Method and device for correcting ring artifacts in CT images and computer program medium

Abstract

The present invention provides a method and device for correcting ring artifacts in CT images, as well as a computer program medium. The correction method includes the steps of preprocessing original detection data, marking defective channels according to the preprocessed original sinogram data, obtaining replacement detection values ​​corresponding to the defective channels through a first averaging process, and performing CT image reconstruction using the sinogram data after the first averaging process. The device includes a preprocessing unit configured to preprocess the original detection data, a marking unit configured to mark defective channels according to the preprocessed original sinogram data, a first averaging unit configured to obtain replacement detection values ​​corresponding to the defective channels through the first averaging process, and an image reconstruction unit. The present invention can effectively remove ring artifacts in CT images while ensuring little loss of spatial resolution of the CT images.

Description

[Technical Field]

[0001] This application claims priority to Chinese Patent Application No. 201911392508.X, filed on December 30, 2019, the entire contents of which are incorporated herein by reference.

[0002] The present invention relates to the field of CT image processing, and more particularly to a method and device for correction of ring artifacts in sinogram-based CT images, and a computer program medium. [Background technology]

[0003] CT is a typical medical imaging device that generally includes a scanning unit, a computer system, and an image display and storage system. The scanning unit includes an X-ray tube, a detector module, a scanning frame, etc. The X-ray tube emits an X-ray beam to scan the object being scanned. The detector module receives the X-rays after passing through the object being scanned and converts them into visible light, which is then converted into an electrical signal by photoelectric conversion. The computer system is responsible for storing and manipulating the electrical signal data collected by the scan. The image display and storage system is responsible for displaying the image processed and reconstructed by the computer system through a TV screen, a multi-frame camera, or a laser camera. In particular, as shown in Figure 1, the scanning principle of CT can be simply summarized as detecting the attenuation of X-rays by an object at multiple angles. For example, in cone-beam CT, as it scans, X-rays emitted from an X-ray tube 1 are irradiated onto the object being scanned 3, which is placed between the X-ray tube 1 and an X-ray detector module 2, which usually includes multiple serially arranged detectors 21 (Figure 2). The detectors are arranged in rows or columns, and each row or column of detectors may be referred to as a channel. The X-ray detector module 2 can detect the intensity of X-rays attenuated by the scanned object 3 at various positions, and the intensity may be compared to the unattenuated intensity of the X-rays to calculate an attenuation coefficient. The X-ray tube 1 and the X-ray detector module 2 are then simultaneously rotated 360° around the z-axis to complete the scan. The attenuation coefficients acquired in the scan at various angles at various positions may further form a projection map for each angle. The projection map may be represented in the form of a sinogram shown in FIG. 3, which is a form for representing raw CT data. Each line of the sinogram represents data sampled from one row of detector channels in the x-direction at one of the detector scan angles (shown in FIG. 1), and the data may be of X-ray intensity, attenuation coefficient, or attenuation coefficient after slope filtering.A two-dimensional matrix may be generated by arranging each line of data in angular order, and each element of the two-dimensional matrix may be represented by I(s,θ), where θ represents the detector scan angle, which should be relative to the direction of X-ray irradiation or other known direction when the scan begins, and s represents one of the positions within the detector row. Since there may be multiple rows of detector channels in the x-direction, and one sinogram may be acquired from each row of channels by scanning multiple angles, multiple sinograms may be acquired in a single scan.

[0004] As shown by the squares 22 with different gray values in Figure 2, a channel is called a bad channel when it cannot read out a signal normally or correctly from one of the channels, i.e., when there is a damaged detector 21 or an abnormal response. When the detector is damaged so that it cannot read out a signal normally, the gray value becomes relatively dark. When it cannot read out a signal correctly from the detector, the gray value becomes relatively light. Because the X-ray tube and X-ray detector module simultaneously rotate 360 degrees during CT scanning, when there are several bad channels in the X-ray detector module or several channels have uneven responses (hereinafter collectively referred to as bad channels), there will generally be ring artifacts appearing in the reconstructed image, which directly affects the quality of the CT image for the reader's review. However, in practical use, it is too costly to replace the detector module only when there are a small number of bad channels. Therefore, image correction methods are usually used to eliminate the effects of bad channels.

[0005] Currently, methods for correcting ring artifacts in CT images can be classified into image-based ring artifact correction, sinogram-based ring artifact correction, and deep learning-based ring artifact correction. Among these, deep learning-based ring artifact correction methods have demonstrated good results but require a large number of medical images as training samples. However, medical image samples are difficult to obtain for various reasons, making the method difficult to popularize and apply. The other two correction methods present a significant dilemma: if a smaller loss of image spatial resolution is achieved, the relatively bright ring artifact cannot be completely removed; if the relatively bright ring artifact is completely removed, a greater loss of image spatial resolution occurs.

[0006] Therefore, it is necessary to develop a method for the correction of ring artifacts in CT images to overcome the above-mentioned technical problems. Summary of the Invention [Problem to be solved by the invention]

[0007] The object of the present invention is to provide a method and device for the correction of ring artifacts in CT images, as well as a computer program medium, thereby solving the problem in the prior art that the correction method must make a compromise between the quality of the correction and the spatial resolution of the image. [Means for solving the problem]

[0008] Provided herein is a method for correcting ring artifacts in CT images, comprising: Step S1: Preprocessing the original detection data; Step S2: Detect detector defects based on the preprocessed original sinogram data. detector marking the Step S3: First averaging process detector obtaining a replacement detection value corresponding to Step S4: Performing CT image reconstruction using the sinogram data after the first averaging process.

