IMAGE PROCESSING APPARATUS, X-RAY CT RADIATION SYSTEM, AND IMAGE PROC

Abstract

To provide an image processing device, an X-ray CT imaging system, and an image processing method capable of accurately removing a metal artifact.SOLUTION: An image processing device 8 includes a communication interface 86, an arithmetic part 84, and a display part 81. The arithmetic part 84 reconfigures X-ray projection data, and calculates, as a gradient vector, a vector in a direction of a change when there is a change in continuity of a structure between each voxel of three-dimensional shape data extracted from three-dimensional CT image data and an adjacent voxel in the case where the three-dimensional CT image data includes a metal artifact A, and executes subtraction processing for the voxel having a gradient vector in which an angular difference for an axially parallel direction is within a predetermined range with a direction parallel to an axial direction of a pivot as the axially parallel direction.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an image processing device, an X-ray CT imaging system, and an image processing method. [Background technology]

[0002] In the dental field (e.g., orthodontic treatment), dentists use an X-ray CT imaging device to capture images of the maxillofacial region of a subject (object). The X-ray CT imaging device irradiates the subject with X-rays to collect projection data, and reconstructs the obtained projection data on a computer to generate three-dimensional CT (Computerized Tomography) image data (CT image cross-sectional image data, volume rendering image data, etc.).

[0003] However, when the dental arch of the subject being imaged contains metal (e.g., a metal dental crown), the metal reflects and absorbs X-rays, forming radial metal artifacts centered on the metal in the reconstructed 3D CT image data. JP2010-201089A (Patent Document 1) discloses a method for automatically removing metal artifacts from 3D CT image data.

[0004] In the method for automatically removing metal artifacts in 3D CT image data described in Patent Document 1, the 3D CT image data reconstructed by an X-ray CT imaging device is displayed, a virtual plane perpendicular to the imaging plane of the X-ray CT image is constructed, and areas that appear on the virtual plane and have a CT value within a predetermined threshold and a number of pixels below a predetermined value are identified as metal artifacts and removed. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2010-201089 A Summary of the Invention [Problem to be solved by the invention]

[0006] However, the image processing method of Patent Document 1 cannot remove metal artifacts larger than a predetermined threshold among those contained in a virtual plane perpendicular to the imaging plane, and these metal artifacts remain. Conversely, for example, if a tooth is photographed while tilted relative to the imaging plane, the cross section of the tooth that appears on the virtual plane falls within the predetermined threshold and is mistakenly removed. Therefore, the image processing method of Patent Document 1 may not be able to remove metal artifacts from the 3D CT image data with high accuracy.

[0007] The present disclosure has been made to solve the above problems, and has an object to provide an image processing device, an X-ray CT imaging system, and an image processing method that are capable of removing metal artifacts with high accuracy. [Means for solving the problem]

[0008] The image processing device according to the present disclosure is an image processing device that reconstructs X-ray projection data acquired by rotating an X-ray detector around a rotation axis in an X-ray CT imaging device having an imaging area of ​​a subject's maxillofacial region. The image processing device includes an input interface that receives input of X-ray projection data from the X-ray CT imaging device, a processing unit that reconstructs the X-ray projection data received by the input interface, and an output interface that outputs the data reconstructed by the processing unit. The processing unit reconstructs the X-ray projection data. Ta 3D CT image data T or 3D shape data extracted from When metal artifacts are included in the 3D shape data, the false image structures that exist as false images are included in the structures of the 3D shape data. Between each voxel and its adjacent voxels, Changes in whether or not the structure contains hard and soft tissues When there is a change, the change To a voxel with Direction Show The vector is calculated as a gradient vector, and a direction parallel to the axial direction of the rotation axis is defined as an axial parallel direction, and subtraction processing is performed on voxels having gradient vectors whose angular difference with respect to the axial parallel direction is within a predetermined range.

[0009] The X-ray CT imaging system according to the present disclosure includes an X-ray CT imaging device that acquires X-ray projection data with an X-ray image sensor that faces an X-ray generator across a subject, and the image processing device described above.

