KR101146006B1 - Method and apparatus to remove ring artifacts in x-ray CT - Google Patents

Abstract

The present invention relates to a method and an apparatus for removing ring artifacts in an X-ray CT that can effectively remove the ring artifacts generated in the cross-sectional image of the X-ray CT to improve the accuracy of the cross-sectional image without compromising the image quality of the cross-sectional image. In the X-ray CT, a method for removing ring artifacts comprises constructing a sinogram from a plurality of projection data obtained through an X-ray detector, and calculating the number of dominant pixel values for each sensing element of the X-ray detector from the sinogram. Determining whether the sensing device is an I-type defective device according to the number of dominant pixel values, correcting pixel values obtained from the I-type defective device when the sensing device is an I-type defective device; Calculating a curve curve error from the sinogram whose pixel value of the type I defect device is corrected, determining whether the sensing device is a type II defect device according to the error of the curve, and the sensing device is a In the case of a type defect device, correcting a pixel value obtained from the type II defect device.

Description

Method and apparatus to remove ring artifacts in x-ray CT}

The present invention relates to a method and apparatus for removing ring artifacts in X-ray CT, and more particularly, to effectively remove ring artifacts generated in cross-sectional images of X-ray CT to improve the accuracy of cross-sectional images without compromising image quality of the cross-sectional images. The present invention relates to a method and apparatus for removing ring artifacts in X-ray CT.

X-ray computed tomography (CT) is a machine that photographs the cross section of an object without cutting the cross section of the object. A conventional X-ray CT system, as shown in Figure 1, generally detects the X-rays passing through the X-ray tube 120 (x-ray tube) projecting the X-ray beam 110 and the object to be photographed (130) The X-ray detector 140, a computer for calculating a cross-sectional image from the signals obtained by the X-ray detector, and a rotating device for rotating the X-ray tube and the X-ray detector about the rotation axis 150. The signal acquired by the X-ray detector 140 is projection data in which X-rays are projected onto the photographing object 130 and attenuated by the inherent X-ray attenuation characteristics of the object. After the projection data is obtained from various angles and the back-projection is performed to project in the reverse direction of the X-ray projection direction, the cross-sectional image can be calculated, and the process of calculating the cross-sectional image is called image reconstruction. The X-ray detector 140 is integrated with high-density X-ray sensing elements. As the X-ray sensing element, a method of bonding a photo diode to a scintillator is used. The scintillator converts X-rays into visible light and a photodiode converts visible light into an electric signal. A typical CT for human body imaging uses an X-ray sensing device composed of hundreds to thousands of scintillator-photodiode pairs. In order to obtain a good quality cross-sectional image, the sensitivity of the X-ray sensing element should be very uniform. Otherwise, as shown in FIG. 2, the shape of the projection data is changed from the original shape due to a defect of a specific sensing element of the X-ray detector 140. Here, the range of the element number j of the X-ray detector sensing element is indicated from 1 to J. FIG. The defects occurring in the projection data may occur in the form of positive peaks occurring in the positive direction or negative peaks occurring in the negative direction. If the peak occurs in the positive direction, the detector sensing element is more sensitive than normal, and the peak occurs in the negative direction and vice versa. Defects of the sensing element appear as ring artifacts of a circular shape in the cross-sectional image of the X-ray CT. The ring artifacts generally appear together. The ring artifact can cause serious problems in making an accurate diagnosis through an image. In order to remove the ring artifact, it is important to make the sensitivity of the X-ray sensing device uniform. However, it is difficult to physically calibrate a large number of sensing elements one by one, and further correction is necessary because the state of the sensing element changes with time even after the calibration. Therefore, it is necessary to remove the ring artifact through software image processing.