[0009] According to an embodiment of the present invention, the method of correction comprises, before step S1, the addition of an additive noise model, namely: I(s,θ) = I r (s,θ) + b(s) + ε(s,θ) establishing I(s, θ) represents the actual detected value at the scan angle θ and detector position s of the original sinogram, and I r The method further includes the step of: (s, θ) represents an ideal detection value at the scan angle θ and detector position s of an ideal sinogram; b(s) represents an offset value of the actual detection value corresponding to the detector position s; and ε(s, θ) represents a random error corresponding to the scan angle θ and detector position s of the original sinogram.

[0010] In an embodiment of the present invention, in step S1, the pre-processing step comprises a specific step, namely: Step S11: selecting actual detection values corresponding to multiple scan angles θ from the original detection data to form an original sinogram; Step S12: filtering the original sinogram in the direction of detector position s; Step S13: obtaining noise of the original sinogram by filtering; Step S14: performing a second averaging of noise in the direction of the scan angle θ; Step S15: Obtaining the standard deviation of the noise in step S13 in the θ direction.

[0011] In the embodiment of the present invention, the number of selected scanning angles θ in step S11 is 50 or more.

[0012] In an embodiment of the present invention, in step S11, the actual detection value is the intensity of the attenuated X-ray measured by the detector, its attenuation coefficient, or the attenuation coefficient after gradient filtering.

[0013] In an embodiment of the present invention, in step S12, the filtering is weighted filtering.

[0014] In an embodiment of the present invention, in step S12, the weighted filtering is performed according to the following equation:

[0015]

number

[0016] is the average filtering using n is a natural number.

[0017] In the embodiment of the present invention, in step S13, the noise is obtained by subtracting the ideal detection value of the filtered ideal sinogram from the actual detection value of the original sinogram.

[0018] In an embodiment of the present invention, in step S14, the second averaging of noise in the direction of the scan angle θ is performed in particular using the following formula:

[0019]

number

[0020] may be performed according to

[0021]

number

[0022] represents the average offset value of the offset values b(s) in the direction of all scan angles θ, and N is the number of scan angles θ selected in step S11.

[0023] In an embodiment of the present invention, in step S15, the standard deviation of the noise at detector position s is calculated using the following formula:

[0024]

number

[0025] is obtained in accordance with N is the number of scan angles θ selected in step S11.

[0026] In the embodiment of the present invention, in step S2, detector The step of marking is performed using the following formula:

[0027]

number

[0028] is performed by determining whether

[0029]

number

[0030] represents the absolute value of the average offset value, d represents the correction coefficient, sqrt(N) represents the root mean square of N, and if equation (6) holds, then the average offset value

[0031]

number

[0032] is determined to be excessive, and the corresponding channel is detector is marked as

[0033] In an embodiment of the present invention, the correction coefficient d is in the range of 1.6 to 2.6, 1.8 to 2.0, or 2.0 to 2.6.

[0034] In the present embodiment, the correction factor d is 1.96.

[0035] In the embodiment of the present invention, in step S3, the first averaging process is performed to calculate the number of defects on the same line. detector The average value of the actual detection values of the 2n adjacent detectors on both sides of detector The present invention includes a specific step in which n is used as a replacement detection value of the step n, where n is a natural number.

[0036] In the embodiment of the present invention, in step S3, when the CT detectors are arranged in a two-dimensional arrangement, the first averaging process is performed to detector The average of the actual detection values of the four adjacent detectors is detector The method includes a specific step of using the detected value as a replacement for the detected value.

[0037] In the embodiment of the present invention, in step S3, when the CT detectors are arranged in a two-dimensional arrangement, the first averaging process is performed to detector The average of the actual detection values of the eight adjacent detectors is used to determine the defect. detector This includes using it as a replacement detection value.

[0038] In the embodiment of the present invention, in step S3, when the CT detectors are arranged in a two-dimensional arrangement, the first averaging process is performed to detector The average of the actual detection values of eight detectors in the direction perpendicular to the detector The method includes a specific step of using the detected value as a replacement for the detected value.

[0039] In the embodiment of the present invention, in step S3, when the CT detectors are arranged in a two-dimensional arrangement, the first averaging process is performed to detector The average of the actual detection values of the 12 adjacent detectors is used to determine the defect. detector The method includes a specific step of using the detected value as a replacement for the detected value.

[0040] In an embodiment of the present invention, in the first averaging process, the actual detection value of each detector is replaced by subtracting the corresponding average offset value from the actual detection value of the detector.

[0041] In the embodiment of the present invention, in step S4, a defect is detected in the sinogram data. detector Only data that is not bad will be replaced by the replacement detection value. detector The sinogram data remains the same as the original sinogram.

[0042] In the embodiment of the present invention, in step S4, a defect is detected in the sinogram data. detector The data is replaced by the replacement detection value and is not bad. detector sinogram data is replaced by subtracting the average offset value from the original sinogram.

