Correction device, system, method, and program

Abstract

To provide a correction device, a system, a method, and a program capable of reducing costs for correcting artifacts by a motion in reconfiguration of a T image.SOLUTION: A correction device for correcting artifacts by a motion in measuring a CT image includes: a projection image acquisition unit for acquiring a projection image of 360° scanning; a motion quantity calculation unit for setting a motion model including a parameter and calculating a motion quantity using the parameter; a relative motion quantity calculation unit for calculating a relative motion quantity in projection data from the projection data in which the projection image is included and its counter data; a fixed point equation creation unit for creating a fixed point equation including the motion quantity and the relative motion quantity; a motion estimation unit for determining the parameter of the motion model by solving the fixed point equation in a self-consistent manner; and a correction unit for correcting the projection image using the motion quantity.SELECTED DRAWING: Figure 6

Description

Technical Field

[0001] The present invention relates to a correction device, system, method, and program for correcting artifacts.

Background Art

[0002] A CT apparatus reconstructs a CT image from a plurality of projection images obtained while rotating a sample or a gantry. In a CT apparatus, movement of the sample or the optical system during measurement is called motion. If reconstruction is performed without correcting the projection image in which motion has occurred, blurring and streak-like artifacts will occur in the reconstructed CT image. Therefore, since the reconstructed image does not accurately reflect the shape of the sample, the quantitative property is lost.

[0003] In order to reduce such artifacts caused by motion, conventionally, imaging has been performed by introducing an apparatus other than a CT apparatus, the imaging method has been devised, or correction has been performed by software. Patent Document 1 aims to provide a CT imaging method and apparatus capable of reducing motion artifacts caused by the body movement of a subject during CT imaging and easily performing imaging without firmly positioning the subject, and discloses a technique for acquiring three-dimensional position information using a belt-shaped laser and correcting a projection image obtained by a CT scan.

[0004] Non-Patent Document 1 discloses a technique for performing a first scan normally, a second scan roughly and quickly, assuming that there is no motion during the second quick scan, and correcting the projection image obtained by the first measurement based on the projection image. Further, Non-Patent Document 1 discloses a technique for gradually and precisely estimating motion in the repetition of projection and backprojection. Non-Patent Document 2 discloses a technique for obtaining relative motion using opposing data, preparing several types of absolute motions estimated from the relative motion, and considering the absolute motion with the largest sharpness of the reconstructed image corrected for that motion as the actual motion.

Prior Art Documents

Patent Document

[0005]

Patent Document 1

Non-Patent Document

[0006]

Non-Patent Document 1

Non-Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0007] However, the technology described in Patent Document 1 requires the introduction of special devices such as lasers and sensors, which incurs high introduction costs. Among the technologies described in Non-Patent Document 1, the technology of performing two measurements still has motions that remain even if the measurement is performed quickly. For example, motions based on tolerance errors derived from the rotation axis cannot be corrected. In addition, two measurements must be performed, which takes time for the measurement. Among the technologies described in Non-Patent Document 1, the technology of sequentially estimating motions requires repeated projection and back-projection calculations, which incurs high computational costs. The technology described in Non-Patent Document 2 requires reconstruction only for combinations of the number of types of motions, which incurs high computational costs. Also, in many cases, the actual motion cannot be understood from relative motions.

[0008] The present invention has been made in view of such circumstances, and an object thereof is to provide a correction device, a system, a method, and a program that can reduce the cost for correcting artifacts caused by motion in CT image reconstruction.

Means for Solving the Problems

[0009] (1) To achieve the above object, a correction device of the present invention is a correction device that corrects artifacts caused by motion during CT image measurement, and includes a projection image acquisition unit that acquires projection images of 360° scan, a motion amount calculation unit that sets a motion model including parameters and calculates a motion amount using the parameters, a relative motion amount calculation unit that calculates a relative motion amount in the projection data from the projection data of the projection image and its opposing data, a fixed point equation creation unit that creates a fixed point equation including the motion amount and the relative motion amount, a motion estimation unit that determines the parameters of the motion model by solving the fixed point equation in a self-consistent manner, and a correction unit that corrects the projection image using the motion amount.

[0010] (2) Further, in the correction device of the present invention, the projection image is an image obtained by a fan beam, and the relative motion amount is obtained from the projection data after fan-parallax conversion and its opposing data.

[0011] (3) Further, in the correction device of the present invention, the motion model is a model representing one or more of the translation of the sample rotation axis on the detector surface or the rotation around a certain axis.

[0012] (4) Further, in the correction device of the present invention, the motion model is represented by a finite-order polynomial having the rotation angle of the sample rotation axis as a variable.

[0013] (5) Further, in the correction device of the present invention, the motion estimation unit determines the parameters of the motion model based on a given number of iterations.

[0014] (6) Further, in the correction device of the present invention, the motion estimation unit determines the parameters of the motion model based on a given threshold value.

[0015] (7) Further, in the correction device of the present invention, the calculation of the relative motion amount is performed based on a partial region of the projection data.

[0016] (8) Further, the correction device of the present invention includes a reconstruction unit that performs reconstruction based on the projection data corrected by the correction unit to generate a CT image, and a display unit that displays the CT image on a display device.