With the development of X-ray detector technology, a detector in which an X-ray sensing device is integrated in a two-dimensional array having a high density or a large area has appeared. The two-dimensional X-ray detector is commonly referred to as an X-ray flat panel detector (FPD). The flat panel detector is produced in various areas according to the purpose of use, and there are 18 cm x 24 cm for mammography and 35 cm x 43 cm for lung imaging. The flat panel detector is used in both general X-ray imaging apparatus and CT system. In particular, the flat panel detector is used in an X-ray c-arm photographing apparatus, and the siam is 200 degrees. Cone beam CT has also emerged, which reconstructs the cross-sectional image from the projection data obtained by rotating it by a degree. Since the cone beam CT acquires 2D projection data, the 3D image may be generated by only one rotation of the X-ray tube and the flat panel detector pair. Since the 3D image provides more accurate information on the human body than the 2D image, the medical application of the cone beam CT has been greatly increased. The flat panel detector of the cone beam CT is composed of X-ray sensing elements composed of millions of two-dimensional arrays. Some of the X-ray sensing elements of the flat panel detector are completely lost in function and thus may emit a constant value regardless of the presence or absence of X-rays, and some may output an error signal that is not proportional to the input X-ray amount. In general, a flat panel detector composed of many two-dimensional arrays of X-ray sensing elements has a lower output uniformity and more ring artifacts in the cross-sectional image because of the larger number of defects than a detector composed of one-dimensional arrays of X-ray sensing elements. Therefore, the cone beam CT using the plate detector also has a problem in that ring artifacts are more severe than in general X-ray CT. In addition, since the sensitivity of the sensing element of the flat panel detector also varies with time, ring artifacts appear in the cone beam CT.

Conventional techniques for removing the ring artifacts include a method of removing ring artifacts through image processing in a cross-sectional image after reconstruction, and a method of removing ring artifacts from projection data before reconstruction. Conventional techniques for removing ring artifacts from cross-sectional images after reconstruction typically use a method of converting cross-sectional images to polar coordinates, converting circular ring artifacts into straight lines, removing them by filtering, and converting them into Cartesian coordinate systems. . However, the ring artifact removal method in the cross-sectional image using the coordinate system transformation has a problem of computational error that is essential in the coordinate system transformation, and thus, there is a problem in that the exact position of the ring artifact is not known. Conventional techniques for eliminating ring artifacts in projection data prior to reconstruction have a defective X-ray at a position having a value larger than the threshold on the basis of a specific threshold in the data accumulated in the direction of rotation of the detector. After determining the position of the sensing element, the ring artifact is removed by filtering at the defective position. However, the method based on the threshold has a problem that it is difficult to find the exact position of the defective sensing element by being overestimated or underestimated when the threshold is too high or too low. In addition, if the location of the defective sensing element is not accurately located and the filtering is performed at the wrong location, there is a problem of damaging the image quality of the cross-sectional image.

The present invention has been made to solve the above problems, by effectively removing the ring artifacts generated in the cross-sectional image of the X-ray CT in X-ray CT that can improve the accuracy of the cross-sectional image without compromising the image quality of the cross-sectional image It is an object of the present invention to provide a ring artifact removal method and apparatus.

According to an aspect of the present invention, there is provided a method for removing ring artifacts in an X-ray CT, comprising constructing a sinogram from a plurality of projection data obtained through an X-ray detector, and detecting the X-ray detector from the sinogram. Calculating the number of dominant pixel values for each element, determining whether the sensing device is an I-type defective device according to the number of dominant pixel values, and if the sensing device is an I-type defective device, pixels obtained from the I-type defective device Compensating a value, Computing the error of the sum curve in the sinogram that the pixel value of the I-type defect device is corrected, and Determining whether the sensing device is a type II defect device according to the error of the curve And correcting the pixel value obtained from the type II defective element when the sensing element is a type II defective element.

