Method for improving metal artifacts in energy spectrum CT image domain

Abstract

The invention provides a method for improving metal artifacts in an energy spectrum CT image domain, and belongs to the technical field of CT imaging, when a target (such as a patient) with a metal implant is imaged, the metal artifacts can still be observed through a virtual monoenergetic diagram obtained by current equipment, and particularly, the virtual monoenergetic diagram is more obvious at low energy (low keV), so that the metal artifacts can be effectively improved. According to the method, a relation model of a base material and an artifact-free (or low-artifact) image (such as a high-energy virtual single-energy diagram) is constructed through extraction of a non-artifact region (or low-artifact region) in a virtual single-energy diagram with artifacts and base material decomposition (such as decomposition images of a water base and a bone base); substituting a pixel (voxel) CT value corresponding to an artifact area in the artifact-containing image in the artifact-free (or low-artifact) single-energy image into the relation model to obtain a new base material decomposition, and further synthesizing an artifact-corrected virtual single-energy image under any energy; according to the method, the metal artifacts in the virtual monoenergy diagram of existing CT equipment which is not limited by manufacturers and models can be effectively inhibited.

Description

Technical Field

[0001] The invention belongs to the technical field of CT imaging, and in particular relates to a method for improving metal artifacts in a virtual monoenergetic image in a spectral CT image domain. Background Art

[0002] Metal artifacts are one of the main artifacts in clinical CT, which will reduce image quality. The causes of metal artifacts include beam hardening, scattering and photon starvation. Existing metal artifact removal methods can be roughly divided into several categories. In terms of scanning method, high-voltage scanning can be used; in terms of algorithm, there is the classic NMAR (Normalized Metal Artifact Reduction). The core of this type of algorithm is projection and interpolation, which can be combined with iteration and other methods to optimize and obtain better images; with the widespread application of deep learning in recent years, deep learning methods have also been applied to metal artifact removal; in addition, virtual monoenergetic images under high energy of dual-energy CT can effectively reduce metal artifacts.

[0003] There are several ways to implement dual-energy CT in the world, including Philips' double-layer detector CT, Siemens' dual-source dual-detector CT, GE Medical's fast high-voltage switching CT, and Canon's 2-time scanning method. All dual-energy CTs produce single-energy spectrum images, also known as virtual monoenergetic images. Virtual monoenergetic images at high energy have good metal artifact suppression effects, but in virtual monoenergetic images at low energy, obvious metal artifacts are still common.

[0004] In response to this phenomenon, different manufacturers have developed corresponding metal artifact correction algorithms (MAR, Metal Artifact Reduction) for their own CT products, and metal artifacts are removed again from virtual monoenergetic images. However, the core of the MAR algorithm is polynomial interpolation. If the shape of the metal is regular and the artifact is small, the MAR algorithm will have a significant correction effect on metal artifacts. On the contrary, for images with irregular metal shapes and relatively complex artifact distribution, the MAR correction effect will be poor, and even new artifacts will be introduced, such as CT after pedicle screw placement, and CBCT or CT after denture placement. Summary of the invention

[0005] In order to solve the defects of the prior art, the present invention provides a method for improving metal artifacts in the spectral CT image domain, which can effectively suppress and improve the metal artifacts in the virtual monoenergetic images of any manufacturer based on the image domain.

[0006] The present invention adopts the following technical solutions to solve the above problems:

[0007] A method for improving metal artifacts in a spectral CT image domain, the method comprising the following steps:

[0008] S1: Input n virtual monoenergetic images of arbitrary energy and select the monoenergetic image with significant artifacts (denoted as Image_Worst);

[0009] S2: Identify the artifact area and the non-artifact area in the above artifact image, and further obtain the artifact area mask (recorded as ImageMask1) and the non-artifact area mask (recorded as ImageMask0);

[0010] S3: Select an image without artifacts or with small artifacts (referred to as Image_Best), and derive two new images from Image_Best according to the two masks obtained in S2, one containing only the pixels or voxels of Image_Best in the selected area of ​​ImageMask1 mask (referred to as Image_Best_Artifact), and the other containing only the pixels or voxels of Image_Best in the selected area of ​​ImageMask0 mask (referred to as Image_Best_noArtifact);