[0017] (9) Further, the system of the present invention includes a CT device including an X-ray source that generates X-rays, a detector that detects X-rays, and a rotation control unit that controls the rotation of the X-ray source and the detector, or a sample, and the correction device according to any one of (1) to (8) above.

[0018] (10) Further, the method of the present invention is a method for correcting artifacts due to motion during CT image measurement, comprising: a step of acquiring projection images of a 360° scan; a step of setting a motion model including parameters and calculating a motion amount using the parameters; a step of calculating a relative motion amount in the projection data from certain projection data of the projection image and its opposing data; a step of creating a fixed-point equation including the motion amount and the relative motion amount; a step of determining the parameters of the motion model by solving the fixed-point equation self-consistently; and a step of correcting the projection image using the motion amount.

[0019] (11) Further, the program of the present invention is a program for correcting artifacts due to motion during CT image measurement, causing a computer to execute: a process of acquiring projection images of a 360° scan; a process of setting a motion model including parameters and calculating a motion amount using the parameters; a process of calculating a relative motion amount in the projection data from certain projection data of the projection image and its opposing data; a process of creating a fixed-point equation including the motion amount and the relative motion amount; a process of determining the parameters of the motion model by solving the fixed-point equation self-consistently; and a process of correcting the projection image using the motion amount.

Brief Description of the Drawings

[0020]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Embodiments for Carrying Out the Invention

[0021] Next, embodiments of the present invention will be described with reference to the drawings. For ease of understanding the description, the same reference numerals are assigned to the same components in each drawing, and duplicate descriptions are omitted.

[0022] [Principle] A CT device irradiates a sample with X-rays of a parallel beam, a fan beam, or a cone beam from all angles, and a detector acquires a distribution of X-ray absorption coefficients, that is, a projection image. In order to irradiate the sample with X-rays from all angles, the CT device is configured to rotate a sample stage with respect to a fixed X-ray source and detector, or to rotate a gantry in which the X-ray source and the detector are integrated.

[0023] In this way, the distribution of the linear absorption coefficient of the sample can be inferred from the shades of the projection images of the sample obtained by performing projections from various angles. And obtaining a three-dimensional linear absorption coefficient distribution from two-dimensional projection images is called reconstruction. Reconstruction basically performs back-projection of the projection image.

[0024] Motion refers to the movement of the sample or the optical system during the measurement of the projection image. Causes of motion include thermal drift, focus shift, tolerance error, poor fixation of the sample, etc. If reconstruction is performed without correcting the projection image in which motion has occurred, blurring and streak-like artifacts will occur in the reconstructed CT image. Such artifacts are called artifacts due to motion. When artifacts due to motion occur, the quantitative property is lost because the reconstructed image does not accurately reflect the shape of the sample.

[0025] Conventionally, in order to suppress artifacts due to motion, imaging using a device other than the CT device, devising an imaging method, or correction by software has been performed. For example, there is a method of correcting using three-dimensional position information measured by introducing a special device such as a laser or a sensor into the CT device. In this case, it costs to introduce.

[0026] Also, as a device of the imaging method, there is a method of measuring precisely for the first time, measuring quickly for the second time, and correcting the projection image measured for the first time based on the projection image measured for the second time (reference scan measurement). However, it takes time for measurement. Also, in such a method, since the rotation axis moves regardless of the measurement time, motion caused by tolerance error derived from the rotation axis cannot be corrected.

[0027] In software-based correction, image processing is performed for correction. For example, there are methods of estimating and correcting motion using the sequential method, and methods of assuming motion, reconstructing multiple times, and searching for the motion with the best index. However, all of them inevitably incur computational costs and practical problems arise.

[0028] The present invention creates a fixed-point equation including a motion amount and a relative motion amount for a projection image of 360° scan, estimates the motion by solving the fixed-point equation self-consistently, and corrects using the motion amount. As a result, it is not necessary to introduce a special device such as a laser or a sensor, so that the introduction cost can be reduced. In addition, since correction is performed on the projection image, the computational cost can be reduced. Therefore, compared with the conventional method, the processing from measurement to CT image acquisition can be performed in a very short time. The correction performed in the present invention is to convert the coordinate values of CU, CV, and θ of the sinogram by the amount of movement by a method corresponding to the type of motion with respect to the assumed type of motion and the given amount of movement.

[0029] The motion model is preferably a function expressed with the rotation angle θ of the sample rotation axis as a variable, assuming the translation of the sample rotation axis on the detector surface or the rotation around a certain axis. Here, a certain axis is either the sample rotation axis or the in-plane rotation axis of the detector. However, depending on the way of the device (measurement), there is a possibility that it is the rotation of the rotation axis of the device rather than the rotation of the sample rotation axis. Since the rotation angle is relative, the rotation angle of the rotation axis of the device can also be applied to the present invention as a variable. FIG. 1 is a schematic diagram showing the axes related to the motion and the way they move. The types of motion include the rotation around the sample rotation axis shown in FIG. 1(a), the translation of the sample rotation axis (CU direction, CV direction) shown in FIG. 1(b), and the rotation around the in-plane rotation axis of the detector. In the present invention, the above motions are the correction targets. By solving the fixed-point equation regarding the parameters of the motion model self-consistently, motion correction can be performed by processing only the projection image, enabling high-speed processing.