[0043] The present invention provides a device for correcting ring artifacts in CT images, the device comprising: a pre-processing unit configured to pre-process original detection data; and a defect correction unit configured to correct the defect according to the pre-processed original sinogram data. detector a marking unit configured to mark the defective area by a first averaging process; detector and a reconstruction unit configured to perform CT image reconstruction using the sinogram data after the first averaging process.

[0044] In an embodiment of the present invention, the pre-processing unit further includes a data acquisition subunit configured to select actual detection values corresponding to multiple scan angles from the original detection data to form an original sinogram; a filtering subunit configured to filter the original sinogram in the direction of the detector position s; a noise acquisition subunit configured to acquire noise of the original sinogram by filtering; a second averaging subunit configured to perform second averaging of the noise in the direction of the scan angle; and a standard deviation acquisition subunit configured to acquire the standard deviation of the noise in the direction of the scan angle.

[0045] In an embodiment of the present invention, the second averaging sub-unit is configured to output an offset value of the actual detection value after the second averaging.

[0046] In the embodiment of the present invention, the first averaging process is performed to compare defects on the same line. detector The average value of the actual detection values of the 2n adjacent detectors on both sides of detector is used as the replacement detection value, where n is a natural number.

[0047] In an embodiment of the present invention, when the CT detectors are arranged in a two-dimensional arrangement, the first averaging process detector The average of the actual detection values of the four adjacent detectors is detector is used as the replacement detection value.

[0048] In an embodiment of the present invention, when the CT detectors are arranged in a two-dimensional arrangement, the first averaging process detector The average of the actual detection values of the eight adjacent detectors is used to determine the defect. detector is used as the replacement detection value.

[0049] In the present invention, it is provided that program instructions, when stored and executed, perform the following functions: pre-processing an original sinogram; and detecting defects in the pre-processed original sinogram. detector The first averaging process is used to mark the defective parts. detector and performing CT image reconstruction using the first averaged data.

[0050] In the present invention, by analyzing the additive noise model, it is possible to effectively detect obvious bad channels using a method based on hypothesis testing of statistical principles, which is a bad detector The detected value of the adjacent detector The obvious ring artifacts in the image are usually caused by a bad channel, so that it may be possible to eliminate the obvious ring artifacts in the image by replacing the detected values of the channels by the average value of the detected values of the channels. The ring artifacts that are not particularly obvious in the image are usually caused by the mismatch of the channel detection efficiency. The present invention shows the effect of suppressing the ring artifacts by subtracting the offset value of the channel to determine its detection efficiency. The present invention maximizes the loss of spatial resolution of the image and ensures the quality of the image by using a small percentage of bad channels. detector All we have to do is replace by the average value of the adjacent channels.

[0051] In summary, the method and device for correcting ring artifacts in CT images and the computer program medium provided in the present invention can essentially remove ring artifacts in CT images without requiring any pre-correction training, while ensuring that there is little loss of spatial resolution in the CT images, which is of positive significance to CT image correction and promotes cost savings. [Brief explanation of the drawings]

[0052] The accompanying drawings related to embodiments or state-of-the-art are described for the purpose of illustrating the embodiments or state-of-the-art of the present invention. It is clear that the illustrated drawings only illustrate some embodiments described in the present disclosure. It should be understood by those skilled in the art that various alternatives to the illustrated drawings may be realized without involving creative work. [Figure 1] FIG. 1 is a schematic diagram illustrating the scanning principle of a conventional CT. [Figure 2] 1 is a schematic diagram of an X-ray detector module of a prior art CT and a faulty detector of the X-ray detector module; [Figure 3] 1 is a schematic diagram of a sinogram obtained by performing a CT scan in the prior art; [Figure 4] 1 is a schematic illustration of the steps of a method for correction of ring artifacts in CT images according to an embodiment of the present invention; [Figure 5] 2 is a schematic illustration of step S1 of a method for correction of ring artifacts in CT images according to an embodiment of the present invention; FIG. [Figure 6] 6 is a schematic diagram of the correction result of the method of correction of ring artifacts in CT images of FIG. 5, where the sinogram has been filtered. [Figure 7] 7 is a schematic illustration of noise due to the method of correction of ring artifacts in CT images according to FIG. 6; [Figure 8] 8 is a schematic diagram of the offset according to the method of correction of ring artifacts in CT images according to FIG. 7; [Figure 9] 8 is a schematic illustration of the standard deviation of the offset according to the method of correction of ring artifacts in CT images according to FIG. 7; [Figure 10] 3 is a schematic illustration of averaging in a method for correction of ring artifacts in CT images according to an embodiment of the present invention; [Figure 11] 3 is a schematic illustration of a type of averaging in a method for correction of ring artifacts in CT images according to an embodiment of the present invention; [Figure 12]3 is a schematic illustration of a type of averaging in a method for correction of ring artifacts in CT images according to an embodiment of the present invention; [Figure 13] 3 is a schematic illustration of a type of averaging in a method for correction of ring artifacts in CT images according to an embodiment of the present invention; [Figure 14] 3 is a schematic illustration of a type of averaging in a method for correction of ring artifacts in CT images according to an embodiment of the present invention; [Figure 15] 3A and 3B are schematic diagrams of the correction effect of a method for correcting ring artifacts in CT images according to an embodiment of the present invention; [Figure 16] 3A and 3B are schematic diagrams of the correction effect of a method for correcting ring artifacts in CT images according to an embodiment of the present invention; [Figure 17] 1 is a schematic diagram of a configuration of a device for correction of ring artifacts in CT images according to an embodiment of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0053] In the following, the present invention will be further described with reference to embodiments, which should be understood as being for illustrative purposes only and not for limiting purposes.

