CT image ring artifact removal method based on data fitting and statistical sorting
Through the method based on data fitting and statistical sorting, the mean projection fitting and one-dimensional filtering technology are used to remove the ring artifacts in CT images, solving the problems of high complexity and poor generalization in the existing technology, and achieving rapid , accurate artifact removal effect.
Patent Information
- Application Number
- CN202411050586.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-01
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-08-01
Smart Images

Figure CN119006324A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of CT imaging, and in particular to a method for removing ring artifacts from CT images based on data fitting and statistical sorting. Background Art
[0002] X-ray computed tomography (CT) scanning obtains CT images of objects by reconstructing projection data. In this process, due to defects in detector units, inconsistent responses and other factors, some abnormal stripe elements that are different from the surrounding elements will appear in the projection image, resulting in ring artifacts in the reconstructed CT image. As the name suggests, a ring artifact is a cluster of abnormal circular elements with different radii centered at the reconstruction center. Ring artifacts reduce the quality of CT images, thereby affecting subsequent image processing and analysis tasks such as segmentation and quantitative analysis. Therefore, accurate and rapid correction of ring artifacts is of great practical significance. At present, according to the specific properties of ring artifacts in projection data and reconstructed images, common techniques for removing ring artifacts in CT images can be divided into two categories: projection domain methods and reconstructed image domain methods.
[0003] The projection domain method removes ring artifacts by correcting abnormal stripes in the projection data. It can be seen from the formation mechanism of ring artifacts that effectively removing abnormal linear elements in the projection data can reduce the impact of ring artifacts on CT images. The idea of transform domain is used to remove ring artifacts, which mainly includes wavelet transform and Fourier transform. Initially, people analyzed the properties of abnormal stripes in the projection data in the frequency domain (obtained by Fourier transform) and designed a corresponding numerical truncation method with a fixed threshold to remove artifacts. In addition, the combination of wavelet transform and frequency domain filtering technology is used to correct abnormal stripes in projection data. Among them, discrete wavelet transform is used to extract feature images in projection data, and Gaussian filter in frequency domain corrects abnormal line elements in a smooth manner. In addition, some researchers are committed to studying the function model corresponding to the numerical value of abnormal stripes and using this model to realize artifact correction. In this type of algorithm, the mean data received by the detector unit is often used to build the model. At the same time, treating abnormal stripe elements as noise and designing a reasonable filtering algorithm is also an idea to remove ring artifacts. If the projection domain method can be successful, the ring artifacts in CT images can be significantly removed. However, the complexity of the ring artifacts is a major challenge for these algorithms.
[0004] The reconstructed image domain method corrects the ring artifact by directly processing the abnormal circular elements in the CT image. On the one hand, the ring artifact is "simple" because it has a very significant geometric property in the CT image, that is, it has a circular property. Based on this property, some works realize the property conversion of abnormal circular elements by converting between Cartesian coordinate system and polar coordinate system, and use efficient algorithms to correct the converted abnormal linear elements. In addition, some research works directly use the circular property of the ring artifact to design related artifact removal algorithms. For example, the directional total variation method implemented by elliptical filter. At the same time, some deep learning methods are also used to remove ring artifacts. However, such methods are heavily dependent on the construction of training data sets, etc., and the generalization of the methods is relatively poor. In addition, the polar coordinate conversion method does not work well when the ring artifact is relatively complex. This is mainly because the numerical values of the reconstructed artifacts are generally complex.
[0005] Therefore, the prior art needs to be improved. Summary of the invention
[0006] The technical problem to be solved by the present invention is to provide a method for removing ring artifacts from CT images based on data fitting and statistical sorting in response to the defects of the prior art, aiming to solve the problems of high complexity, low generalization and unsatisfactory effect of the correction algorithm for removing ring artifacts from CT images in the prior art.
[0007] The technical solution adopted by the present invention to solve the problem is as follows:
[0008] In a first aspect, an embodiment of the present invention provides a method for removing ring artifacts in CT images based on data fitting and statistical sorting, wherein the method comprises:
[0009] Acquiring original projection data, and pre-correcting the original projection data using a fitting algorithm based on mean projection to obtain pre-corrected projection data;
[0010] A one-dimensional filtering algorithm based on statistical sorting is used to perform secondary correction on the pre-corrected projection data to obtain corrected projection data.