[0030] Figs. 2(a) and 2(b) are schematic diagrams showing projection data at a rotation angle θ of the sample rotation axis and its opposing data when a motion occurs in which the sample rotation axis translates in the CU direction. In an actual projection image, the position of the sample rotation axis is not detected, but for convenience, the rotation axis is described. In the case of the example of Fig. 2, in the projection image at each rotation angle θ, the difference between the axis passing through the center of the sample rotation axis and the detector can be expressed as a motion amount Δx(θ). The motion amount in the opposing data of the projection image is Δx(θ + π). When there is translation in the CV direction or rotation within the detector plane, a model may be established assuming each motion. The relative motion amount is the relative movement amount obtained from the data of the projection image at a rotation angle θ and the data of the projection image at a rotation angle θ + π (a pair of projection images with a 180° difference in the rotation angle of the sample rotation axis) that opposes it. The relative motion amount r(θ) at a rotation angle θ can be expressed by the following formula (1).

[0031]

Equation

[0032] The calculation of the relative motion amount may be limited to a partial region of the projection data of the projection image and a partial region of its opposing data. Thereby, the calculation cost for calculating the relative motion amount can be reduced. Fig. 2 shows a state in which a frame A is set in (a) and a frame B is set in (b), and the relative motion amount is calculated from the degree of coincidence between the inside of the frame A and the inside of the frame B.

[0033] The relative motion amount is preferably determined using as an index the degree of coincidence of frames defined in the projection image at a rotation angle θ of the sample rotation axis and the projection image at a rotation angle θ + π of the sample rotation axis that opposes it. Depending on the type of motion assumed, the way of moving the frame defined on the projection image is changed, and the movement amount of the frame with the highest degree of coincidence is calculated. This movement amount can be taken as the relative motion amount r(θ).

[0034] The projection image cannot be corrected directly from the relative motion amount. Also, the motion amount cannot be directly obtained from the projection image. On the other hand, if the motion model is known as a function, the motion amount can be calculated, and the projection image can be corrected using this. The motion model is a function that can obtain the motion amount at the rotation angles of all sample rotation axes. Here, for example, when setting a motion model assuming the motion of the sample rotation axis translating in the CU direction, the value obtained by adding the relative motion amount r(θ) to the motion amount Δx(θ) at the rotation angle θ of a certain sample rotation axis obtained from the motion model can infer the motion amount d(θ + π) at the rotation angle (θ + π) of the opposite sample rotation axis obtained from the motion model. This relationship can be expressed by the following equation (2).

[0035]

Number

[0036] To determine the parameters of the motion model, an equation (fixed-point equation) regarding a finite number of parameters is set. Using equation (2), d(θ + π) is calculated for a plurality of angles θ. Let the column vector obtained by arranging these calculated values in order be D. Let the column vector C obtained by arranging the parameters of the motion model in order and the matrix A with the coefficients of each parameter at the rotation angle of the sample rotation axis of the motion model as elements, and an equation is set such that their product becomes D. This equation is represented by the following equation (3). θ + π where the rotation angle of the sample rotation axis of the motion model

[0037]

Number

[0038] Since D is a function of C, Equation (3) can be regarded as a fixed-point equation with C as the fixed point. By solving this equation self-consistently (self-collision-free), C can be obtained. For example, C can be obtained iteratively. Obtaining C iteratively means estimating C using the above equation for the C assumed at the k-th step, and repeating the process of estimating C from the above equation with this as the C assumed at the (k + 1)-th step. Specifically, a numerical value is given to the parameter C of the motion model of the left side D as C k and the parameter C of the motion model of the right side AC is assumed to be undetermined as C k+1 and C k+1 is obtained and used again as the parameter C of the left side D. Note that the method of solving the fixed-point equation self-consistently is not limited to this.

[0039] Using the function obtained by applying the thus obtained C to the motion model, the motion amount is calculated. The calculated motion amount can be used as a correction amount to correct the projection image. By reconstructing the corrected projection image, an image with reduced artifacts due to motion during CT image measurement can be obtained.

[0040] [Embodiment] Hereinafter, the correction method of the present invention will be described in detail. Hereinafter, assuming that the rotation axis of the sample moves by Δx(θ) and Δz(θ) in the CU direction and the CV direction, respectively, with reference to the detector origin, a motion model is set. In this way, the motion model preferably represents one or more of the translation of the sample rotation axis on the detector surface and the rotation around the in-plane rotation axis of the sample rotation axis or the detector, which facilitates the setting of the motion model.

[0041] For the function representing the motion model, any function can be used as long as it can appropriately represent the motion. Assuming that, as described above, with the detector origin as the reference, when the rotation axes of the sample move by Δx(θ) and Δz(θ) in the CU direction and CV direction respectively, since the motion is a function of the rotation angle θ of the sample rotation axis, for example, the motion model can be given by a power series expansion with respect to θ. In this case, the actually set motion model is preferably represented by a finite-order polynomial with the rotation angle of the sample rotation axis as a variable. Additionally, for example, when the motion oscillates, the motion model may be given by a Fourier series expansion. When there is no motion, Δx = 0 and Δz = 0 at any angle.

