration number is determined based on the x-axis component, the y-axis component, the z-axis component and the plurality of normalization factors of the coordinates of the target ellipsoid in the target light source coordinate system under the current iteration number.

[0024] In an embodiment, the plurality of normalization factors include a first normalization factor and a plurality of normalization factors;

[0025] The plurality of normalization factors of the target ellipsoid in the target light source coordinate system under the current iteration number are determined based on the x-axis component, the y-axis component and the z-axis component of the coordinates of the target ellipsoid in the target light source coordinate system under the current iteration number, comprising:

[0026] The first sum of squares of the target ellipsoid in the target light source coordinate system under the current iteration number is determined according to the x-axis component and the y-axis component of the coordinates of the target ellipsoid in the target light source coordinate system under the current iteration number;

[0027] The second sum of squares of the target ellipsoid in the target light source coordinate system under the current iteration number is determined according to the first sum of squares and the z-axis component of the coordinates of the target ellipsoid in the target light source coordinate system under the current iteration number;

[0028] The first normalization factor of the target ellipsoid in the target light source coordinate system under the current iteration number is determined based on the first sum of squares of the target ellipsoid in the target light source coordinate system under the current iteration number;

[0029] The second normalization factor of the target ellipsoid in the target light source coordinate system under the current iteration number is determined based on the second sum of squares of the target ellipsoid in the target light source coordinate system under the current iteration number.

[0030] In an embodiment, the rotation matrix in the target light source coordinate system at the current iteration number of the target ellipsoid is determined based on the x-axis component, the y-axis component, the z-axis component of the coordinate in the target light source coordinate system at the current iteration number of the target ellipsoid, and a plurality of normalization factors, comprising:

[0031] The x-axis intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid is determined according to the x-axis component of the coordinate in the target light source coordinate system at the current iteration number of the target ellipsoid and the first normalization factor.

[0032] The y-axis intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid is determined according to the y-axis component of the coordinate in the target light source coordinate system at the current iteration number of the target ellipsoid and the first normalization factor.

[0033] The comprehensive intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid is determined according to the z-axis component of the coordinate in the target light source coordinate system at the current iteration number of the target ellipsoid and the second normalization factor.

[0034] The target intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid is determined according to the comprehensive intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid.

[0035] The rotation matrix in the target light source coordinate system at the current iteration number of the target ellipsoid is determined according to the x-axis intermediate value, the y-axis intermediate value, the comprehensive intermediate value and the target central value in the target light source coordinate system at the current iteration number of the target ellipsoid.

[0036] In an embodiment, the reconstructed X-ray image at the current iteration number of the target light source is obtained based on the real X-ray image captured by the target camera, the density at the current iteration number of all 3D Gaussian ellipsoids, and the rotated covariance matrix at the current iteration number of all 3D Gaussian ellipsoids in the target light source coordinate system, using integration and rendering operations, comprising:

[0037] The projection direction of the rotated target ellipsoid at the current iteration number in the target light source coordinate system along the target light source is integrated to obtain the 2D Gaussian ellipse of the target ellipsoid at the current iteration number of the target light source, and the 2D Gaussian ellipse of the target ellipsoid at the current iteration number of the target light source is determined as the target ellipse at the current iteration number.

[0038] Any pixel of the real X-ray image captured by the target camera is determined as a target pixel, and the coordinate system of the target camera is determined as the target camera coordinate system.

[0039] determine the probability density of the target pixel in the target light source coordinate system in the current iteration number of the target ellipse based on the coordinates of the target pixel in the target camera coordinate system and the rotated covariance matrix of the target ellipsoid in the target light source coordinate system in the current iteration number;

[0040] render based on the density of all 3D Gaussian ellipsoids in the current iteration number and the probability density of all 2D Gaussian ellipses of the target pixel in the target light source coordinate system in the current iteration number to obtain the gray value of the target pixel under the target light source in the current iteration number;

[0041] obtain the reconstructed X-ray image under the target light source in the current iteration number based on the gray value of all pixels in the real X-ray image captured by the target camera under the target light source in the current iteration number.