[0011] In one implementation, the original projection data includes original sub-projection data received by a plurality of detector units, and the pre-correcting the original projection data using a mean projection-based fitting algorithm to obtain pre-corrected projection data includes:
[0012] Calculating a mean projection of the raw sub-projection data received by each detector unit;
[0013] Performing function fitting using the mean projection of the original sub-projection data received by each detector unit as a control point set to obtain a fitting function;
[0014] calculating an ideal mean projection for each detector cell using the fitting function;
[0015] Determining a mean projection error of each detector unit based on the mean projection of the original sub-projection data of each detector unit and the ideal mean projection data;
[0016] Correcting the original sub-projection data based on the mean projection error of each detector unit to obtain pre-corrected sub-projection data of each detector unit;
[0017] The pre-corrected projection data is obtained based on the pre-corrected sub-projection data of each detector unit.
[0018] In one implementation, the adopting a one-dimensional filtering algorithm based on statistical sorting to perform secondary correction on the pre-corrected projection data to obtain corrected projection data includes:
[0019] sorting the pre-correction sub-projection data of each detector unit;
[0020] Performing one-dimensional smoothing filtering on the sorted pre-correction sub-projection data of each detector unit to obtain filtered sub-projection data of each detector unit;
[0021] The sorting transformation matrix is used to perform data recovery on the filtered sub-projection data of all detector units to obtain corrected projection data.
[0022] In one implementation, before performing one-dimensional smoothing filtering on the pre-correction sub-projection data of each detector unit after the sorting, the method further includes:
[0023] Obtaining raw indexes of pre-corrected sub-projection data of each detector unit;
[0024] A sorting transformation matrix is generated based on the sorting result of the pre-correction sub-projection data and the original index of each detector unit, and the sorting transformation matrix is saved.
[0025] In one implementation, the function fitting is a local weighted linear function fitting.
[0026] In one implementation, the direction of performing the one-dimensional smoothing filtering is the direction of the detector unit.
[0027] In one implementation, the one-dimensional smoothing filter uses a one-dimensional Gaussian filter or a median filter.
[0028] In a second aspect, an embodiment of the present invention further provides a CT image ring artifact removal system based on data fitting and statistical sorting, wherein the system comprises:
[0029] A pre-correction module, used to obtain raw projection data, and pre-correct the raw projection data using a fitting algorithm based on mean projection to obtain pre-corrected projection data;
[0030] The secondary correction module is used to perform secondary correction on the pre-corrected projection data by using a one-dimensional filtering algorithm based on statistical sorting to obtain corrected projection data.
[0031] In one implementation, the pre-correction module includes:
[0032] a mean projection calculation unit, used for calculating the mean projection of the original sub-projection data received by each detector unit;
[0033] A function fitting unit, used for performing function fitting by taking the mean projection of the original sub-projection data received by each detector unit as a control point set to obtain a fitting function;
[0034] a data calculation unit, configured to calculate an ideal mean projection of each detector unit using the fitting function;
[0035] an error calculation unit, configured to determine a mean projection error of each detector unit based on a mean projection of the original sub-projection data of each detector unit and an ideal mean projection;
[0036] A data correction unit, configured to correct the original sub-projection data based on the mean projection error of each detector unit to obtain pre-corrected sub-projection data of each detector unit;
[0037] A data generating unit is used to obtain the pre-corrected projection data based on the pre-corrected sub-projection data of each detector unit.
[0038] In one implementation, the secondary correction module includes:
[0039] A sorting unit, used for sorting the pre-correction sub-projection data of each detector unit;
[0040] A filtering unit, used for performing one-dimensional smoothing filtering on the sorted pre-correction sub-projection data of each detector unit to obtain filtered sub-projection data of each detector unit;
[0041] The data recovery unit is used to recover the filter projection data of all detector units by using the sorting transformation matrix to obtain the corrected projection data.
[0042] In one implementation, the secondary correction module further includes:
[0043] An index acquisition unit, used to acquire an original index of the pre-correction sub-projection data of each detector unit;
[0044] The sorting transformation matrix generating unit is used to generate a sorting transformation matrix based on the sorting result of the pre-correction sub-projection data and the original index of each detector unit, and save the sorting transformation matrix.
[0045] In a third aspect, an embodiment of the present invention further provides a terminal device. The terminal device is a commercial display terminal, and the terminal device includes a memory, a processor, and a program of a method for removing ring artifacts from CT images based on data fitting and statistical sorting, which is stored in the memory and can be run on the processor. When the processor executes the program of the method for removing ring artifacts from CT images based on data fitting and statistical sorting, the steps of the method for removing ring artifacts from CT images based on data fitting and statistical sorting in any one of the above schemes are implemented.
[0046] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein a program of a method for removing ring artifacts from CT images based on data fitting and statistical sorting is stored on the computer-readable storage medium, and when the program of the method for removing ring artifacts from CT images based on data fitting and statistical sorting is executed by a processor, the steps of the method for removing ring artifacts from CT images based on data fitting and statistical sorting described in any one of the above schemes are implemented.