[0042] Assuming that the motion model is represented by a finite-order polynomial with the rotation angle θ of the sample rotation axis as a variable, for example, the motion model Δx(θ) is expressed as in the following formula (4). deg is the degree of the motion model. Since the motion is small, the degree deg of the motion model is preferably low. Thereby, the number of parameters of the motion model can be reduced, and the cost required for processing can be reduced. For example, the degree deg of the motion model can be 10 or less. Also, the motion Δz(θ) in the CV direction and the rotational motion Δφ(θ) around an axis can be expressed in the same way as formula (2).

[0043]

Equation

[0044] Next, substitute (θ + π) for multiple angles θ in formula (4), and arrange multiple results obtained by substituting d(θ + π) into the left side of each formula. When represented as a matrix equation, it becomes formula (5). Here, the left side corresponds to D, the column vector obtained by arranging the parameters on the right side in order corresponds to C, and the matrix on the right side corresponds to A. The maximum value of the subscript of θ is n p Let it be. n p represents the total number of projection images. Since the number of equations M of the system of linear equations is represented by the number of pairs of opposing data used for motion estimation, the maximum value of M is n pIt becomes / 2. The m-th equation of the simultaneous equations is obtained by substituting the projection angle selected as the m-th plus π into (4). The elements of the matrix on the right side are given by the following equation (6).

[0045] [Number]

[0046] [Number]

[0047] Calculate the relative motion amount in the projection data from the projection data with a projection image and its opposing data. The projection data with a projection image and its opposing data almost coincide except for inversion when there is no motion. Therefore, examine how much and in which direction the data obtained by inverting the opposing data with respect to the projection data is translated or rotated, and the motion amount at which the degree of coincidence is maximized is taken as the relative motion amount.

[0048] The calculation of the relative motion amount may be performed not from all projection images but from a part of the projection data of the projection image and its opposing data. This is because since the parameters of the motion model are finite, it is sufficient to calculate the number of relative motion amounts necessary to obtain those parameters. Thereby, the calculation cost required for calculating the relative motion amount can be reduced.

[0049] Let the equation specifically giving the initial value of the parameter of the motion model Δx(θ) be the motion amount at the angle θ (0 ≤ θ < π). Let D be a column vector obtained by arranging in order the sums of the relative motion amounts at θ calculated above. Then, create an equation assuming that D is equal to the right side AC of equation (5). This becomes the fixed point equation corresponding to equation (3) in the present embodiment.

[0050] By solving this fixed-point equation self-consistently, the parameters of the motion model Δx(θ) can be determined. For example, by first using the obtained C as the value of the parameter for calculating the next motion amount to create a new fixed-point equation and then repeatedly solving it, the fixed-point equation can be solved self-consistently.

[0051] When iteratively obtaining the parameters of the motion model Δx(θ) from the fixed-point equation, it is preferable to determine the parameters of the motion model based on a given number of iterations in advance. Thereby, the parameters can be determined within a predetermined time.

[0052] Also, when iteratively obtaining the parameters of the motion model Δx(θ) from the fixed-point equation, it is preferable to determine the parameters of the motion model based on a given threshold value in advance. For example, for the residual square (index) between the initial C k and the updated C k+1 at each step, a threshold value can be set. Thereby, parameters with a predetermined accuracy can be determined.

[0053] Since the absolute amount of movement can be known from the motion model Δx(θ) and the determined parameters, the motion amount Δx(θ) at each rotation angle θ of the sample rotation axis can be calculated using the motion model Δx(θ) and the determined parameters, and the projection image can be corrected.

[0054] The above description is a method when the X-ray is a parallel beam. However, when the X-ray is a fan beam, the relative motion amount cannot be calculated as it is. Therefore, the correction points when the X-ray is a fan beam will be described below.

[0055] When the projection data of the projection image obtained when the X-ray is a fan beam is not the data in which the opposing data exactly corresponds due to the magnification factor. Therefore, a fan para conversion is performed to convert the projection image obtained by the fan beam method into the projection image of the parallel beam method. Further, the fan para conversion for the projection image with motion needs to be performed after correcting the motion. Therefore, the parameter of the initial value is set with a motion model including parameters, the amount of motion is calculated, and the projection image is corrected (temporarily corrected) using the calculated amount of motion. Then, the corrected projection image is fan para converted, and the relative amount of motion is calculated.

[0056] FIG. 3(a) and (b) are schematic diagrams showing the sinogram of the measured data viewed in the CU-θ plane before the fan para conversion and the sinogram viewed in the CU-θ plane after the temporary correction and the fan para conversion, respectively. Each pixel of the projection image at the angle θ + π in FIG. 3(a) corresponds to the pixel on the cross section represented by the dotted line near the angle θ + π of the sinogram in FIG. 3(b). The opposing position of each pixel of the projection image after the fan para conversion is on the cross section represented by the straight line near the angle θ in FIG. 3(b). When calculating the amount of motion, the coordinate position of each pixel of the projection image at the angle θ + π in FIG. 3(a) is temporarily shifted, and the degree of coincidence between its luminance value and the luminance value of the pixel on the straight line near the angle θ in FIG. 3(b) is compared. At this time, the temporarily shifted amount with the highest degree of coincidence is used as the relative amount of motion. Here, in FIG. 3(b), the CU after the fan para conversion is replaced with the coordinate (CU') of the virtual detection surface. The relative amount of motion obtained in such a way is considered to be the absolute amount of motion because it is the relative amount from the sinogram near the angle θ after the motion correction. Therefore, when creating the fixed point equation, Δx on the right side of Equation (2) is set to 0.