[0054] In particular, when a component or element is said to be "disposed on" another component or element, the component or element may be directly disposed on the other component or element, or intermediate components or elements may be present. When a component or element is said to be "connected or coupled" to another component or element, the component or element may be directly connected or coupled to the other component or element, or intermediate components or elements may be present. As used herein, the term "connection or coupling" may include electrical and / or mechanical or physical connections or couplings. As used herein, the term "comprise" or "include" refers to the presence of a feature, step, component, or element, but does not exclude the presence or addition of one or more additional features, steps, components, or elements. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0055] Unless otherwise indicated, all technical and scientific terms used herein have the common meaning commonly understood by one of ordinary skill in the art to which this disclosure pertains. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0056] Furthermore, the terms "first," "second," "third," etc., as used herein are for descriptive purposes only and to distinguish similar objects from one another and do not represent an ordering of those objects, nor can they be understood as an indication or suggestion of relative importance. Additionally, in the description of this disclosure, unless otherwise specified, "plurality" means two or more than two.

[0057] In the present invention, before processing the correction of ring artifacts in CT images, it first uses an additive noise model I(s,θ) = I r (s,θ) + b(s) + ε(s,θ) (Equation 1) may be established. In the formula, I(s, θ) represents the actual detection value at the scan angle θ and detector position s of the original sinogram obtained by CT scanning. The scan angle θ and detector position s may be customized according to various application scenarios. Generally, it may be possible to select the initial irradiation direction of the X-ray generator as the reference direction of the scan angle θ. The detector position s may be represented by the matrix arrangement of the detectors in the X-ray detector module. I r (s, θ) represents the ideal detection value at the scan angle θ and detector position s of the ideal sinogram, b(s) represents the offset value of the actual detection value of the detector channel corresponding to the detector position s, this offset value is independent of the angle but is only related to whether the detector is damaged and the influence of the detector output signal itself, and therefore is the source of the ring artifact, and ε(s, θ) represents the random error corresponding to the scan angle θ and detector position s of the original sinogram, and the expected value of the random error should be 0.

[0058] 4 is a schematic diagram illustrating steps of a method for correcting ring artifacts in CT images according to an embodiment of the present invention. As can be seen from FIG. 4, the method for correcting ring artifacts in CT images provided in the present invention includes at least the following steps: Step S1: Preprocessing the original detection data; Step S2: Detect detector defects based on the preprocessed original sinogram data. detector marking the Step S3: First averaging process detector obtaining a replacement detection value corresponding to Step S4: Reconstructing a CT image using the sinogram data after the first averaging process.

[0059] In particular, as shown in FIG. 5, the above step S1 includes the following steps: Step S11: selecting actual detection values corresponding to multiple scan angles θ from the original detection data to form an original sinogram; Step S12: filtering the original sinogram formed in step S11 in the direction of detector position s; Step S13: obtaining noise of the original sinogram by filtering; Step S14: performing a second average of the noise obtained in step S13 in the θ direction; Step S15: obtaining the standard deviation of the noise of step S13 at the detector position s.

[0060] In particular, in step S11, an original sinogram is formed by selecting actual detection values, i.e., I(s, θ), corresponding to multiple scan angles θ. Generally, the number of scan angles θ is 50 or more. In the original sinogram, the actual detection values may be the intensity of X-rays after attenuation, the attenuation coefficient, or the attenuation coefficient after gradient filtering. It should be understood by those skilled in the art that gradient filtering is a filtering method commonly used in the art. Gradient filtering may multiply a slope function in frequency domain space. Gradient filtering includes standard slope filtering or window slope filtering, which may use a convolution method to replace gradient filtering in frequency domain space, which will not be described in detail herein. The attenuation coefficient is the ratio of the intensity of the attenuated X-rays detected by the detector to the intensity of the original X-rays emitted by the X-rays. The sinogram formed in step S11 is shown in FIG. 3, where the horizontal direction represents the detector position s and the vertical direction represents the scan angle θ. For example, the black horizontal line in FIG. 3 represents the actual detection value I(s, θ) of the detector corresponding to a particular scan angle θ at a certain position s.

[0061] Furthermore, in the above step S12, the filtering process of the original sinogram in the s direction may be an average filtering, which may be performed according to the following formula:

[0062]

number

[0063] where n is a natural number. That is, the filtering process is performed by averaging 2n+1 adjacent actual detection values I(s,θ) with the ideal detection value I(s,θ) at the corresponding position. rFor example, the filtering process may select the average of three (n = 1) adjacent actual detection values of the channel as the ideal detection value, in which case the above equation (2) becomes: I r (s,θ) = [I(s - 1,θ) + I(s,θ) + I(s + 1,θ)] / 3 (Equation 3) This becomes:

[0064] It should be understood by those skilled in the art that the filtering process in step S12 can also be weighted filtering.Average filtering is generally considered to be weighted filtering using equal weights, but weighted filtering can also be Gaussian weighted filtering or other weighted filtering using large weights in the center and small weights on both sides, and these filtering can be easily understood by those skilled in the art according to the technical teachings of the present invention combined with mathematical knowledge, and this will not be described in detail in this specification.