[0047] Beneficial effects of the present invention: Compared with the prior art, the present invention provides a method for removing ring artifacts from CT images based on data fitting and statistical sorting. The present invention first obtains original projection data, and uses a fitting algorithm based on mean projection to pre-correct the original projection data to obtain pre-corrected projection data; then, a one-dimensional filtering algorithm based on statistical sorting is used to perform secondary correction on the pre-corrected projection data to obtain corrected projection data. The present invention first utilizes the numerical mutation properties of abnormal stripes in the projection data, and adopts the idea of local weighted linear function fitting to weaken the abnormal mutation, thereby removing the pseudo-texture information in the projection data caused by the abnormal stripes to a certain extent, and providing a basis for the subsequent filtering algorithm to accurately correct the projection data, and then uses a smoothing filtering algorithm based on data sorting to perform detail correction on the pre-corrected projection data, and finally achieves rapid and accurate correction of ring artifacts in CT images. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments and the prior art description. Obviously, the drawings described below are only some embodiments recorded in the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0049] Figure 1It is a flow chart of a method for removing ring artifacts from CT images based on data fitting and statistical sorting provided in an embodiment of the present invention.
[0050] Figure 2 It is a diagram of the CT scanning and reconstruction process of the CT image ring artifact removal method based on data fitting and statistical sorting provided by an embodiment of the present invention.
[0051] Figure 3 It is a projection process diagram of a CT scan of a method for removing ring artifacts in CT images based on data fitting and statistical sorting provided by an embodiment of the present invention.
[0052] Figure 4 It is a schematic diagram of projection data and detector mean projection curves of a CT image ring artifact removal method based on data fitting and statistical sorting provided in an embodiment of the present invention.
[0053] Figure 5 It is a diagram showing the effect of ring artifact removal of a CT image ring artifact removal method based on data fitting and statistical sorting provided by an embodiment of the present invention.
[0054] Figure 6 The present invention provides a ring artifact correction flow chart of a CT image ring artifact removal method based on data fitting and statistical sorting.
[0055] Figure 7 It is a principle block diagram of a CT image ring artifact removal system based on data fitting and statistical sorting provided by an embodiment of the present invention.
[0056] Figure 8 It is a block diagram of the internal structure principle of a CT image ring artifact removal device based on data fitting and statistical sorting provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0057] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0058] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.
[0059] In the prior art, the CT scanning and reconstruction process can be modeled as Figure 2As shown in the figure, X represents the object projection data obtained by the detector during the scanning process, and Y represents the reconstructed CT image obtained by using the reconstruction algorithm P and the projection data X. The process can be specifically expressed as formula (1):
[0060] Y=P(X)(1)
[0061] However, during the projection process, due to the inconsistent response of the detector units and other reasons, there are some abnormal elements in the projection data X, which in turn leads to the appearance of ring artifacts in the reconstructed CT image Y, thus affecting the quality of the CT image. Figure 3 The figure shows the CT scan projection process of an object, and the abnormal detector unit (such as Figure 3 The data received by the detector unit (the dots in the image) has a fixed rule at all rotation angles. Therefore, the detector unit has problems in the entire scanning process, which leads to abnormal stripes in the projection data along the rotation angle direction, and the characteristics of these elements along the angle direction are relatively obvious. Obviously, when the abnormal stripes in the projection data are relatively simple, such as the difference between the values of the abnormal stripes and the normal values of the adjacent normal stripes is relatively small, and the number of abnormal stripes is relatively small, these abnormal elements can be filtered out by some filtering algorithms, thereby achieving the removal of weak artifacts, because the abnormal stripes at this time are not enough to form pseudo-structural information that affects the data structure, and there is more effective information at this time, which can ensure the normal operation of the filtering algorithm. However, when the abnormal stripes in the projection data are more complex, such as the difference between the values of the abnormal stripes and the normal values around them is large, and the number of abnormal stripes is relatively large, the filtering algorithm under normal circumstances cannot effectively correct the abnormal elements, because the abnormal stripes at this time will cause a large amount of pseudo-texture information in the data, and there is less effective data.
[0062] In order to solve the problems in the prior art, this embodiment provides a method for removing ring artifacts in CT images based on data fitting and statistical sorting, by which the ring artifacts in CT images can be quickly and accurately removed. In specific implementation, first, the original projection data is acquired, and the original projection data is pre-corrected by a fitting algorithm based on mean projection to obtain pre-corrected projection data; then, the pre-corrected projection data is secondary corrected by a one-dimensional filtering algorithm based on statistical sorting to obtain corrected projection data. The present invention first utilizes the numerical mutation properties of abnormal stripes in the projection data, and adopts the idea of local weighted linear function fitting to weaken the abnormal mutation of the data, thereby removing the pseudo-texture information of the projection data caused by the abnormal stripes to a certain extent, and providing a basis for the subsequent filtering algorithm to accurately correct the projection data, and then adopts a smoothing filtering algorithm based on data sorting to perform detail correction on the pre-corrected projection data, and finally achieves the rapid and accurate correction of the ring artifacts in the CT image.