[0057] When the projection image is an image obtained by a fan beam, the relative motion amount is obtained from the fan-parallax-transformed projection data and its opposing data. In this way, by using the fan-parallax-transformed projection image, the relative motion amount can be calculated. Therefore, artifacts due to motion can be corrected. That is, the present invention can also be applied when the X-ray is a fan beam. When the X-ray is a cone beam, by setting a frame near the central cross-section in the CV direction of the projection image, the cone beam can be regarded as a fan beam, so the present invention can be applied.

[0058] Also, as another application method of the present invention, the following method can also be considered. When artifacts due to motion occur during a 180° scan with a parallel beam, the present invention cannot be applied as it is. However, if the apparatus can perform a 360° scan, the present invention can be applied as follows. First, by halving the exposure time and doubling the number of projection images, a 360° scan is performed instead of a 180° scan. Then, motion correction of the present invention is performed, and by adding the opposing data, a projection image for 180° can be obtained.

[0059] By using such a method, a reconstructed image with the same noise level can be obtained while performing motion correction. Also, the measurement time and the reconstruction time are the same as those in the case of a 180° scan.

[0060] [Overall system] FIG. 4 is a schematic diagram showing the configuration of an overall system 100 including a CT apparatus 200, a processing apparatus 300, a correction apparatus 400, an input apparatus 510, and a display apparatus 520 connected thereto. Here, the CT apparatus 200 shown in FIG. 4 is configured to rotate a sample with respect to an X-ray source 260 and a detector 270, but is not limited thereto, and may be configured to rotate a gantry in which the X-ray source and the detector are integrated. Also, the CT apparatus 200 can be used as an apparatus that uses any of a parallel beam, a fan beam, or a cone beam. However, for any beam, a 360° scan is required.

[0061] The processing device 300 is connected to the CT device 200 and controls the CT device 200 and processes the acquired data. The correction device 400 corrects the projection images. The processing device 300 and the correction device 400 may be a PC terminal or a server on the cloud. The input device 510 is, for example, a keyboard or a mouse and inputs to the processing device 300 and the correction device 400. The display device 520 is, for example, a display and displays projection images and the like.

[0062] In FIG. 4, in order to emphasize the correction function of the correction device 400, the processing device 300 and the correction device 400 are shown as separate components. However, as shown in FIG. 5, the correction device 400 may be configured as a part of the functions included in the processing device 300, or the correction device 400 and the processing device 300 may be configured as an integrated unit. FIG. 5 is a schematic diagram showing a modification of the configuration of the entire system. By using such a system, the cost for correcting artifacts due to motion in the reconstruction of CT images can be reduced.

[0063] [CT Device] As shown in FIG. 4, the CT device 200 includes a rotation control unit 210, a sample stage 250, an X-ray source 260, a detector 270, and a drive unit 280. The sample stage 250 installed between the X-ray source 260 and the detector 270 is rotated to perform X-ray CT imaging. Note that the X-ray source 260 and the detector 270 may be installed on a gantry (not shown), and the gantry may be rotated with respect to a sample fixed to the sample stage 250.

[0064] The CT device 200 drives the sample stage 250 at the timing instructed by the processing device 300 and acquires projection images of the sample. The measurement data is transmitted to the processing device 300. The CT device 200 is suitable for use in precision industrial products such as semiconductor devices, and can be applied not only to industrial devices but also to animal devices.

[0065] The X-ray source 260 irradiates X-rays toward the detector 270. The detector 270 has a light-receiving surface for receiving X-rays and can measure the intensity distribution of X-rays transmitted through the sample by a large number of pixels. The rotation control unit 210 rotates the sample stage 250 at a speed set during CT imaging by the drive unit 280.

[0066] [Processing device] FIG. 6 is a block diagram showing the configurations of the processing device 300 and the correction device 400. The processing device 300 is composed of a computer in which a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and a memory are connected to a bus. The processing device 300 is connected to the CT device 200 and receives information.

[0067] The processing device 300 includes a measurement data storage unit 310, a device information storage unit 320, a reconstruction unit 330, and a display unit 340. Each unit can send and receive information via the control bus L. The input device 510 and the display device 520 are connected to the CPU via appropriate interfaces.

[0068] The measurement data storage unit 310 stores the measurement data acquired from the CT device 200. The measurement data includes rotation angle information and the corresponding projection images. The device information storage unit 320 stores the device information acquired from the CT device 200. The device information includes the device name, beam shape, geometry during measurement, scan method, etc.

[0069] The reconstruction unit 330 reconstructs a CT image from the target projection images. The display unit 340 causes the reconstructed CT image and the projection images before and after correction to be displayed on the display device 520. Thereby, the user can confirm the CT image based on the corrected projection image and the projection images before and after correction. Also, the user can give instructions and designations to the processing device, the correction device, etc. based on the CT image and the projection images before and after correction.