[0010] The image processing method according to the present disclosure is an image processing method for subtracting metal artifacts from 3D CT image data reconstructed from X-ray projection data acquired by rotating an X-ray detector around a rotation axis in an X-ray CT imaging device having a subject's maxillofacial region as an imaging region. When metal artifacts are included in the 3D shape data, the false image structures that exist as false images are included in the structures of the 3D shape data. Between each voxel and its adjacent voxels, Changes in whether or not the structure contains hard and soft tissues When there is a change, the change To a voxel with Direction Show The method includes a step of calculating a vector as a gradient vector, and a step of performing a subtraction process on voxels having gradient vectors whose angular difference with respect to the axial direction of the rotation axis is within a predetermined range, with the direction parallel to the axial direction being taken as the axial parallel direction. Effect of the Invention

[0011] In the image processing device according to the present disclosure, subtraction processing is performed on voxels having gradient vectors whose angular differences with respect to the axial parallel direction are within a predetermined range, so that metal artifacts can be accurately removed from the three-dimensional shape data. [Brief description of the drawings]

[0012] [Figure 1] FIG. 1 is a schematic diagram showing the configuration of an X-ray CT imaging system. [Diagram 2] FIG. 2 is a schematic diagram for explaining the axial directions of the X-ray CT imaging system. [Diagram 3] FIG. 1 is a diagram conceptually showing the configuration of an X-ray CT imaging apparatus. [Figure 4] FIG. 1 is a block diagram showing the configuration of an X-ray CT imaging system. [Diagram 5] FIG. 2 is a schematic diagram showing how the lower jaw of a subject is photographed using an X-ray CT imaging device. [Figure 6]FIG. 1 is a diagram showing three-dimensional shape data of a part of a subject's lower jaw photographed by an X-ray CT scanner. [Figure 7] FIG. 4 is a diagram illustrating a method for removing metal artifacts in the image processing device according to the first embodiment. [Figure 8] 4 is a flowchart showing a method for removing metal artifacts in the image processing device according to the first embodiment. [Figure 9] FIG. 4 is a schematic diagram for explaining a method for deleting voxels of metal artifacts in the image processing device according to the first embodiment. [Figure 10] 11 is a flowchart showing a method of removing and interpolating metal artifacts in an image processing device according to a second embodiment. [Figure 11] FIG. 13 is a diagram showing three-dimensional shape data including a noisy portion. [Figure 12] FIG. 11 is a schematic diagram for explaining a method for deleting voxels of metal artifacts in the image processing device according to the second embodiment. [Figure 13] FIG. 13 is a diagram showing three-dimensional shape data from which a part of a structure of a subject has been deleted. [Figure 14] 13 is a diagram showing a schematic diagram of a state in which a part of a structure of a subject is deleted in the image processing device according to the second embodiment. FIG. [Figure 15] FIG. 11 is a schematic diagram for explaining a method for interpolating voxels of a structure of a subject that has been deleted in the image processing device according to the second embodiment. [Figure 16] 13A and 13B are diagrams showing three-dimensional shape data for explaining the degree of expansion in dilation processing. [Figure 17] FIG. 13 is a diagram showing another three-dimensional shape data for explaining the degree of expansion in the dilation process. [Figure 18] FIG. 1 shows 3D CT image data with metal artifacts removed. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. (Embodiment 1) When the maxillofacial region of a subject (object) is imaged using an X-ray CT imaging device, if the dental arch of the subject being imaged contains metal (e.g., a metal dental crown), a radial metal artifact centered on the metal is formed in the reconstructed three-dimensional CT image data due to X-ray reflection and absorption by the metal part. In the X-ray CT imaging system according to this embodiment, the X-ray projection data of the object imaged by the X-ray CT imaging device is reconstructed by an image processing device, and the image processing device can perform subtraction processing such as accurately removing the metal artifact from the obtained three-dimensional shape data. First, the configuration of the X-ray CT imaging system according to this embodiment will be described below. [X-ray CT imaging system configuration] Fig. 1 is a schematic diagram showing the configuration of an X-ray CT imaging system 100. Fig. 3 is a diagram conceptually showing the configuration of an X-ray CT imaging device 1. Fig. 4 is a block diagram showing the configuration of the X-ray CT imaging system 100.