[0063] For example, if the ring artifacts of CT images are to be removed, the original projection data is first obtained, and the original projection data is preliminarily corrected using a pre-correction algorithm to remove the abnormal stripes therein, and the pre-corrected projection data is obtained, wherein the pre-correction algorithm is generated based on the local weighted linear function fitting of the mean projection; after the abnormal stripes are filtered out by pre-correction, the pre-corrected projection data is then corrected twice using a one-dimensional filtering algorithm, that is, detail correction is performed to obtain the final corrected projection data. The filtering algorithm is generated based on statistical sorting. The present invention first utilizes the numerical mutation properties of the abnormal stripes in the original projection data, and adopts the idea of local weighted linear function fitting to weaken the abnormal mutation of the data, thereby removing the pseudo-texture information in the original projection data caused by the abnormal stripes to a certain extent, and providing a basis for the subsequent filtering algorithm to accurately correct the projection data, and then adopts a smoothing filtering algorithm based on data sorting to perform detail correction on the original projection data, and finally realizes the rapid and accurate correction of the ring artifacts of the CT image.
[0064] Exemplary Methods
[0065] This embodiment provides a method for removing ring artifacts from CT images based on data fitting and statistical sorting. The method can be applied to a terminal device for removing ring artifacts from CT images. Figure 1 As shown, the method includes:
[0066] Step S100: acquiring original projection data, and pre-correcting the original projection data by using a fitting algorithm based on mean projection to obtain pre-corrected projection data.
[0067] In this embodiment, it is necessary to remove the ring artifacts from the CT image. By using the geometric properties of the ring artifacts in the projection domain, the mutation characteristics of the data received by the abnormal detector unit are studied, and a pre-correction algorithm is designed to perform preliminary correction on the acquired original projection data to obtain the pre-processed original projection data. In the specific implementation, the original projection data includes original sub-projection data received by several detector units, and the pre-correction of the original projection data by the fitting algorithm based on the mean projection is performed to obtain the pre-corrected projection data, including: calculating the mean projection of the original sub-projection data received by each detector unit; performing function fitting with the mean projection of the original sub-projection data received by each detector unit as the control point set to obtain the fitting function; using the fitting function to calculate the ideal mean projection of each detector unit; determining the mean projection error of each detector unit based on the mean projection and the ideal mean projection of the original sub-projection data of each detector unit; correcting the original sub-projection data based on the mean projection error of each detector unit to obtain the pre-corrected sub-projection data of each detector unit; obtaining the pre-corrected projection data based on the pre-corrected sub-projection data of each detector unit. The pre-correction algorithm can weaken the mutation data caused by abnormal stripes to a certain extent by utilizing the smoothing property of the fitting function, thereby preliminarily realizing the correction of abnormal stripes in the projection image and obtaining the preprocessed corrected projection data, namely the pre-corrected projection data.
[0068] In the specific implementation, since the abnormal data that causes the ring artifact in the original projection data usually has line attributes in the direction of the rotation angle, visually, these lines have a relatively uniform numerical coverage based on the ideal value. Therefore, the average projection value received by each detector unit at all scanning angles is studied, such as Figure 4 The figure shows the projection data of CT scan and the detector mean projection curve. The background is the projection data, the vertical axis is the rotation angle, and the horizontal axis is the detector unit. Figure 4 The curve in is the average projection value received by all detector units. It can be seen that the abnormal fringes of the projection data cause a sudden peak in the curve of the detector average projection value (such as Figure 4), which results in the average projection curve being not smooth enough, and the degree of mutation of the peak is positively correlated with the numerical value of the corresponding abnormal stripes. In addition, if the curve disturbance caused by the abnormal peak is not considered, the change process of the mean projection curve of the detector unit can well describe the complexity of the information received by each detector unit. Therefore, before using the filtering algorithm to correct the abnormal stripes in the original projection data, it is reasonable to weaken the pseudo texture in the data caused by the abnormal stripes to a certain extent. At the same time, according to the analysis of the influence of the strip elements on the mean projection curve of the detector, it can be seen that the pre-correction of the stripe elements can be achieved by compressing the abnormal peaks of the curve. In order to accurately and effectively remove the abnormal peaks in the mean projection curve so that the corrected mean projection curve is smoother, the mean projection value of the detector unit can be used to perform local weighted linear function fitting, so as to obtain a function that is close to and smooth to the mean projection data. Then the fitting function can be approximated as the ideal mean projection curve, so as to find the difference between the abnormal mean projection curve and the ideal mean projection curve, and use the difference to remove the mutation peaks in the original mean projection curve, so as to achieve pre-correction of the abnormal stripes. For example, the original projection data obtained is set as Wherein, x represents the index of the detector unit, and y represents the angle of the detector unit during scanning. That is to say, the original projection data is composed of the original sub-projection data obtained by several detector units. The average projection data of the original sub-projection data of each detector unit at all angles is calculated, as shown in formula (2):
