Decoupled ray-dependent spectral ct image reconstruction method, device, medium, and product

Abstract

The embodiment of the application discloses a kind of decoupling ray-dependent spectral CT image reconstruction method, equipment, medium and product, belong to image processing technical field.The method includes: determining residual projection combination for current base image combination;Determine weight matrix and constant approximation matrix of each target data block in weight matrix, weight matrix includes the weight of each element in Jacobian matrix at current base image combination, target data block and spectral base substance combination one-to-one correspondence;Based on current base image combination, residual projection combination and each constant approximation matrix, update current base image combination;Determine the residual projection combination corresponding to the updated current base image combination, in the case where the predetermined element of residual projection combination meets the iteration condition, return the step of determining the residual projection combination for current base image combination, until iteration ends, obtain target base image combination.The embodiment of the application can give consideration to image reconstruction quality and efficiency.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of image processing, and in particular to a method, device, medium and product for decoupling X-ray dependent spectral CT image reconstruction. Background Technology

[0002] Compared to traditional CT (Computed Tomography), spectral CT is a more advanced imaging method. It reconstructs spectral images by analyzing projection data from multiple X-ray energy spectra, resulting in images that reflect richer physical information about the sample's interior. However, existing spectral CT reconstruction methods struggle to balance image reconstruction efficiency and quality. Summary of the Invention

[0003] This invention provides a decoupled X-ray dependent spectral CT image reconstruction method, device, medium, and product to achieve the technical effect of simultaneously ensuring image reconstruction efficiency and quality.

[0004] According to one aspect of the present invention, a method for decoupling ray-dependent energy spectral CT image reconstruction is provided, the method comprising: Determine the residual projection combination for the current base image combination, wherein the residual projection combination is the difference between the measured projection data and the forward projection data of the current base image combination; Determine the weight matrix and the constant approximation matrix of each target data block in the weight matrix. The weight matrix includes the weight of each element in the Jacobian matrix at the current base image combination. The target data block corresponds one-to-one with the energy spectrum base material combination. The current base image combination is updated based on the current base image combination, the residual projection combination, and each of the constant approximation matrices; Determine the residual projection combination corresponding to the updated current base image combination. If the predetermined elements of the residual projection combination meet the iteration conditions, return to the step of determining the residual projection combination for the current base image combination until the predetermined elements do not meet the iteration conditions. Then, use the updated current base image combination as the target base image combination.

[0005] According to another aspect of the present invention, a decoupled X-ray dependent spectral CT image reconstruction apparatus is provided, the apparatus comprising: The residual projection module is used to determine the residual projection combination for the current base image combination, wherein the residual projection combination is the difference between the measured projection data and the forward projection data of the current base image combination; The weight matrix module is used to determine the weight matrix and the constant approximation matrix of each target data block in the weight matrix. The weight matrix includes the weight of each element in the Jacobian matrix at the current base image combination. The target data block corresponds one-to-one with the energy spectrum base material combination. The base image update module is used to update the current base image combination based on the current base image combination, the residual projection combination, and each of the constant approximation matrices; The judgment module is used to determine the residual projection combination corresponding to the updated current base image combination. If the predetermined elements of the residual projection combination meet the iteration conditions, the module returns to the step of determining the residual projection combination for the current base image combination until the predetermined elements do not meet the iteration conditions, and then the updated current base image combination is used as the target base image combination.

[0006] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: One or more processors; Storage device for storing one or more programs. When one or more programs are executed by one or more processors, the one or more processors implement the decoupled ray-dependent energy spectrum CT image reconstruction method as described in any embodiment of the present invention.

[0007] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores computer instructions for causing a processor to execute and implement the decoupled ray-dependent energy spectrum CT image reconstruction method according to any embodiment of the present invention.

[0008] According to another aspect of the present invention, a computer program product is provided, which, when executed by a processor, implements the decoupled X-ray dependent energy spectrum CT image reconstruction method described in any embodiment of the present invention.

[0009] The technical solution provided by the embodiments of the present invention includes a weight matrix that includes target data blocks corresponding to each energy spectrum base material combination. The constant approximation matrix can decouple the dependency relationship between each weight and ray in the corresponding target data block. Therefore, the combination result of all constant approximation matrices can significantly reduce the complexity of the weight matrix. Therefore, image iterative reconstruction based on all constant approximation matrices can reduce the amount of data calculation in the image iterative process and improve the efficiency of image iterative solution while ensuring the image reconstruction effect.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A schematic flowchart of the decoupled X-ray dependent energy spectrum CT image reconstruction method provided in an embodiment of the present invention; Figure 2 This is another flowchart illustrating the decoupled X-ray dependent energy spectrum CT image reconstruction method provided in this embodiment of the invention. Figure 3 This is another flowchart illustrating the decoupled X-ray dependent spectral CT image reconstruction method provided in this embodiment of the invention. Figure 4A A graph showing the change of the relative error of the base image with the number of iterations, provided in an embodiment of the present invention. Figure 4B A graph showing the change of relative error of projection data with the number of iterations provided in an embodiment of the present invention; Figure 5 This is a comparative schematic diagram of the image combination results provided in an embodiment of the present invention; Figure 6A This is another graph showing the change of the relative error of the base image with the number of iterations, provided in an embodiment of the present invention. Figure 6B Another graph showing the change of the relative error of the projection data with the number of iterations provided in the embodiments of the present invention; Figure 7A Another graph showing the change of the relative error of the base image over time, provided in an embodiment of the present invention; Figure 7B Another graph showing the change of relative error of projection data over time, provided in an embodiment of the present invention; Figure 8 This is another comparative schematic diagram of the image combination results provided in an embodiment of the present invention; Figure 9A This is another graph showing the change of the relative error of the base image with the number of iterations, provided in an embodiment of the present invention. Figure 9B Another graph showing the change of the relative error of the projection data with the number of iterations provided in the embodiments of the present invention; Figure 10A This is another graph showing how the difference between adjacent steps of the base image changes with the number of iterations, provided in an embodiment of the present invention. Figure 10BThis is another graph showing how the difference between adjacent steps in the projection data changes with the number of iterations, as provided in this embodiment of the invention. Figure 11 This is another comparative schematic diagram of the image combination results provided in an embodiment of the present invention; Figure 12 A schematic diagram of the structure of the decoupled ray-dependent energy spectrum CT image reconstruction device provided in an embodiment of the present invention; Figure 13 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0013] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0014] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0015] The imaging principle of CT is to perform tomographic imaging using the absorption rate of X-rays by the target material, thereby obtaining structural information about the target material's interior. Physically, this information is expressed by the spatial distribution of the linear attenuation coefficient, which corresponds to the pixel value (or CT value) of the CT image. The linear attenuation coefficient of a sample is related not only to its spatial location but also to the energy of the photons it comes into contact with. The core breakthrough of energy-spectral CT lies in its use of multi-energy spectral X-rays to scan objects and, through precise imaging model construction, fully utilizes the variation information of attenuation characteristics in the energy dimension. In a typical energy-spectral CT imaging scenario, if Q different energy spectra are used to scan the object, and the object is modeled as a combination of D base materials, then the linear attenuation coefficient of the object at different energies can be expressed as the following linear combination: ; in, Representing the d The first basic substance in the first m The mass decay coefficient within a specific energy range, for example, assuming 50 keV is in the 5th energy range. Representing 50keV, the first i The linear attenuation coefficient of the sample in a pixel block / voxel block, also known as A virtual monoenergetic image at 50 keV; Representing the d The first basic substance in the first i The distribution density of each pixel / voxel, forming a vector. Called the first Each base image; Indicates the object in the first position Within the energy range of the first The linear attenuation coefficients at each pixel (or voxel), forming a vector. The object is called the first Virtual monoenergetic images within a specific energy range. The core of the spectral CT reconstruction problem is to solve... This is equivalent to solving for the joint vector formed by all basis images. .