The ring artifact removing device in the X-ray CT according to the present invention comprises constructing a sinogram from a plurality of projection data obtained through the X-ray detector, and calculates the number of dominant pixel values for each sensing element of the X-ray detector from the sinogram, It is determined whether the sensing device is an I-type defective device according to the number of the dominant pixel values. When the sensing device is an I-type defective device, the pixel value obtained from the I-type defective device is corrected, and the pixel value of the I-type defective device is corrected. Calculate the error of the curve from the corrected sinogram, determine whether the sensing element is a type II defect element according to the error of the curve, and if the sensing element is a type II defect element from the type II defect element And a controller for correcting the acquired pixel value.

Ring artifact removal method and apparatus in the X-ray CT according to the present invention has the following effects.

By finding the position of the defective pixel by the defective element of the X-ray detector and correcting the pixel value only at that position, it is possible to effectively eliminate the ring artifacts that occur in the cross-sectional image of the X-ray CT to improve the accuracy of the cross-sectional image without compromising the image quality of the cross-sectional image. It becomes possible.

1 is a block diagram of an X-ray CT system according to the prior art.
2 is a view showing projection data of the X-ray CT according to the prior art.
3 is a flowchart of a ring artifact removal method in an X-ray CT in accordance with the present invention.
4 is a diagram illustrating a sinogram configuration method according to an embodiment of the present invention.
5 is a flowchart illustrating a process of calculating the number of dominant pixel values according to an embodiment of the present invention.
6 illustrates a histogram according to an embodiment of the present invention.
7 is a diagram illustrating a number of dominant pixel values and a filter passing signal according to an exemplary embodiment of the present invention.
8 is a view showing a method of processing a continuous defective device according to an embodiment of the present invention.
9 is a diagram showing a curve before and after correction of an I-type defect according to an embodiment of the present invention.
10A and 10B are reference views showing cross-sectional images before and after removing ring artifacts according to an embodiment of the present invention.

Hereinafter, a method and a device for removing ring artifacts in an X-ray CT according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

3 is a flowchart illustrating a ring artifact removal method in X-ray CT according to the present invention. As shown in FIG. 3, the ring artifact removal method in the X-ray CT according to the present invention comprises constructing a sinogram from a plurality of projection data acquired through the X-ray detector (S310), and the sino. Calculating a number of a dominant pixel-value for each sensing element of the X-ray detector from a gram (S320), and determining whether the sensing element is an I-type defective element according to the dominant pixel value (S330). ), Correcting the pixel value obtained from the I-type defect element when the sensing element is an I-type defect element (S340), and a cross curve in the sinogram in which the pixel value of the I-type defect element is corrected. sum curve) calculating an error (S350), determining whether the sensing device is a type II defective device according to the error of the curve (S360), and the type II when the sensing device is a type II defective device With defective element Emitter and a step (S370) of correcting the obtained pixel value. Each step will be described in detail as follows.

A step S310 of constructing a sinogram from the plurality of pieces of projection data acquired through the X-ray detector is a first step of removing ring artifacts from the X-ray CT, which will be described with reference to FIG. 4. The sinogram is used for projection data P (1, j), P (2, J), ..., P (N, j) received at various rotation angles n (1, 2, ..., N). , The horizontal axis is the axis of the sensing element number j (1, 2, ..., J) of the X-ray detector, the vertical axis is the axis of the rotation angle n, and the projection data is displayed in gray level. This refers to a two-dimensional image, which is represented by P (n, j). Here, the device number j has a value from 1 to J, and usually has a value between 1000 and 2000. The rotation angle n has a value from 1 to N and usually has a value between 300 and 1000. One value of the projection data is a value received from one sensing element of the X-ray detector, and forms one pixel of the sinogram. In the case where the specific sensing element of the X-ray detector is defective, the influence of the defective sensing element on the sinogram is indicated by the arrow in the sinogram of FIG. It appears in the form of a vertical line defect 410 shown. When a tomographic image is created using the defective sinogram, ring artifacts appear.