[0014] 1, the X-ray CT imaging system 100 includes an X-ray CT imaging device 1 that executes X-ray CT imaging and collects projection data, and an image processing device 8 that processes the projection data collected by the X-ray CT imaging device 1 to generate X-ray CT image data and the like. The X-ray CT imaging device 1 is preferably housed in a hollow, vertically elongated rectangular parallelepiped-shaped X-ray-proof room 70, and is connected to the image processing device 8 arranged outside the X-ray-proof room 70 by a connection cable 83.

[0015] The X-ray CT imaging device 1 is composed of an X-ray generator 10 that emits an X-ray cone beam Bx or an X-ray slit beam composed of a bundle of X-rays toward a subject M1, an X-ray detector 20 having an X-ray detection surface that detects the X-rays emitted by the X-ray generator 10, a rotating arm 30 that supports the X-ray generator 10 and the X-ray detector 20, a vertically extending support column 50, a rotating arm lifting unit 40 that suspends the rotating arm 30 and is movable up and down in the vertical direction relative to the support column 50, and a control unit 60. The X-ray generator 10, the X-ray detector 20, and the beam shaping mechanism 13 arranged on the X-ray detector 20 side of the X-ray generator 10 constitute an imaging mechanism 3.

[0016] The X-ray generator 10 and the X-ray detector 20 are fixed in a suspended state to both ends of a rotating arm 30, and supported so as to face each other. The rotating arm 30 is fixed in a suspended state to a rotating arm lifting section 40 via a rotating shaft 31 extending in the vertical direction.

[0017] The rotating arm 30 has a generally inverted U-shape when viewed from the front, and rotates around a rotating shaft 31 at the upper end as a rotation center Sc. The rotating arm 30 is composed of a rotating unit 30R that rotates around the axis of the rotating shaft 31, an X-ray generating unit 30G provided below one end of the rotating unit 30R and provided with an X-ray generator 10, and an X-ray detecting unit 30D provided below the other end of the rotating unit 30R and provided with an X-ray detector 20. Note that the shape of the rotating arm 30 is not limited to this, and for example, the X-ray generator 10 and the X-ray detector 20 may be supported so as to face each other on a member that rotates around the center of a circular ring portion as a rotation center.

[0018] Hereinafter, the direction parallel to the axial direction of the pivot shaft 31 (in the illustrated embodiment, the vertical direction) is referred to as the "Z-axis direction", the direction intersecting with the Z-axis is referred to as the "X-axis direction", and the direction intersecting with the X-axis direction and the Z-axis direction is referred to as the "Y-axis direction". The direction parallel to the axial direction (Z-axis direction) may be referred to as the axial parallel direction. The direction parallel to the axial direction (Z-axis direction) may be referred to as the axial parallel direction. Note that the X-axis and Y-axis directions may be arbitrarily determined, but in the illustrated embodiment, the left-right direction of the subject, which is the subject M1, when the subject is positioned in the X-ray CT imaging device 1, more specifically, the left-right direction of the subject's head MH facing the support 50, is defined as the X-axis direction, and the front-back direction of the subject, more specifically, the front-back direction of the subject's head MH, is defined as the Y-axis direction. In this embodiment, the X-axis direction, the Y-axis direction, and the Z-axis direction are mutually orthogonal. The X-axis direction, the Y-axis direction, and the Z-axis direction may be simply expressed as the X-direction, the Y-direction, and the Z-direction, respectively.