[0016] On the other hand, the rays emitted by an X-ray tube typically contain photons of multiple energies; the distribution of photons of different energies within an X-ray is called the energy spectrum of that X-ray. Traditional CT is equivalent to scanning a sample using only single-energy X-rays, assuming that all X-rays contain photons of only one energy. Spectral CT, however, uses X-rays with different energy spectra to scan the sample, obtaining multiple sets of observational data. For example, dual-energy CT uses two types of X-rays with different energy spectra (referred to as low-energy and high-energy spectra, respectively) to scan the sample.

[0017] The embodiments of the present invention use Energy spectrum The energy distribution vector, then Energy spectrum In the m The proportion of photons within each energy range to the total number of photons. In summary, without considering noise and other physical factors, the forward process of energy spectrum CT can be modeled according to Beer's Law: ; in, Indicates the first In the energy spectrum, the first The proportion of photons within a specific energy range to the total number of photons in that energy spectrum; set Called the first The energy distribution vector of the energy spectrum; For the first The projection matrix corresponding to all projected (or scanned) data under each energy spectrum. Here is the discretized X-ray transformation vector, where It can represent the first The first energy spectrum X-ray and the first The length of the intersection line of each pixel (or voxel); Indicates the first The number of energy ranges under each energy spectrum. The scanning geometry here can be a two-dimensional parallel beam, a two-dimensional fan-shaped beam, or a three-dimensional parallel beam, a three-dimensional cone beam, a spiral cone beam, etc.

[0018] In addition, base image In the The projection data under each energy spectrum can be expressed as follows: , Also known as the basis sine diagram. The basis sine diagram can be understood as the projection data of the basis image corresponding to the basis material.

[0019] Spectral CT equipment in the The detection data collected under each energy spectrum are denoted as The image reconstruction problem of spectral CT can then be transformed into solving the following system of equations: .

[0020] In dual-energy CT scanning and data acquisition systems, the mainstream equipment architectures currently include single-source single-detector fast kilovolt switching CT, dual-source dual-detector CT, and single-source dual-layer detector CT, etc.

[0021] Figure 1 This is a flowchart illustrating the decoupled ray-dependent energy-spectral CT image reconstruction method provided in this embodiment of the invention. This embodiment decouples the dependence between the weights of each element in the Jacobian matrix and the rays by using a constant approximation matrix for each target data block in the target weight matrix, simplifying the iterative solution process for energy-spectral CT images. This method can be executed by a decoupled ray-dependent energy-spectral CT image reconstruction device, which can be implemented in hardware and / or software and can be configured in electronic devices such as computers or servers. Figure 1 As shown, the method in this embodiment includes: S110. Determine the residual projection combination for the current base image combination. The residual projection combination is the difference between the measured projection data and the forward projection data of the current base image combination.

[0022] The current base image combination in the first iteration is the initial base image combination. Each initial base image in the initial base image combination can be a random image at a predetermined resolution or a zero-value image at a predetermined resolution. The predetermined resolution is the target resolution for image reconstruction, i.e., the resolution of the target base image.

[0023] The measured projection data is the actual projection data, that is, the projection data observed by the detector.

[0024] The residual projection combination includes the residual projections corresponding to each current basis image in the current basis image combination, and the residual projection combination is equal to the difference between the measured projection data and the forward projection data of the current basis image combination. The forward projection data of the current basis image combination includes the sub-forward projection data of the current basis image combination under each energy spectrum. The forward projection data of the current basis image combination can be determined using existing techniques, such as using the following formula: ; in, It is identified as q The projection matrix of the energy spectrum scanning geometry, q The range of values ​​is It can be a two-dimensional parallel beam, a two-dimensional fan-shaped beam, a three-dimensional cone beam, or a spiral cone beam, etc. For the first d Each base image, To be identified as d The base material is identified as the corresponding energy spectrum. m The mass decay coefficient within the energy range, It is identified as q The energy spectrum is in the energy range m Energy distribution below; To be identified as The total number of energy ranges under the energy spectrum. D The total number of basic substances; To be identified as q Sub-projection data under the energy spectrum, This is the forward projection operator.

[0025] Understandably, before determining the forward projection data corresponding to the current base image combination, it is necessary to obtain... D Mass decay coefficient of seed material as well as Q Energy distribution under each energy spectrum .

[0026] The residual projection combination corresponding to the current base image combination is determined by the following formula: ; in, Energy spectrum The measured sub-projection data below; Indicates the number of iterations. Time, energy spectrum The formula shows that the residual projection combination includes the residual projections under each energy spectrum.

[0027] The method described in this embodiment of the invention is applicable to scenarios where the number of energy spectra is greater than or equal to the number of base materials.

[0028] S120. Determine the weight matrix and the constant approximation matrix of each target data block in the weight matrix. The weight matrix includes the weight of each element in the Jacobian matrix at the current base image combination. The target data block corresponds one-to-one with the energy spectrum base material combination.

[0029] The Jacobian matrix is ​​the derivative of the forward projection operator at the current combination of base images. The forward projection operator is used to describe the physical mapping process from the image domain space to the projected data domain space.