[0042] In an embodiment, the loss value in the current iteration number is determined based on the reconstructed X-ray image under each light source in the current iteration number and the corresponding real X-ray image, comprising:

[0043] The loss value in the current iteration number is determined according to the reconstructed X-ray image under each light source in the current iteration number and the corresponding real X-ray image by using a loss function; the loss function is:

[0044]

[0045] wherein, loss is the loss value; M is the number of light sources; N is the number of pixels in the reconstructed X-ray image and the real X-ray image; I' mn is the gray value of the nth pixel in the reconstructed X-ray image under the mth light source; I mn is the gray value of the nth pixel in the real X-ray image captured by the camera at the position corresponding to the mth light source.

[0046] In a second aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned 3D Gaussian sputtering based cone beam X-ray three-dimensional reconstruction method.

[0047] In a third aspect, the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the above-mentioned 3D Gaussian sputtering based cone beam X-ray three-dimensional reconstruction method.

[0048] In a fourth aspect, the present application provides a computer program product comprising a computer program, wherein the computer program is executed by a processor to implement the above-mentioned 3D Gaussian sputtering based cone beam X-ray three-dimensional reconstruction method.

[0049] According to the specific embodiments provided in the application, the application discloses the following technical effects:

[0050] The application discloses a cone beam X-ray three-dimensional reconstruction method and device based on 3D Gaussian sputtering, a medium and product, which utilizes a rotation matrix and integration, performs multiple iteration updates on the properties of each 3D Gaussian ellipsoid based on the initial properties of each 3D Gaussian ellipsoid, obtains the optimized properties of each 3D Gaussian ellipsoid, and thus obtains the three-dimensional reconstruction result of the object to be reconstructed. In the traditional 3D Gaussian sputtering (3DGS) method, when processing cone beam X-ray projection, a Jacobian matrix is used for approximate transformation, which may introduce errors. Compared with the traditional 3DGS method, the application uses a rotation matrix instead of a Jacobian matrix to accurately align the Gaussian ellipsoid, avoids such approximation, and thus improves the accuracy of projection and ultimately improves the accuracy of cone beam X-ray three-dimensional reconstruction. Further, the application also integrates the 3D Gaussian ellipsoid to simulate the actual X-ray imaging process and more realistically simulate the physical process of X-ray penetrating the object and producing attenuation. Such integration operation considers the density distribution inside the Gaussian ellipsoid, and thus can more accurately calculate the projection value and improve the accuracy of cone beam X-ray three-dimensional reconstruction. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the application or the related art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only constitute some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0052] Figure 1 A flowchart of a cone beam X-ray three-dimensional reconstruction method based on 3D Gaussian sputtering provided by an embodiment of the application is shown in the figure.

[0053] Figure 2 A schematic diagram of the positions of a light source and an object to be reconstructed is shown in the figure.

[0054] Figure 3 A structural schematic diagram of a computer device provided by an embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the application will be described clearly and completely in combination with the drawings in the embodiments of the application. Obviously, the described embodiments only constitute some of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0056] The purpose of the present application is to provide a cone beam X-ray three-dimensional reconstruction method, device, medium and product based on 3D Gaussian sputtering, aiming to improve the accuracy of cone beam X-ray three-dimensional reconstruction.

[0057] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below in combination with the drawings and specific embodiments.

[0058] In an exemplary embodiment, as shown in Figure 1 A cone beam X-ray three-dimensional reconstruction method based on 3D Gaussian sputtering is provided, comprising:

[0059] Step 1: initialize the properties of a plurality of 3D Gaussian ellipsoids corresponding to the object to be reconstructed, and obtain the initial properties of each 3D Gaussian ellipsoid; the properties include: center position, covariance matrix and density.

[0060] Step 2: using a rotation matrix and integration, based on the initial properties of each 3D Gaussian ellipsoid, the properties of each 3D Gaussian ellipsoid are updated multiple times, and the optimized properties of each 3D Gaussian ellipsoid are obtained.

[0061] As an optional implementation, in the process of executing step 2, the determination process of the properties of each 3D Gaussian ellipsoid at any current iteration number includes:

[0062] Step 201: determine any 3D Gaussian ellipsoid as a target ellipsoid, determine any light source as a target light source, determine the coordinate system of the target light source as the target light source coordinate system, and determine the camera at the position of the target light source as the target camera.

[0063] Specifically, as shown in Figure 2 All light sources are on a horizontal plane and form a circle around the three-dimensional object to be reconstructed ②. A camera is also provided at each light source position. The light source and the camera form a cone beam X-ray projection system ①.

[0064] Step 202: determine the updated properties of the target ellipsoid at the last iteration number as the initial properties of the target ellipsoid at the current iteration number.