[0019] Further, the axial directions of the X-ray CT imaging system 100 will be described with reference to the drawings. FIG. 2 is a schematic diagram for explaining the axial directions of the X-ray CT imaging system 100. FIG. 2(a) shows three-dimensional coordinates of the X-axis, Y-axis, and Z-axis in an orthogonal coordinate system XYZC. In this coordinate system, the parietal side of the head MH is set to the upper side, the neck side of the head MH is set to the lower side, and the upper side of the Z axis is set to the +Z side, and the lower side is set to the -Z side. On the X axis, the right side of the head MH is set to the +X side, and the left side is set to the -X side. On the Y axis, the front side of the head MH is set to the +Y side, and the rear side is set to the -Y side. The intersection of the three-dimensional coordinates is indicated by an intersection IS.

[0020] In the case of this embodiment, the Z-axis direction is sometimes called the vertical direction, and the direction on the plane defined two-dimensionally in the X-axis and Y-axis directions is sometimes called the horizontal direction. A composite direction of these directions may be expressed by characters of multiple directions of the components. For example, a direction along a two-dimensional spread consisting of the X-axis and Y-axis directions may be expressed as an XY direction. Both the X-direction and the Y-direction are examples of the XY direction. Furthermore, a surface that spreads in two of these directions may be expressed by characters of multiple directions of the components. For example, a surface that spreads in two dimensions consisting of the X-direction and the Y-direction may be expressed as an XY surface.

[0021] Furthermore, the composite direction of two of the X-axis, Y-axis, and Z-axis directions may be expressed as S, T, or U. A two-dimensional plane extending in the X- and Y-directions is defined as the SS plane. The composite direction of the X- and Y-directions on the SS plane is the S-direction. Figure 2(a) shows the S1 direction as an example of an S-direction and the S2 direction as another example. The SSI plane, which is the SS plane that passes through the intersection point IS, is the reference plane of the SS plane. The SS plane does not have to be on the same plane as the SSI plane as long as it extends parallel to the SSI plane, and the S3 direction on the SS plane as shown in the figure is also an example of an S-direction. The S-direction may coincide with the X- or Y-direction.

[0022] As shown in Figure 2(c), the two-dimensional plane extending in the X and Z directions is the TT plane. The combined direction of the X and Z directions on the TT plane is the T direction. The TTI plane, which is the TT plane passing through the intersection point IS, is the reference plane for the TT plane. The concept is the same as for the SS plane and S direction, so a detailed explanation is omitted.

[0023] As shown in Figure 2(d), the two-dimensional plane extending in the Y and Z directions is the UU plane. The composite direction of the Y and Z directions on the UU plane is the U direction. The UUI plane, which is the UU plane that passes through the intersection point IS, is the reference plane of the UU plane. The concept is the same as the SS plane and S direction, so a detailed explanation will be omitted.

[0024] A two-dimensional plane extending in the Z and S directions may be referred to as a ZS plane. The combined direction of the Z and S directions on the ZS plane may be referred to as the ZS direction. On a ZS plane, the S direction may be differentiated to become the S direction, and a surface such as the ZS1 plane in Figure 2(b) may also be considered a ZS plane. Similarly, a two-dimensional plane extending in the Y and T directions may be referred to as a YT plane. The combined direction of the Y and T directions on the YT plane may be referred to as the YT direction. Similarly, a two-dimensional plane extending in the X and U directions may be referred to as an XU plane. The combined direction of the X and U directions on the XU plane may be referred to as the XU direction.

[0025] On the other hand, for the three-dimensional coordinates on the rotating arm 30, the direction in which the X-ray generator 10 and the X-ray detector 20 face each other is defined as the "y-axis direction", the horizontal direction perpendicular to the y-axis direction is defined as the "x-axis direction", and the vertical direction perpendicular to these x and y-axis directions is defined as the "z-axis direction". In this embodiment and the following embodiments, the Z-axis direction is the same as the z-axis direction. Furthermore, the rotating arm 30 of this embodiment rotates around the vertically extending rotation axis 31 as the axis of rotation. Therefore, the xyz Cartesian coordinate system rotates around the Z-axis (=z-axis) with respect to the XYZ Cartesian coordinate system. The x-axis direction, the y-axis direction, and the z-axis direction may be simply expressed as the x-direction, the y-direction, and the z-direction, respectively. The composite direction of these directions may be expressed by the letters of the multiple directions of the components. For example, the direction along a two-dimensional spread consisting of the x-direction and the y-direction can be expressed as the xy direction. Both the x-direction and the y-direction are examples of the xy direction. A surface that extends in two of these directions may also be represented by letters in multiple directions of the component parts. For example, a surface that extends in two dimensions, consisting of the x and y directions, can be represented as an xy surface.