Defects of the X-ray detector may be classified into two types. In the first case, a constant output value is produced regardless of the input X-ray dose, which is called an I-type defect. Secondly, the output value is proportional to the input X-ray amount, but an offset value of a certain magnitude is added to the output value, and this is called a type II defect. Since the type I defects and the type II defects make ring artifacts having different characteristics, the ring artifacts can be effectively removed only by distinguishing and correcting them. However, the conventional ring artifact removal method did not distinguish between type I and type II. The sinogram having the type I defect, P _I (n, j) is defined as in Equation 1 below.

That is, the sinogram value at the detector sensing element position without the I type defect is a normal value, but a constant value A _j is output at the defective element position having the I type defect regardless of the input X-ray dose.

The type II defective sinogram, P _II (n, j) is defined as in Equation 2 below.

That is, although the sinogram value at the detector sensing element position without the type II defect is a normal value, the offset value a _j of a certain size appears in addition to the normal sinogram pixel value at the defect element position having the type II defect.

Calculating the number of dominant pixel values for each sensing element of the X-ray detector from the sinogram (S320) is a step for finding a position of a defective element having an I type defect. 5 is a flowchart illustrating a process of calculating the number of dominant pixel values according to the present invention. The calculating of the number of dominant pixel values may include obtaining a vertical axis size N and a horizontal axis size J of the sinogram P (n, j) (S511), and an element number of the X-ray detector sensing element. initializing j ₀ ) to 1 (S512), initializing the variable k of the number of dominant pixel values to 0 (S513), and projecting data of all rotation angles at the selected element number j ₀ . Obtaining a histogram h (m) of the made sinogram P (n, j ₀ ) (S514) and a histogram h _s (m _s ) in which the histogram h (m) is arranged in descending order. ) (S515), increasing the variable k of the dominant pixel value by 1 (S516), and calculating the cumulative frequency (S) for the sorted histogram (h _s (m _s )). Step S517, determining the case where the cumulative frequency S is greater than or equal to 0.5N, and step S518; and when the cumulative frequency S is greater than or equal to 0.5N, And pear pixel determining a parameter (k) of gapsu a controlled pixel gapsu (H (j ₀₎₎ of the selected element number (j _{0) (S519),} step adds 1 to the selected element number (j ₀₎ And performing an iterative loop until the selected device number j ₀ becomes the maximum device number J (S520). Here, m is a pixel value and m _s is an independent variable of the histogram h _s (m _s ) arranged in descending order.

The flowchart of calculating the number of dominant pixel values is described in detail as follows. The step S511 of obtaining the vertical axis size N and the horizontal axis size J of the sinogram P (n, j) is used in the step S518 and S521 of determining the number of dominant pixel values. It is a step to find 0.5N and J which are the judgment indicators. Initializing the device number j ₀ of the X-ray detector sensing device to 1 (S512) to performing a repeating loop until the device number j ₀ becomes the maximum device number J (S521). The process of repeats the above step since the number of dominant pixel values H (j ₀ ) for each selected device number j ₀ is calculated while increasing the device number j ₀ of the X-ray detector sensing element from 1 to J one by one. After the step S521 of performing the loop, the dominant pixel value number H (j) of the entire device number j may be obtained. For the selected element number (j ₀₎ controlled pixel gapsu (H (j ₀₎₎ device number (j ₀₎ No-gram _{(P (n, j 0)} ) between the total rotational angle in order to calculate for A step S514 of obtaining a histogram h (m) is performed. The histogram shows a frequency value for each pixel value with the pixel value being the horizontal axis and the frequency of the pixel value being the vertical axis. If the digital output value of the X-ray detector sensing element is composed of B bits, the pixel value ranges from 0 to 2 ^B −1, resulting in a total of 2 ^B cases. Pixel values at device locations without defects may have various values while pixel values at device locations with defects of type I will have very few cases. 6 is a diagram illustrating a histogram according to an embodiment of the present invention. Here, the selected device number j ₀ is the 889th device immediately adjacent to the 888th device. As shown in FIG. 6, only three pixel values show an extremely high frequency in the 888th device, while various pixel values show similar frequencies in the 889th device. The characteristic that only a small number of pixel values shown in the histogram of the 888th element has an extremely high frequency is a characteristic of a typical type I defective element. A similar frequency of various pixel values shown in the histogram of the 889th device is a characteristic of a normal device. Therefore, even between immediately adjacent devices, it is possible to clearly distinguish the type I defect from the histogram characteristics.