[0065] Step 203: based on the coordinates of the center position of the target ellipsoid at the current iteration number in the target light source coordinate system, determine the rotation matrix of the target ellipsoid in the target light source coordinate system at the current iteration number.

[0066] As an optional implementation, step 203 includes:

[0067] Step 2031: determining, based on the x-axis component, the y-axis component and the z-axis component of the coordinate of the target ellipsoid at the current iteration number in the target light source coordinate system, a plurality of normalization factors of the target ellipsoid at the current iteration number in the target light source coordinate system.

[0068] As an optional implementation, the plurality of normalization factors comprises a first normalization factor and a plurality of normalization factors. Step 2031 comprises:

[0069] Step 20311: determining, according to the x-axis component and the y-axis component of the coordinate of the target ellipsoid at the current iteration number in the target light source coordinate system, a first sum of squares of the target ellipsoid at the current iteration number in the target light source coordinate system.

[0070] Specifically, at any iteration number, the calculation formula of the first sum of squares is:

[0071]

[0072] wherein, is the first sum of squares of the i th 3D Gaussian ellipsoid in the coordinate system of the m th light source; x i,m is the x-axis component of the coordinate of the i th 3D Gaussian ellipsoid in the coordinate system of the m th light source; y i,m is the y-axis component of the coordinate of the i th 3D Gaussian ellipsoid in the coordinate system of the m th light source.

[0073] Step 20312: determining, according to the first sum of squares and the z-axis component of the target ellipsoid at the current iteration number in the target light source coordinate system, a second sum of squares of the target ellipsoid at the current iteration number in the target light source coordinate system.

[0074] Specifically, at any iteration number, the calculation formula of is:

[0075]

[0076] wherein, is the second sum of squares of the i th 3D Gaussian ellipsoid in the coordinate system of the m th light source; z i,m is the z-axis component of the coordinate of the i th 3D Gaussian ellipsoid in the coordinate system of the m th light source.

[0077] Step 20313: determining, based on the first sum of squares of the target ellipsoid at the current iteration number in the target light source coordinate system, a first normalization factor of the target ellipsoid at the current iteration number in the target light source coordinate system.

[0078] Specifically, at any iteration number, the calculation formula of the first normalization factor is:

[0079]

[0080] wherein r xy,i,m is the first normalization factor of the i-th 3D Gaussian ellipsoid in the coordinate system of the m-th light source.

[0081] Step 20314: determining, based on the second sum of squares of the target ellipsoid in the target light source coordinate system at the current iteration number, a second normalization factor of the target ellipsoid in the target light source coordinate system at the current iteration number.

[0082] Specifically, the calculation formula of the second normalization factor at any iteration number is:

[0083]

[0084] wherein r xyz,i,m is the second normalization factor of the i-th 3D Gaussian ellipsoid in the coordinate system of the m-th light source.

[0085] Step 2032: determining, based on the x-axis component, the y-axis component, the z-axis component of the coordinate of the target ellipsoid in the target light source coordinate system at the current iteration number and the plurality of normalization factors, a rotation matrix of the target ellipsoid in the target light source coordinate system at the current iteration number.

[0086] As an optional implementation, step 2032 comprises:

[0087] Step 20321: determining, based on the x-axis component of the coordinate of the target ellipsoid in the target light source coordinate system at the current iteration number and the first normalization factor, an x-axis intermediate value of the target ellipsoid in the target light source coordinate system at the current iteration number.

[0088] Specifically, the calculation formula of the x-axis intermediate value at any iteration number is:

[0089]

[0090] wherein k x,i,m is the x-axis intermediate value of the i-th 3D Gaussian ellipsoid in the coordinate system of the m-th light source.

[0091] Step 20322: determining, based on the y-axis component of the coordinate of the target ellipsoid in the target light source coordinate system at the current iteration number and the first normalization factor, a y-axis intermediate value of the target ellipsoid in the target light source coordinate system at the current iteration number.

[0092] Specifically, the calculation formula of the y-axis intermediate value at any iteration number is:

[0093]

[0094] wherein k y,i,mis the intermediate value of the i-th 3D Gaussian ellipsoid in the y-axis of the coordinate system of the m-th light source.