[0030] The Jacobian matrix can be represented as: ; in, , This refers to the combination of the current base images in the corresponding iteration round; Q For the number of energy spectra, For energy spectrum q The total number of all rays below; D As the amount of basic substances, I This represents the number of pixels in each base image in the current base image combination.

[0031] As can be seen from the above formula, the Jacobian matrix includes Jacobian submatrices corresponding to each energy spectrum, where the energy spectrum... q The corresponding Jacobian submatrix can be represented as: ; in, For energy spectrum , base material and radiation The weight of the corresponding element, For energy spectrum In the corresponding projection matrix, and the ray The corresponding projection data.

[0032] As can be seen from the expression of the Jacobian submatrix, each element of the Jacobian matrix is ​​the product of the weight and the projection data under the corresponding ray in the projection matrix.

[0033] The weight of each element in the weight matrix, i.e., the weight of the projection data under each ray in the Jacobian matrix, can be expressed as: ; in, For the current iteration round In the middle, energy spectrum Base image The corresponding base sine curve; This indicates the first sinusoid in the basic sine diagram. Each element.

[0034] Because of the complex structure of the Jacobian matrix, it is extremely difficult to directly invert it, and the traditional Newton method cannot directly solve this problem.

[0035] Meanwhile, during the reconstruction of spectral CT images, the combination of spectral base materials can be considered as a decomposition unit. Therefore, in this embodiment, the weight matrix is ​​divided into target data blocks corresponding to each combination of spectral base materials; wherein, one combination of spectral base materials corresponds to one energy spectrum and one base material. For example, Q individual energy spectrum and D Each basic substance corresponds to A combination of energy spectrum-based materials.

[0036] All elements in the constant approximation matrix are identical, representing the target constants. This decouples the dependencies between weights and rays in the corresponding target data block. By concatenating all constant approximation matrices into the target weight matrix, a block-based constant matrix is ​​formed. This simplifies the weight matrix and reduces the complexity of iterative image solving using the Jacobian matrix.

[0037] In this embodiment, the position range of the target data block corresponding to each energy spectrum and base material combination in the weight matrix can be predetermined based on the number of energy spectra, the number of base materials, and the number of rays. After the weight matrix is ​​updated, each target data block can be directly located in the weight matrix according to the predetermined position range.

[0038] The target constant can be determined based on the global statistics of the corresponding target data block, or it can be a specified element value in the corresponding target data block.

[0039] Understandably, if the target constants are determined based on the global statistics of the corresponding target data blocks, then the target weight matrix, which includes the approximate matrices of the constants for each target data block, discards unnecessary detail fluctuations in the weight matrix while retaining the overall information trend of the weight matrix, thus reducing the complexity of the weight matrix. Using this target weight matrix instead of the weight matrix for iterative image reconstruction can significantly reduce the amount of data computation in the image reconstruction process while ensuring the accuracy of the image reconstruction results, thereby significantly shortening the image reconstruction time.

[0040] In one embodiment, the central trend aggregation result of all elements in the target data block is determined; each element in the target data block is replaced with the central trend aggregation result to obtain a constant approximation matrix of the target data block.

[0041] Aggregation refers to the process of combining multiple data values ​​into a representative value. This representative value can reflect certain characteristics of the multiple data values ​​in a simplified form; it is an information compression process. Central tendency emphasizes that the aggregation result reflects the center or typical level of the multiple data values, rather than the degree of dispersion or extreme values. Therefore, the central tendency aggregation result of all elements in the target data block is a numerical value that can reflect the overall characteristics of all elements.

[0042] The central trend aggregation result in this embodiment can be selected as the mean, median, mode, truncated mean, truncated median, etc. of all elements in the target data block.

[0043] For example, when the central trend aggregation result is the mean of all elements in the target data block, the central trend aggregation result of the target data block can be represented as: .

[0044] When the central trend aggregation result is the median of all elements in the target data block, the central trend aggregation result of the target data block can also be expressed as: .

[0045] Replace all elements in the weight matrix corresponding to the target data block with the above central trend aggregation results. After this approximation, the weights No longer dependent on rays Therefore, it can be written as .

[0046] Since the determination processes of the constant approximation matrices for each target data block are independent, the efficiency of determining the target weight matrix can be improved by determining the constant approximation matrices for each target data block in the weight matrix in parallel.

[0047] S130. Update the current base image combination based on the current base image combination, residual projection combination, and each constant approximation matrix.

[0048] In one embodiment, the current base image combination, residual projection combination, and each constant approximation matrix are input into a pre-trained machine learning model to obtain the updated current base image combination. This embodiment completes the update of the current base image combination through a pre-trained machine learning model, which is simple and fast.

[0049] In one embodiment, a residual image combination corresponding to the residual projection combination is determined; the base image combination update amount is determined based on the residual image combination and all constant approximation matrices; and the sum of the current base image combination and the base image combination update amount is used as the updated current base image combination.

[0050] Specifically, using energy spectrum Taking the corresponding Jacobian submatrix as an example, it can be approximately expressed as: .

[0051] Therefore, the Jacobian matrix can be approximately simplified to the following block matrix: .

[0052] Here It is a block diagonal matrix composed of sub-projection matrices under each energy spectrum; It is a constant matrix independent of rays and is relatively small in size. Operators This represents block matrix multiplication. In this case, the inverse of the Jacobian matrix can be approximated as: ; in, It is a combination of CT reconstruction operators. Let be the inverse of the target weight matrix, which can be represented as: The corresponding Newton's iteration formula can be expressed as: .

[0053] in, It is a combination of residual projections.

[0054] Specifically, after the base image combination, residual projection combination, and target weight matrix are determined, the residual projection combination can be reconstructed based on the reconstruction operator to obtain the residual image. The expression is as follows: ; in, Energy spectrum The CT reconstruction operator corresponding to the scanning geometry can be selected based on the actual scanning geometry type. For example, for two-dimensional scenes, algorithms such as filtered back projection (FBP), algebraic reconstruction (ART), optimization model-based reconstruction, or deep learning reconstruction can be used. For three-dimensional cone-beam reconstruction, adaptive methods such as the FDK (Feldkamp-Davis-Kress) algorithm can be used. It is the energy spectrum The residual projection.

[0055] The specific update formula for the current base image combination can be expressed as: ;or, .

[0056] in, Represents the combination of residual images. For energy spectrum The residual image below, This represents the current base image combination, including the base images corresponding to each base substance. For the updated base image combination; The inverse matrix of a constant approximation matrix is ​​the first... OK, The element at the column.