[0070] [Correction device] The correction device 400 is configured by a computer in which a CPU, a ROM, a RAM, and a memory are connected to a bus. The correction device 400 may be directly connected to the CT device 200, or may be connected to the CT device 200 via the processing device 300. Further, the correction device 400 may receive information from the CT device 200, or may receive information from the processing device 300. As shown in FIG. 7, the correction device 400 may be configured as a part of the functions included in the processing device 300, or as shown in FIG. 8, the correction device 400 and the processing device 300 may be configured as an integrated unit. Further, the correction device 400 may have a part of the functions of the processing device 300.

[0071] The correction device 400 includes a projection image acquisition unit 410, a motion amount calculation unit 420, a relative motion amount calculation unit 430, a fixed point equation creation unit 440, a motion estimation unit 450, and a correction unit 460. Each unit can transmit and receive information via the control bus L. When the correction device 400 and the processing device 300 have separate configurations, the input device 510 and the display device 520 are also connected to the CPU of the correction device 400 via appropriate interfaces. In this case, the input device 510 and the display device 520 may be different from those connected to the processing device 300.

[0072] The projection image acquisition unit 410 acquires projection images of a 360° scan from the CT device 200 or the processing device 300. The projection images may be images obtained with any of a parallel beam, a fan beam, and a cone beam.

[0073] The motion amount calculation unit 420 sets a motion model including parameters, and calculates the motion amount based on the motion model. The motion model stores the function form in advance. Also, it may be configured to be arbitrarily set by the user selecting the correction target (translation CU: Δx, translation CV: Δz, rotation: Δφ), the function form (power series, Fourier series), and the maximum degree (maximum number of terms). Thereby, the number of motion models and the number of parameters to be estimated can be arbitrarily determined. Also, the motion model may be predetermined. The parameters for calculating the initial motion amount may be specified by the user. Also, it may be configured such that the initial parameters are all set to 0 and the motion amount is calculated. The motion amount calculation unit 420 calculates the motion amount to be corrected (correction amount) using the model with the parameters determined by the motion estimation unit 450 substituted therein.

[0074] The relative motion amount calculation unit 430 calculates the relative motion amount in the projection data from the projection data with a projection image and its opposing data. The relative motion amount calculation unit 430 preferably calculates the relative motion amount in the projection data based on the degree of coincidence between the projection data and the opposing data. When the projection image is an image obtained by a fan beam or a cone beam, the relative motion amount calculation unit 430 fan-parity converts the acquired projection image and calculates the relative motion amount from the projection data and its opposing data.

[0075] The relative motion amount may be calculated not from all the projection images but from a part of the projection data of the projection images and their opposing data. In this case, for example, the relative motion amount can be calculated every 10 projections from the projection images. The setting of the number (number of sets) of opposing data used for the calculation may be configured by the user. Also, it may be configured to be automatically set by the computer based on the degree of the motion model. Also, it may be predetermined.

[0076] The calculation of the relative motion amount may be performed based on a partial region of the projection data of the projection image and a partial region of the opposing data thereof. The setting of the partial region of the projection data may be configured to be performed by the user. For example, the region can be set by specifying the widths in the CU direction and CV direction and the center position of the frame. Also, it may be configured to be automatically set by the computer based on a characteristic structure. Further, it may be predetermined.

[0077] The fixed-point equation creation unit 440 creates a fixed-point equation for determining the parameters of the motion model. The fixed-point equation is created according to the number of set motion models. For example, when the motion model represents the translation in the CU direction of the rotation axis of the sample, from the relative motion amount r(θ) calculated by the relative motion amount calculation unit and the motion amount Δx(θ) calculated by the motion amount calculation unit, the rotation angles θ + π of a plurality of sample rotation axes are calculated and set for d(θ + π). The matrix A with the coefficients of the respective parameters at the rotation angle θ of the rotation axis of the motion model as elements is automatically set when the functional form of the rotation angle and the motion model is determined.

[0078] The motion estimation unit 450 determines the parameters of the motion model by solving the fixed-point equation self-consistently. The determined parameters may be output to the motion amount calculation unit 420.

[0079] The correction unit 460 corrects the projection image using the motion amount (correction amount) calculated by the motion amount calculation unit 420. Thereby, the motion of the projection image can be corrected. The corrected projection image is finally output to the reconstruction unit 330 and converted into a CT image. When the projection image is an image obtained by a fan beam or a cone beam, the corrected projection image may be output to the relative motion amount calculation unit 430, and after fan parameter conversion, it may be used for the calculation of the relative motion amount.

[0080] When the above settings are specified by the user, for example, it is preferable to use a UI function that can make various settings by mouse operations or keyboard operations. FIG. 9 is a conceptual diagram showing an example of the UI when the user makes various settings. In the example of FIG. 9, for example, a frame for calculating the degree of coincidence with the image of the opposing data can be set on the display screen of a certain projection data displayed on the left side of the screen.

[0081] The search range can set how much the frame in the left figure can be moved up, down, left, and right at most. The step can set at what interval the frame is moved to calculate the degree of coincidence. The order can set the maximum order of the motion model. The number of sets can set the number of sets of projection data and opposing data used in the calculation. The number of iterations can set the number of iterations when the fixed-point equation is solved iteratively. Note that the setting items shown in FIG. 9 are just examples. Even when the user sets these, only some of them may be settable, or all of them may be settable. Also, there may be setting items not shown in FIG. 9.