[0065] The sinogram filtered by the above step S12 is the ideal sinogram, i.e., I shown in FIG. r The sinogram formed by (s, θ) may be regarded as an ideal sinogram. However, the filtered sinogram at this point cannot be directly used for CT image reconstruction, since it will lead to a significant reduction in the spatial resolution of the CT image. It should be understood by those skilled in the art that other filtering methods may be used for the processing in step S12, which will not be described in this specification. Meanwhile, the number n may be selected according to actual needs or spatial resolution requirements, which will not be described in detail in this specification.

[0066] In the above step S13, the noise of the original sinogram is obtained by subtracting the ideal detection value data of the filtered ideal sinogram from the actual detection value data of the original sinogram, i.e., I(s,θ) - I r (s, θ) gives the sum b(s)+ε(s, θ) of the offset value and random error output by the detector, collectively referred to as noise as shown in FIG. 7.

[0067] In the above step S14, a second averaging of the noise in the direction of the scan angle θ may be performed in particular according to the following formula:

[0068]

number

[0069] where N is the number of scan angles θ selected in step S11;

[0070]

number

[0071] represents the average offset value of all offset values b(s) in all scan angle θ directions. According to the additive noise model established in the present invention, there exists an ideal offset value b(s) for the actual detection value I(s, θ) of each channel. Therefore, in this application, the ideal offset value b(s) is defined as the average offset value of all ideal offset values b(s).

[0072]

number

[0073] may be replaced by the average offset value

[0074]

number

[0075] can be made equal to the ideal offset value b(s) corresponding to each detection channel.

[0076] For a detector channel, the average value of its random error ε(s,θ) becomes 0, i.e., when the detector position s is set, the average value of the random error ε(s,θ) in the direction of the scan angle θ becomes 0. Therefore, by the second averaging, it is possible to obtain the average offset value output by the detector, as shown in FIG.

[0077]

number

[0078] That is, the average offset value corresponding to the detector position s by the second averaging

[0079]

number

[0080] where the abscissa represents the detector position s and the ordinate represents the average offset value

[0081]

number

[0082] Represents.

[0083] In the above step S15, the standard deviation of the noise at the detector position s may be obtained by the following formula:

[0084] JPEG0007725073000016.jpg23154

[0085] where N is the number of scan angles θ selected in step S11;

[0086]

number

[0087] represents the average offset value of all offset values b(s). Since the offset value b(s) output by the detector is constant in the θ direction, the standard deviation of the noise acquired at the detector position s is essentially the random error of the detector position s, as shown in Figure 9, where the abscissa represents the detector position s and the ordinate represents the standard deviation std[ε(s)], and is the standard deviation std[ε(s)] of the random error ε(s,θ) associated only with the detector position s, regardless of the scan angle θ.

[0088] In the above step S2, detector Marking is performed by determining whether the following formula holds:

[0089]

number

[0090] During the ceremony,

[0091]

number

[0092] is the average offset value

[0093]

number

[0094] where d is the correction coefficient, and sqrt(N) is the root mean square of N. The offset value b(s) is the source of the ring artifact, so the excess average offset value

[0095]

number

[0096] is the average offset value for which channel is excessive.

[0097]

number

[0098] For each detector position s, the average offset value is determined depending on whether equation (6) holds.

[0099]

number

[0100] If equation (6) holds, the average offset value

[0101]

number

[0102] is determined to be excessive, and therefore the channel corresponding to detector position s is defective. detector Marked as defective. detector is shown in Figure 2, where all blocks with different gray values are considered to be defective. detector Represents.

[0103] The theoretical basis of the judgment method is hypothesis testing. The distribution pattern of the offset value b(s) is unknown, but the average offset value

[0104]

number

[0105] It can be seen from the central limit theorem that satisfies a Gaussian distribution. For normal detectors, their offset value b(s) will be 0, i.e., the calculated average offset value

[0106]

number

[0107] The average value of is 0. Because there is an error, the average offset value

[0108]

number

[0109] In fact, satisfies a Gaussian distribution with mean value 0 and standard deviation std[ε(s)]÷sqrt(N). In a Gaussian distribution, the probability that abs[b(s)] < d×std[ε(s)]÷sqrt(N) is relatively high; for example, when d = 1.96, the probability that abs[b(s)] < d×std[ε(s)]÷sqrt(N) is 95%. Therefore, the probability that equation (6) is satisfied is only 5%, which is a very low probability. Usually, there is a reason for the occurrence of a low probability event, i.e., the offset value b(s) is not 0 and there is a defect at a specific position s of the detector. detector indicates the existence of

[0110] Furthermore, in the above equation (6), the correction coefficient d corresponds to the probability used to determine whether a low-probability event exists. For example, when d = 1.96, the corresponding probability that the above equation (6) holds is only 5%. In such a case, 5% is considered a low-probability event. Generally, the correction coefficient is in the range of 1.6 to 2.6. When the correction coefficient d is in the range of 1.6 to 1.8, almost all ring artifacts can be removed by the correction method proposed in the present invention, but the spatial resolution is slightly reduced. When the correction coefficient d is in the range of 1.8 to 2.0 (preferably 1.96), almost all ring artifacts can be removed by the correction method proposed in the present invention, while the spatial resolution is largely unaffected. When the correction coefficient d is in the range of 2.0 to 2.6, most of the ring artifacts can be removed by the correction method proposed in the present invention, while the spatial resolution is barely affected.

