A method for removing metal artifacts from CT images

Abstract

The present invention discloses a method for removing metal artifacts from CT images, comprising the following steps: (1) acquiring original projection sequence images; (2) reconstructing the original projection sequence images to obtain a reconstructed image containing metal artifacts; (3) extracting metal regions in the reconstructed image; (4) orthogonally projecting the metal region onto a projection plane corresponding to each image in the original projection sequence images to obtain a corresponding metal projection sequence image; (5) correcting each metal region in the metal projection sequence image, and linearly interpolating the corrected metal region on the original projection sequence image to obtain an interpolated image, and removing the metal image; (6) reconstructing the interpolated image to obtain an image with metal and metal artifacts removed, and overlapping the interpolated image with the metal region image obtained in step (3) to obtain a projection sequence image with metal artifacts finally removed. The present invention uses the information of the original image to correct the reprojected image, and can completely remove metal artifacts.

Description

Technical Field

[0001] The invention relates to the field of image processing, and in particular to a method for removing metal artifacts from CT images. Background Art

[0002] CBCT technology uses an X-ray light source and a flat-panel detector to rotate and scan an object, obtain projections of the object at different angles, and then use a cone-beam back-projection reconstruction algorithm to obtain a three-dimensional image of the object.

[0003] If there are metal objects in the reconstructed object, due to the high attenuation characteristics of metal itself, radial or annular metal artifacts will be introduced in the back-projection reconstruction process, affecting the final reconstruction effect.

[0004] The traditional method of removing metal artifacts is to first segment the metal part through methods such as thresholding, then project the metal part back to the original acquired data through Redon transform, eliminate the metal part on the original data through methods such as interpolation, and then perform back-projection reconstruction to obtain an image without metal and metal artifacts; finally, the segmented metal part is reintegrated into the reconstructed image to obtain a reconstructed image without metal artifacts.

[0005] This method has extremely high requirements on the accuracy of the reconstructed metal. During reprojection (Redon transform), the reprojected position of the metal area will be offset and distorted due to inaccurate metal position and system distortion. The metal part cannot be completely eliminated, resulting in residual metal artifacts. Summary of the invention

[0006] Purpose of the invention: In view of the above-mentioned deficiencies, the present invention proposes a method for removing metal artifacts from CT images, introduces an image restoration method to compare the reprojected metal image with the original image, and uses the information of the original image to correct the reprojected image so that it can cover the real metal area as much as possible to completely remove the metal artifacts.

[0007] Technical solution:

[0008] A method for removing metal artifacts from CT images, comprising the steps of:

[0009] (1) Obtaining original projection sequence images through CT imaging equipment;

[0010] (2) reconstructing the original projection sequence images to obtain a reconstructed image containing metal artifacts;

[0011] (3) extracting the metal area in the reconstructed image obtained in step (2);

[0012] (4) projecting the metal area obtained in step (3) onto the projection plane corresponding to each image in the original projection sequence image to obtain the corresponding metal projection sequence image;

[0013] (5) correcting each metal region in the metal projection sequence image obtained in step (4), and performing linear interpolation on the corrected metal region on the original projection sequence image to obtain an interpolation image, thereby removing the metal region;

[0014] (6) Reconstructing the interpolated image obtained in step (5) to obtain an image with metal and metal artifacts removed, and superimposing it with the metal area image obtained in step (3) to obtain a final projection sequence image with metal artifacts removed.

[0015] In the steps (2) and (6), the FDK algorithm is used to perform image reconstruction.

[0016] In the step (3), a threshold method is used to extract the metal area in the reconstructed image.

[0017] The correction of step (5) is specifically as follows:

[0018] (51) using the image connected domain segmentation method to segment and extract each metal region of each image in the metal projection sequence image as a template, and recording the coordinates L of each metal region of each image;

[0019] (52) performing template matching on the metal area obtained in step (51) in the original projection sequence image to determine a position correction coefficient;

[0020] (53) The position of the metal area is corrected according to the position correction coefficient obtained in step (52), and expansion processing is performed on it.