[0057] It should be noted that different energy spectra... The corresponding scan geometry may or may not be the same; during reconstruction, the energy spectrum is used. The CT reconstruction operator (inverse projection operator) corresponding to the scan geometry is automatically corrected, and the final result is... It includes residual image components from different energy spectra, but is consistent across the energy spectrum. Consistency across the energy spectrum means that the mass attenuation coefficient vector distribution of the residual image components under each energy spectrum is theoretically identical, unaffected by differences in scanning geometry. Thus, each base image combination update does not require consideration of whether the scanning geometry is consistent.

[0058] Among them, the geometric paths of fast kilovolt switching CT and dual-source dual-detector CT differ significantly when scanning at different energy levels; similarly, in dual-energy CT systems with a cone-beam dual-panel detector structure, scanning geometry often exhibits inconsistencies under different energy spectra. Therefore, geometric inconsistency is not a phenomenon unique to a particular device, but rather one of the core problems commonly found in spectral CT technology. More importantly, the method described in this invention, as a reconstruction method capable of effectively handling geometric inconsistency scenarios, naturally possesses compatibility and applicability to geometric consistency problems, thus having broader engineering practical value.

[0059] S140. Determine the residual projection combination corresponding to the updated current base image combination. If the predetermined elements of the residual projection combination meet the iteration conditions, return to the step of determining the residual projection combination for the current base image combination until the predetermined elements do not meet the iteration conditions. Then, take the updated current base image combination as the target base image combination.

[0060] The predetermined elements of the residual projection combination include the total norm of the residual projection and the number of iterations corresponding to the residual projection. Therefore, if the total norm of the residual projection combination is greater than a predetermined residual threshold and the number of iterations corresponding to the residual projection combination is less than a predetermined step threshold, it is determined that the predetermined elements of the residual projection combination meet the iteration conditions, and the process returns to S110; until the predetermined elements of the residual projection combination no longer meet the iteration conditions, the latest current base image combination is taken as the target base image combination.

[0061] The technical solution provided by the embodiments of the present invention includes a weight matrix that includes target data blocks corresponding to each energy spectrum base material combination. The constant approximation matrix can decouple the dependency relationship between each weight and ray in the corresponding target data block. Therefore, the combination result of all constant approximation matrices can significantly reduce the complexity of the weight matrix. Therefore, image iterative reconstruction based on all constant approximation matrices can reduce the amount of data calculation in the image iterative process and improve the efficiency of image iterative solution while ensuring the image reconstruction effect.

[0062] Figure 2 This is another schematic flowchart of the decoupled ray-dependent energy-spectral CT image reconstruction method provided in this embodiment of the invention. This embodiment refines the steps for determining the constant approximation matrix in the aforementioned embodiments. The technical solution of this embodiment can be combined with other embodiments. For the same or related parts, they can be described in conjunction with the descriptions of other embodiments, and will not be repeated here. Figure 2 As shown, the method in this embodiment may specifically include: S210. Determine the residual projection combination for the current base image combination. The residual projection combination is the difference between the measured projection data and the forward projection data of the current base image combination.

[0063] S2201. Determine the weight matrix and the basis sine diagram combination corresponding to the current basis image combination. The basis sine diagram combination includes the basis sine diagram corresponding to each energy spectrum basis material combination.

[0064] Each basis sine curve is determined based on the basis images in the current basis image combination. Specifically, the basis sine curve for each energy spectrum basis material combination can be represented as follows: .

[0065] S2202. For each base sine graph in the base sine graph combination, determine the central trend aggregation result of all elements of the base sine graph.

[0066] In one embodiment, the central trend aggregation result is the mean of all elements in the base sine graph, and in this case, the central trend aggregation result of the target data block is... .

[0067] In one embodiment, the central trend aggregation result is the median of all elements in the base sine graph, and in this case, the central trend aggregation result of the target data block is... .

[0068] It should be noted that the central trend aggregation result can also be selected as the mode, truncated mean, truncated median, etc. of all elements of the base sine graph.

[0069] S2203. For each target data block, determine the target constant based on the expression of each weight of the target data block and the aggregation result of each central trend; replace each element of the target data block with the target constant to obtain the constant approximation matrix of the target data block.

[0070] When the central trend aggregation result is the mean of all elements in the basis sine plot corresponding to energy spectrum q and basis material d, the target constants of the target data block corresponding to energy spectrum q and basis material d are: .

[0071] When the central trend aggregation result is the median of all elements in the basis sine plot corresponding to energy spectrum q and basis material d, the target constants for the target data block corresponding to energy spectrum q and basis material d are: .

[0072] Once the target constant for each target data block is determined, each element of the block is replaced with the target constant to obtain the corresponding constant approximation matrix.

[0073] S230. Update the current base image combination based on the current base image combination, residual projection combination, and each constant approximation matrix.

[0074] S240. If the residual projection combination of the updated current base image combination meets the iteration condition, return to the step of determining the residual projection combination for the current base image combination until the latest residual projection combination does not meet the iteration condition, and take the updated current base image combination as the target base image combination.

[0075] The technical solution provided by the embodiments of the present invention has two advantages. First, since the amount of data in the base sine graph is relatively small, the central trend aggregation result corresponding to each base sine graph can be determined relatively quickly. Since the central trend aggregation result is a definite value, the target constant for the corresponding target data block can be determined quickly based on it, thereby quickly determining the constant approximation matrix of the corresponding target data block. Second, since the base sine graph is closer to the current base image and reflects the projection value of the ray after passing through the object under the current base image combination, its central trend aggregation result represents the overall projection intensity under the corresponding energy spectrum base material combination. Therefore, the constant approximation matrix of the target data block determined based on the central trend aggregation result of the base sine graph has high accuracy.

[0076] It should be noted that the embodiments of the present invention only exemplify the scheme for determining the target constant for the corresponding target data block based on the center aggregation result of the base sine graph. However, in actual use, other methods can be used to process the base sine graph to determine the target constant for the corresponding target data block. For example, the value at a predetermined position in the base sine graph can be used as the target constant, or the center trend aggregation result of all elements within a predetermined area of ​​the base sine graph can be used as the target constant, etc.

[0077] Figure 3 This is another schematic flowchart illustrating the decoupled X-ray dependent energy-spectral CT image reconstruction method provided in this embodiment of the invention. Parts that are the same as or related to the foregoing embodiments can be described in conjunction with the descriptions of other embodiments, and will not be repeated here. Figure 3 As shown, the method in this embodiment may specifically include: S310. Obtain the initial base image combination and the measured projection data, and use the initial base image combination as the current base image combination.

