A method and apparatus for correcting image hardening artifacts

Abstract

The application discloses a method and device for correcting image hardening artifacts, which comprises the following steps: obtaining an original image, segmenting the original image to obtain a high-density tissue image, performing orthographic projection, data fusion and inverse projection on the image to obtain first fusion data and second fusion data, and finally solving a correction coefficient array according to the original image, the high-density tissue image, the first fusion image and the second fusion image, and obtaining a corrected image by combining the correction coefficient array. The correction method is based on the original image, and does not involve the steps of obtaining and processing of the ray energy in the correction process, so that additional prior knowledge such as energy-dependent attenuation coefficient, ray source energy and detector related information is not required. The corrected image can be obtained only by simple calculation based on the original image, the speed is relatively faster than that of the iterative algorithm, the efficiency of artifact correction is improved, and the method is more suitable for engineering application.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a method and device for correcting image hardening artifacts. BACKGROUND

[0002] Beam hardening artifact is a common artifact in CT three-dimensional reconstruction images. Since the X-ray emitted by the ball tube is a mixed energy ray, when the mixed energy ray passes through the object, the low energy photons are absorbed, and the high energy photons pass through, forming a "hardened" beam, which increases the average energy of the ray. Since the attenuation coefficient is greatly dependent on the energy of the ray, when the "hardened" beam continues to pass through other structures with higher density, the attenuation coefficient decreases significantly, and a low-density band shadow is formed on the image. The existence of the artifact affects the imaging of the low-density tissue between the high-density structures, resulting in incomplete structure of the image.

[0003] At present, in order to eliminate or weaken the influence of the artifact, various hardening artifact correction techniques have been developed. Among them, the hardening artifact correction method based on high-order iteration has large calculation amount and long time consumption in engineering implementation. The method based on projection correction needs to consider the energy-dependent attenuation coefficient. Therefore, both methods have limitations for engineering batch application. SUMMARY

[0004] The technical problem to be solved by the present application is to provide a method for correcting image hardening artifacts and improve the efficiency of artifact correction.

[0005] To solve the above technical problems, the technical scheme adopted by the present application is:

[0006] A method for correcting image hardening artifacts, comprising the steps of:

[0007] Obtaining an original image, segmenting high-density tissue in the original image to obtain a high-density tissue image;

[0008] Performing forward projection on the original image and the high-density tissue image to obtain original forward projection data and high-density forward projection data;

[0009] Fusing the original forward projection data and the high-density forward projection data to obtain first fused data, and self-fusing the high-density forward projection data to obtain second fused data;

[0010] Performing back projection on the first fused data and the second fused data to obtain first fused image and second fused image;

[0011] Solving a correction coefficient array according to the original image, the high-density tissue image, the first fused image and the second fused image, and obtaining a corrected image combined with the correction coefficient array.

[0012] To solve the above technical problems, another technical solution adopted by the present application is:

[0013] A device for correcting image hardening artifacts, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements each step of the method for correcting image hardening artifacts as described above when executing the computer program.

[0014] The present application has the beneficial effect that: after obtaining the original image, the original image is segmented to obtain a high-density tissue image, and then first fusion data and second fusion data are obtained through forward projection, data fusion and back projection of the image, and finally a correction coefficient group is solved according to the original image, the high-density tissue image, the first fusion image and the second fusion image, and a corrected image is obtained in combination with the correction coefficient group. The correction method is based on the original image, and the correction process does not involve the acquisition and processing of the ray energy, so it does not require additional prior knowledge such as energy-dependent attenuation coefficient, ray source energy and detector related information, and only a simple calculation based on the original image can obtain the corrected image. The speed is relatively faster than the iterative algorithm, the efficiency of artifact correction is improved, and it is more suitable for engineering application. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 A step flow chart of the method for correcting image hardening artifacts in an embodiment of the present application;

[0016] Figure 2 Another step flow chart of the method for correcting image hardening artifacts in an embodiment of the present application;

[0017] Figure 3 Another step flow chart of the method for correcting image hardening artifacts in an embodiment of the present application;

[0018] Figure 4 A first image comparison example before and after correction in an embodiment of the present application;

[0019] Figure 5 A second image comparison example before and after correction in an embodiment of the present application;

[0020] Figure 6 A third image comparison example before and after correction in an embodiment of the present application;

[0021] Figure 7 A fourth image comparison example before and after correction in an embodiment of the present application;

[0022] Figure 8 A structure schematic diagram of the device for correcting image hardening artifacts in an embodiment of the present application. DETAILED DESCRIPTION

[0023] To describe the technical solutions in the present application more clearly, achieve the purposes and obtain the effects, the following will be described in detail in connection with the embodiments and with reference to the drawings.