[0021] In the step (52), the position correction coefficient is determined as follows:

[0022] a. Frame the search area at the corresponding metal area on the original projection sequence image;

[0023] b. Slide the corresponding template obtained in step (51) in the search area to obtain the corresponding area and calculate the matching coefficient between it and the template. The matching coefficient R (x, y) is the sum of the matching coefficients of all corresponding points. The calculation formula is as follows:

[0024]

[0025] Where Mm(i,j) represents the coordinates of a point on the metal region Mm, and B(x,y) represents the coordinates of the corresponding point on the search region B;

[0026] c. Obtain the coordinate Lm of the template in the search area when the minimum matching coefficient is obtained, and obtain the position correction coefficient C=Lm-L accordingly.

[0027] The expansion process in step (53) is specifically to expand the edge of the metal area by 3 pixels.

[0028] The step (5) further includes filtering the interpolated image:

[0029] Compare the pixel values ​​of the corresponding points on the interpolated image and the original projection sequence image. If the pixel value of a point on the interpolated image is less than the pixel value of the corresponding point on the original projection sequence image, the pixel value of the corresponding point on the original projection sequence image is used to replace the pixel value of the point on the interpolated image.

[0030] Beneficial effects: The present invention introduces an image restoration method to compare the reprojected metal image with the original image, and uses the information of the original image to correct the reprojected image so that it can cover the real metal area as much as possible to completely remove the metal artifacts. At the same time, to prevent over-correction, after the metal area is removed by image interpolation, it is compared with the original projection image again to correct the non-metal area that is incorrectly covered. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is a flow chart of the method for removing metal artifacts of the present invention;

[0032] Figure 2 This is a schematic diagram of metal area position correction;

[0033] Figure 3 Schematic diagram of template matching. DETAILED DESCRIPTION

[0034] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments.

[0035] Figure 1 is a flow chart of the method for removing metal artifacts of the present invention, as shown in Figure 1 As shown, the method for removing metal artifacts in cone beam CT of the present invention comprises the following steps:

[0036] (1) Acquire images through cone beam CT to obtain the original projection sequence image W;

[0037] (2) The original projection sequence image W is reconstructed using the FDK algorithm to obtain a reconstructed image D containing metal artifacts;

[0038] (3) extracting the metal region image Dm from the reconstructed image D by using a threshold method; wherein the threshold method is to extract pixels whose grayscale values ​​in the reconstructed image D are greater than a set threshold h, thereby obtaining the metal region image Dm; wherein the set threshold h is a fixed value, generally more than 10 times the pixel value of the non-metal region;

[0039] (4) projecting the metal region image Dm obtained in step (3) onto the detector plane corresponding to each image in the original projection sequence image W through Redon transformation to obtain the corresponding metal projection sequence image Wm;

[0040] (5) Correcting each metal region Mm in the metal projection sequence image Wm obtained in step (4) to eliminate metal artifacts caused by reconstruction distortion or insufficient precision, and obtaining a corrected metal region Mm′; the correction steps are as follows:

[0041] (51) Figure 2 As shown, the image connected domain segmentation method is used to segment and extract each metal region Mm of each image in the metal projection sequence image Wm as a template, and each metal region coordinate set L of each image is recorded, wherein the metal region coordinate set L is a set of coordinates of all pixel points on the metal region;

[0042] (52) Using the metal area Mm obtained in step (51) in the original projection sequence image W to perform template matching, determine the position correction coefficient; the specific operation method is:

[0043] a. Frame the search area B at the corresponding metal area on the original projection sequence image W;

[0044] b. Slide the corresponding template obtained in step (51) in the search area B and calculate the matching coefficient (i.e., traverse the search area B to obtain the corresponding area and calculate its matching coefficient with the template). The matching coefficient R (x, y) is the sum of the matching coefficients of all corresponding points, and the calculation formula is as follows:

[0045]

[0046] Wherein, Mm(i,j) represents the coordinates of a point on the metal area Mm, and B(x,y) represents the coordinates of the corresponding point on the search area; in this step, other methods can also be used to calculate the template matching coefficient;

[0047] c. Obtain the coordinate Lm of the template on the search area B when the minimum matching coefficient is obtained, and obtain the position correction coefficient C = Lm-L;

[0048] (53) Correcting the position of the metal area Mm according to the position correction coefficient C obtained in step (52) to obtain the metal area Mm′ in the corrected metal projection sequence image Wm, and performing expansion processing on it (e.g., expanding the edge by 3 pixels) to cover the real metal area as much as possible;

[0049] (6) Performing linear interpolation on the metal region Mm′ obtained in step (53) on the original projection sequence image W to remove it and obtain an interpolated image Wo, thereby eliminating the metal region;

[0050] (7) Compare the pixel values ​​of each pixel point of the interpolated image Wo and the original projection sequence image W one by one. If the pixel value of a pixel on the interpolated image Wo is smaller than the pixel value of the corresponding pixel on the original projection sequence image W, it means that this is not a metal projection area. Replace the interpolated value with the pixel value corresponding to the original projection sequence image W to obtain the final filtered projection sequence image Wr.

