A method for eliminating truncation artifacts in limited-viewing-angle tomography
By acquiring and processing projected data in finite viewing CT imaging, and using gradient information to complete the data, the problem of truncation artifacts at the edges is solved, and high-quality CT reconstruction images are achieved.
Patent Information
- Application Number
- CN202210145816.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-02-17
- Publication Date
- 2025-06-06
- Estimated Expiration
- 2042-02-17
AI Technical Summary
In finite viewing CT imaging, discontinuous strip-like truncation artifacts often appear at the edges, affecting image quality.
By placing an X-ray emitter and receiver up and down the target tissue, projection data is collected within a limited angle, reconstruction and projection transformation are performed to obtain complete projection data, gradient information is taken to complete the original projection data, and finally the completed data is reconstructed to eliminate truncation artifacts.
Effectively eliminates truncation artifacts at the edges in finite viewing CT imaging, improves image quality, makes the reconstructed image more realistic, retains the information covered by artifacts, and avoids jagged artifacts at the edges of the image.
Smart Images
Figure CN114549680B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of image processing, and in particular to a method for eliminating truncation artifacts in limited viewing angle tomographic imaging. Background Art
[0002] Traditional CT (Computed Tomography) imaging technology uses a conical X-ray emitter to emit X-rays at a certain angle around the target tissue at 180°, and receives the signal on the other side of the target tissue through an X-ray receiver, thereby obtaining several sets of projection data. The internal structure of the target tissue is reconstructed based on the different absorption or attenuation coefficients of X-rays in different parts of the target tissue.
[0003] In actual use, due to the differences in the target tissue structure and the constraints of the application environment, it is usually difficult to scan the target tissue in a full angle range. Therefore, limited-angle CT imaging technology is often used in actual applications. Unlike traditional CT technology that requires the acquisition of a complete 180° data set, limited-angle CT imaging technology usually reconstructs images based on projection data obtained within a limited angle range of 15° to 60°. Traditional limited-angle CT reconstruction methods include iterative algorithms and non-iterative algorithms. Non-iterative algorithms include BP (Back Projection) algorithms and FBP (Filtered BackProjection) algorithms. The BP algorithm has a fast reconstruction speed, but the reconstructed image quality is poor. The FBP algorithm uses a filter on the basis of the BP algorithm to filter out noise information in the BP reconstruction process, which can provide satisfactory image quality while providing fast calculation speed. However, this method is limited by the geometric shape of the imaging. For example, for some systems that use a very narrow projection angle range, non-uniform angle sampling schemes, and fixed detectors, the calculation of FBP is more complicated, and the calculation error will also reduce the image quality. Iterative algorithms, such as SART (Simultaneous Algebraic Reconstruction Technique), and statistical model-based algorithms such as ML-convex (Maximum Likelihood convex) can adapt to limited-viewing tomography without being limited to geometric shapes, and can effectively enhance the edges and contrast of target tissue images. With the upgrade of computer hardware and the improvement of parallel computing efficiency, the reconstruction speed of iterative algorithms has also been improved to a certain extent.
[0004] However, due to the limited size of the detector and the point light source, the receiver cannot receive complete projection data during some large-angle projections, and the corresponding PVs (projection views) are truncated. After updating the truncated PVs, the voxel values outside the FOV (field of view) are not updated, resulting in discontinuous edge voxel values. If no correction is performed, the reconstructed image will have TPA (truncated projection artifacts). TPA obviously affects the quality of the reconstructed image. Summary of the invention
[0005] Purpose of the invention: The technical problem to be solved by the present invention is to provide a method for eliminating truncation artifacts in limited-view CT imaging results in view of the problem of discontinuous strip-shaped truncation artifacts appearing at the edges.
[0006] To achieve the above object, the specific steps of the present invention are as follows:
[0007] Step 1, placing an X-ray transmitter and a receiver above and below the target tissue respectively;
[0008] Step 2, using an X-ray transmitter to move around the tissue within a limited angle and emit X-rays, while the receiver collects raw projection data;
[0009] Step 3, reconstructing the original projection data to obtain a reconstructed image;
[0010] Step 4, performing projection transformation on the reconstructed image to obtain complete projection data; obtaining gradient information of the complete projection data;
[0011] Step 5, using the gradient information to complete the original projection data to obtain completed projection data;
[0012] Step 6: reconstruct the completed projection data to obtain a limited angle CT reconstructed image with truncation artifacts eliminated.