[0082] [Measurement method] A sample is installed in the CT apparatus 200, and by repeating the movement of the rotation axis and the projection of X-rays under predetermined conditions, a projection image is acquired while irradiating the sample with X-rays. The CT apparatus 200 transmits apparatus information such as the scan method and the acquired projection image as measurement data to the processing apparatus 300 or the correction apparatus 400.

[0083] [Correction method] (Explanation of the flow in the case of a parallel beam) FIG. 10 is a flowchart showing an example of the operation of the correction device 400. First, the correction device 400 acquires a projection image (step S1). Next, a motion model is set (step S2). Next, the relative motion amount is calculated (step S3). A set of opposing data is acquired at a plurality of specified angles, and the opposing data is moved by a predetermined amount at each angle, and the movement amount that maximizes their degree of coincidence is defined as the relative motion amount. When a frame is set in the projection image, the degree of coincidence is calculated only for the area within the frame. Next, the parameters of the motion model are set (step S4). Here, all of the initial parameters may be 0. Also, if there are parameters specified or estimated as the initial parameters, those may be used. Next, using the set parameters, the motion amount is calculated (step S5). Next, a fixed-point equation is created (step S6). The relative motion amount and the motion amount calculated at each rotation angle θ are added together and substituted into the left side. These equations are arranged in matrix form to create a fixed-point equation. Next, the fixed-point equation is solved (step S7). By solving the fixed-point equation self-consistently, the parameters of the motion model can be estimated.

[0084] And when the set conditions are not satisfied (step S8 - NO), the process returns to step S4, the estimated parameters are set, and the processing up to step S7 is performed again. On the other hand, when the set conditions are satisfied (step S8 - YES), next, the motion amount (correction amount) is calculated from the estimated parameters (step S9). Then, the projection image is corrected using the estimated motion amount (correction amount) (step S10). In this way, the projection image can be corrected. Note that it does not matter whether the acquisition of the projection image or the setting of the motion model is performed first.

[0085] (Explanation of the flow in the case of a fan beam) FIG. 11 is a flowchart showing a modification of the operation of the correction device 400. FIG. 11 is an example of the operation when correcting a projection image obtained by a fan beam. First, the correction device 400 acquires a projection image (step U1). Next, a motion model is set (step U2). Next, parameters of the motion model are set (step U3). Here, all of the initial parameters may be 0. If there are parameters specified or estimated as the initial parameters, those may be used. Next, using the parameters, the amount of motion is calculated (step U4). Next, using the amount of motion, the projection image is corrected (step U5). Next, the sinogram is fan-par-transformed (step U6). Next, the relative amount of motion is calculated (step U7). The projection image of the opposing data is moved by a predetermined amount, and the amount of movement with the highest degree of coincidence with the opposing data in the corrected sinogram is taken as the relative amount of motion. When a frame is set in the projection image, the degree of coincidence is calculated only for the area within the frame. Next, a fixed-point equation is created (step U8). Next, the fixed-point equation is solved (step U9). By solving the fixed-point equation self-consistently, the parameters of the motion model can be estimated.

[0086] And when the set conditions are not satisfied (step U10-NO), the process returns to step U3, the estimated parameters are set, and the processes up to step U9 are performed again. On the other hand, when the set conditions are satisfied (step U10-YES), next, the amount of motion (correction amount) is calculated from the estimated parameters (step U11). Then, the projection image is corrected using the estimated amount of motion (correction amount) (step U12). In this way, even a projection image obtained by a fan beam can be corrected. Note that, as described above, the acquisition of the projection image and the setting of the motion model may be performed in either order without any problem.

[0087] Note that the conditions for step S8 and step U10 can be the number of loop iterations, whether the determined parameters satisfy a predetermined threshold value, or the like.

[0088] [Correction and Reconstruction Method] The flowcharts of FIGS. 10 and 11 show only the operation of the correction device 400. Although this is sufficient for the correction of the projection image, for the sake of clarity in comparison with the prior art, the operations including the measurement and reconstruction of the projection image will also be described.

[0089] FIG. 12 is a flowchart showing an example of the operation of the system 100. First, the CT device 200 performs a CT measurement (step V1). The CT measurement includes the repetition of the movement of the rotation axis and the projection of X-rays. Next, the motion is estimated and corrected (step V2). The flow for estimating the motion may adopt, for example, one of the flows in FIGS. 10 or 11 according to the type of the optical system of the acquired projection image. Also, the optical system may be discriminated from the device information so that an appropriate flow can be selected. Then, the processing device 300 or the correction device 400 reconstructs a CT image using the corrected projection image (step V3). In this way, a CT image reconstructed using the corrected projection image can be obtained.

[0090] In the prior art, the reconstruction of the CT image and the correction for artifact reduction were repeated. In contrast, in the present invention, the projection image before reconstruction is corrected for artifact reduction. By doing so, the cost for correction can be reduced.