[0026] In addition, when the X-ray generator 10 and X-ray detector 20 shown in Figure 1 are viewed in a planar view from above, the direction from the X-ray generator 10 to the X-ray detector 20 is defined as the (+y) direction, and the horizontal direction perpendicular to this (+y) direction, i.e., when the head MH is positioned between the X-ray generator 10 and the X-ray detector 20 with the top side vertically upward and the neck side of the head MH vertically downward and looking toward the X-ray detector 20, is defined as the (+x) direction, and the vertical upward direction is defined as the (+z) direction.

[0027] The swivel arm lifting section 40 is composed of an upper frame 41 and a lower frame 42, and protrudes in one direction from an engagement base end that engages with a support pillar 50 erected in the vertical direction, i.e., in the illustrated example, generally toward the right front side when the support pillar 50 is viewed from the front.

[0028] The upper frame 41 has a rotating shaft 31 of the rotating arm 30 attached thereto, and the rotating arm lifting section 40 moves vertically along the support column 50, thereby allowing the rotating arm 30 to move up and down.

[0029] The lower frame 42 has a subject holding arm 422 extending from the support 50, and the subject holding arm 422 is provided with a subject fixing part 423 consisting of a head holder for fixing the subject M1 (here, the head MH of a human body) from the left and right, a chin rest for fixing the chin, etc. Note that ear rods having portions to be inserted into the left and right ear holes of the human head may also be used for the subject fixing part 423. The subject fixing part 423 is configured to fix the head MH so that the axial direction of the body axis in the head and the axial direction of the rotation shaft 31 are approximately parallel.

[0030] As shown in Fig. 3, a pivot movement mechanism 34 is provided inside the upper frame 41. This pivot movement mechanism 34 is connected to one end of the pivot 31. The other end of the pivot 31 is connected to a rotation drive mechanism 35 provided on the rotating arm 30. The rotation drive mechanism 35 includes the pivot 31 and a belt (not shown) wound around a motor (not shown), and the rotation arm 30 rotates about the pivot 31 when driven by the motor of the rotation drive mechanism 35. The pivot 31 is connected to the rotating arm 30 via a bearing (not shown), and when the rotating arm 30 rotates, only the rotating arm 30 rotates without rotating the pivot 31.

[0031] The pivot movement mechanism 34 is composed of, for example, a bolt, a motor, a guide mechanism, etc. By driving the pivot movement mechanism 34, the pivot 31 moves in a horizontal plane, and the pivot arm 30 can be moved to any position within a predetermined range.

[0032] Although not shown, the rotation drive mechanism 35 may be provided in the rotation shaft moving mechanism 34, so that the rotation drive mechanism 35 drives the rotation shaft 31 to rotate, thereby rotating the rotation arm 30 fixed to the rotation shaft 31. In this case, the rotation shaft moving mechanism 34 can be configured to move the rotation drive mechanism 35 together with the rotation shaft 31 two-dimensionally within a horizontal plane.