From the step (S515) of obtaining the sorted histogram (h _s (m _s )) to the step (S519) of determining the dominant pixel value H (j ₀ ), it is explicit and mathematical by a criterion for determining the dominant pixel value. In this method, the dominant pixel value is determined. First, the histogram (h (m)) is sorted in descending order, which is the order of the large values to the small values, to form the sorted histogram h _s (m _s ). Since the sorted histogram h _s (m _s ) is sorted in descending order, the independent variable m _s of h _s is also rearranged from the order of pixel values m to the order of h _s , starting from 1 for convenience of calculation. Map to a natural number incrementing by one. Then increments the variable (k) of the representative pixel gapsu by one, the value of h _s (m _s) of the alignment of the histogram (h _s (m _s)) from the m _s = 1 with respect to the up to m _s = k The cumulative frequency S is obtained by adding them all together. Here, the maximum value of the cumulative frequency (S) is equal to the maximum rotation angle (N). A decision criterion for determining the dominant pixel value H (j ₀ ) is 0.5N, and the variable k of the dominant pixel value when the cumulative frequency S is greater than or equal to 0.5N is determined as the dominant pixel value H (j ₀ ). do. Herein, 0.5 N, which is the criterion, is a value corresponding to 50% of the total rotation angle N, and S must be at least 50% of N to become a dominant pixel value. Determining a k of not less than the cumulative total S is controlled by the pixel gapsu 0.5N H (j ₀₎ of the selected element number (j _0), and j is ₀ from step S513 to step S521 until a maximum number of elements J By performing the iterative loop, the dominant pixel value number H (j) for the entire element number j can be calculated.

In step S330 of determining whether the sensing device is an I-type defective device, the characteristic of the dominant pixel value number H (j) is used. Since the number of the dominant pixel values at the position of the type I defect element is significantly smaller than the peripheral value, the position of the type I defect element can be known. The small value may vary depending on the state of the X-ray detector, and may typically be determined in the range of 1 to 20 degrees. In an embodiment, when H (j) is 100 when the frequency of the total rotation angle is 360, since the pixel values of the sinogram have various values, it means that the j th element is a normal element, and H (j) is 5 In this case, since the pixel value has only about five specific values, that is, the j th element is a defective element.

To easily locate the defective device, a filter pass signal H _d (j) obtained by passing the dominant pixel value H (j) through a high pass filter can be additionally used as shown in Equation 3 below. have.

The filter passing signal H _d (j) further enhances the peak of the H (j) in the negative direction so as to have a strong negative value. Therefore, the position of the element number j when the value of H _d (j) is smaller than the appropriate negative threshold can be determined as the position of the type I defective element. FIG. 7 is a diagram illustrating a dominant pixel value H (j) and a filter passing signal H _d (j) according to an exemplary embodiment of the present invention. As shown in the lower magnified area of Fig. 7, H (j) represents a peak in the negative direction at the position of the defective element, and H _d (j) represents a stronger negative value with the peak being further strengthened.