[0024] Please refer to Figure 1 A method for correcting image hardening artifacts, comprising the steps of:

[0025] Obtaining an original image, segmenting high-density tissues in the original image to obtain a high-density tissue image;

[0026] Performing forward projection on the original image and the high-density tissue image to obtain original forward projection data and high-density forward projection data;

[0027] Fusing the original forward projection data and the high-density forward projection data to obtain first fused data, and self-fusing the high-density forward projection data to obtain second fused data;

[0028] Performing back projection on the first fused data and the second fused data to obtain first fused images and second fused images;

[0029] Solving a correction coefficient group according to the original image, the high-density tissue image, the first fused images and the second fused images, and obtaining a corrected image in combination with the correction coefficient group.

[0030] As can be seen from the above description, after obtaining an original image, the original image is segmented to obtain a high-density tissue image, and then first fused data and second fused data are obtained through forward projection, data fusion and back projection on the image. Finally, a correction coefficient group is solved according to the original image, the high-density tissue image, the first fused images and the second fused images, and a corrected image is obtained in combination with the correction coefficient group. The correction method is based on the original image, and no steps such as acquisition and processing of ray energy are involved in the correction process. Therefore, no additional prior knowledge such as energy-dependent attenuation coefficient, ray source energy and detector information is required. The corrected image can be obtained through simple calculation based on the original image. The speed is relatively faster than that of the iterative algorithm, the efficiency of artifact correction is improved, and the method is more suitable for engineering application.

[0031] Further, the segmenting of the high-density tissues in the original image to obtain the high-density tissue image comprises:

[0032] Obtaining a CT value difference of the original image, and segmenting the original image into a high-density tissue region and a low-density tissue region according to the CT value difference;

[0033] Obtaining the high-density tissue image according to the high-density tissue region.

[0034] As can be seen from the above description, since different densities of tissues have differences in CT values, segmentation of the original image based on the CT value differences in the original image can effectively segment the original image into high-density tissue regions and low-density tissue regions.

[0035] Further, the solving of the correction coefficient array according to the original image, the high-density tissue image, the first fusion image and the second fusion image comprises:

[0036] obtaining a target function according to the original image, the high-density tissue image, the first fusion image and the second fusion image;

[0037] obtaining an evaluation function, and solving the correction coefficient array according to the evaluation function and the target function.

[0038] As can be seen from the above description, the target function is obtained through the original image, the high-density tissue image, the first fusion image and the second fusion image, and the target function is obtained by fully utilizing each obtained image, thereby improving the processing effect of the target function.

[0039] Further, the obtaining of the target function according to the original image, the high-density tissue image, the first fusion image and the second fusion image comprises:

[0040] obtaining a first threshold, and dividing a region of interest in the original image, the high-density tissue image, the first fusion image and the second fusion image according to the first threshold;

[0041] obtaining a target function according to the region of interest in the original image, the region of interest in the high-density tissue image, the region of interest in the first fusion image and the region of interest in the second fusion image.

[0042] As can be seen from the above description, after the first threshold is obtained, the regions of interest in the original image, the high-density tissue image, the first fusion image and the second fusion image are divided, and the target function is obtained from the region of interest in each image, which not only improves the solving speed but also improves the accuracy of the correction coefficient array.

[0043] Further, the target function is:

[0044] min S(c 01 ,c 11 ,c 02 )=min∫std(f0+c 01 f 01 +c 11 f 11 +c 02 f 02 )dxdy;

[0045] wherein c01 , c 11 , and c 02 are correction coefficient arrays, f0is the region of interest of the original image, f 01 is the region of interest of the high-density tissue image, f 11 is the region of interest of the first fusion image, f 02 is the region of interest of the second fusion image.

[0046] As can be seen from the above description, since there are dark band streaks in the image containing the hardening artifact, the standard deviation of the image is increased, and by setting the objective function, the standard deviation minimization of the four images, i.e., the region of interest of the original image, the region of interest of the tissue image, the region of interest of the first fusion image, and the region of interest of the second fusion image, is solved when the hardening artifact correction is performed, so that the standard deviation is reduced, and thus the image can be made smooth after the artifact correction.