[0051] (8) reconstructing the projection sequence image Wr obtained in step (7) using the FDK algorithm to obtain an image Do with metal and metal artifacts removed;

[0052] (9) The metal area image Dm obtained in step (3) is overlapped with the image Do obtained in step (8) after removing metal and metal artifacts to obtain the final result Dr, which is the projection sequence image after removing metal artifacts.

[0053] The present invention is applicable to cone beam CT and spiral CT. By introducing the image restoration method, the reprojected metal image is compared with the original image, and the reprojected image is corrected using the information of the original image so that it can cover the real metal area as much as possible to completely remove the metal artifacts. At the same time, to prevent over-correction, after the metal area is removed by image interpolation, it is compared with the original projection image again to correct the non-metal area that is incorrectly covered.

[0054] The preferred embodiments of the present invention are described in detail above, but the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations (such as quantity, shape, position, etc.) can be made to the technical scheme of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A method for removing metal artifacts from CT images, characterized in that: Includes steps: (1) Obtaining original projection sequence images through CT imaging equipment; (2) reconstructing the original projection sequence images to obtain a reconstructed image containing metal artifacts; (3) extracting the metal area in the reconstructed image obtained in step (2); (4) projecting the metal area obtained in step (3) onto the projection plane corresponding to each image in the original projection sequence image to obtain the corresponding metal projection sequence image; (5) correcting each metal region in the metal projection sequence image obtained in step (4), and performing linear interpolation on the corrected metal region on the original projection sequence image to obtain an interpolation image, thereby removing the metal region; The correction of step (5) is specifically as follows: (51) using the image connected domain segmentation method to segment and extract each metal region of each image in the metal projection sequence image as a template, and recording the coordinates L of each metal region of each image; (52) performing template matching on the metal area obtained in step (51) in the original projection sequence image to determine a position correction coefficient; (53) Correcting the position of the metal area according to the position correction coefficient obtained in step (52) and performing expansion processing on it; (6) Reconstructing the interpolated image obtained in step (5) to obtain an image with metal and metal artifacts removed, and superimposing it with the metal area image obtained in step (3) to obtain a final projection sequence image with metal artifacts removed.

2. The method for removing metal artifacts from CT images according to claim 1, characterized in that: In the steps (2) and (6), the FDK algorithm is used to perform image reconstruction.

3. The method for removing metal artifacts from CT images according to claim 1, characterized in that: In the step (3), a threshold method is used to extract the metal area in the reconstructed image.

4. The method for removing metal artifacts from CT images according to claim 1, characterized in that: In the step (52), the position correction coefficient is determined as follows: a. Frame the search area at the corresponding metal area on the original projection sequence image; b. Slide the corresponding template obtained in step (51) in the search area to obtain the corresponding area and calculate the matching coefficient between it and the template. The matching coefficient R (x, y) is the sum of the matching coefficients of all corresponding points. The calculation formula is as follows: Where Mm(i,j) represents the coordinates of a point on the metal region Mm, and B(x,y) represents the coordinates of the corresponding point on the search region B; c. Obtain the coordinate Lm of the template in the search area when the minimum matching coefficient is obtained, and obtain the position correction coefficient C=Lm-L accordingly.

5. The method for removing metal artifacts from CT images according to claim 1, characterized in that: The expansion process in step (53) is specifically to expand the edge of the metal area by 3 pixels.

6. The method for removing metal artifacts from CT images according to claim 1, characterized in that: The step (5) further includes filtering the interpolated image: Compare the pixel values ​​of the corresponding points on the interpolated image and the original projection sequence image. If the pixel value of a point on the interpolated image is less than the pixel value of the corresponding point on the original projection sequence image, the pixel value of the corresponding point on the original projection sequence image is used to replace the pixel value of the point on the interpolated image.

Images