In the step S330 of determining whether the sensing device is an I-type defective device, even when the defective device does not appear alone, but continuously, the I device is characterized by the characteristics of the dominant pixel value number H (j) and the filter passing signal H _d (j). The location of the defective element can be determined. When the defective device appears continuously, ring artifacts also appear continuously, resulting in thick ring artifacts or band artifacts. When the defective device occurs continuously, it is possible to determine whether the defective device by separating the H (j) and H _d (j) by the number of consecutive defective devices. 8 is a view showing a method of processing a continuous defective device according to an embodiment of the present invention. As shown in Fig. 8, the 887th and 888th elements have continuous defects. If there are two consecutive defective devices, the odd and even positions of H (j) are separated to form new H ₁ (j) and H ₂ (j). In other words, H ₁ (j) = H (2j-1), H ₂ (j) = H (2j) separated by passing the high-pass filter through each of the H ₁ (j), H ₂ (j) H _1d (j), H _2d (j) is obtained. In the graphs of H ₁ (j) and H _1d (j) for the odd number of elements separated in this manner, only one element of the 887th element shows defect characteristics and H ₂ (j for even number of elements. ) And H _2d (j), since only one of the 888th devices shows the characteristics of the defective device, the position of the defective device can be determined more accurately. If L defective pixels appear in succession, H ₁ (j), H ₂ (j),... … The position of the defective element can be found from H _L (j), respectively.

Correcting the pixel value obtained from the type I defective element (S340) is performed on the sinogram that finds the position of the type I defective element. The sinogram may be corrected by estimating and filling defective pixel values of the position of the defective device of type I from adjacent normal pixel values. The method of filling the bad pixel value with the surrounding normal value is called inpainting. The inpainting is performed by convolution of a partial sinogram centered on an inpainting matrix and a defective pixel, and is defined as Equation 4 below.

Where c is the corrected pixel value, H is the inpainting matrix, P ¹ is the partial sinogram centered on the defective pixel, a and b represent the rows and columns of the matrix, and a is 1 to A and b is It has a range from 1 to B. The matrix sizes of H and P ¹ are the same, and when the defective pixels are represented by a single defective element in the form of a vertical line, a 3 × 3 matrix of A = 3 and B = 3 is required. In the case of two or more continuous vertical lines by two or more consecutive defective elements, a matrix of 5 × 5 or more is required. The convolution is a process of calculating a weighted average that weights and adds defective pixel values to the surrounding normal values by an inpainting matrix. Here, the sum of all components of the inpainting matrix should be 1. The convolution is performed two-dimensionally, and is performed only at the position of the defective pixel, so that the computation speed does not matter much. An embodiment of the inpainting matrix when the defective pixel is in the form of a single vertical line is expressed by Equation 5 below.

The calculating of the curve curve error in the sinogram in which the pixel value of the type I defective device is corrected (S350) is a step for finding the position of the type II defective device. The defective pixel due to the type II defective element is difficult to find on a general sinogram because a small abnormal offset is added to the normal pixel. The abnormal offset may accumulate when the sum curve, y (j), is obtained by adding the sinogram P (n, j) in the vertical direction, which is the direction of the rotation angle n. The definition of the curve is shown in Equation 6 below.

Here, the curve y (j) is a function of element number j of the detector sensing element. When the detector detecting element is defective, a sharp peak appears on the curve having a continuous shape. This peak may be in the positive or negative direction. In other words, when the defective element emits a value larger than the normal value, the peak direction becomes positive, and in the opposite case, the peak direction becomes negative. 9 is a diagram showing a curve before and after correcting type I defects according to an exemplary embodiment of the present invention. Where y (j) is the summation curve for the sinogram P (n, j) before correcting type I defects, and y '(j) is the sinogram P' (n, after correcting type I defects j) for j). And y _s '(j) is a curve created by smoothing y' (j). The smoothing may be implemented by applying a moving average filter or a low pass filter. Comparing the curves before and after correcting the type I defects, y (j) and y '(j), it can be seen that type I defects with a very large peak are corrected. Although a large peak does not appear in the curved line y '(j) in which the type I defect is corrected, as shown in the enlarged area in FIG. 9, an error due to the type II defective element still exists.

The chamfer error is used to find the position of the type II defective element, and the chamfer error e (j) is defined as in Equation 7 below.