[0013] In the present invention, preferably, in step 4, the SART reconstructed image is subjected to projection transformation to obtain complete projection data; the complete projection data is not truncated. Obtaining gradient information from the complete projection data includes:
[0014] According to the relative positions of the complete projection data and the original projection data, the complete projection data and the original projection data are centrally aligned, and the required part of the complete projection data is extracted. This part consists of two blocks, the left and the right, which are respectively the truncation point of the original projection data relative to the complete projection data to the non-zero points on both sides of the complete projection data, recorded as the left reconstructed projection data f L and the right reconstructed projection data fR ;
[0015] Reconstruct the projection data f on the left L and the right reconstructed projection data f R Take the x-direction gradient information and reconstruct the projection data f on the left L In the figure, the pixel (x l ,y n ) The gradient value g in the x direction L (x l ,y n )for:
[0016] g L (x l ,y n ) = f L (x l+1 ,y n )-f L (x l ,y n )
[0017] Reconstruct the projection data f on the right side R , pixel (x r ,y n )The gradient value in the x direction is:
[0018] g R (x r ,y n ) = f R (x r ,y n )-f R (x r-1 ,y n )
[0019] Where l represents the left reconstructed projection data f L The position index in the x direction, r represents the right reconstructed projection data f R The position index in the x direction, n represents the position index in the y direction.
[0020] By performing a projection transformation on the reconstructed image, the area of the receiver can be artificially enlarged so that the projection data is no longer truncated, thereby fundamentally avoiding the generation of artifacts. In addition, the projection image obtained by this method has more commonalities with the original projection data, which can better complement the original projection data.
[0021] In the present invention, preferably, in step 5, according to the left reconstructed projection data f L and the right reconstructed projection data f R The gradient information of the original projection data is used to complete the missing parts of the original projection data from the inside to both sides to obtain the completed projection data.
[0022] In the present invention, preferably, step 5 comprises:
[0023] The completed projection data is recorded as f 0 , the completed projection data f 0 In the figure, the middle is the original projection data; the original projection data is completed from the middle to both sides, and the parts that need to be completed on both sides are iterated as follows:
[0024] For the left area y n The row data includes:
[0025] f 0 (x l+1 ,y n ) = f 0 (x l ,y n )+g L (x l ,y n )
[0026] Where l = l 0 is the original projection data, and the completed value is obtained according to the above formula According to the above formula, we can get The value of , and so on, until the end of the iteration;
[0027] For the right area y n The row data includes:
[0028] f 0 (x r-1 ,y n ) = f 0 (x r ,y n )-g R (x r ,y n )
[0029] Here, r = r 0 is the original projection data, and the completed value is obtained according to the above formula According to the above formula, we can get The value of , and so on, until the end of the iteration;
[0030] The above-mentioned left area and right area are iterated for each row to complete the original projection data.
[0031] In step 5, gradient information is used to complete the projection data to avoid discontinuity at the truncation of the projection data. Compared with the existing projection data expansion method, the data completion method we use can better restore the truncated projection data, and the restored data is closer to the true value, so that a good image can be reconstructed in the end. In terms of eliminating artifacts, compared with the existing method of using pixel compensation, the image reconstructed by this application is more realistic, and the information covered by the artifact can be completely retained, and at the same time, jagged artifacts will not be caused at the edge of the image.
[0032] In the present invention, preferably, step 7 is further included, and steps 4 to 6 are cycled more than once to obtain a limited angle CT reconstructed image with truncation artifacts eliminated. During the cycle, the reconstructed image input to step 4 is the limited angle CT reconstructed image with truncation artifacts eliminated obtained in step 6 in the previous cycle. When the slice at a lower position is concerned, steps 1 to 6 can effectively eliminate the strip truncation artifacts at the edge of the image; when the slice at a higher position is concerned, step 7 is executed to cycle steps 4 to 6 multiple times, which can effectively reduce the strip truncation artifacts at the edge of the high-level image.
[0033] In the present invention, preferably, in step 1, the X-ray transmitter should be located above the receiver so that the X-ray beam covers the target tissue as much as possible.
[0034] In the present invention, preferably, in step 2, a straight line in the receiver plane is used as the rotation axis, and the X-ray transmitter moves in a limited angle in an arc shape around the axis in a motion plane perpendicular to the receiver plane, and the angle range is represented by the center angle determined by the boundary points of the transmitter moving left and right, wherein the left and right boundary points are symmetrical about the center of the trajectory. During the acquisition process, the X-ray transmitter moves from the initial endpoint to the other motion endpoint, and emits X-rays to the receiver every time it moves a certain angle, and the receiver is deemed to have acquired a set of projection data after receiving the data. The scanning process, i.e., step 2, ends, and several sets of original projection data are acquired.