[0011] S4: Select two virtual monoenergetic images with energy differences from S1 to perform dual-energy decomposition to obtain two or more base material images;

[0012] S5: obtaining an image composed of pixels or voxels in the non-artifact area of ​​two or more base material images by comparing with ImageMask0;

[0013] S6: Based on the Image_Best_noArtifact image in S3 and the image composed of pixels or voxels in the non-artifact area obtained in S5, a relationship model between the Image_Best_noArtifact image and two or more of the base material graphs is constructed;

[0014] S7: Substitute the Image_Best_Artifact image obtained in S3 into the relationship model established in S6 to obtain a component map of the artifact area on two or more base materials, that is, a corrected image;

[0015] S8: synthesize the non-artifact area image of the base material obtained in S5 and the corrected artifact area image of the base material obtained in S7. Through the corrected base material image, an artifact-corrected energy spectrum image at any energy can be synthesized.

[0016] Furthermore, in S8, the energy spectrum image includes but is not limited to virtual monochrome images, material density images, effective atomic number images, electron density images, virtual plain scan images, iodine images and other energy spectrum images, and can also improve the accuracy of non-image energy spectrum decomposition results such as energy spectrum curves and scatter plots.

[0017] Furthermore, in S1, the energy range of the virtual monoenergetic image can be but is not limited to 40 to 140 keV. Any artifact correction algorithm can be used to perform artifact correction on the image to obtain a mask of the artifact area, or the artifact area or non-artifact area of ​​the image can be directly identified. The method includes but is not limited to traditional metal artifact removal algorithms based on interpolation and reconstruction, such as MAR, fsMAR, iMAR, OMAR, sMAR, etc., and is not limited to MAR algorithms based on deep learning or directly using deep learning and other methods to identify artifact areas or non-artifact areas.

[0018] Furthermore, when extracting the artifact area, a metal threshold is set for image threshold segmentation, and the image is traversed point by point. The image CT value greater than the threshold is set to 1, and less than the threshold is set to 0, thereby segmenting the metal area and the non-metal area.

[0019] Furthermore, in S2, a threshold segmentation method is used to obtain an artifact area mask.

[0020] Furthermore, in S3, the image without artifacts or with relatively small artifacts (Image_Best) is selected from n virtual monoenergetic images of arbitrary energy in S1, or an image reconstructed by other means, including but not limited to images reconstructed by other sources (such as different image layers, data after coarse correction of artifacts, etc.) or other means.

[0021] Furthermore, in S4, a relationship model between CT value and water / bone base image can be constructed, which can be extended to the relationship model between CT value and any base material pair image, as well as the relationship model between any other different types of energy spectrum images.

[0022] Furthermore, this algorithm selects (but is not limited to) 6 images, and the input single energy images are 50keV, 70keV, 80keV, 100keV, 120keV, and 140keV.

[0023] Furthermore, the method of creating an artifact-free base material image and establishing a model using two single-energy images of the non-artifact area can be extended to using other images in the image sequence (either images containing metal or not) and establishing a model, and the two are equivalent.

[0024] Furthermore, the method is applicable to improving artifacts of two-dimensional images, and can also be directly extended to improve artifacts of three-dimensional images.

[0025] Furthermore, in addition to constructing a correction model for the current data for each set of input data, the method can also pre-construct a universal model relationship between the monoenergetic image and the basis material component. In this case, one correction model can be applied to all input data (including but not limited to constructing a universal model for different scanning sites, the presence or absence of contrast agents, etc.).

[0026] Furthermore, the method for establishing or optimizing the model includes but is not limited to polynomial fitting, deep learning (neural network), pattern recognition, etc.

[0027] Furthermore, the model in the method can be used to optimize artifact types, including but not limited to metal artifacts, bone artifacts, water hardening artifacts, etc.

[0028] Furthermore, the model in the method is not limited to the energy spectrum image modeling, and the same modeling can also be performed on non-energy spectrum images.