[0078] S320. Determine the forward projection data of the current base image combination based on the forward projection operator, and use the difference between the measured projection data and the forward projection data as the residual projection combination.

[0079] S330. Determine the combination of residual images corresponding to the combination of residual projections.

[0080] S340. Determine the weight matrix and the constant approximation matrix corresponding to each target data block of the weight matrix. The weight matrix includes the weights of each element in the Jacobian matrix at the current base image combination.

[0081] S350. Based on the current base image combination, residual projection combination, and all constant approximation matrices, update the current base image combination.

[0082] S360. Determine whether the predetermined elements of the residual projection combination meet the iteration conditions. If they do, increment the iteration count and return to S320 and S340. If they do not meet the conditions, execute S370.

[0083] S370. Use the current base image combination as the target base image combination.

[0084] In this embodiment, steps S320 and S340 can be executed in parallel or sequentially. Therefore, when the total norm of the residual projection combination is detected to be greater than a predetermined norm threshold, and the corresponding iteration count is less than a predetermined step threshold, the iteration condition is deemed met, the iteration count is incremented by 1, and the process returns to steps S320 and S340, i.e., steps S320 and S340 are executed in parallel. Then, step S360 determines whether the predetermined elements of the residual projection combination determined by step S320 meet the iteration condition. This process continues until the predetermined elements of the residual projection combination no longer meet the iteration condition, at which point the current base image combination is taken as the target base image combination.

[0085] The technical solution provided in this embodiment of the invention allows S320 and S340 to be executed in parallel. The residual projections under each energy spectrum can be executed in parallel during the determination of the residual projection combination. S340 can determine the constant approximation matrix under each energy spectrum base material combination in parallel. Therefore, this embodiment can quickly and accurately complete the determination of the target base image combination.

[0086] Taking the noiseless dual-energy spectral projection data of the Forbild head phantom as an example, the performance of the method described in this embodiment of the invention in terms of convergence efficiency and reconstruction accuracy is verified. The projection data acquisition conditions include: a low-energy spectrum with a tube voltage of 80 kVp and a high-energy spectrum with a tube voltage of 140 kVp; the high-energy X-ray is supplemented with a 1 mm thick copper filter; the phantom is composed of water-based and bone-based components, with a spatial distribution range of [-5 cm, 5 cm] × [-5 cm, 5 cm], and an image resolution of 128 × 128 pixels. Both the low-energy and high-energy spectra use parallel beam scanning geometry, and the specific parameter settings are as follows: the high-energy spectrum has 384 scanning angles, uniformly distributed in the range of 0° to 180°; the number of detector elements is 384, uniformly arranged in the interval of [7.05 cm, 7.05 cm]. The detector configuration parameters for the low-energy spectrum are the same as those for the high-energy spectrum, but its scanning angle is offset from the high-energy spectrum by half the minimum angle, and is uniformly distributed between 0.2344° and 180.2344°. The detector has 320 elements, arranged in the same range as the high-energy spectrum.

[0087] By substituting the energy spectrum distribution, the mass attenuation coefficient of the matrix material, the discrete X-ray transformation operator, and the real image corresponding to the head phantom into the imaging model, noise-free dual-energy projection data is finally generated, and this dual-energy projection data is used as the measured projection data.

[0088] In this embodiment, the real images, the corresponding bone-based images, and the water-based images are all from publicly available image sets. The 60keV virtual monoenergetic images and 100keV virtual monoenergetic images are based on the bone-based images, the water-based images, and the aforementioned formulas. Sure.

[0089] Figure 4A A graph showing the change of the relative error of the base image with the number of iterations, provided in an embodiment of the present invention. Figure 4B A graph showing the relative error of the projection data as a function of the number of iterations, provided in an embodiment of the present invention.

[0090] Specifically, Figure 4A The graphs show the relative error of the base image as a function of the number of iterations during the first 500 iterations for four energy spectrum CT image reconstruction methods corresponding to four constant approximation matrices. Figure 4BThe graphs show the relative error of the projection data as a function of the number of iterations in the first 500 iterations for four energy spectrum CT image reconstruction methods corresponding to four constant approximation matrices. Figure 4A and Figure 4B The red dashed line, blue solid line, black dashed line, and cyan solid line in the diagram correspond to the first constant approximation matrix, the second constant approximation matrix, the third constant approximation matrix, and the fourth constant approximation matrix, respectively. These matrices are determined based on the first central trend aggregation results, the second central trend aggregation results, the third central trend aggregation results, and the fourth central trend aggregation results, respectively. The first central trend aggregation result is the mean of all elements in the base sine graph; the second central trend aggregation result is the median of all elements in the base sine graph; the third central trend aggregation result is the mean of all weights in the target data block; and the fourth central trend aggregation result is the median of all weights in the target data block.

[0091] from Figure 4A and Figure 4B It can be seen that the energy spectrum CT image reconstruction results corresponding to these four constant approximation matrices all exhibit stable convergence characteristics. Specifically, the relative error of the base image reaches 10. -6 The relative error of the projected data reached 10 on the order of magnitude. -7 The magnitudes are as follows. Among them, the first constant approximation matrix, the second constant approximation matrix, and the third constant approximation matrix all converge within 40 iterations, while the fourth constant approximation matrix converges within 500 iterations.

[0092] The relative error of the base image is determined by the following formula: ; in, For the first The base image determined by the next iteration, and the virtual monoenergetic image at the corresponding energy. For the real base image, the real virtual monoenergetic image under the corresponding energy; This represents the Frobenius norm of the matrix.

[0093] The relative error of the projected data is determined by the following formula: ; in, For energy spectrum The measured projection data below, For the number of iterations Time, energy spectrum Sub-forward projection data; For energy spectrum The residual data corresponding to the current base image.

[0094] Figure 5 This is a schematic diagram illustrating the energy spectrum CT image reconstruction results corresponding to the four constant approximation matrices provided in this embodiment of the invention. Specifically, Figure 5 The first row in the figure is the real image, and the second to fifth rows are the energy spectrum CT image reconstruction results generated by 500 iterations based on the first constant approximation matrix, the second constant approximation matrix, the third constant approximation matrix and the fourth constant approximation matrix, respectively. In the figure, from left to right, there are water-based image, bone-based image, 60keV virtual monoenergetic image and 100keV virtual monoenergetic image.