[0035] In the present invention, preferably, in step 3, the SART reconstruction algorithm is used to reconstruct the reconstruction area so that the final result meets the optimization conditions and obtains the first limited viewing angle CT imaging result. At this time, the imaging result has strip truncation artifacts at the edge.
[0036] In the present invention, preferably, in step 6, the completed projection data is reconstructed using a SART algorithm to obtain a limited angle CT reconstructed image that eliminates truncation artifacts.
[0037] Beneficial effects:
[0038] The present invention proposes a method for eliminating truncation artifacts in limited-viewing angle tomography. The method is based on a traditional SART reconstruction algorithm, uses the gradient information of the projection value of the reconstructed image, complements the truncated original projection data, and performs SART reconstruction on the complemented projection data to obtain a limited-viewing angle CT reconstruction result with truncation artifacts eliminated. The method solves the problem of strip-shaped truncation artifacts at the edge of the reconstructed image caused by the receiver being unable to receive all X-rays in CT image reconstruction under limited viewing angle. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments, and the above and / or other advantages of the present invention will become more clear.
[0040] Figure 1 is a flow chart of the method of the present invention.
[0041] Figure 2 It is a schematic diagram of the original projection data acquisition process of the present invention.
[0042] Figure 3 It is the original projection data in the embodiment of the present application.
[0043] Figure 4 This is a diagram obtained by performing the initial SART algorithm reconstruction on the original projection data in the embodiment of the present application.
[0044] Figure 5 It is a projection data diagram of the initial reconstruction diagram in the embodiment of the present application.
[0045] Figure 6 It is a schematic diagram of the completion process in the embodiment of the present application.
[0046] Figure 7 It is a projection data diagram after completion in the embodiment of the present application.
[0047] Figure 8 This is a diagram obtained by reconstructing the completed projection data using the SART algorithm in the embodiment of the present application. DETAILED DESCRIPTION
[0048] The embodiments of the present invention will be described below in conjunction with the accompanying drawings.
[0049] like Figure 1 As shown, the present invention proposes a method for eliminating truncation artifacts in limited viewing angle tomographic imaging, comprising the following steps:
[0050] Step 1, placing an X-ray transmitter and a receiver above and below the target tissue respectively;
[0051] Step 2, using an X-ray transmitter to move around the tissue within a limited angle and emit X-rays, while the receiver collects and obtains raw projection data;
[0052] Step 3, reconstructing the original projection data to obtain a reconstructed image;
[0053] Step 4, performing projection transformation on the reconstructed image to obtain complete projection data; obtaining gradient information of the complete projection data;
[0054] Step 5, using the gradient information to complete the original projection data to obtain completed projection data;
[0055] Step 6: reconstruct the completed projection data to obtain a limited angle CT reconstructed image with truncation artifacts eliminated.
[0056] In this embodiment, in step 1, the X-ray receiver is 23.04 cm long and 19.20 cm wide, and the size of each pixel is a square with a side length of 0.1 mm. Therefore, the projection data received each time is a plane image with a resolution of 1920×2304. The target tissue is placed between the X-ray transmitter and the receiver, close to the receiver plane, with a distance of 2 cm between the bottom surface of the object and the receiver plane, and a distance of 64 cm from the X-ray transmitter to the rotation fulcrum. The placement of the X-ray transmitter, receiver and object is as follows: Figure 2 shown.
[0057] In this embodiment, step 2, as Figure 2 As shown, the center of the arc motion trajectory is located at point P, the rotation fulcrum P′ is the projection of point P on the rotation axis, the initial position of the X-ray emitter is located at the left end point L of the motion trajectory, and the end point of the motion is the right end point R. In the motion plane, the X-ray emitter takes point P′ as the rotation center and makes an arc motion around this point. The central angle of the arc is 60°. The left end point L and the right end point R are symmetrical about PP′ and each forms an angle of 30° with the central axis. The X-ray emitter makes an arc motion from the left end point L to the right end point R, and emits an X-ray to the X-ray receiver every 3°. The data received by the receiver is used as a set of projection values of the angle. When the emitter reaches point R, the scanning process ends, and a total of 21 sets of projection data are collected. The original projection data in this embodiment are shown as follows. Figure 3 shown.