[0029] The beneficial effects of the present invention are:

[0030] 1. The present invention has developed an algorithm that can effectively suppress and improve metal artifacts in virtual mono-energetic images of any manufacturer based on the image domain, and can make the metal artifact suppression effect of mono-energetic images at low energy consistent with that at high energy;

[0031] 2. This algorithm does not introduce new artifacts. It is particularly effective for situations with complex artifacts and metal shapes, making up for the shortcomings of traditional MAR algorithms (but not limited to MAR);

[0032] 3. The case of the present invention shows the establishment of a model between base materials and CT values. Based on the existing results, a relationship model between base materials can also be directly established. The two are equivalent.

[0033] 4. There can be n kinds of decomposition base materials, but the case presented in the present invention uses two kinds of decomposition base materials in order to explain the problem concisely. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the specific embodiments of the present invention, the drawings required for use in the description of the specific embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 It is the algorithm flow chart of the present invention;

[0036] Figure 2 An example of the complete artifact image extraction process;

[0037] Figure 3 A schematic diagram of the steps for model construction;

[0038] Figure 4 It is a virtual monoenergetic image of the phantom test result;

[0039] Figure 5 Virtual monoenergetic images of pedicle screw placement patient data test results;

[0040] Figure 6 Virtual monoenergetic image of the test results of the patient data for oral prosthesis implants. DETAILED DESCRIPTION

[0041] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0042] It should be noted that the method for extracting artifacts in the example of the present invention is MAR, but is not limited to MAR; in addition, the case of the present invention shows the establishment of a model between base materials and CT values. Based on the existing results, a relationship model between base materials can also be directly established, and the two are equivalent.

[0043] It should be noted that the method of creating an artifact-free water-bone image and establishing a model using two single-energy images of the non-artifact area in the example of the present invention can be extended to using other images in the image sequence (either images containing metal or not) and establishing a model, and the two are equivalent.

[0044] It should be noted that the relationship model between the CT value and the water / bone base image in the present invention can be extended to the relationship model between the CT value and any base material pair image, as well as the relationship model between any other different types of energy spectrum images.

[0045] It should be noted that this method is applicable to artifact improvement of two-dimensional images, and can also be directly extended to artifact improvement of three-dimensional images.

[0046] It should be noted that in addition to constructing a correction model for the current data for each set of input data, this method can also pre-construct a universal model relationship between the monoenergetic image and the basis material component. In this case, one correction model can be applied to all input data (including but not limited to constructing a universal model for different scanning sites, the presence or absence of contrast agents, etc.).

[0047] It should be noted that the model building or optimization methods in this method include but are not limited to polynomial fitting, deep learning (neural network), pattern recognition, etc.

[0048] It should be noted that the model in this method can be used to optimize artifact types, including but not limited to metal artifacts, bone artifacts, water hardening artifacts, etc.

[0049] It should be noted that the model in this method is not limited to energy spectrum image modeling, and the same modeling can also be performed on non-energy spectrum images.

[0050] For ease of understanding, the specific meanings of the proper nouns used in this application are shown in the proper noun comparison table:

[0051]

[0052] like Figure 1 As shown, dual-energy CT can usually generate n virtual single-energy images of any energy at 40-140keV. Here, it is assumed that the single-energy images output by the product are 50keV, 70keV, 80keV, 100keV, 120keV, and 140keV, a total of 6 images. The present invention proposes a method for improving metal artifacts in the energy spectrum CT image domain, comprising the following steps:

[0053] S1: Input n virtual monoenergetic images of arbitrary energy and select the monoenergetic image with significant artifacts (denoted as Image_Worst);