[0033] The X-ray CT imaging device 1 having such a configuration collects X-ray projection data for the imaging area (area to be imaged) CA by rotating the rotating arm 30 about the rotating shaft 31 while irradiating the imaging area CA of the subject M1 with the X-ray cone beam Bx. X-ray CT imaging may be performed by fixing the axis of the rotating shaft 31 at the center position of the imaging area CA as viewed in the Z direction and rotating the rotating arm 30 by the rotating drive mechanism 35, but for example, X-ray CT imaging may be performed by rotating the rotating arm 30 by the rotating drive mechanism 35 in synchronization with the movement of the rotating shaft 31 on a circular orbit by the rotating shaft moving mechanism 34. Also, X-ray CT imaging may be performed by irregular control, such as moving the X-ray detector 20 closer to the imaging area CA by the rotation axis moving mechanism 34 to reduce the magnification at a specific rotation angle and simultaneously moving the X-ray generator 10 away from the imaging area CA, or moving the X-ray detector 20 away from the imaging area CA and simultaneously moving the X-ray generator 10 closer to the imaging area CA to increase the magnification at a specific rotation angle. In either case, X-ray CT imaging is performed by rotating the X-ray generator 10 and the X-ray detector 20 around the axis of the rotation axis 31 with the center of rotation in the imaging area CA. The collected X-ray projection data is then transmitted to the image processing device 8. The rotation axis moving mechanism 34 and the rotation drive mechanism 35 constitute the rotation arm drive unit 30K.

[0034] In the X-ray CT imaging device 1, the rotating arm driving unit 30K is connected to the control unit 60 as shown in Fig. 4, and is driven according to a predetermined program to move the rotating arm 30 in the XY directions while rotating. This allows the rotating shaft 31 to be two-dimensionally moved and controlled in the XY directions (front / back, left / right).

[0035] The control unit 60 can control the operation of each component of the X-ray CT imaging device 1, and is disposed inside the X-ray detection unit 30D as shown in Fig. 1. In detail, as shown in Fig. 4, the control unit 60 is connected to the X-ray generator 10, the X-ray detector 20, the rotary arm driving unit 30K, the display unit 61, the operation unit 62, the communication interface 63, the storage unit 64, and the notification unit 90, and communicates with and controls each component.

[0036] In this embodiment, the display unit 61 is configured with a touch panel or the like, and thus has some of the functions of the operation unit 62. That is, the display unit 61 also functions as the operation unit 62. Furthermore, a switch unit 65 is connected to the operation unit 62. For example, by pressing the switch unit 65 before CT imaging, the rotating arm 30 can be moved to an imaging start position.

[0037] The communication interface 63 is connected to the connection cable 83 and is an interface for communicating with the image processing device 8. The storage unit 64 stores a control program for controlling each component by the control unit 60, X-ray projection data, etc. The notification unit 90 can execute a notification operation such as emitting a predetermined buzzer sound.

[0038] The image processing device 8 is composed of an image processing main body 80, a display unit 81 consisting of a display device such as an LCD monitor, an operation unit 82 consisting of a keyboard, mouse, etc., a calculation unit 84, a memory unit 85 that stores programs to be executed by the calculation unit 84, and a communication interface 86.

[0039] The display unit 81, the operation unit 82, the storage unit 85, and the communication interface 86 are connected to the calculation unit 84, and are controlled by the calculation unit 84 executing a control program stored in the storage unit 85. The calculation unit 84 also executes an image processing program stored in the storage unit 85 to perform calculation processing such as back projection on the X-ray projection data to generate three-dimensional CT image data (volume data) representing the three-dimensional area of ​​the imaging area CA. That is, the calculation unit 84 is an arithmetic processing device that reconstructs the X-ray projection data. Furthermore, the calculation unit 84 can generate three-dimensional shape data by extracting the three-dimensional shape of the imaging area CA from the three-dimensional CT image data, and can also perform calculation processing to remove metal artifacts from the three-dimensional shape data as described later. Here, the three-dimensional CT image data is data in which the value of the strength of X-ray absorption (CT value) is stored in each voxel. On the other hand, the three-dimensional shape data is data in which "1" is stored in voxels where the structure of the subject M1 is present, and "0" is stored in voxels where no structure is present. In other words, the three-dimensional shape data is data obtained by deleting the internal information of the subject M1 from the three-dimensional CT image data and extracting only the shape information of the subject M1. In this way, "1" or "0" may be assigned according to the presence or absence of a structure, but since the structures include biological hard tissue structures (hereinafter simply referred to as hard tissue structures) and biological soft tissue structures (hereinafter simply referred to as soft tissue structures), which will be described later, real numbers or continuous values, such as "1" for hard tissue structures and "0.5" for biological soft tissue structures, may be stored, and whether or not to process them may be selected according to a threshold value.