[0058] In this embodiment, in step 3, the collected raw projection data are initially reconstructed using the SART reconstruction algorithm. The SART algorithm iteratively finds the solution x of the following equation:
[0059] Ax=b (1)
[0060] In formula (1), A is an M×N matrix, which is the system matrix of the projection transformation in the image reconstruction process. x is an N-dimensional column vector, which is composed of the values of the voxels to be reconstructed, representing the attenuation coefficient of each voxel in the target tissue to X-rays, N is the number of reconstructed voxels, b is an M-dimensional column vector, representing the projection value, and M is the total number of collected projection pixels. In limited-angle CT imaging, the value of M is generally much smaller than N, so this equation has an array solution. The SART algorithm first gives an initial estimate x′ for x, then projects this estimate to obtain an estimate b′ of the projection vector b, and then uses the projection value b measured at the first angle 1 The difference b from b′ 1 -b′ is used as the correction value to correct x′. For all 21 sets of projection values, the correction of x′ is completed as one iteration, and finally x′ converges to The smallest x is the least square solution. According to the SART algorithm theory, the introduction of the weighted matrix W can effectively reduce artifacts.
[0061] Therefore, we need to find the value of x that minimizes the following loss function L(x), as shown in formula (2):
[0062]
[0063] Using the gradient descent method to find the x that minimizes L(x) gives the iterative formula of the SART algorithm:
[0064]
[0065] in, represents the value of the jth voxel after the k+1th iteration, i is the number of the projection value taken, λ is the relaxation factor, and changing λ affects the convergence speed and reconstruction results. J represents the number of voxels passed by each X-ray, 1≤j≤J; m represents the number of times the jth voxel is updated during this iteration, M i Indicates the total number of times the jth voxel is updated during this iteration, 1≤m≤M i , A mj,i It represents the weight of the jth voxel when the projection value number is i during the mth update in this iteration. To ensure convergence, λ is set to 0.1 in this example and iterates 5 times. To speed up the calculation, the original projection data is proportionally reduced to a resolution of 480×576.
[0066] like Figure 4 As shown, the initial reconstruction result is obtained. In this example, the size of the reconstruction result is 480×576×100. The reconstructed image is a three-dimensional image, and the three-dimensional image is cut into 100 two-dimensional images for processing. Figure 4 It can be seen that there are obvious truncated projection artifacts at the edges.
[0067] In this embodiment, in step 4, projection transformation is performed on the initial reconstruction result to obtain a complete projection data map, such as Figure 5 As shown. The projection transformation of the initial reconstruction result is calculated using the existing technology known to those skilled in the art, and can be completed by a computer. Therefore, compared with the scanning stage in step 2, there is no restriction on the size of the receiver, and complete projection data is obtained. Therefore, the relative position of the original projection data and the projection data of the reconstructed image is known. The new projection image and the original projection data are centrally aligned, and the required part of the projection data of the reconstructed image is extracted. This part is composed of two blocks, left and right, which are the truncation point of the original projection image relative to the reconstructed projection image to the non-zero points on both sides of the reconstructed projection image, and the x-direction gradient information is obtained for this part. As shown Figure 6 As shown in the figure, the gradient is calculated by subtracting the pixel value of a point to the right from the pixel value of that point. L In the middle, point (x l ,y n )The gradient value in the x direction is:
[0068] g L (x l ,y n ) = f L (x l+1 ,y n )-f L (x l ,y n ) (4)
[0069] In formula (4), g L Represents the gradient information of the pixels in the left area. It is a scalar. The subscript L indicates that it is in the left area.
[0070] For the right reconstructed projection data f R , point (x r ,y n )The gradient value in the x direction is:
[0071] g R (x r ,y n ) = f R (x r ,y n )-f R (x r-1 ,y n ) (5)
[0072] In formula (5), g R Represents the gradient information of the pixels in the right area. It is a scalar. The subscript R indicates that it is in the right area.
[0073] In this embodiment, in step 5, after obtaining the gradient information, the original projection data is supplemented from the middle to both sides. Reconstructing the area f 0 In the figure, the center is the original projection data, and the parts that need to be completed on both sides are iterated as follows:
[0074] For the left area y n The row data includes:
[0075] f 0 (x l+1 ,y n ) = f 0 (x l ,y n )+g L (x l ,y n ) (6)
[0076] Where l = l 0 The original projection data is shown in the figure. The completed value is obtained according to formula (6): According to formula (6), we can get And so on until the end of the iteration.
[0077] For the right area y n The row data includes:
[0078] f 0 (x r-1 ,y n ) = f 0 (x r ,y n )-g R (x r ,y n ) (7)
[0079] Where r = r 0 The original projection data is shown in the figure. The completed value is obtained according to formula (7): According to formula (7), we can get And so on until the end of the iteration.
[0080] Reconstruction area f 0 Repeat the above steps for each row of to complete the original projection data. In this example, the completed projection image is as follows Figure 7 shown.
[0081] In this embodiment, in step 6, the completed projection data is reconstructed using the SART algorithm to obtain a new reconstructed image.