[0069]
[0070] Among them, N y Indicates the number of times the ray source rotates during scanning, N x Represents the number of units in the detector; after the mean projection of the original sub-projection data of all detector units is calculated, it can be used as the control point set, that is, As a control point set, a local weighted linear function fitting is performed on it to obtain a smooth fitting function f(t), as shown in formula (3):
[0071]
[0072] Among them, Fit represents the local weighted linear fitting function adopted in the data fitting process. The fitting of the mean projection curve in this application uses the mean projection of all detector units as the control point. This fitting method can detect abnormal detector units and their corresponding disturbance values; the obtained fitting function can be used as an ideal mean projection curve to estimate the approximate ideal mean projection data of each detector unit. The mean projection error of each detector unit can be obtained based on the approximate ideal mean projection data and the mean projection of the original sub-projection data, that is, the potential error in the mean projection of the original sub-projection data is calculated using the fitting function f(t), as shown in formula (4),
[0073]
[0074] Where f(x) represents the value of the fitting function f(t) at point x, which is regarded as the approximate ideal projection data. The difference Res(x) reflects the difference between the abnormal stripes and the ideal data mean projection to a certain extent. Finally, the error of each detector unit is used to correct the original sub-projection data, as shown in formula (5):
[0075]
[0076] Where S′ represents the pre-corrected projection data, and the definition of S′(x, y) indicates that for a detector unit x, at all angles y∈{0,…,N y}, the estimated error Res(x) is used for data correction, thereby weakening the pseudo-texture properties caused by abnormal lines, and removing the abnormal mutation peaks in the original mean projection curve, making it have a relatively smooth property; finally, all pre-corrected sub-projection data are combined to obtain pre-corrected projection data. At this point, the pre-correction process of abnormal stripes based on mean projection fitting is completed. Through this processing method, the error information caused by complex abnormal stripes can be weakened to a certain extent. However, pre-correction cannot completely complete the correction of ring artifacts, and subsequent filtering algorithms are still required to correct small abnormal elements.
[0077] In one implementation, the function fitting is a local weighted linear function fitting. The fitting function f(t) is usually smooth, does not contain sharp peaks, and can describe the change form of the control point set as accurately as possible. At the same time, the participation of abnormal data in the fitting process is reduced, while the participation of normal data in the fitting process is increased. Therefore, the fitting function can be regarded as an approximation of the ideal mean projection curve.
[0078] Step S200: performing secondary correction on the pre-corrected projection data using a one-dimensional filtering algorithm based on statistical sorting to obtain corrected projection data.
[0079] In this embodiment, after the original projection data is preprocessed to obtain the preprocessed projection data, in order to further correct the small amount of abnormal elements with small values remaining in the preprocessed projection data, the present application uses a one-dimensional filtering algorithm based on statistical sorting to achieve the final correction of the ring artifact. Since the pre-correction algorithm based on data fitting is a unified supplement or deletion operation on the data of all angles received by each detector unit, the abnormal elements in the pre-corrected data still exist in the form of lines, but the difference between the abnormal stripes and their adjacent data is not prominent at this time. Since normal projection data usually contains a large amount of weak detail information, in order to prevent the loss of data texture caused by the smoothing algorithm processing the lines, a one-dimensional filtering algorithm based on sorting is used to perform the final correction on the pre-corrected data, thereby effectively avoiding the loss of important data structure information caused by the smoothing operation. The filter can better restore the structural information in the image while removing the ring artifacts. In a specific implementation, the one-dimensional filtering algorithm based on statistical sorting is used to perform secondary correction on the pre-corrected projection data to obtain corrected projection data, including: sorting the pre-corrected sub-projection data of each detector unit; performing one-dimensional smoothing filtering on the sorted pre-corrected sub-projection data of each detector unit to obtain filtered sub-projection data of each detector unit; and using a sorting transformation matrix to perform data recovery on the filtered sub-projection data of all detector units to obtain corrected projection data. The specific processing steps of the one-dimensional filtering algorithm based on statistical sorting are: first, sort the pre-corrected sub-projection data of each detector unit at all angles in the manner shown in the following formula (6),
[0080] S′ s (x,y)=Sort(S′(X,y),2),x∈{0,…,N x} (6)
[0081] Among them, S′ s Represents the sorted pre-corrected projection data, Sort(S′(x, y), 2) is the formal expression of the sorting operation, where 2 means that only all the data of the second dimension of S′ is sorted, that is, each column vector of the matrix S′ (corresponding to the data at all angles) is sorted respectively. By sorting the pre-processed projection data at all angles received by the detector unit, similar data can be clustered. By clustering, more similar data can be involved in the filtering process, thereby effectively avoiding data loss and structural ambiguity caused by the presence of abnormal data. Then, a one-dimensional smoothing filter is performed on each pre-corrected sub-projection data. In this application, a one-dimensional filter along the detector direction is used to perform a one-dimensional smoothing filter on the pre-corrected sub-projection data to obtain the filtered sub-projection data of each detector unit, as shown in formula (7).