[0091] [Example 1] Using the system 100 configured as described above, the cross-section of a bamboo skewer (sample 1) was observed. The CT device 200 was measured using Rigaku nano3DX (parallel beam). FIGS. 13(a) and (b) are the CT images of sample 1 reconstructed using the uncorrected projection image and the CT images of sample 1 reconstructed using the corrected projection image, respectively. The reconstruction used the FDK method.

[0092] By comparing FIGS. 13(a) and (b), it was found that the artifacts were reduced by the correction.

[0093] [Example 2] Next, the cross-section of another skewer (sample 2) was observed. The CT apparatus 200 was measured using Rigaku HX (fan beam). FIGS. 14(a) and (b) are the CT images of sample 2 reconstructed using the uncorrected projection images and the CT images of sample 2 reconstructed using the corrected projection images, respectively. The reconstruction was performed using the FDK method as in sample 1.

[0094] By comparing FIGS. 14(a) and (b), it was found that the artifacts were reduced by the correction. It was confirmed that the correction method of the present invention can also be applied to the projection images measured by the fan beam.

[0095] From the above results, it was confirmed that the correction apparatus, system, method, and program of the present invention can effectively correct the artifacts due to motion in the reconstruction of CT images and can reduce the calculation cost.

Explanation of Signs

[0096] 100 System 200 CT apparatus 210 Rotation control unit 250 Sample stage 260 X-ray source 270 Detector 280 Driving unit 300 Processing apparatus 310 Measurement data storage unit 320 Apparatus information storage unit 330 Reconstruction unit 340 Display unit 400 Correction apparatus 410 Projection image acquisition unit 420 Motion amount calculation unit 430 Relative motion amount calculation unit 440 Fixed point equation creation unit 450 Motion estimation unit 460 Correction unit 510 Input device 520 Display device

Claims

1. A correction device for correcting artifacts due to motion during CT image measurement, comprising: a projection image acquisition unit that acquires projection images of a 360° scan; a motion amount calculation unit that sets a motion model including parameters and calculates a motion amount using the parameters; a relative motion amount calculation unit that calculates a relative motion amount in the projection data from certain projection data of the projection image at the rotation angle θ of the sample rotation axis and opposing data at the rotation angle θ + 180°; a fixed-point equation creation unit that includes the motion amount and the relative motion amount, estimates the motion amount at θ + 180° for a plurality of the rotation angles θ, and creates a fixed-point equation with the parameters of the motion model as fixed points; a motion estimation unit that determines the parameters of the motion model by solving the fixed-point equation self-consistently; and a correction unit that corrects the projection image using the motion amount. The correction device is characterized by comprising these components.

2. The projection image is an image obtained by a fan beam, and the relative motion amount is obtained from the projection data after fan parameter conversion and its opposing data. The correction device according to claim 1 is characterized by this.

3. The motion model is a model representing one or more of the translation of the sample rotation axis on the detector surface or rotation around a certain axis. The correction device according to claim 1 or claim 2 is characterized by this.

4. The motion model is represented by a finite-order polynomial with the rotation angle of the sample rotation axis as a variable. The correction device according to any one of claims 1 to 3 is characterized by this.

5. The motion estimation unit determines the parameters of the motion model based on a given number of iterations. The correction device according to any one of claims 1 to 4 is characterized by this.

6. The motion estimation unit determines the parameters of the motion model based on a given threshold value. The correction device according to any one of claims 1 to 4 is characterized by this.

7. The calculation of the relative motion amount is performed based on a partial region of the projection data. The correction device according to any one of claims 1 to 6 is characterized by this.

8. a reconstruction unit that performs reconstruction based on the projection image corrected by the correction unit to generate a CT image; A display unit that causes the CT image to be displayed on a display device, wherein the correction device according to any one of claims 1 to 7 is characterized by comprising the display unit.

9. A CT apparatus comprising: an X-ray source that generates X-rays; a detector that detects X-rays; and a rotation control unit that controls the rotation of the X-ray source and the detector, or a sample. A system, characterized by comprising: the correction device according to any one of claims 1 to 8.

10. A method for correcting artifacts due to motion during CT image measurement, comprising: acquiring projection images of a 360° scan; setting a motion model including parameters and calculating a motion amount using the parameters; calculating a relative motion amount in the projection data from certain projection data of the projection image at a rotation angle θ of a sample rotation axis and opposing data at a rotation angle θ + 180°; including the motion amount and the relative motion amount, estimating the motion amount at θ + 180° for a plurality of the rotation angles θ, and creating a fixed-point equation with the parameters of the motion model as fixed points; determining the parameters of the motion model by solving the fixed-point equation self-consistently; correcting the projection image using the motion amount.

11. A program for correcting artifacts due to motion during CT image measurement, comprising: a process of acquiring projection images of a 360° scan; a process of setting a motion model including parameters and calculating a motion amount using the parameters; a process of calculating a relative motion amount in the projection data from certain projection data of the projection image at a rotation angle θ of a sample rotation axis and opposing data at a rotation angle θ + 180°; a process of including the motion amount and the relative motion amount, estimating the motion amount at θ + 180° for a plurality of the rotation angles θ, and creating a fixed-point equation with the parameters of the motion model as fixed points; a process of determining the parameters of the motion model by solving the fixed-point equation self-consistently; a process of causing a computer to execute correcting the projection image using the motion amount.

Images