[0054] In this embodiment, Figure 3 As shown in (a), Image_Worst is a 50keV monoenergetic image. Any artifact correction algorithm can be used to correct the image artifacts and obtain a mask of the artifact area, or directly identify the artifact area or non-artifact area of ​​the image. The method includes but is not limited to traditional metal artifact removal algorithms based on interpolation and reconstruction, such as MAR, fsMAR, iMAR, OMAR, sMAR, etc., and is not limited to MAR algorithms based on deep learning or directly using deep learning and other methods to identify artifact areas or non-artifact areas; Figure 2As shown in the figure, it is a way to extract artifact areas: ① Image threshold segmentation sets a metal threshold, such as HU is 3000, traverses the image point by point, and sets the image CT value greater than 3000 to 1, and less than 3000 to 0, thereby segmenting the metal area and the non-metal area; ② Mean smoothing of the metal image. The operator used in this case is a 7×7 mean filter operator with all 1s. The sliding window is multiplied with the corresponding points of the image to obtain a smoothed and slightly enlarged metal area image; ③ The input image is forward projected using a projection algorithm, and the metal image is forward projected at the same time. In the projection data, the position of the metal projection data can be used to determine which positions in the projection of the original image are metal areas, and then polynomial interpolation is performed on these areas to obtain the corrected Projection data; ④ Reconstruct the corrected projection data to obtain a preliminary corrected image; ⑤ Use a 7×7 size all-one mean filter operator to smooth the image again; ⑥ Use threshold segmentation. In this embodiment, the threshold interval of soft tissue is set to 800-1200, extract the soft tissue area, and calculate the average CT value of the soft tissue; ⑦ Fill the soft tissue area except the metal area to obtain the mean image; ⑧ Use the projection algorithm to forward project the mean image, use the projection data of the input image and the projection data of the mean image to make a difference and record it as Pjr_diff, and perform polynomial interpolation on this difference again and record it as Pjr_inter; ⑨ Use the reconstruction algorithm to reconstruct the result of Pjr_inter minus Pjr_diff to obtain the final artifact image.

[0055] S2: Identify the artifact area and the non-artifact area in the above artifact image, and further obtain the artifact area mask (recorded as ImageMask1) and the non-artifact area mask (recorded as ImageMask0);

[0056] In this embodiment, the algorithm selected is the threshold segmentation method, which uses the threshold to perform segmentation calculation on the artifact image, and the CT values ​​of corresponding positions are added and reassigned to obtain an artifact area mask.

[0057] The threshold selected in this embodiment is -800, which is intended to extract the air region (the air CT value is -1000), and the image with the lowest energy ( Figure 3 (a)) is split, and the points are traversed point by point. If the value is greater than -800, it is set to 1, and if it is less than -800, it is set to 0. ( Figure 3 (d)). Using the same method, the threshold is set to 100 to extract the main artifact area. Figure 3 (c) Perform threshold segmentation to obtain Figure 3 (e), and finally Figure 3 (e) and Figure 3(d) The CT values ​​of the corresponding positions are added, and the points with a CT value of 2 are reset to 1, and the final artifact area mask is obtained, which is recorded as ImageMask1( Figure 3 (f)).

[0058] S3: Select an image without artifacts or with small artifacts (referred to as Image_Best), and derive two new images from Image_Best according to the two masks obtained in S2, one containing only the pixels or voxels of Image_Best in the selected area of ​​ImageMask1 mask (referred to as Image_Best_Artifact), and the other containing only the pixels or voxels of Image_Best in the selected area of ​​ImageMask0 mask (referred to as Image_Best_noArtifact);

[0059] It should be noted that the image without artifacts or with smaller artifacts (Image_Best) can be selected from n virtual monoenergetic images of any energy, or images reconstructed by other methods, including but not limited to other sources (such as different image layers, data after coarse correction of artifacts, etc.) or images reconstructed by other methods.

[0060] In this embodiment, Image_Best is selected from n virtual monoenergetic images of arbitrary energy in S1; Figure 3 As shown in (b), an algorithm is used to automatically identify the image with the smallest artifact among the n input monoenergetic images, such as the TV (Total Variation) algorithm. In this case, the cost function of the theoretical TV function is used (as shown in formula (1), i, j represent the pixel index of the image, f i,j represents the image pixel value at position (i, j), u i,j The TV value represents a pixel value. By traversing each pixel point of the image, the TV value of the current point can be calculated. Finally, the TV values ​​of all points are summed to obtain the final TV value of the image). The cost function values ​​of n images are calculated in sequence. The image with the smallest cost function value is considered to be the image with the smallest artifact. The image with the smallest artifact is multiplied by the corresponding position of the mask ImageMask0 of the non-artifact area to obtain the non-artifact area of ​​the current image (denoted as Image_Best_noArtifact). The image with the smallest artifact is multiplied by the mask ImageMask1 of the artifact area to obtain the artifact area of ​​the current image (denoted as Image_Best_Artifact).