[0040] The three-dimensional CT image data and the three-dimensional shape data generated by the calculation unit 84 are stored in the storage unit 85. In the image processing device 8, an operator (surgeon) can input various commands to the image processing device 8 via the operation unit 82. The display unit 81 can be configured as a touch panel, in which case the display unit 81 has part or all of the functions of the operation unit 82 and also functions as the operation unit 82. Images of various buttons, etc. may be displayed on the display unit 81 so that they can be turned on by operating a pointer with a mouse, etc. In this case, the display unit 81 also functions as the operation unit 82. The three-dimensional CT image data and the three-dimensional shape data generated by the calculation unit 84 are displayed on the display unit 81. The display unit 81 is also an output interface that outputs data reconstructed by the calculation unit 84.

[0041] The image processing device 8 is composed of, for example, a computer or a workstation, and connects a communication interface 86 to the communication interface 63 via a connection cable 83, which is a communication cable, to transmit and receive various data between the image processing device 8 and the X-ray CT imaging device 1. The communication interface 86 is an input interface that accepts input of X-ray projection data from the X-ray CT imaging device 1. However, the communication between the image processing device 8 and the X-ray CT imaging device 1 may be performed wirelessly instead of wired. [Metal artifact image processing method] FIG. 5 is a schematic diagram showing how the lower jaw of the subject M1 is imaged by the X-ray CT imaging device 1. As shown in FIG. 5, the dental arch of the lower jaw of the subject M1 includes a metal dental crown R. The X-ray CT imaging device 1 collects X-ray projection data by rotating the rotating arm 30 around the rotating axis 31 while irradiating the lower jaw of the subject M1 with an X-ray cone beam Bx from the X-ray generator 10 and detecting X-rays with the X-ray detector 20. In FIG. 5, for the sake of simplicity, only the X-ray irradiation axis Xax of the X-ray cone beam Bx is shown. The geometric locus of the X-ray irradiation axis Xax forms an X-ray irradiation rotation plane XSF by rotating. The X-ray irradiation axis Xax may be defined as passing through the center of gravity of the cross section of the xz plane of the X-ray cone beam, for example. In the 3D CT image data (or 3D shape data extracted from the 3D CT image data) obtained by reconstructing the collected X-ray projection data using the image processing device 8, radial metal artifacts are formed centered on the tooth crown R due to a lack of photon information in the tooth crown R and beam hardening.

[0042] Fig. 6 is a diagram showing three-dimensional shape data 3FD of a part of the lower jaw of subject M1 imaged by the X-ray CT imaging device 1. The three-dimensional shape data shown in Fig. 6 is the part of the lower jaw of subject M1 included in the area S shown in Fig. 5. In Fig. 6, in addition to the three-dimensional shape of the teeth S1 and the three-dimensional shape of the jawbone S2 of subject M1, metal artifacts A extending radially from the tooth crown R are also displayed.