[0082] In this embodiment, step 7 is also included, and steps 4 to 6 are repeated multiple times to reduce the truncation artifacts of the high-level part of the image and make the contour of the internal tissue clearer. The SART reconstructed image input in step 4 during the loop is the limited angle CT reconstructed image obtained in step 6 in the previous loop with the truncation artifacts eliminated. According to the residual situation of the truncation artifacts, when the slice with a higher position is concerned, this embodiment is repeated 3 times, and finally a limited angle CT reconstruction result with the truncation projection artifacts at the edge is obtained, such as Figure 8 shown.
[0083] The present invention provides a method for eliminating truncation artifacts in limited-viewing-angle tomography. There are many methods and ways to implement the technical solution. The above is only a specific implementation of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the protection scope of the present invention. All components not specified in this embodiment can be implemented by existing technologies.
Claims
1. A method for eliminating truncation artifacts in limited-viewing-angle tomographic imaging, It is characterized in that The following steps are involved: Step 1, placing an X-ray transmitter and a receiver above and below the target tissue respectively; Step 2, using an X-ray transmitter to move around the tissue within a limited angle and emit X-rays, while the receiver collects raw projection data; Step 3, reconstructing the original projection data to obtain a reconstructed image; Step 4, performing projection transformation on the reconstructed image to obtain complete projection data; Extracting the required partial gradient information from the complete projection data; Step 5, using the gradient information to complete the original projection data to obtain completed projection data; Step 6, reconstructing the completed projection data to obtain a limited angle CT reconstructed image with truncation artifacts eliminated; The gradient information of the required part of the complete projection data extracted in step 4 includes: According to the relative positions of the complete projection data and the original projection data, the complete projection data and the original projection data are centrally aligned, and the required part of the complete projection data is extracted. This part consists of two blocks, the left and the right, which are respectively the truncation point of the original projection data relative to the complete projection data to the non-zero points on both sides of the complete projection data, recorded as the left reconstructed projection data. And the right side reconstructed projection data ; Reconstruct projection data on the left side And the right side reconstructed projection data Take the x-direction gradient information and reconstruct the projection data on the left In the pixel The gradient value in the x direction for: ， Reconstruct projection data on the right side , pixel The gradient value in the x direction is: ， Where l represents the left reconstructed projection data The position index in the x direction, r represents the right side reconstruction projection data The position index in the x direction, n represents the position index in the y direction; In step 5, the projection data is reconstructed based on the left side. And the right side reconstructed projection data The gradient information of the original projection data is used to fill in the missing parts from the inside to both sides to obtain the filled projection data; The step 5 comprises: The completed projection data is recorded as , the completed projection data In the figure, the middle is the original projection data; the original projection data is completed from the middle to both sides, and the parts that need to be completed on both sides are iterated as follows: For the left area The row data includes: ， Among them, is the original projection data, and the completed value is obtained according to the above formula , and then according to the above formula we can get until the iteration ends; For the right area The row data includes: ， Among them, is the original projection data, and the completed value is obtained according to the above formula , and then according to the above formula we can get until the iteration ends; The above-mentioned left area and right area are iterated for each row to complete the original projection data.
2. A method for eliminating truncation artifacts in limited viewing angle tomographic imaging according to claim 1, It is characterized in that The method further includes step 7, looping step 4 to step 6 for more than one time to obtain a limited angle CT reconstructed image with truncation artifacts eliminated, wherein the reconstructed image input into step 4 during the loop is the limited angle CT reconstructed image with truncation artifacts eliminated obtained in step 6 in the previous loop.
3. A method for eliminating truncation artifacts in limited viewing angle tomographic imaging according to claim 2, It is characterized in that In step 1, the X-ray transmitter is located above the receiver, so that the X-rays emitted by the transmitter can basically cover the entire target tissue.
4. A method for eliminating truncation artifacts in limited viewing angle tomographic imaging according to claim 3, It is characterized in that In step 2, the X-ray transmitter moves in an arc around the receiver in a plane perpendicular to the receiver, and transmits X-rays at regular angles within a limited angle range, while the receiver receives the X-rays to obtain original projection data.
5. A method for eliminating truncation artifacts in limited viewing angle tomographic imaging according to claim 4, It is characterized in that The step 3 uses the SART reconstruction algorithm to perform three-dimensional reconstruction on the collected original projection data to obtain an initial SART reconstructed image.
6. A method for eliminating truncation artifacts in limited viewing angle tomographic imaging according to claim 5, It is characterized in that In step 6, the SART algorithm is used to reconstruct the completed projection data to obtain a limited angle CT reconstructed image with truncation artifacts eliminated.