[0082]
[0083] Among them, S′ sf is the sorted projection data after filtering, Cov 1 It is a one-dimensional filter, and the filter has a radius only in the direction of the detector unit. This step can reduce the participation of abnormal fringe data in data recovery and improve the effect of data recovery. Finally, the sorted data after filtering is restored, and the sorting transformation matrix is used to restore the filtered sub-projection data of all detector units to obtain the final corrected projection data. The sorting transformation matrix is represented by Rank, as shown in formula (8),
[0084]
[0085] Among them, S" represents the projection data with normal order after correction, Rank -1 Is a formal expression, indicating the data S′ sf Restore in the original order.
[0086] In one implementation, before performing one-dimensional smoothing filtering on the pre-correction sub-projection data of each detector unit after the sorting, the method further includes: obtaining the original index of the pre-correction sub-projection data of each detector unit; generating a sorting transformation matrix based on the sorting result of the pre-correction sub-projection data and the original index of each detector unit, and saving the sorting transformation matrix. In a specific implementation, when sorting the pre-correction sub-projection data of each detector unit, the original index corresponding to each detector unit is obtained at the same time, and according to the sorting result, the original index is generated into a sorting transformation matrix, which is saved using Rank to facilitate the order restoration of the sorted data of the subsequent filter core.
[0087] In one implementation, the direction of the one-dimensional smoothing filter is the detector unit direction. The main reasons for using only the one-dimensional filter along the detector unit direction to correct abnormal data are: first, the data in the angle direction may be damaged by the abnormal data, and second, the sorting operation makes more similar data exist in the detector unit direction. Therefore, by performing one-dimensional smoothing filtering along the detector unit direction, effective correction of abnormal elements in the one-dimensional area can be achieved.
[0088] In one implementation, the one-dimensional smoothing filter uses a one-dimensional Gaussian filter and a median filter, both of which achieve good results.
[0089] In one implementation, due to special circumstances, a large number of control points can easily lead to overfitting problems. Therefore, according to the actual situation, the mean projections of some continuous detector units can be selected as control points, and segmented fitting can be performed to finally achieve the fitting and correction of all mean projection curves. This method is suitable for situations where the scanned object has complex information. The mean projection data can be fitted more accurately through segmented fitting, and the fitting error near the detector units that change dramatically can be effectively avoided. Therefore, when the projection data is relatively simple, the global mean projection data in this application can be used as the control point for fitting; when the projection data is relatively complex, the segmented fitting method can be used to improve the effect of pre-correction.
[0090] Figure 5 This is the experimental effect of the ring artifact correction achieved by this application, where: Figure 5 a is the original projection data containing abnormal fringe elements, Figure 5 b is the effect of removing the linear elements in the original projection data using the present invention. Figure 5 a and Figure 5 It can be seen from the comparison between FIG. 1 and FIG. 2 that the present invention can effectively correct and remove abnormal fringe elements in the projection data, and other valid information in the projection data is also effectively preserved and restored. Figure 5 c is the original projection data ( Figure 5 a) Reconstructed CT image, Figure 5 d is a CT image reconstructed using the corrected projection data. By comparison, it can be seen that the present invention can effectively remove the ring artifacts in the CT image, and the structural information in the image is also effectively restored.
[0091] In the present invention, Figure 6 The flowchart of the overall framework is a combination of the pre-correction of abnormal stripes based on data fitting and the one-dimensional filtering algorithm based on statistical sorting. Among them, the pre-correction algorithm of abnormal stripes based on data fitting provides a more reasonable data basis for the one-dimensional filtering algorithm based on statistical sorting. The pre-correction algorithm of abnormal stripes based on data fitting utilizes the property of data mutation caused by abnormal stripe elements, and effectively weakens the pseudo image structure in the projection data caused by these abnormal data to a certain extent, thereby providing more reliable and effective data for subsequent filtering algorithms. The one-dimensional filtering algorithm based on statistical sorting uses more similar data to filter out abnormal elements by sorting, thereby effectively avoiding the loss of image structure information during the filtering process. Through the combination of the above technologies, the effective removal of ring artifacts in CT images can be achieved.