As shown in the enlarged area on the right side of FIG. 9, it can be seen that the signal difference between the two waveforms of y '(j) and y _s ' (j) is relatively large at the position where the type II defective element is present. Accordingly, in the determining of whether the sensing element of the X-ray detector is a type II defect device (S360), the curve error e (j) may be used. If the magnitude of the curve error is larger than a specific threshold, it is determined as the location of the type II defective device. The threshold may vary depending on the characteristics of the detector.

Correcting the pixel value obtained from the type II defective element (S370) is performed at the position of the type II defective element of the sinogram that corrects the type I defective pixel. The sinogram P " (n, j) correcting the defective pixel value of the type II defective element position has an average sum curve error e (j) in the sinogram P '(n, j) correcting the type I defective pixel. ) / N is calculated by subtracting and defined as Equation 8 below.

Here, N denotes the total number of rotation angles, which is equal to the number of vertical directions of the sinogram. Since the curve error e (j) is an error accumulated in the vertical direction, the average curve curve e (j) / N divided by N is subtracted.

The ring artifact removing device in the X-ray CT according to an embodiment of the present invention further includes a control unit having the following features in the X-ray CT system of FIG. 1. The control unit configures a sinogram from a plurality of projection data acquired through the X-ray detector, calculates the number of dominant pixel values for each sensing element of the X-ray detector from the sinogram, and detects the dominant pixel values according to the dominant pixel values. Determining whether it is an I-type defective device, correcting the pixel value obtained from the I-type defective device when the sensing device is an I-type defective device, and combining the curve error in the sinogram whose pixel value of the I-type defective device is corrected. And calculating whether the sensing device is a type II defective device according to the error of the curve, and correcting the pixel value obtained from the type II defective device when the sensing device is a type II defective device. .

In the above, the method and apparatus for removing ring artifacts in X-ray CT according to an embodiment of the present invention have been described. For reference, as shown in FIG. 10, it can be seen that the ring artifact removing method and apparatus in the X-ray CT of the present invention effectively remove the ring artifact. FIG. 10A is a cross-sectional image before removing a ring artifact according to an embodiment of the present invention, and FIG. 10B is a cross-sectional image after removing a ring artifact according to an embodiment of the present invention. The experiment obtained the cross-sectional image is taken of the abdomen of the experimental rat. The X-ray CT used in the experiment was a flat plate detector (model name: C7943CP-02) having 1248 × 1248 pixels produced by Hamamatsu, Japan. The pixel size of the flat plate detector was 100 μm and the scintillator plate was made of CsI. X-ray tubes are microfocus tubes operating in the band between 20kV and 150kV and have a focal size ranging from 5μm to 50μm. The maximum current of the X-ray tube is 0.5 mA. Comparing the cross-sectional image before the ring artifact removal with the cross-sectional image after the removal, it can be seen that the ring artifact is effectively removed, thereby greatly improving the resolution and contrast of the image.

<Description of the symbols for the main parts of the drawings>
110: X-ray beam
120: X-ray tube
130: object to be photographed
140: X-ray detector
150: rotation axis
410 vertical line defect
<Explanation of Symbols for Equation>
n: rotation angle (1, 2, ……, N)
N: maximum rotation angle
j: Element number of sensing element (1, 2, ……, J)
J: Maximum device number
P (n, j): sinogram
P _I (n, j): Sinogram with type I defect
P _II (n, j): sinogram with type II defects
A _j : Type I defect value
a _j : Offset value due to type II defect
j ₀ : Selected device number
k: variable of dominant pixel values
m: pixel value
h (m): histogram
m _s : Independent variable of sorted histogram
h _s (m _s ): sorted histogram
S: cumulative frequency
H (j ₀ ): Number of dominant pixel _values for the selected device number j ₀
H (j): Number of dominant pixel values for the entire device number j
H _d (j): Filter pass signal
H ₁ (j): H (j) for odd element number
H _1d (j): H _d (j) for odd element number
H ₂ (j): H (j) for even device number
H _2d (j): H _d (j) for even element numbers
c: corrected pixel value
H: inpainting matrix
P ¹ : partial sinogram
a: row of matrix
b: column of the matrix
y (j): Sum curve for P (n, j)
P '(n, j): sinogram after correcting type I defect
P '' (n, j): sinogram after correcting type II defects
y '(j): The curve for P' (n, j)
y _s '(j): Smoothed curve of y' (j)
e (j): y '(j) minus y _s ' (j)