[0095] Step 20323: determining the comprehensive intermediate value of the target ellipsoid in the target light source coordinate system at the current iteration number of the target ellipsoid according to the z-axis component of the coordinate of the target ellipsoid in the target light source coordinate system at the current iteration number of the target ellipsoid and the second normalization factor.

[0096] Specifically, the calculation formula of the comprehensive intermediate value at any iteration number is:

[0097]

[0098] wherein, c i,m is the comprehensive intermediate value of the i-th 3D Gaussian ellipsoid in the coordinate system of the m-th light source.

[0099] Step 20324: determining the target intermediate value of the target ellipsoid in the target light source coordinate system at the current iteration number of the target ellipsoid according to the comprehensive intermediate value of the target ellipsoid in the target light source coordinate system at the current iteration number of the target ellipsoid.

[0100] Specifically, the calculation formula of the target intermediate value at any iteration number is:

[0101]

[0102] wherein, e i,m is the target intermediate value of the i-th 3D Gaussian ellipsoid in the coordinate system of the m-th light source.

[0103] Step 20325: determining the rotation matrix of the target ellipsoid in the target light source coordinate system at the current iteration number of the target ellipsoid according to the x-axis intermediate value, the y-axis intermediate value, the comprehensive intermediate value and the target central value of the target ellipsoid in the target light source coordinate system at the current iteration number of the target ellipsoid.

[0104] Specifically, the calculation formula of the rotation matrix at any iteration number is:

[0105]

[0106] wherein, R i,m is the rotation matrix of the i-th 3D Gaussian ellipsoid in the coordinate system of the m-th light source.

[0107] Step 204: rotating the target ellipsoid by using the rotation matrix of the target ellipsoid in the target light source coordinate system at the current iteration number of the target ellipsoid to obtain the rotated target ellipsoid in the target light source coordinate system at the current iteration number of the target ellipsoid, and determining the covariance matrix of the rotated target ellipsoid in the target light source coordinate system at the current iteration number of the target ellipsoid as the rotated covariance matrix of the target ellipsoid in the target light source coordinate system at the current iteration number of the target ellipsoid.

[0108] Step 205: Obtain the real X-ray image captured by the target camera.

[0109] Step 206: Obtain the reconstructed X-ray image under the target light source at the current iteration number based on the real X-ray image captured by the target camera, the density of all 3D Gaussian ellipsoids at the current iteration number, and the rotated covariance matrix of all 3D Gaussian ellipsoids in the coordinate system of the target light source at the current iteration number, by using the integration and rendering operation.

[0110] As an optional implementation, step 206 includes:

[0111] Step 2061: Integrate the rotated target ellipsoid in the coordinate system of the target light source at the current iteration number along the projection direction of the target light source to obtain the 2D Gaussian ellipse of the target ellipsoid under the target light source at the current iteration number, and determine the 2D Gaussian ellipse of the target ellipsoid under the target light source at the current iteration number as the target ellipse at the current iteration number.

[0112] Step 2062: Determine any pixel of the real X-ray image captured by the target camera as a target pixel, and determine the coordinate system of the target camera as the target camera coordinate system.

[0113] Step 2063: Determine the target ellipse of the target pixel at the current iteration number and the probability density in the coordinate system of the target light source based on the coordinates of the target pixel in the target camera coordinate system and the rotated covariance matrix of the target ellipsoid in the coordinate system of the target light source at the current iteration number.

[0114] Specifically, at any iteration number, the calculation formula of the probability density is:

[0115]

[0116] wherein f n,i,m is the probability density of the nth pixel in the coordinate system of the ith 3D Gaussian ellipsoid and the mth light source; x' n,m is the x-axis component of the coordinates of the nth pixel in the coordinate system of the camera at the mth light source position; y' n,m is the y-axis component of the coordinates of the nth pixel in the coordinate system of the camera at the mth light source position; α, β and χ are intermediate parameters; σ 33,i,m , σ 11,i,m , σ 13,i,m , σ 12,i,m , σ 13,i,m , σ 23,i,m and σ 22,i,m are elements in the covariance matrix of the ith 3D Gaussian ellipsoid in the coordinate system of the mth light source.

[0117] Step 2064: rendering based on the density of all 3D Gaussian ellipsoids at the current iteration number and the probability density of the target pixel at the current iteration number of all 2D Gaussian ellipses and the target light source coordinate system to obtain the gray value of the target pixel at the current iteration number under the target light source.