[0061]

[0062] S4: Select two virtual monoenergetic images with energy differences from S1 to perform dual-energy decomposition to obtain two or more base material images;

[0063] S5: obtaining an image composed of pixels or voxels in the non-artifact area of ​​two or more base material images by comparing with ImageMask0;

[0064] It should be noted that there can be n kinds of decomposed base materials, but in order to explain the problem concisely, the embodiment shown in the present invention uses two decomposition base materials. The first base material is water-based and the second base material is bone-based. The density images of the water-based material and the bone-based material are multiplied by the mask of the non-artifact area to obtain the water-based material image and the bone-based material image without artifacts.

[0065] The image with the smallest artifact (Image_Best) and the image with the lowest energy (Image_Worst) identified in S3 are used for dual energy decomposition to obtain the water base image (denoted as imgWater) and the bone base image (denoted as imgBone). The dual energy decomposition process is as follows (Formula (2)), where E1 and E2 represent two types of energy. and Corresponding to Image_Best and Image_Worst, ρ w represents the density of water, m w (E) represents the mass attenuation coefficient of water material under energy E, m b (E) represents the mass attenuation coefficient of the bone material under energy E. The mass attenuation coefficients of water at different energies and the mass attenuation coefficients of bone materials at different energies are both known, and relevant information can be found on public websites. The water-based image and the bone-based image can be solved by this equation; finally, the water-based image and the bone-based image are multiplied by the mask ImageMask0 of the non-artifact area, respectively, to obtain the water-based image and the bone-based image without artifacts (denoted as imgWater_noArtifact and imgBone_noArtifact);

[0066]

[0067] S6: Based on the Image_Best_noArtifact image in S3 and the image composed of pixels or voxels in the non-artifact area obtained in S5, a relationship model between the Image_Best_noArtifact image and the two or more base material graphs is constructed;

[0068] Based on the CT value of the image without artifacts (Image_Best_noArtifact) and the water-based image and bone-based image without artifacts (imgWater_noArtifact and imgBone_noArtifact), we traverse each pixel of Image_Best_noArtifact, imgWater_noArtifact, and imgBone_noArtifact in turn. The sizes of the three images are the same. By taking out the pixel values ​​at the same position, we can form two sets of data pairs, namely CT value-water-based image value and CT value-bone-based image value; that is, multiple scattered points are formed, corresponding to Figure 3 (g) and (h) are the scattered point distributions. Finally, a polynomial fitting is performed on these scattered points to obtain the red model curve. So far, the relationship model between CT value and water / bone base map has been constructed.

[0069] It should be pointed out that the algorithm builds a model of the current data to be processed for a certain piece of data, so the model form is not unique.

[0070] S7: Substitute the Image_Best_Artifact image obtained in S3 into the relationship model established in S6 to obtain a component map of the artifact area on two or more base materials, that is, a corrected image;

[0071] Use the artifact area image obtained by S3, that is, Image_Best_Artifact, to traverse each pixel in the image, obtain the CT value, and bring it into step 5 to establish the red curve fitting formula, and you will get the water-based image value and bone-based image value corresponding to each CT value (recorded as imgWater_ArtifactCorrect and imgBone_ArtifactCorrect). Finally, imgWater_ArtifactCorrect+imgWater_noArtifact is recorded as imgWater_Correct, and imgBone_ArtifactCorrect+imgBone_noArtifact is recorded as imgBone_Correct.

[0072] S8: The artifact-corrected energy spectrum image at any energy can be synthesized through the base material images (imgWater_Correct and imgBone_Correct) obtained in S7;

[0073] It should be noted that energy spectrum images include but are not limited to virtual monochrome images, material density images, effective atomic number images, electron density images, virtual plain scan images, iodine images and other energy spectrum images, and can also improve the accuracy of non-image energy spectrum decomposition results such as energy spectrum curves and scatter plots.

[0074] In this embodiment, based on the results imgWater_Correct and imgBone_Correct of step 6, dual-energy decomposition (Formula (3)) is performed again to obtain a virtual monoenergetic image with artifact correction at any energy.