[0092] The method disclosed in the present invention designs a reasonable fringe element pre-correction algorithm based on the geometric properties of the ring artifacts in the projection data and the smoothing properties of the fitting function, thereby solving the problems of poor effect and data information loss caused by directly using the filtering algorithm to correct the artifacts when the ring artifacts are serious. The method is applicable to a variety of CT image reconstruction methods and can achieve fast and accurate correction of CT image ring artifacts.
[0093] Exemplary Systems
[0094] Based on the above embodiments, the present invention also provides a CT image ring artifact removal system based on data fitting and statistical sorting, such as Figure 7 As shown, the system in this embodiment includes a pre-correction module 10 and a secondary correction module 20. Specifically, the pre-correction module 10 is used to obtain original projection data, and pre-correct the original projection data using a pre-correction algorithm based on mean projection fitting to obtain pre-corrected projection data; the secondary correction module 20 is used to perform secondary correction on the pre-corrected projection data using a one-dimensional filtering algorithm based on statistical sorting to obtain corrected projection data.
[0095] In one implementation, the pre-correction module includes:
[0096] a mean projection calculation unit, used for calculating the mean projection of the original sub-projection data received by each detector unit;
[0097] A function fitting unit, used for performing function fitting by taking the mean projection of the original sub-projection data received by each detector unit as a control point set to obtain a fitting function;
[0098] a data calculation unit, configured to calculate an ideal mean projection of each detector unit using the fitting function;
[0099] an error calculation unit, configured to determine a mean projection error of each detector unit based on a mean projection of the original sub-projection data of each detector unit and an ideal mean projection;
[0100] A data correction unit, configured to correct the original sub-projection data based on the mean projection error of each detector unit to obtain pre-corrected sub-projection data of each detector unit;
[0101] A data generating unit is used to obtain the pre-corrected projection data based on the pre-corrected sub-projection data of each detector unit.
[0102] In one implementation, the secondary correction module includes:
[0103] A sorting unit, used for sorting the pre-correction sub-projection data of each detector unit;
[0104] A filtering unit, used for performing one-dimensional smoothing filtering on the sorted pre-correction sub-projection data of each detector unit to obtain filtered sub-projection data of each detector unit;
[0105] The data recovery unit is used to recover the filter projection data of all detector units by using the sorting transformation matrix to obtain the corrected projection data.
[0106] In one implementation, the secondary correction module further includes:
[0107] An index acquisition unit, used to acquire an original index of the pre-correction sub-projection data of each detector unit;
[0108] The sorting transformation matrix generating unit is used to generate a sorting transformation matrix based on the sorting result of the pre-correction sub-projection data and the original index of each detector unit, and save the sorting transformation matrix.
[0109] The working principles of each module in the CT image ring artifact removal system based on data fitting and statistical sorting in this embodiment are the same as the principles of each step in the above method embodiment, and will not be repeated here.
[0110] Based on the above embodiments, the present invention further provides a terminal device, the principle block diagram of the terminal device can be as follows: Figure 8 The terminal device may include one or more processors 100 ( Figure 8 Only one is shown in the figure), a memory 101 and a computer program 102 stored in the memory 101 and executable on one or more processors 100, for example, a program for a method for removing ring artifacts from CT images based on data fitting and statistical sorting. When one or more processors 100 execute the computer program 102, the various steps in the method embodiment of the method for removing ring artifacts from CT images based on data fitting and statistical sorting can be implemented. Alternatively, when one or more processors 100 execute the computer program 102, the functions of the various modules / units in the device embodiment of the method for removing ring artifacts from CT images based on data fitting and statistical sorting can be implemented, which is not limited here.
[0111] In one embodiment, the processor 100 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.
[0112] In one embodiment, the memory 101 may be an internal storage unit of an electronic device, such as a hard disk or memory of the electronic device. The memory 101 may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device. Further, the memory 101 may also include both an internal storage unit of the electronic device and an external storage device. The memory 101 is used to store computer programs and other programs and data required by the terminal device. The memory 101 may also be used to temporarily store data that has been output or is to be output.
[0113] Those skilled in the art will understand that Figure 8 The principle block diagram shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the terminal device to which the solution of the present invention is applied. The specific terminal device includes Figure 8 More or fewer components may be shown, or certain components may be combined, or may have a different arrangement of components.