[0118] Specifically, at any iteration number, the calculation formula of the gray value is:

[0119]

[0120] wherein, I' nm is the gray value of the nth pixel under the mth light source, that is, the gray value of the nth pixel in the reconstructed X-ray image under the mth light source; p i is the density of the ith 3D Gaussian ellipsoid; j is the number of 3D Gaussian ellipsoids and 2D Gaussian ellipses.

[0121] Step 2065: obtaining the reconstructed X-ray image under the current iteration number under the target light source based on the gray value of all pixels in the real X-ray image captured by the target camera under the current iteration number under the target light source.

[0122] Step 207: determining the loss value at the current iteration number based on the reconstructed X-ray image under the current iteration number under each light source and the corresponding real X-ray image.

[0123] As an optional implementation, step 207 comprises:

[0124] adopting a loss function to determine the loss value at the current iteration number according to the reconstructed X-ray image under the current iteration number under each light source and the corresponding real X-ray image; the loss function is:

[0125]

[0126] wherein, loss is the loss value; M is the number of light sources; N is the number of pixels in the reconstructed X-ray image and the real X-ray image; I' mn is the gray value of the nth pixel in the reconstructed X-ray image under the mth light source; I mn is the gray value of the nth pixel in the real X-ray image captured by the camera at the position corresponding to the mth light source.

[0127] Step 208: determining whether a stop condition is reached; the stop condition is that a preset iteration number is reached or the loss value at the current iteration number is less than a loss threshold.

[0128] Step 209: if yes, determining the initial attribute of each 3D Gaussian ellipsoid at the current iteration number as the optimized attribute of each 3D Gaussian ellipsoid.

[0129] Step 210: If no, the initial attribute of each 3D Gaussian ellipsoid at the current iteration number is updated to obtain the updated attribute of each 3D Gaussian ellipsoid at the current iteration number, and the current iteration number is updated to the next iteration number for the next iteration until the stopping condition is reached to obtain the optimized attribute of each 3D Gaussian ellipsoid.

[0130] Step 3: Each 3D Gaussian ellipsoid with the optimized attribute is determined as the three-dimensional reconstruction result of the object to be reconstructed.

[0131] In an exemplary embodiment, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to implement the cone-beam X-ray three-dimensional reconstruction method based on 3D Gaussian sputtering.

[0132] In an exemplary embodiment, a computer readable storage medium is provided, having stored thereon a computer program, the computer program being executed by a processor to implement the cone-beam X-ray three-dimensional reconstruction method based on 3D Gaussian sputtering.

[0133] In an exemplary embodiment, a computer program product is provided, comprising a computer program, the computer program being executed by a processor to implement the cone-beam X-ray three-dimensional reconstruction method based on 3D Gaussian sputtering.

[0134] In an exemplary embodiment, a computer device is provided, which can be a server or a terminal, and its internal structure diagram can be as shown in Figure 3 The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. 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, a computer program, and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to implement a cone-beam X-ray three-dimensional reconstruction method based on 3D Gaussian sputtering.

[0135] Those skilled in the art can understand that, Figure 3It should be noted that the structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0136] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiments. Any reference to memory, database or other medium used in the embodiments provided by 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 storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric 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 but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0137] The database involved in the embodiments provided by the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided by the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0138] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0139] Any combination of the technical features in the above embodiments can be made. For the sake of brevity, the foregoing description has not described all possible combinations of the technical features in the above embodiments, however, it is understood that any combination of the technical features is within the scope of the present disclosure as long as there is no contradiction.

[0140] The principles and implementation manners of the present application are described herein by using specific examples, and the above embodiments are only used to help understand the method of the present application and its core idea; meanwhile, according to the idea of the present application, the specific implementation manners and application scopes will be changed by those skilled in the art. In conclusion, the content of the present description should not be understood as a limitation of the present application.

Claims

1. A cone-beam X-ray three-dimensional reconstruction method based on 3D Gaussian sputtering, characterized in that, The cone-beam X-ray three-dimensional reconstruction method based on 3D Gaussian sputtering comprises: initializing attributes of a plurality of 3D Gaussian ellipsoids corresponding to the object to be reconstructed, to obtain initial attributes of each 3D Gaussian ellipsoid; the attributes include: center position, covariance matrix and density; updating the attributes of each 3D Gaussian ellipsoid based on the initial attributes of each 3D Gaussian ellipsoid by using a rotation matrix and integration, to obtain optimized attributes of each 3D Gaussian ellipsoid; determining each 3D Gaussian ellipsoid with the optimized attributes as the three-dimensional reconstruction result of the object to be reconstructed.