[0075]

[0076] The above algorithm verification is performed. The red arrows indicate the artifacts in the image itself, the yellow arrows indicate the artifacts introduced by the correction algorithms (two types, the product MAR algorithm and the currently proposed algorithm) relative to the original image, and the yellow box indicates the changes in the image structure caused by the correction algorithm relative to the original image. The results are shown in Figures 4 to 6 As shown:

[0077] Verification Example 1: A self-developed CT performance phantom with a diameter of 200 mm and two titanium cylindrical phantoms with a diameter of 10 mm embedded inside was placed in a dual-energy CT of an internationally renowned manufacturer for scanning.

[0078] like Figure 4 As shown in the figure, the first row of results is the virtual monoenergetic images at different energies provided by the manufacturer's software without MAR correction. It can be seen that at low energy, such as monoenergetic images below 70keV, there are obvious artifacts between the two metals, and the artifacts of high-energy monoenergetic images are corrected better; the second row of results is the result of MAR correction by the manufacturer's software. It can be seen that the artifacts of the overall image have been suppressed to a certain extent, but many new artifacts have been introduced, and the correction effect of the 50keV image is not ideal; the third row is the correction result of the algorithm described in the present invention based on the first row of images. The overall correction effect is better than the correction effect of the manufacturer's MAR algorithm, and no new artifacts have been introduced. The algorithm artifact suppression effect is significant, indicating the effectiveness of the algorithm; the fourth row is the enlarged image of the 70keV image in the first three rows of results. It can be clearly seen that the proposed algorithm effectively improves the artifacts in the area indicated by the red arrow without introducing new artifacts.

[0079] Verification Example 2: A postoperative CT scan of a patient undergoing pedicle screw placement was performed using a dual-energy CT scan from an internationally renowned manufacturer.

[0080] like Figure 5 As shown, similar to Figure 4The first row is the manufacturer's original monoenergetic image, the second row is the manufacturer's monoenergetic image after MAR, the third row is the correction result of the current algorithm based on the first row of images, and the fourth row is the enlarged image of the 70keV image in the first three rows of results; the product MAR correction result obviously introduces dark band artifacts (as shown by the yellow arrow), the metal is blurred, and causes changes in tissue structure (as shown by the yellow box), while the current algorithm correction effectively improves the artifacts of images at all energies, and the metal boundary is clearer.

[0081] Verification Example 3: A postoperative CT scan of a patient with oral denture implants was performed using a dual-energy CT scan from an internationally renowned manufacturer.

[0082] like Figure 6 As shown, similarly, the first row is the manufacturer's original monoenergetic image, the second row is the manufacturer's monoenergetic image after MAR, the third row is the correction result of the current algorithm based on the first row of images, and the fourth row is the enlarged image of the 70keV image in the first 3 rows of results; the product MAR correction result, although it improves the artifacts in the original image, it also introduces many new artifacts. The current correction algorithm is significantly better than the product's MAR result, does not introduce new artifacts, and also effectively improves the artifacts in the low keV image.

[0083] The present invention is described in detail through the embodiments above, but the contents described are only preferred embodiments of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of application of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for improving metal artifacts in spectral CT image domain, characterized by: The method comprises the following steps: S1: Input n virtual monoenergetic images of arbitrary energy and select the monoenergetic image with significant artifacts (denoted as Image_Worst); S2: Identify the artifact area and the non-artifact area in the above artifact image, and further obtain the artifact area mask (recorded as ImageMask1) and the non-artifact area mask (recorded as ImageMask0); S3: Select an image without artifacts or with small artifacts (referred to as Image_Best), and derive two new images from Image_Best according to the two masks obtained in S2, one containing only the pixels or voxels of Image_Best in the selected area of ​​ImageMask1 mask (referred to as Image_Best_Artifact), and the other containing only the pixels or voxels of Image_Best in the selected area of ​​ImageMask0 mask (referred to as Image_Best_noArtifact); S4: Select two virtual monoenergetic images with energy differences from S1 to perform dual-energy decomposition to obtain two or more base material images; S5: obtaining an image composed of pixels or voxels in the non-artifact area of ​​two or more base material images by comparing with ImageMask0; S6: Based on the Image_Best_noArtifact image in S3 and the image composed of pixels or voxels in the non-artifact area obtained in S5, a relationship model between the Image_Best_noArtifact image and two or more of the base material graphs is constructed; S7: Substitute the Image_Best_Artifact image obtained in S3 into the relationship model established in S6 to obtain a component map of the artifact area on two or more base materials, that is, a corrected image; S8: synthesize the non-artifact area image of the base material obtained in S5 and the corrected artifact area image of the base material obtained in S7. Through the corrected base material image, an artifact-corrected energy spectrum image at any energy can be synthesized.