[0111] In the embodiment of the present invention, in the above step S3, detector After obtaining the information, obtaining a replacement detection value of the detector that is a bad channel through a first averaging process includes the following specific method.

[0112] Regarding the one-dimensional channel arrangement of the CT detector shown in Fig. 10, the defect in the s direction detector The actual detection values of the detectors corresponding to the 2n adjacent channels are detector For example, in FIG. 10, there are nine detectors arranged in a one-dimensional arrangement, and one detector in the middle is defective. detector When detector The actual detection values of all other eight detectors in the row except detector The replacement detection value may be averaged as detector On both sides of the fault, the actual detection values of the four detectors are obtained. detector Defects within the line except detector The actual detection values of the other four detectors on the left and right of detector The replacement detection value may be averaged as detector The actual detection values of the two detectors are obtained on both sides of the channel. It should be understood by those skilled in the art that the averaging method shown in Figure 10 is also applicable to two-dimensional channels.

[0113] Regarding 2D channel placement, detector The actual detection value of the detector corresponding to the adjacent or neighboring channel is detector The "next to" in this application refers to the bad channel. becomes This means the detector directly adjacent to the detector, i.e., the bad channel. becomes There are no other detectors between the detector and the neighboring detector, while "close" or "adjacent" means that there is no bad channel. becomes This means a detector close to the detector, i.e., a bad channel becomes It should be understood by those skilled in the art that there may be other detectors between the detector and the adjacent or neighboring detector. For example, in FIG. detector The actual detection values of the eight detectors surrounding detector In FIG. 12, the defective detector The actual detection values of the four adjacent detectors on the top, bottom, left and right are detector In FIG. 13, the defective detector The actual detection values of the eight detectors in the direction perpendicular to the detector In FIG. 14, the defective detector The actual detection values of the 12 detectors adjacent to detector may be averaged as replacement detection values.

[0114] In a further embodiment of the present invention, before performing the first averaging process in the above step S3, it uses an approximate value obtained by subtracting an average offset value from the actual detection value of each detector as the basis of the first averaging process.

[0115]

number

[0116] For example, if the CT detectors are arranged in a one-dimensional arrangement, the first averaging process may be performed using a detector The average value of the approximation values of the 2n adjacent detectors on both sides of detector It should be understood by those skilled in the art that to perform the first averaging process, it may use the weighted actual detection value as the basis of the first averaging process, which will not be described in detail herein.

[0117] In the above-mentioned step S4, the reconstruction of the CT image using the data processed by the first averaging may correspond to prior art image reconstruction methods, which will not be described in detail in this specification.

[0118] 15 is a schematic diagram of the correction effect of a method for correcting ring artifacts in CT images according to an embodiment of the present invention. In FIG. 15, the portion indicated by the arrow on the left indicates the ring artifact, while the arrow on the right indicates the corresponding area of the CT image after the ring artifact has been removed. At first glance, the ring artifact has been almost completely removed by the present invention, and the spatial resolution of the CT image is barely affected.

[0119] 16 is a schematic diagram of the correction effect of a method for correcting ring artifacts in CT images according to an embodiment of the present invention. In FIG. 16, the portion indicated by the arrow on the left shows the ring artifact, while the arrow on the right shows the corresponding area of the CT image after the ring artifact has been removed. At first glance, the ring artifact has been almost completely removed by the present invention, and the spatial resolution of the CT image is barely affected.

[0120] In the present invention, by analyzing the additive noise model, it is possible to effectively detect obvious bad channels using a method based on hypothesis testing of statistical principles, which is a bad detector The detected value of the adjacent detector The obvious ring artifacts in the image are usually caused by bad channels, so that it may be possible to eliminate the obvious ring artifacts in the image by replacing the channel offset value with the average value of the detection value of the channel. The ring artifacts that are not particularly obvious in the image are usually caused by the mismatch of channel detection efficiency. The present invention shows the effect of suppressing the ring artifacts by subtracting the offset value of the channel to determine its detection efficiency. The present invention maximizes the loss of spatial resolution of the image and ensures the quality of the image by substituting a small percentage of bad channels into the adjacent channels. detector All we need to do is replace it by the average value of

[0121] In an embodiment of the present application, a device for correcting ring artifacts in CT images is provided. As shown in FIG. 17, the correction device includes at least: a pre-processing unit 110, which may be configured to pre-process the original detection data and may output offset values of the actual detection values of the detectors at various positions; Detector failure according to the preprocessed original sinogram data detector a marking unit 120 that may be configured to mark the The first averaging process detector a first averaging unit 130 configured to obtain a replacement detection value corresponding to the offset value of each detector, or to subtract each offset value from the actual detection value of each detector before the first averaging process; and a reconstruction unit 140 that may be configured to perform CT image reconstruction using the first averaged sinogram data.

[0122] Furthermore, the above-mentioned pre-processing unit a data acquisition subunit 111, which may be configured to select actual detection values corresponding to a plurality of scan angles from the original detection data to form an original sinogram; a filtering subunit 112 that may be configured to filter the original sinogram in the direction of the detector position s; a noise acquisition subunit 113, which may be configured to acquire noise of the original sinogram by filtering; a second averaging sub-unit 114 that may be configured to perform a second averaging of noise in the direction of the scan angle θ and may output offset values of the actual detection values of the detector at various positions; and a standard deviation obtaining subunit 115 that may be configured to obtain the standard deviation of the noise in the direction of the scan angle θ.