2. The 3D Gaussian sputter-based cone-beam X-ray three-dimensional reconstruction method according to claim 1, characterized in that, In the process of updating the attributes of each 3D Gaussian ellipsoid based on the initial attributes of each 3D Gaussian ellipsoid by using a rotation matrix and integration, to obtain optimized attributes of each 3D Gaussian ellipsoid, the determination process of the attributes of each 3D Gaussian ellipsoid at any current iteration number comprises: determining any 3D Gaussian ellipsoid as a target ellipsoid, determining any light source as a target light source, determining the coordinate system of the target light source as a target light source coordinate system, and determining the camera at the position of the target light source as a target camera; determining the updated attributes of the target ellipsoid at the last iteration number as the initial attributes of the target ellipsoid at the current iteration number; determining the rotation matrix of the target ellipsoid in the target light source coordinate system at the current iteration number based on the coordinates of the center position of the target ellipsoid in the target light source coordinate system at the current iteration number; rotating the target ellipsoid by using the rotation matrix of the target ellipsoid in the target light source coordinate system at the current iteration number, to obtain the rotated target ellipsoid in the target light source coordinate system at the current iteration number, and determining the covariance matrix of the rotated target ellipsoid in the target light source coordinate system at the current iteration number as the rotated covariance matrix of the target ellipsoid in the target light source coordinate system at the current iteration number; acquiring a real X-ray image collected by the target camera; obtaining the reconstructed X-ray image under the target light source at the current iteration number based on the real X-ray image collected by the target camera, the densities of all 3D Gaussian ellipsoids at the current iteration number, and the rotated covariance matrices of all 3D Gaussian ellipsoids in the target light source coordinate system at the current iteration number by using integration and rendering operations; determining the loss value at the current iteration number based on the reconstructed X-ray image under each light source at the current iteration number and the corresponding real X-ray image; determining whether a stop condition is reached; the stop condition is that a preset iteration number is reached or the loss value at the current iteration number is less than a loss threshold; if yes, determining the initial attributes of each 3D Gaussian ellipsoid at the current iteration number as the optimized attributes of each 3D Gaussian ellipsoid; if no, updating the initial attributes of each 3D Gaussian ellipsoid at the current iteration number to obtain the updated attributes of each 3D Gaussian ellipsoid at the current iteration number, updating the current iteration number to the next iteration number, and performing the next iteration until the stop condition is reached, to obtain the optimized attributes of each 3D Gaussian ellipsoid.

3. The 3D Gaussian sputter-based cone-beam X-ray three-dimensional reconstruction method according to claim 1, characterized in that, determining the rotation matrix of the target ellipsoid in the target light source coordinate system at the current iteration number based on the coordinates of the center position of the target ellipsoid in the target light source coordinate system at the current iteration number comprises: determining, based on the x-axis component, the y-axis component, the z-axis component and the plurality of normalization factors of the coordinates in the target light source coordinate system at the current iteration number of the target ellipsoid, a rotation matrix in the target light source coordinate system at the current iteration number of the target ellipsoid; determining, based on the x-axis component, the y-axis component, the z-axis component and the plurality of normalization factors of the coordinates in the target light source coordinate system at the current iteration number of the target ellipsoid, a rotation matrix in the target light source coordinate system at the current iteration number of the target ellipsoid.

4. The 3D Gaussian sputter-based cone beam X-ray three-dimensional reconstruction method according to claim 3, characterized in that, The plurality of normalization factors comprises a first normalization factor and a plurality of normalization factors. determining, based on the x-axis component, the y-axis component, the z-axis component and the plurality of normalization factors of the coordinates in the target light source coordinate system at the current iteration number of the target ellipsoid, a rotation matrix in the target light source coordinate system at the current iteration number of the target ellipsoid, comprises: determining, based on the x-axis component and the y-axis component of the coordinates in the target light source coordinate system at the current iteration number of the target ellipsoid, a first sum of squares in the target light source coordinate system at the current iteration number of the target ellipsoid; determining, based on the first sum of squares and the z-axis component in the target light source coordinate system at the current iteration number of the target ellipsoid, a second sum of squares in the target light source coordinate system at the current iteration number of the target ellipsoid; determining, based on the first sum of squares in the target light source coordinate system at the current iteration number of the target ellipsoid, a first normalization factor in the target light source coordinate system at the current iteration number of the target ellipsoid; determining, based on the second sum of squares in the target light source coordinate system at the current iteration number of the target ellipsoid, a second normalization factor in the target light source coordinate system at the current iteration number of the target ellipsoid.