[0114] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, operating database or other media used in the embodiments provided by the present invention may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double operational data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0115] In summary, the present invention discloses a method for removing ring artifacts in CT images based on data fitting and statistical sorting, the method comprising: first, obtaining original projection data, and pre-correcting the original projection data using a fitting algorithm based on mean projection to obtain pre-corrected projection data; then, performing secondary correction on the pre-corrected projection data using a one-dimensional filtering algorithm based on statistical sorting to obtain corrected projection data. The present invention first utilizes the numerical mutation properties of abnormal stripes in the projection data, and uses the idea of data fitting to weaken the abnormal mutation of the data, thereby removing the pseudo-texture information in the projection data caused by the abnormal stripes to a certain extent, and providing a data basis for the subsequent filtering algorithm to accurately correct the projection data, and then uses a smoothing filtering algorithm based on data sorting to perform detail correction on the original projection data, and finally achieves fast and accurate correction of ring artifacts in CT images.
[0116] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for removing ring artifacts from CT images based on data fitting and statistical sorting, characterized in that: The method comprises: Acquiring original projection data, and pre-correcting the original projection data using a fitting algorithm based on mean projection to obtain pre-corrected projection data; A one-dimensional filtering algorithm based on statistical sorting is used to perform secondary correction on the pre-corrected projection data to obtain corrected projection data.
2. The method for removing ring artifacts from CT images based on data fitting and statistical sorting according to claim 1, characterized in that: The original projection data includes original sub-projection data received by a plurality of detector units, and the pre-correcting the original projection data using a fitting algorithm based on mean projection to obtain pre-corrected projection data includes: Calculating a mean projection of the raw sub-projection data received by each detector unit; Performing function fitting using the mean projection of the original sub-projection data received by each detector unit as a control point set to obtain a fitting function; calculating an ideal mean projection for each detector cell using the fitting function; determining a mean projection error for each detector element based on the mean projection and the ideal mean projection of the raw sub-projection data of each detector element; Correcting the original sub-projection data based on the mean projection error of each detector unit to obtain pre-corrected sub-projection data of each detector unit; The pre-corrected projection data is obtained based on the pre-corrected sub-projection data of each detector unit.
3. The method for removing ring artifacts from CT images based on data fitting and statistical sorting according to claim 2, characterized in that: The method of performing secondary correction on the pre-corrected projection data by using a one-dimensional filtering algorithm based on statistical sorting to obtain corrected projection data includes: sorting the pre-correction sub-projection data of each detector unit; Performing one-dimensional smoothing filtering on the sorted pre-correction sub-projection data of each detector unit to obtain filtered sub-projection data of each detector unit; The sorting transformation matrix is used to perform data recovery on the filtered sub-projection data of all detector units to obtain corrected projection data.
4. The method for removing ring artifacts from CT images based on data fitting and statistical sorting according to claim 3, characterized in that: Before performing one-dimensional smoothing filtering on the pre-correction sub-projection data of each detector unit after sorting, the method further includes: Obtaining raw indexes of pre-corrected sub-projection data of each detector unit; A sorting transformation matrix is generated based on the sorting result of the pre-correction sub-projection data and the original index of each detector unit, and the sorting transformation matrix is saved.
5. The method for removing ring artifacts from CT images based on data fitting and statistical sorting according to claim 2, characterized in that: The function fitting is a local weighted linear function fitting.
6. The method for removing ring artifacts from CT images based on data fitting and statistical sorting according to claim 3, characterized in that: The direction of performing the one-dimensional smoothing filtering is the direction of the detector unit.
7. The method for removing ring artifacts from CT images based on data fitting and statistical sorting according to claim 6, characterized in that: The one-dimensional smoothing filter adopts a one-dimensional Gaussian filter or a median filter.
8. A CT image ring artifact removal system based on data fitting and statistical sorting, characterized in that: The system comprises: A pre-correction module, used to obtain raw projection data, and pre-correct the raw projection data using a fitting algorithm based on mean projection to obtain pre-corrected projection data; The secondary correction module is used to perform secondary correction on the pre-corrected projection data by using a one-dimensional filtering algorithm based on statistical sorting to obtain corrected projection data.
9. A terminal device, characterized in that: The terminal device includes a memory, a processor, and a CT image ring artifact removal method program based on data fitting and statistical sorting, which is stored in the memory and can be run on the processor. When the processor executes the CT image ring artifact removal method program based on data fitting and statistical sorting, the steps of the CT image ring artifact removal method based on data fitting and statistical sorting as described in any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program for a method for removing ring artifacts from CT images based on data fitting and statistical sorting. When the program for the method for removing ring artifacts from CT images based on data fitting and statistical sorting is executed by a processor, the steps of the method for removing ring artifacts from CT images based on data fitting and statistical sorting as described in any one of claims 1 to 7 are implemented.
Citation Information
Patent Citations
Method for generation of correction coefficient configured to eliminate CT artifacts and method for elimination of CT artifacts based on correction coefficient
CN108109185A
KR20210147384A