[0123] For details of each unit and subunit, you may refer to the description of steps S1 to S4 and steps S11 to S15 of the above method embodiment, which will not be described in detail herein.

[0124] The correction device may be a server, an electronic device, etc., or any device capable of performing data processing based on a sinogram, which is not limited herein. It should be noted that the functions performed by the units of the correction device described above may be implemented by executing instructions stored in a memory by a processor of a computer.

[0125] The present invention provides a program that, when stored and executed, performs the following functions: preprocessing the original sinogram and outputting offset values of the actual detection values of the detector at various positions; and detecting defects in the preprocessed original sinogram. detector The first averaging process is used to mark the defective parts. detector Alternatively, each offset value is subtracted from the actual detection value of each detector, and then a first averaging process is performed to obtain a replacement detection value corresponding to the detected value. detectorand performing CT image reconstruction using the first averaged data.

[0126] The above-described program instructions may be executed by a processor or other processing device.

[0127] It should be understood by those skilled in the art that all or part of the steps in the above-described method embodiments may be stored in a non-volatile computer-readable storage medium and may be executed by instructing associated hardware through a computer program that, when executed, may include the process of the method in the above-described embodiment. In particular, all references to memory, storage medium, database, or other medium used in various embodiments of 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 demonstration and not limitation, RAM may be in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct Rambus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM).

[0128] The devices, units, subunits, etc. described in the above embodiments may be implemented, in particular, by computer chips and / or components, or by products having specific functions. For convenience, the device descriptions are each made up of individual units for each function. Of course, when implementing the present disclosure, the functions of the individual units may be embodied in the same or different computer chips.

[0129] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of the present invention. Various alternatives to the above embodiment of the present invention may be made. In this regard, simple or equivalent changes or modifications made in accordance with the claims and the description fall within the scope of the present invention as defined in the claims. What is not described in detail in this disclosure is ordinary.

Claims

1. 1. A method for correction of ring artifacts in CT images, comprising: Step S1: Pre-processing the original detection data, including the following steps S11 to S15: Step S11: selecting actual detection values corresponding to multiple scan angles θ from the original detection data to form an original sinogram; Step S12: filtering the original sinogram in the direction of detector position s, where the filtering is weighted filtering or average filtering; Step S13: obtaining noise of the original sinogram by filtering; Step S14: performing a second averaging of the noise in the direction of the scan angle θ; and Step S15: Obtaining the standard deviation of the noise in step S13 in the θ direction; Step S2: Marking defective detectors based on the pre-processed original sinogram data, including the data obtained in steps S14 and S15 of step S1; Step S3: obtaining a replacement detection value corresponding to the faulty detector by a first averaging process of actual detection values of detectors adjacent to or adjacent to the faulty detector in the pre-processed and marked sinogram; Step S4: Reconstructing a CT image using the sinogram data after the first averaging process.

2. Before step S1, an additive noise model is provided, i.e., I(s,θ) = I r (s,θ) + b(s) + ε(s,θ) where I(s, θ) represents the actual detected value at scan angle θ and detector position s of the original sinogram, and I ｒ 2. The method for correcting ring artifacts in CT images of claim 1, further comprising the step of: (s, θ) representing an ideal detected value at the scan angle θ and the detector position s of an ideal sinogram; b(s) representing an offset value of the actual detected value corresponding to the detector position s; and ε(s, θ) representing a random error corresponding to the scan angle θ and the detector position s of the original sinogram.

3. 2. The method for correcting ring artifacts in CT images according to claim 1, wherein the number of the selected scan angles θ is 50 or more in step S11.

4. 2. The method for correcting ring artifacts in CT images according to claim 1, wherein in step S11, the actual detection value is the intensity of the attenuated X-ray measured by the detector, the attenuation coefficient, or the attenuation coefficient after gradient filtering.

5. In step S12, the mean filtering is performed according to the following equation: [Equation 1] 2. The method for correcting ring artifacts in CT images according to claim 1, wherein n is a natural number.

6. 2. The method for correcting ring artifacts in CT images according to claim 1, wherein in step S13, the noise is obtained by subtracting the ideal detection value of the filtered ideal sinogram from the actual detection value of the original sinogram.

7. In step S14, a second averaging of the noise in the direction of the scan angle θ is carried out in particular according to the following formula: [Equation 2] may be performed according to [Equation 3] 7. The method for correcting ring artifacts in CT images according to claim 6, wherein N represents the average offset value of the offset values b(s) in the direction of all the scan angles θ, and N is the number of the scan angles θ selected in step S11.

8. In step S15, the standard deviation of the noise at the detector position s is calculated using the following formula: [Equation 4] 8. The method for correcting ring artifacts in CT images according to claim 7, wherein N is the number of the scan angles θ selected in step S11.

9. In step S2, the step of marking the bad detectors is performed using the following formula: [Equation 5] is performed by determining whether [0060] is the average offset value [Equation 7] where d represents the absolute value of N, d represents a correction coefficient, sqrt(N) represents the root mean square of N, and if equation (6) holds, then the average offset value [Equation 8] 9. The method for correcting ring artifacts in CT images according to claim 8, wherein: is determined to be excessive and the corresponding channel is marked as the bad detector.