[0095] from Figure 5 It can be seen that the energy spectrum CT image reconstruction results determined by the constant approximation matrices are visually highly consistent with the real images, with accurate detail restoration and no obvious artifacts.

[0096] To further quantify the quality of image reconstruction results, determine Figure 5 The root mean square error, structural similarity, and peak signal-to-noise ratio of each image are shown in Table 1.

[0097] As shown in Table 1, the method described in this embodiment of the invention demonstrates outstanding performance in both computational efficiency and reconstruction accuracy. Regarding computational efficiency, the single iteration time corresponding to each constant approximation matrix is ​​only 0.04-0.05 seconds. The image reconstruction methods corresponding to the first, second, and third constant approximation matrices only require 40 iterations to converge, and the total time for reconstructing a complete set of images is approximately 2 seconds, demonstrating excellent real-time processing potential. In terms of reconstruction accuracy, the normalized root mean square error of the image reconstruction results corresponding to the four constant approximation matrices is as low as 10. -7 The reconstruction quality is near perfect, with structural similarity close to the ideal value of 1 and peak signal-to-noise ratio generally exceeding 120 dB. In summary, the method described in this invention achieves rapid iterative convergence while ensuring high reconstruction accuracy, making it a practical solution that combines high precision and high efficiency.

[0098] Table 1. Results of Evaluation Indicator Data Display The normalized root mean squared error (NRMSE) is defined as follows: ; in, For real image vectors, To reconstruct the image vector.

[0099] The local similarity index is defined as: ; in, and windows respectively The mean and variance, for and covariance, constant .

[0100] Structural similarity (SSIM) is defined as: ; in, Corresponding to and No. indivual Local window, Indicates the total number of windows.

[0101] Peak signal-to-noise ratio (PSNR) is defined as: ; in, The total number of pixels in the image. For real image vectors The infinite norm (i.e., the maximum pixel value).

[0102] Next, another energy spectrum CT image reconstruction experiment was conducted to compare the advantages and disadvantages of the method described in this embodiment of the invention with existing methods. The phantom images and energy spectrum parameters used in the experiment remained consistent with the previous experiment, but the geometric scanning settings were adjusted. Specifically, the low-energy spectrum had 384 scanning angles, uniformly distributed within the range of 0° to 180°, with 384 detector elements uniformly arranged in the interval [-7.05cm, 7.05cm]. The high-energy spectrum had 128 scanning angles, uniformly distributed within the range of 0.2344° to 180.2344°, with the same detector configuration as the low-energy spectrum. By substituting the energy spectrum distribution, the mass attenuation coefficient distribution of the substrate material, the discrete X-ray transformation operator, and the actual phantom image into the forward imaging model, corresponding noise-free dual-energy spectrum projection data was generated, and this dual-energy spectrum projection data was used as the measured projection data.

[0103] To comprehensively evaluate the performance of this method in terms of geometric applicability, reconstruction accuracy, numerical stability and computational efficiency, two representative advanced algorithms were selected for comparative analysis: the Iterative Filter Back Projection (IFBP) algorithm and the Accurate Fast Image Reconstruction (AFIRE) algorithm.

[0104] In this experiment, the third constant approximation matrix was selected as an example for comparing image reconstruction results. During the intermediate image reconstruction process using each algorithm, 1000 iterations were performed, and curves showing the relative error of the base image and the relative error of the projected data as a function of the number of iterations were plotted. Figure 6A and 6B As shown, the image reconstruction process based on the third constant approximation matrix and the image reconstruction process based on the AFIRE algorithm exhibit similar performance in terms of the relative error of the final base image and the relative error of the projection data. Specifically, the relative error of the base image in both processes can reach 10. -6 The relative error of the projected data can reach 10 on the order of magnitude. -7 In terms of convergence efficiency, the image reconstruction algorithm based on the third constant approximation matrix is ​​significantly better than AFIRE. When reconstructing images based on the third constant approximation matrix, it tends to stabilize after approximately 500 iterations, while AFIRE requires approximately 700 iterations to converge. In contrast, the IFBP algorithm experiences a brief decrease in error at the beginning of the iterations, followed by a rapid increase and maintenance at a high level, failing to achieve effective convergence.

[0105] Figure 7A The graph shows the relative error of the base image over time, as provided in an embodiment of the present invention. Figure 7B A graph showing the relative error of projection data over time, provided in an embodiment of the present invention.

[0106] Depend on Figure 7A and Figure 7B It can be seen that the image reconstruction time required based on the third constant approximation matrix is ​​about 20 seconds, which is significantly less than the 27 seconds required for AFIRE image reconstruction. When reconstructing the image based on the third constant approximation matrix, it takes about 13 seconds to complete the convergence of the projection data, which is significantly less than the 20 seconds required to complete the convergence of the projection data using AFIRE. Therefore, the image reconstruction algorithm based on the third constant approximation matrix is ​​significantly better than the AFIRE algorithm in terms of computational efficiency.

[0107] The reconstruction results of each algorithm are as follows Figure 8 As shown in the figure, the first row is the real image; the second to fourth rows correspond to the image reconstruction results of the image reconstruction algorithms corresponding to the third constant approximation matrix after 1000 iterations, namely the IFBP algorithm and the AFIRE algorithm; each row from left to right corresponds to the water-based image, the bone-based image, the 60 keV virtual monoenergetic image, and the 100 keV virtual monoenergetic image.

[0108] As can be seen from the figure, the image reconstruction results corresponding to the third constant approximation matrix and the AFIRE algorithm are extremely close to the real images. In contrast, although the image reconstruction results of the IFBP algorithm roughly outline the structural contours, they suffer from severe stripe artifacts and deviate significantly from the real images.

[0109] To further quantify and evaluate the reconstruction quality, calculations were performed separately. Figure 8 The normalized root mean square error, structural similarity, and peak signal-to-noise ratio of each image reconstruction result are calculated and shown in Table 2.

[0110] Table 2 shows the results of the evaluation index data. As shown in Table 2, the normalized root mean square error of the image reconstruction result determined based on the third constant approximation matrix reaches 10. -6 The magnitude of the structural similarity is very close to the ideal value of 1, and the peak signal-to-noise ratio is higher than 118dB, placing it on the same order of magnitude as the AFIRE algorithm. Both are significantly better than the IFBP algorithm. This fully demonstrates that the energy spectrum CT image reconstruction method based on the third constant approximation matrix still possesses excellent numerical stability and anti-interference capabilities even when data is relatively scarce. In terms of computational efficiency, the image reconstruction method corresponding to the third constant approximation matrix takes 0.042 seconds per iteration, comparable to AFIRE's 0.036 seconds per iteration, but requires fewer iterations to converge, resulting in a shorter overall reconstruction time. In contrast, IFBP fails to converge effectively, and its reconstruction results show significant shortcomings in accuracy.