[0047] Further, the solving the correction coefficient arrays according to the evaluation function and the objective function comprises:

[0048] solving the objective function according to the evaluation function to obtain the correction coefficient arrays;

[0049] The evaluation function is:

[0050]

[0051] wherein c 01 , c 11 , and c 02 are correction coefficient arrays, F = f0+ c 01 f 01 + c 11 f 11 + c 02 f 02 , is the mean value of F, and N is the number of pixels in the region of interest of each image.

[0052] As can be seen from the above description, by setting the evaluation function to solve the correction coefficient arrays in the objective function, the value of the three coefficients of the linear combination of the region of interest standard deviation minimum time can be obtained based on the standard deviation minimum function, i.e., the correction coefficient arrays are obtained, without iterative calculation, thereby improving the calculation efficiency.

[0053] Further, the solving the correction coefficient arrays according to the evaluation function and the objective function comprises:

[0054] obtaining a preset step size;

[0055] According to the preset step, the correction coefficient array is iteratively calculated, and the optimized correction coefficient array is obtained through reverse search.

[0056] From the above description, the result of the iteratively calculated correction coefficient array is continuously optimized, and the image correction effect is further improved.

[0057] Further, the iteratively calculating the correction coefficient array according to the preset step, and obtaining the optimized correction coefficient array through reverse search comprises:

[0058]

[0059]

[0060]

[0061] Wherein, k is the iteration number; sc 01 , sc 11 and sc 02 are preset steps of correction coefficient arrays c 01 , c 11 and c 02 , and are correction coefficient arrays in the iteration process, and are correction coefficient arrays obtained after one iteration. and are correction coefficient arrays obtained after one iteration.

[0062] From the above description, the result of the iteratively calculated correction coefficient array is continuously optimized, and the image correction effect is further improved.

[0063] Further, the solving the correction coefficient array according to the original image, the high-density tissue image, the first fusion image and the second fusion image, and obtaining the corrected image combining the correction coefficient array comprises:

[0064] According to the correction coefficient array, the original image, the high-density tissue image, the first fusion image and the second fusion image are weighted and fused to obtain the corrected image.

[0065] The weighted fusion comprises:

[0066]

[0067] Wherein, is the corrected image, f0' is the original image, f1'0 is the high-density tissue image, f1'1 is the first fusion image, f0'2 is the second fusion image; c01 , c 11 and c 02 is a correction coefficient array.

[0068] As can be seen from the above description, the correction coefficient array is calculated based on the region of interest of the original image, the region of interest of the high-density tissue image, the region of interest of the first fusion image and the region of interest of the first fusion image, the weighted fusion coefficient is obtained, and the original image is corrected according to the weighted fusion coefficient, so that the modified image with weakened hardening artifacts is obtained.

[0069] Please refer to Figure 8 , another embodiment of the present application provides a device for correcting image hardening artifacts, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements each step of the above-mentioned method for correcting image hardening artifacts when executing the computer program.

[0070] The above-mentioned method and device for correcting image hardening artifacts can be applied to the processing of medical X-ray CT images, especially for the correction of hardening artifacts in CT images, which will be described in detail in the following specific embodiments:

[0071] Embodiment one

[0072] Please refer to Figure 1 , a method for correcting image hardening artifacts, comprising the steps of:

[0073] S1, obtaining an original image, segmenting high-density tissue in the original image to obtain a high-density tissue image; specifically, obtaining the CT value difference of the original image, segmenting the original image into high-density tissue region and low-density tissue region according to the CT value difference, and obtaining the high-density tissue image according to the high-density tissue region; if the original image includes bone, metal, soft tissue and air, the tissue in the original image can be divided into two parts, one part of high-density tissue includes bone and metal, and the other part of low-density tissue includes soft tissue and air; wherein the segmentation method can include single threshold segmentation method, multi-threshold segmentation method or soft threshold segmentation method; if the soft threshold weighted segmentation method is used, the threshold value of the bone is empirically set to a fixed value t; in order to more effectively separate the bone and soft tissue, a small threshold value is needed to segment the values between the bone and soft tissue in the picture according to linear weighting, which is about 0.85-0.9 of the bone threshold value, and the small threshold value can be adjusted according to the actual situation; at the same time, hardening artifacts mainly appear around high-density materials, so when segmenting, the region of interest R, i.e. the part of the region around the high-density tissue, can be expanded outward along the contour of the high-density tissue by 30-60 pixel points;

[0074] S2, orthographic projection is performed on the original image and the high-density tissue image to obtain original orthographic projection data and high-density orthographic projection data;