Claims (5)

Constructing a sinogram (P (n, j)) from a plurality of projection data obtained through the X-ray detector,
Calculating a number of dominant pixel values for each sensing element of the X-ray detector from the sinogram P (n, j);
Determining whether the sensing device is an I-type defective device according to the number of dominant pixel values;
Correcting the pixel value obtained from the I-type defective device when the sensing device is an I-type defective device;
A compound curve y '(j) obtained by adding the sinograms P' (n, j) in which the pixel values of the I-type defect element are corrected in the rotation angle n direction, and the compound curve y '( calculating the curve error (e (j) = y '(j) -y _s ' (j)) by the difference between the curve (y _s ' (j)) produced by smoothing j)); ,
Determining the sensing device as a type II defective device when the magnitude of the curve error e (j) is larger than a threshold set in the X-ray detector;
And correcting the pixel value obtained from the type II defective device when the sensing device is a type II defective device. The method of claim 1, wherein the calculating of the number of dominant pixel values comprises:
Obtaining a vertical axis size (N) and a horizontal axis size (J) of the sinogram (P (n, j)),
Initializing the device number j ₀ of the X-ray detector sensing device to 1;
Initializing the variable k of the number of dominant pixel values to 0;
Obtaining a histogram h (m) of the sinogram P (n, j ₀ ) consisting of projection data of the total rotation angles at the selected device number j ₀ ;
Obtaining a histogram h _s (m _s ) in which the histogram h (m) is arranged in descending order;
Increasing the variable k of the number of dominant pixel values by 1;
Obtaining a cumulative frequency S for the sorted histogram h _s (m _s ),
Determining a case where the cumulative frequency S is greater than or equal to 0.5N;
Determining the variable k of the dominant pixel value when the cumulative frequency S is greater than or equal to 0.5N as the dominant pixel value H (j ₀ ) in the selected device number j ₀ ;
Adding 1 to the selected device number j ₀ ;
And performing a repeating loop until the selected device number (j ₀ ) becomes the maximum device number (J). The method of claim 1, wherein the determining of whether the sensing device is an I-type defective device comprises:
And determining the defect type I element when the filter pass signal passing the number of dominant pixel values through the high pass filter is smaller than a negative threshold. 4. The method of claim 3, further comprising: separating the filter pass signals by the number of the plurality of filter pass signals when the filter pass signal is smaller than a negative threshold, and determining the defect type I device. Method for removing ring artifacts in x-ray CT. The sinogram P (n, j) is constructed from a plurality of projection data obtained through the X-ray detector,
The number of dominant pixel values is calculated for each sensing element of the X-ray detector from the sinogram P (n, j),
It is determined whether the sensing device is an I-type defective device according to the number of dominant pixel values.
If the sensing device is an I-type defective device, correct the pixel value obtained from the I-type defective device,
A compound curve y '(j) obtained by adding the sinograms P' (n, j) in which the pixel values of the I-type defect element are corrected in the rotation angle n direction, and the compound curve y '( Calculate the curve error (e (j) = y '(j) -y _s ' (j)) by the difference between the curve (y _s ' (j)) made by smoothing j)),
When the magnitude of the curve error e (j) is larger than the threshold set in the X-ray detector, the sensing device is determined as a type II defective device.
And a control unit for correcting a pixel value obtained from the type II defective element when the sensing element is a type II defective element.

Images