[0111] Finally, the robustness of the reconstruction method described in this embodiment under noise interference is illustrated through a noisy dual-energy CT image reconstruction experiment. The phantom images, energy spectrum parameters, and geometric configurations used in the experiment were consistent with those in Experiment 2. First, the energy spectrum distribution, the mass attenuation coefficient distribution of the substrate material, the discrete X-ray transform operator, and the real image were substituted into the imaging model to generate noise-free dual-energy projection data. Then, Gaussian white noise was added to the noise-free data to finally obtain noisy (noise level of 30 dB) dual-energy projection data, which was used for subsequent reconstruction performance verification.

[0112] In this experiment, the third constant approximation matrix was selected as a representative matrix and compared with the IFBP and AFIRE algorithms. All algorithms involved in the comparison were iterated for 400 steps, and the curves showing the relative error of the base image as a function of the number of iterations were plotted. Figure 9A The curve showing the relative error between the projection data and the number of iterations. Figure 9B The graph also shows the variation of the difference metric between two adjacent steps of the base image with the number of iterations. Figure 10A ), and a graph showing the change in the difference between adjacent steps of the projected data with the number of iterations ( Figure 10B ).

[0113] from Figure 9A and Figure 9BIt can be seen that the energy spectrum CT image reconstruction results based on the third constant approximation matrix are similar to those determined by the AFIRE algorithm in terms of final relative error accuracy. Specifically, the relative error of the base images for both can reach 10. -2 The relative error of the projected data can reach 10 on the order of magnitude. -3 In terms of convergence efficiency, the energy spectrum CT image reconstruction method based on the third constant approximation matrix outperforms AFIRE. The energy spectrum CT image reconstruction method based on the third constant approximation matrix stabilizes after approximately 60 iterations, while AFIRE requires approximately 100 iterations to converge. In contrast, the IFBP algorithm experiences a brief decrease in error at the beginning of the iterations, followed by a rapid increase, failing to achieve effective convergence.

[0114] from Figure 10A and Figure 10B It can be seen that both the energy spectrum CT image reconstruction algorithm based on the third constant approximation matrix and the AFIRE algorithm eventually achieved stable convergence, and the difference metric between adjacent steps of the base image reached 10. -6 The difference between adjacent steps of the projected data both reached 10. -7 This indicates that the energy spectrum CT image reconstruction algorithm based on the third constant approximation matrix possesses excellent numerical stability and noise resistance. The IFBP algorithm exhibits a high level of difference between adjacent steps and does not show a clear convergence trend.

[0115] Figure 11 This is another schematic diagram of the image combination result provided in an embodiment of the present invention. In this figure, the first row is the real image; the second to fourth rows are, respectively, the image reconstruction result determined based on the third constant approximation matrix, the reconstruction result after 100 iterations of the IFBP algorithm and the AFIRE algorithm; each row corresponds from left to right to the water-based image, the bone-based image, the 60keV virtual monoenergetic image, and the 100keV virtual monoenergetic image. Figure 11 As can be seen, the image reconstruction results determined by the third constant approximation matrix and the AFIRE reconstruction results are almost completely consistent with the real image, with excellent noise suppression and clear detail restoration. However, the reconstructed image obtained by the IFBP algorithm has obvious stripe artifacts.

[0116] Using normalized root mean square error and other quantitative indicators to... Figure 11 The reconstruction results of each image were evaluated. Specific quantitative indicators are shown in Table 3.

[0117] Table 3. Results of Evaluation Indicator Data Display As shown in Table 3, under noisy imaging conditions, the energy spectrum image reconstruction algorithm corresponding to the third constant approximation matrix still exhibits robust reconstruction performance highly similar to the AFIRE algorithm, while maintaining an advantage in computational efficiency. Regarding noise resistance and reconstruction accuracy, the normalized root mean square error of the energy spectrum image reconstruction algorithm corresponding to the third constant approximation matrix remains at 10. -3 Up to 10 -2 The order of magnitude, the structural similarity compensation term is in the range of 10. -3 Up to 10 -1 The peak signal-to-noise ratio (SNR) is between 36 dB and 56 dB, and all indicators are almost identical to the AFIRE algorithm, with both significantly outperforming the IFBP algorithm. This indicates that the energy spectrum image reconstruction algorithm corresponding to the third constant approximation matrix can maintain stable decomposition capability and image quality even under noise interference. In terms of time efficiency, the energy spectrum image reconstruction algorithm corresponding to the third constant approximation matrix has a single iteration time of 0.042 seconds, which is comparable to the 0.039 seconds of the AFIRE algorithm. Moreover, thanks to its faster convergence characteristics, the overall reconstruction time is more advantageous.

[0118] In summary, in noisy real-world imaging scenarios, the algorithm of this invention can rival the current advanced AFIRE method in reconstruction accuracy, while exhibiting superior convergence efficiency, achieving a technical effect that balances image reconstruction quality and efficiency.

[0119] Figure 12 This is a schematic diagram of the structure of a decoupled X-ray dependent spectral CT image reconstruction device provided in an embodiment of the present invention. This decoupled X-ray dependent spectral CT image reconstruction device is disposed within an electronic device and is executed by the electronic device. The electronic device can be a workstation or server of a spectral CT device, or other data processing equipment. Figure 12 As shown, the decoupled X-ray dependent spectral CT image reconstruction device includes: The residual projection module 410 is used to determine the residual projection combination for the current base image combination, wherein the residual projection combination is the difference between the measured projection data and the forward projection data of the current base image combination; The weight matrix module 420 is used to determine the weight matrix and the constant approximation matrix of each target data block in the weight matrix. The weight matrix includes the weight of each element in the Jacobian matrix at the current base image combination. The target data block corresponds one-to-one with the energy spectrum base material combination. The base image update module 430 is used to update the current base image combination based on the current base image combination, the residual projection combination, and each of the constant approximation matrices; The judgment module 440 is used to determine the residual projection combination corresponding to the updated current base image combination. If the predetermined elements of the residual projection combination meet the iteration conditions, the module returns to the step of determining the residual projection combination for the current base image combination until the predetermined elements do not meet the iteration conditions, and then the updated current base image combination is used as the target base image combination.