[0075] S3, the original orthographic projection data and the high-density orthographic projection data are fused to obtain first fusion data, and the high-density orthographic projection data is self-fused to obtain second fusion data; the above two kinds of fusion data can be obtained by product calculation;

[0076] S4, back projection is performed on the first fusion data and the second fusion data to obtain a first fusion image and a second fusion image; that is, image reconstruction is performed on the first fusion data and the second fusion data to obtain the first fusion image and the second fusion image;

[0077] S5, correction coefficient groups are solved according to the original image, the high-density tissue image, the first fusion image and the second fusion image, and a corrected image is obtained by combining the correction coefficient groups; specifically:

[0078] S51, a target function is obtained according to the original image, the high-density tissue image, the first fusion image and the second fusion image; in an optional embodiment, a first threshold value is obtained, the original image, the high-density tissue image, the first fusion image and the second fusion image are divided into regions of interest according to the first threshold value, and a target function is obtained according to the regions of interest of the original image, the regions of interest of the high-density tissue image, the regions of interest of the first fusion image and the regions of interest of the second fusion image;

[0079] The target function is obtained as follows:

[0080] min S(c 01 ,c 11 ,c 02 )=min∫std(f0+c 01 f 01 +c 11 f 11 +c 02 f 02 )dxdy;

[0081] Wherein, c 01 , c 11 and c 02 are correction coefficient groups, f0 is a region of interest of the original image, f 01 is a region of interest of the high-density tissue image, f 11 is a region of interest of the first fusion image, and f 02 is a region of interest of the second fusion image; min S(c 01 , c 11 , c 02) represents the minimum standard deviation function, std() represents the calculation of f0+c 01 f 01 +c 11 f 11 +c 02 f 02 standard deviation;

[0082] S52, obtaining an evaluation function, and solving the correction coefficient array according to the evaluation function and the target function;

[0083] Solving the target function according to the evaluation function to obtain the correction coefficient array;

[0084] The evaluation function is:

[0085]

[0086] Wherein, c 01 , c 11 and c 02 are correction coefficient arrays, F=f0+c 01 f 01 +c 11 f 11 +c 02 f 02 , is the mean value of F, and N is the number of pixels in the region of interest in each image. The number of pixels is different according to the region of interest, and the size of the region of interest defined in each original image is constant, so N is a constant value in other images under the same original image. If it is necessary to improve the solving speed, the obtained c 01 , c 11 and c 02 are correction coefficient arrays directly used for image correction;

[0087] Further, the derivative of E is obtained as:

[0088]

[0089] F'(c 01 ) = f 01

[0090] F'(c 11 ) = f 11

[0091] F'(c 02 ) = f 02

[0092] Wherein is the mean value of F in the region of interest R, is f01 the mean value of the region of interest R,

[0093] the value of f is f 11 the mean value of the region of interest R, is f 02 the mean value of the region of interest R;

[0094] Further, the partial derivatives of c 01 , c 11 and c 02 are respectively obtained as follows:

[0095]

[0096]

[0097]

[0098] Further, in an optional embodiment, the correction coefficient array is iteratively calculated to obtain an optimized correction coefficient array, specifically: a preset step size is obtained, the correction coefficient array is iteratively calculated according to the preset step size, and the optimized correction coefficient array is obtained through reverse search.

[0099] wherein the iterative calculation of the correction coefficient array comprises:

[0100]

[0101]

[0102]

[0103] wherein k is the iteration number; sc 01 , sc 11 and sc 02 are preset step sizes of the correction coefficient arrays c 01 , c 11 and c 02 , and are correction coefficient arrays in the iteration process, and are correction coefficient arrays obtained after one iteration, and are correction coefficient arrays obtained after one iteration; an iteration termination condition is set, such as stopping the iteration when N times are reached, c k+1 is the correction coefficient array obtained after the iteration is stopped; through multiple iterations, the value of the three coefficients in the linear combination when the standard deviation of the region of interest is the smallest can be finally obtained.

[0104] S53、according to the correction coefficient array, performing weighted fusion on the original image, the high-density tissue image, the first fusion image and the second fusion image to obtain the corrected image;

[0105] The weighted fusion comprises:

[0106]

[0107] wherein, is the corrected image, f'0 is the original image, f' 10 is the high-density tissue image, f' 11 is the first fusion image, f' 02 is the second fusion image; c 01 , c 11 and c 02 are the correction coefficient array.