5. The 3D Gaussian sputter-based cone beam X-ray three-dimensional reconstruction method according to claim 4, characterized in that, determining, based on the x-axis component, the y-axis component, the z-axis component and the plurality of normalization factors of the coordinates in the target light source coordinate system at the current iteration number of the target ellipsoid, a rotation matrix in the target light source coordinate system at the current iteration number of the target ellipsoid, comprises: determining, based on the x-axis component and the first normalization factor of the coordinates in the target light source coordinate system at the current iteration number of the target ellipsoid, an x-axis intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid; determining, based on the y-axis component and the first normalization factor of the coordinates in the target light source coordinate system at the current iteration number of the target ellipsoid, a y-axis intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid; determining, based on the z-axis component and the second normalization factor of the coordinates in the target light source coordinate system at the current iteration number of the target ellipsoid, a comprehensive intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid; determining, based on the comprehensive intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid, a target intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid; determining, based on the x-axis intermediate value, the y-axis intermediate value, the comprehensive intermediate value and the target intermediate value in the target light source coordinate system at the current iteration number of the target ellipsoid, a rotation matrix in the target light source coordinate system at the current iteration number of the target ellipsoid.

6. The 3D Gaussian sputter-based cone beam X-ray three-dimensional reconstruction method according to claim 2, characterized in that, The integral and rendering operations are used to obtain the reconstructed X-ray image under the target light source at the current iteration number based on the real X-ray image collected by the target camera, the density of all 3D Gaussian ellipsoids at the current iteration number, and the rotated covariance matrix of all 3D Gaussian ellipsoids at the current iteration number in the target light source coordinate system, including: Integrating the rotated target ellipsoid at the current iteration number in the target light source coordinate system along the projection direction of the target light source to obtain the 2D Gaussian ellipse of the target ellipsoid at the current iteration number under the target light source, and determining the 2D Gaussian ellipse of the target ellipsoid at the current iteration number under the target light source as the target ellipse at the current iteration number; Determining any pixel of the real X-ray image collected by the target camera as a target pixel, and determining the coordinate system of the target camera as the target camera coordinate system; Based on the coordinates of the target pixel in the target camera coordinate system and the rotated covariance matrix of the target ellipsoid at the current iteration number in the target light source coordinate system, determining the target ellipse of the target pixel at the current iteration number and the probability density in the target light source coordinate system; Based on the density of all 3D Gaussian ellipsoids at the current iteration number and the probability density of all 2D Gaussian ellipses of the target pixel at the current iteration number in the target light source coordinate system, rendering is performed to obtain the gray value of the target pixel at the current iteration number under the target light source; Based on the gray values of all pixels in the real X-ray image collected by the target camera at the current iteration number under the target light source, the reconstructed X-ray image at the current iteration number under the target light source is obtained.

7. The 3D Gaussian sputter-based cone beam X-ray three-dimensional reconstruction method according to claim 2, characterized in that, The loss value at the current iteration number is determined based on the reconstructed X-ray image at the current iteration number under each light source and the corresponding real X-ray image, including: A loss function is used to determine the loss value at the current iteration number according to the reconstructed X-ray image at the current iteration number under each light source and the corresponding real X-ray image; the loss function is: wherein loss is a loss value; M is the number of light sources; N is the number of pixels in the reconstructed X-ray image and the real X-ray image; I ' mn is the gray value of the nthpixel in the reconstructed X-ray image under the mthlight source; I mn is the gray value of the nthpixel in the real X-ray image captured by the camera at the position corresponding to the mthlight source.

8. A computer apparatus comprising: A memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that the processor executes the computer program to implement the 3D Gaussian sputtering-based cone beam X-ray three-dimensional reconstruction method of any one of claims 1-7.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the 3D Gaussian sputtering-based cone beam X-ray three-dimensional reconstruction method of any one of claims 1-7.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the 3D Gaussian sputtering-based cone beam X-ray three-dimensional reconstruction method of any one of claims 1-7.

Images