2. The method for improving metal artifacts in spectral CT image domain according to claim 1, characterized in that: In S8, the energy spectrum image includes but is not limited to virtual monochrome images, material density images, effective atomic number images, electron density images, virtual plain scan images, iodine images and other energy spectrum images, and can also improve the accuracy of non-image energy spectrum decomposition results such as energy spectrum curves and scatter plots.

3. The method for improving metal artifacts in spectral CT image domain according to claim 1, characterized in that: In S1, the energy range of the virtual monoenergetic image can be but is not limited to 40 to 140 keV. Any artifact correction algorithm can be used to perform artifact correction on the image to obtain a mask of the artifact area, or the artifact area or non-artifact area of ​​the image can be directly identified. The method includes but is not limited to traditional metal artifact removal algorithms based on interpolation and reconstruction, such as MAR, fsMAR, iMAR, OMAR, sMAR, etc., and is not limited to MAR algorithms based on deep learning or directly using deep learning and other methods to identify artifact areas or non-artifact areas.

4. The method for improving metal artifacts in spectral CT image domain according to claim 3, characterized in that: When extracting artifact areas, a metal threshold is set for image threshold segmentation, and the image is traversed point by point. Image CT values ​​greater than the threshold are set to 1, and values ​​less than the threshold are set to 0, thereby segmenting metal areas and non-metal areas.

5. The method for improving metal artifacts in spectral CT image domain according to claim 1, characterized in that: In S2, a threshold segmentation method is used to obtain an artifact area mask.

6. The method for improving metal artifacts in spectral CT image domain according to claim 1, characterized in that: In S3, the image without artifacts or with relatively small artifacts (Image_Best) is selected from n virtual monoenergetic images of arbitrary energy in S1, or an image reconstructed by other means, including but not limited to other sources (such as different image layers, data after rough correction of artifacts, etc.) or images reconstructed by other means.

7. The method for improving metal artifacts in spectral CT image domain according to claim 1, characterized in that: In S4, a relationship model between CT value and water / bone base image can be constructed, which can be extended to the relationship model between CT value and any base material pair image, as well as the relationship model between any other different types of energy spectrum images.

8. The method for improving metal artifacts in spectral CT image domain according to any one of claims 1 to 7, characterized in that: This algorithm selects (but is not limited to) 6 images, and the input single energy images are 50keV, 70keV, 80keV, 100keV, 120keV, and 140keV.

9. The method for improving metal artifacts in spectral CT image domain according to any one of claims 1 to 7, characterized in that: The method of creating an artifact-free base material image and building a model using two single-energy images of the non-artifact area can be extended to using other images in the image sequence (either images containing metal or not) and building models, and the two are equivalent.

10. The method for improving metal artifacts in spectral CT image domain according to any one of claims 1 to 7, characterized in that: The method is applicable to improving artifacts of two-dimensional images, and can also be directly extended to improving artifacts of three-dimensional images.

11. The method for improving metal artifacts in spectral CT image domain according to any one of claims 1 to 7, characterized in that: In addition to constructing a correction model for the current data for each set of input data, the method can also pre-construct a universal model relationship between monoenergetic images and base material components. In this case, one correction model can be applied to all input data (including but not limited to constructing a universal model for different scanning sites, the presence or absence of contrast agents, etc.).

12. The method for improving metal artifacts in spectral CT image domain according to any one of claims 1 to 7, characterized in that: The method for establishing or optimizing the model in the method includes but is not limited to polynomial fitting, deep learning (neural network), and pattern recognition.

13. The method for improving metal artifacts in spectral CT image domain according to any one of claims 1 to 7, characterized in that: The model in the method can be used to optimize artifact types, including but not limited to metal artifacts, bone artifacts, water hardening artifacts, etc.

14. The method for improving metal artifacts in spectral CT image domain according to any one of claims 1 to 7, characterized in that: The model in the method is not limited to the energy spectrum image modeling, and the same modeling can also be performed on non-energy spectrum images.

Images