[0108] Embodiment Two

[0109] This embodiment specifically describes how to establish the weighted fusion formula:

[0110] Since the X-ray is not a single energy, but a mixed energy ray, the multi-energy spectrum attenuation formula can be expressed as follows:

[0111]

[0112] wherein, L is the line integral corresponding to the path, which can be expressed as μ(E, s+λθ), and the subsequent formula uses r to represent L, s is the ray source vector, and θ is the projection angle;

[0113] μ(E, s+λθ) is the attenuation of the multi-energy spectrum ray after passing through the object, and such attenuation is the joint action of the object and the energy, so the attenuation value can also be represented by the object and the energy, i.e., f0(r) represents the object and ψ0(E) represents the energy, i.e.,

[0114] μ(E, s+λθ) = μ(E, r) = f0(r) ψ0(E) (2);

[0115] wherein, ψ0(E) is the energy-dependent term of the material in the object;

[0116]

[0117] wherein, q is the measurement value received by the detector; the above formula is equivalent to replacing a parameter μ related to energy and ray passing distance with a parameter ψ0(E) related to energy and a parameter p0 related to ray passing distance; in clinical practice, ψ0(E) is usually replaced with ψ water(E), that is, the image is pre-corrected by water; but after pre-corrected by water, hardening artifacts will also appear, because the object contains substances with large difference in density from water, and the energy dependence is not the same; the water pre-correction is performed in the step of reconstructing the image, and generally, the CT image needs to be pre-corrected by water, and the method proposed in the embodiment is used for the reconstructed image which has been pre-corrected by water; it is assumed that the object contains two materials, water and bone, and because the density of the two is greatly different, the dependence of the substance and energy will have two, that is, formula (2) can be written as:

[0118]

[0119] Finally, what needs to be solved is the water equivalent base image that is, the image after hardening correction, according to formula (3), formula (4) can be written as:

[0120]

[0121] It is further assumed that the X-ray is a single energy ray, and then let ψ1(E0) = ψ2(E0) = 1, at this time The remaining is the projection of the image without hardening artifacts based on water pre-correction;

[0122] The multi-level expansion method is used to solve

[0123]

[0124] Because the part exceeding the second order has almost no effect on the image, it can be discarded; the projection data p0, p2, p0p2 and in formula (6) are back-projected to obtain four images f'0, f' 10 , f' 11 and f' 02 , that is, to obtain:

[0125]

[0126] Then formula (7) is the composition of the corrected image , after solving the correction coefficient groups in formula (7), the four images obtained are combined to obtain the corrected image;

[0127] Please refer to Figure 2 , that is, after inputting the original image, the correction coefficient groups c 01 , c 11 and c 02 are solved after the hardening artifact correction, and the original image is corrected according to the correction coefficient groups and the corrected image is output.

[0128] Embodiment three

[0129] This embodiment provides a specific application scenario to illustrate the steps in embodiment one;

[0130] Please refer to Figure 3 , S1, obtaining the original image f0, and segmenting the high-density tissue in the original image f0 to obtain the bone tissue image f2;

[0131] S2, orthographic projection is performed on the original image f0 and the bone tissue image f2 to obtain the original orthographic projection data p0 and the bone tissue orthographic projection data p2;

[0132] S3, fusion is performed on the original orthographic projection data p0 and the bone tissue orthographic projection data p2, and the product of the original orthographic projection data p0 and the bone tissue orthographic projection data p2 is calculated to obtain the first fusion data p 11 ,

[0133] Self-fusion is performed on the bone tissue orthographic projection data p2, and the self-product of the bone tissue orthographic projection data p2 is calculated to obtain the second fusion data p 02 ;

[0134] S4, back projection is performed on the first fusion data p 11 and the second fusion data p 02 to obtain the first fusion image f 11 and the second fusion image f 02 ;

[0135] According to the original image f0, the bone tissue image f2, the first fusion image f 11 and the second fusion image f 02 , the correction coefficient groups c 01 , c 11 and c 02 are solved, and the corrected image f1 is obtained in combination with the correction coefficient groups;

[0136] Please refer to Figures 4-7 , for comparison between the images before and after correction, it can be seen that the hardened artifacts in the original image are obviously eliminated or weakened after correction.

[0137] Embodiment four

[0138] Please refer to Figure 8 , a device for correcting image hardened artifacts, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements each step in the method for correcting image hardened artifacts according to any one of embodiments one, two or three when executing the computer program.