[0120] The technical solution provided by the embodiments of the present invention includes a weight matrix that includes target data blocks corresponding to each energy spectrum base material combination. The constant approximation matrix can decouple the dependency relationship between each weight and ray in the corresponding target data block. Therefore, the combination result of all constant approximation matrices can significantly reduce the complexity of the weight matrix. Therefore, image iterative reconstruction based on all constant approximation matrices can reduce the amount of data calculation in the image iterative process and improve the efficiency of image iterative solution while ensuring the image reconstruction effect.

[0121] In one embodiment, the constant approximation matrix is ​​determined based on global statistics of the corresponding target data block.

[0122] In one embodiment, the weight matrix module 420 determines the constant approximation matrix of each target data block in the weight matrix through a constant approximation unit. The constant approximation unit is used for: For the target data block corresponding to each of the energy spectrum base material combinations in the weight matrix, determine the central trend aggregation result of all elements in the target data block; replace each element in the target data block with the central trend aggregation result to obtain the constant approximation matrix of the target data block.

[0123] In one embodiment, the weight matrix module 420 determines the constant approximation matrix of each target data block in the weight matrix through a constant approximation unit. The constant approximation unit is used for: Determine the basis sine diagram combination corresponding to the current basis image combination, wherein the basis sine diagram combination includes the basis sine diagrams corresponding to each of the energy spectrum basis material combinations; For each of the base sine graphs in the base sine graph combination, determine the central trend aggregation result of all elements of the base sine graph; For each target data block, a target constant is determined based on the expression of each weight of the target data block and the aggregation result of each central trend; each element of the target data block is replaced with the target constant to obtain the constant approximation matrix of the target data block.

[0124] In one embodiment, the central trend aggregation result is the mean or median.

[0125] In one embodiment, the base image update module 430 is used for: Determine the residual image combination corresponding to the residual projection combination; The base image combination update amount is determined based on the residual image combination and all the constant approximation matrices; The sum of the current base image combination and the update amount of the base image combination is used as the updated current base image combination.

[0126] In one embodiment, the predetermined elements of the residual projection combination include the total norm of the residual projection and the number of iterations corresponding to the residual projection combination.

[0127] The decoupled X-ray dependent spectral CT image reconstruction device provided in the embodiments of the present invention can execute the decoupled X-ray dependent spectral CT image reconstruction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0128] It is worth noting that the various units and modules included in the above-mentioned decoupled X-ray dependent spectral CT image reconstruction device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of the present invention.

[0129] Figure 13 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0130] like Figure 13As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0131] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0132] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as decoupling X-ray dependent spectral CT image reconstruction methods.

[0133] In some embodiments, the decoupled X-ray-dependent spectral CT image reconstruction method can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via read-only memory (ROM) 12 and / or communication unit 19. When the computer program is loaded into random access memory (RAM) 13 and executed by processor 11, one or more steps of the decoupled X-ray-dependent spectral CT image reconstruction method described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the decoupled X-ray-dependent spectral CT image reconstruction method by any other suitable means (e.g., by means of firmware).

[0134] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0135] Computer programs for implementing the decoupled X-ray-dependent energy-spectral CT image reconstruction method described in embodiments of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer programs can be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0136] This invention provides a computer-readable storage medium storing computer instructions for causing a processor to execute the decoupled X-ray dependent energy spectrum CT image reconstruction method described in any embodiment.

[0137] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0138] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0139] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0140] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0141] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication unit 19, or installed from storage unit 18, or installed from ROM 12. When the computer program is executed by processor 11, it performs the functions defined in the methods of the embodiments of the present invention.

[0142] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the decoupled X-ray dependent energy spectrum CT image reconstruction method according to any embodiment of the invention.

[0143] In implementing the computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0144] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0145] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for decoupling X-ray dependent energy-spectral CT image reconstruction, characterized in that, The method includes: Determine the residual projection combination for the current base image combination, wherein the residual projection combination is the difference between the measured projection data and the forward projection data of the current base image combination; Determine the weight matrix and the constant approximation matrix of each target data block in the weight matrix. The weight matrix includes the weight of each element in the Jacobian matrix at the current base image combination. The target data block corresponds one-to-one with the energy spectrum base material combination. The current base image combination is updated based on the current base image combination, the residual projection combination, and each of the constant approximation matrices; Determine the residual projection combination corresponding to the updated current base image combination. If the predetermined elements of the residual projection combination meet the iteration conditions, return to the step of determining the residual projection combination for the current base image combination until the predetermined elements do not meet the iteration conditions. Then, use the updated current base image combination as the target base image combination.

2. The method according to claim 1, characterized in that, The constant approximation matrix is ​​determined based on the global statistics of the corresponding target data block.

3. The method according to claim 2, characterized in that, The constant approximation matrix for each target data block in the weight matrix is ​​determined by the following steps: For the target data block corresponding to each of the energy spectrum base material combinations in the weight matrix, determine the central trend aggregation result of all elements in the target data block; replace each element in the target data block with the central trend aggregation result to obtain the constant approximation matrix of the target data block.

4. The method according to claim 2, characterized in that, The constant approximation matrix for each target data block in the weight matrix is ​​determined by the following steps: Determine the basis sine diagram combination corresponding to the current basis image combination, wherein the basis sine diagram combination includes the basis sine diagrams corresponding to each of the energy spectrum basis material combinations; For each of the base sine graphs in the base sine graph combination, determine the central trend aggregation result of all elements of the base sine graph; For each target data block, a target constant is determined based on the expression of each weight of the target data block and the aggregation result of each central trend; each element of the target data block is replaced with the target constant to obtain the constant approximation matrix of the target data block.

5. The method according to claim 3 or 4, characterized in that, The central trend aggregation result is the mean or median.

6. The method according to claim 1, characterized in that, The step of updating the current base image combination based on the current base image combination, the residual projection combination, and each of the constant approximation matrices includes: Determine the residual image combination corresponding to the residual projection combination; The base image combination update amount is determined based on the residual image combination and all the constant approximation matrices; The sum of the current base image combination and the update amount of the base image combination is used as the updated current base image combination.

7. The method according to claim 1, The predetermined elements of the residual projection combination include the total norm of the residual projection combination and the number of iterations corresponding to the residual projection combination.

8. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the decoupled ray-dependent energy spectral CT image reconstruction method as described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the decoupled ray-dependent energy spectrum CT image reconstruction method according to any one of claims 1-7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the decoupled ray-dependent energy spectrum CT image reconstruction method according to any one of claims 1-7.

Images