10. 10. The method for correcting ring artifacts in CT images according to claim 9, wherein the correction factor d is in the range of 1.6 to 2.6, 1.8 to 2.0, or 2.0 to 2.

6.

11. 10. The method for correction of ring artifacts in CT images according to claim 9, wherein the correction factor d is 1.

96.

12. 2. The method for correcting ring artifacts in CT images according to claim 1, wherein in step S3, the first averaging process includes a specific step of using an average value of actual detection values of 2n detectors adjacent to both sides of the defective detector on the same line as the replacement detection value of the defective detector, where n is a natural number.

13. 2. The method for correcting ring artifacts in CT images according to claim 1, wherein in step S3, when the CT detectors are arranged in a two-dimensional arrangement, the first averaging process includes a specific step of using an average value of actual detection values of four detectors adjacent to the defective detector as the replacement detection value of the defective detector.

14. 2. The method for correcting ring artifacts in CT images according to claim 1, wherein in step S3, when the CT detectors are arranged in a two-dimensional arrangement, the first averaging process includes using an average value of actual detection values of eight detectors adjacent to the defective detector as the replacement detection value of the defective detector.

15. 2. The method for correcting ring artifacts in CT images according to claim 1, wherein in step S3, when the CT detectors are arranged in a two-dimensional arrangement, the first averaging process includes a specific step of using an average value of actual detection values of eight of the detectors in a direction perpendicular to the defective detector as the replacement detection value of the defective detector.

16. 2. The method for correcting ring artifacts in CT images according to claim 1, wherein in step S3, when the CT detectors are arranged in a two-dimensional arrangement, the first averaging process includes a specific step of using an average value of actual detection values of 12 detectors adjacent to the defective detector as the replacement detection value of the defective detector.

17. 17. The method for correcting ring artifacts in CT images according to claim 12, wherein in the first averaging process, the actual detection value of each detector is replaced by subtracting a corresponding average offset value from the actual detection value of the detector.

18. 2. The method for correcting ring artifacts in CT images according to claim 1, wherein in step S4, only the data of the faulty detector is replaced by the replacement detection value in the sinogram data, and the sinogram data of non-faulty detectors remains the same as the original sinogram.

19. 8. The method for correcting ring artifacts in CT images according to claim 7, wherein in step S4, in the sinogram data, only the data of the faulty detector is replaced by the replacement detection value, and the sinogram data of non-faulty detectors is replaced by subtracting the average offset value from the original sinogram.

20. a pre-processing unit configured to pre-process the original detection data, a data acquisition subunit configured to select actual detection values corresponding to a plurality of scan angles from the original detection data to form an original sinogram; a filtering subunit configured to filter the original sinogram in the direction of detector position s, wherein the filtering is weighted filtering or average filtering; a noise obtaining subunit configured to obtain noise of the original sinogram by filtering; a second averaging subunit configured to perform a second averaging of the noise in the direction of the scan angle; and a standard deviation obtaining subunit configured to obtain a standard deviation of the noise in the direction of the scan angle; the pre-treatment unit comprising: a marking unit configured to mark faulty detectors according to the data of the pre-processed original sinogram, including the data obtained in steps S14 and S15 of step S1; a first averaging unit configured to obtain a replacement detection value corresponding to the faulty detector by first averaging actual detection values of detectors adjacent to or adjacent to the faulty detector in the pre-processed and marked sinogram; a reconstruction unit configured to perform a CT image reconstruction using the sinogram data after the first averaging process.

21. The device for correcting ring artifacts in CT images according to claim 20, characterized in that the second averaging sub-unit is configured to output an offset value of the actual detection value after the second averaging.

22. 21. The device for correcting ring artifacts in CT images according to claim 20, wherein the first averaging process uses the average value of the actual detection values of 2n detectors adjacent to both sides of the faulty detector on the same line as the replacement detection value of the faulty detector, where n is a natural number.

23. 21. The device for correcting ring artifacts in CT images according to claim 20, wherein when CT detectors are arranged in a two-dimensional arrangement, the first averaging process uses the average value of the actual detection values of the four detectors adjacent to the defective detector as the replacement detection value of the defective detector.

24. 21. The device for correcting ring artifacts in CT images according to claim 20, wherein, when CT detectors are arranged in a two-dimensional arrangement, the first averaging process uses the average value of the actual detection values of eight detectors adjacent to the defective detector as the replacement detection value of the defective detector.

25. In a computer program storage medium, program instructions are stored and, when executed, perform the following functions: Pre-processing the original sinogram, including steps S11 to S15: Step S11: selecting actual detection values corresponding to multiple scan angles θ from the original detection data to form the original sinogram; Step S12: filtering the original sinogram in the direction of detector position s, where the filtering is weighted filtering or average filtering; Step S13: obtaining noise of the original sinogram by filtering; Step S14: performing a second averaging of the noise in the direction of the scan angle θ; and Step S15: Obtaining the standard deviation of the noise in step S13 in the θ direction; marking bad detectors of the preprocessed original sinogram based on data in the preprocessed original sinogram, including data obtained in steps S14 and S15 in the preprocessing; obtaining a replacement detection value corresponding to the faulty detector by first averaging actual detection values of detectors next to or adjacent to the faulty detector in the pre-processed and marked sinogram; and performing CT image reconstruction using the data after the first averaging process.

Images