[0139] In summary, the present application provides a kind of method and device for correcting image hardening artifact, comprising the following steps: after obtaining original image, the original image is segmented to obtain high-density tissue image, then first fusion data and second fusion data are obtained by carrying out orthographic projection, data fusion and inverse projection to the image, finally, correction coefficient group is solved according to original image, high-density tissue image, first fusion image and second fusion image, and the correction coefficient group is combined to obtain corrected image, and the interested region in original image, high-density tissue image, first fusion image and second fusion image is further segmented to obtain target function, which not only improves the solving efficiency and improves the solving precision;At the same time, the correction method is corrected on the basis of original image, and the steps such as acquisition and processing of ray energy are not involved in the correction process, so additional prior knowledge such as energy-dependent attenuation coefficient, ray source energy and detector and related information is not required, and corrected image can be obtained by simple calculation on the basis of original image, which is relatively faster than iterative algorithm, improves the efficiency of artifact correction, and is more suitable for engineering application.

[0140] The above is only an embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent transformation or direct or indirect application in related technical fields using the content of the specification and drawings of the present application are also included in the patent protection scope of the present application.

Claims

1. A method for correcting image hardening artifacts, characterized in that, Including the following steps: Obtain the original image, segment the high-density tissue in the original image to obtain the high-density tissue image; The original image and the high-density tissue image are orthographically projected to obtain the original orthographic projection data and the high-density orthographic projection data. The original orthographic projection data and the high-density orthographic projection data are fused to obtain the first fused data, and the high-density orthographic projection data are self-fused to obtain the second fused data; Back-projecting the first fused data and the second fused data yields a first fused image and a second fused image. The correction coefficient set is calculated based on the original image, the high-density tissue image, the first fused image, and the second fused image, and the corrected image is obtained by combining the correction coefficient set. The step of solving for the correction coefficient set based on the original image, the high-density tissue image, the first fused image, and the second fused image includes: The objective function is obtained based on the original image, the high-density tissue image, the first fused image, and the second fused image. Obtain the evaluation function, and solve the set of correction coefficients based on the evaluation function and the objective function; The objective function is: ； in, , and For the correction coefficient group, The region of interest in the original image. Region of interest in high-density tissue images The region of interest in the first fused image. The region of interest is the second fused image; The step of solving the set of correction coefficients based on the evaluation function and the objective function includes: The objective function is solved based on the evaluation function to obtain the set of correction coefficients; The evaluation function is: ； in, , and For the correction coefficient group, , for The mean, This represents the number of pixels in the region of interest for each image. The step of solving the objective function based on the evaluation function to obtain the set of correction coefficients further includes: Get the preset step size; The correction coefficient set is iteratively calculated according to the preset step size, and the optimized correction coefficient set is obtained by reverse search.

2. The method for correcting image hardening artifacts according to claim 1, characterized in that, The process of segmenting high-density tissue from the original image to obtain a high-density tissue image includes: The CT value difference of the original image is obtained, and the original image is segmented into high-density tissue regions and low-density tissue regions based on the CT value difference; The high-density tissue image is obtained based on the high-density tissue region.

3. The method for correcting image hardening artifacts according to claim 1, characterized in that, The objective function obtained based on the original image, the high-density tissue image, the first fused image, and the second fused image includes: A first threshold is obtained, and the region of interest in the original image, high-density tissue image, first fused image, and second fused image is divided according to the first threshold. The objective function is obtained based on the region of interest (ROI) of the original image, the region of interest of the high-density tissue image, the region of interest of the first fused image, and the region of interest of the second fused image.

4. The method for correcting image hardening artifacts according to claim 1, characterized in that, The step of iteratively calculating the correction coefficient set according to the preset step size and obtaining the optimized correction coefficient set through reverse search includes: ； Where k is the number of iterations; , and Correction coefficient group , and The preset step size, , and This refers to the set of correction coefficients during the iteration process. , and During the iteration process , and The set of correction coefficients obtained after one iteration.

5. The method for correcting image hardening artifacts according to claim 1, characterized in that, The step of solving for the correction coefficient set based on the original image, the high-density tissue image, the first fused image, and the second fused image, and then combining the correction coefficient set to obtain the corrected image, includes: The original image, high-density tissue image, first fused image, and second fused image are weighted and fused according to the correction coefficient group to obtain the corrected image; The weighted fusion includes: ； in, To correct the image, For the original image, For high-density tissue images, For the first fused image, For the second fused image; , and This is a set of correction coefficients.

6. An apparatus for correcting image hardening artifacts, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements each step of the method for correcting image hardening artifacts as described in any one of claims 1-5.

Images