[0043] In the dental field, X-ray CT imaging is performed with the maxillofacial region of the subject M1 as the imaging region. In this X-ray CT imaging, a certain regularity can be found between the structure of the subject M1 and the metal artifact A. Specifically, there is a relationship in which the direction in which the metal artifact A extends from the metal region is generally along the X-ray irradiation rotation plane XSF, and the main surfaces of the structures of the subject M1, such as the teeth S1 and the jawbone S2, have an extension generally along the ZS plane, and as a result, there is a regularity in which the normal direction of the surface of the structures of the subject M1, such as the teeth S1 and the jawbone S2, and the normal direction of the surface of the metal artifact A differ by more than a predetermined angle. For example, as shown in Fig. 6, when X-ray CT imaging is performed with an X-ray imaging device in which X-rays irradiated from an X-ray generator 10 proceed in the S direction and are received by an X-ray detector 20, the surface along the X-ray irradiation rotation plane XSF formed by the metal artifact A becomes a surface along the SS plane, and the normal direction of the surface of the structure of the subject M1 is mainly in the XY direction (S direction), while the normal direction of the surface of the metal artifact A is mainly in the Z direction. The main surfaces of the structures of the subject M1, such as the teeth S1 and the jawbone S2, generally extend along the ZS plane, because, as described above, the head MH is positioned so that the axial direction of the body axis in the head and the axial direction of the rotation axis 31 are generally parallel. Therefore, by utilizing this regularity in the normal direction, the metal artifact A can be deleted from the three-dimensional shape data of the subject M1. Regarding whether or not the metal artifact A has a spread in the S direction, for example, if the target voxel and its adjacent voxels in the S direction are both at levels within the same threshold value, it may be determined that there is a spread in the S direction.

[0044] In this embodiment, in addition to an X-ray imaging device in which the mounting positions of the X-ray generator 10 and the X-ray detector 20 are at the same height, for example, the X-ray generator 10 is mounted at a low position and the X-ray detector 20 is mounted at a high position in the relationship of launch irradiation can be handled. The spread of the metal artifact A is along the X-ray irradiation rotation plane XSF of the launch irradiation, but since the spread is generally in the S direction even in the above-mentioned case, it may be processed as the normal direction of the surface of the metal artifact A being mainly the Z direction. Of course, it is also possible to strictly calculate the normal direction of the X-ray irradiation rotation plane XSF. Or, it is also possible to simply process the direction within a range of a predetermined angle from the S direction as the spread direction of the metal artifact A.

[0045] Fig. 7 is a diagram showing a schematic diagram of a method for removing a metal artifact A in the image processing device 8 according to the first embodiment. Fig. 8 is a flowchart showing a method for removing a metal artifact A in the image processing device 8 according to the first embodiment. First, Fig. 7(a) shows a cross-sectional view of a structure of a subject M1 on which a metal artifact A is formed. The structure of the subject M1 includes teeth S1 and a jawbone S2. The cross-sectional view shown in Fig. 7(a) is a cross-sectional view of three-dimensional shape data of the subject M1.

[0046] The image processing device 8 performs the following process to obtain the three-dimensional shape data. First, the calculation unit 84 of the image processing device 8 determines whether or not X-ray projection data has been acquired from the X-ray CT imaging device 1 in step S101 shown in Fig. 8. If X-ray projection data has not been acquired (NO in step S101), the calculation unit 84 returns the process to step S101 and waits for the X-ray projection data to be transmitted from the X-ray CT imaging device 1.

[0047] When X-ray projection data is acquired (YES in step S101), the calculation unit 84 performs calculation processing such as back projection on the acquired X-ray projection data to reconstruct three-dimensional CT image data (step S102). Furthermore, the calculation unit 84 extracts three-dimensional shape data containing only shape information of the subject M1 from the reconstructed three-dimensional CT image data (step S103). The cross-sectional view shown in FIG. 7(a) is a cross-sectional view of the three-dimensional shape data of the subject M1 extracted in step S103.

[0048] In order to remove the metal artifact A included in this three-dimensional shape data, the regularity that the normal direction of the surface of the structure of the subject M1 differs from the normal direction of the surface of the metal artifact A by a predetermined angle or more is utilized. Therefore, the calculation unit 84 calculates the gradient vector of each voxel of the three-dimensional shape data (step S104). Here, the gradient vector is a vector of the direction of change when there is a change in the continuity of the structure between each voxel and an adjacent voxel. For example, the calculation unit 84 may calculate the gradient vector based on whether or not the adjacent voxel includes a structure, or may calculate the gradient vector when observing a change from a voxel including hard tissue to a voxel including soft tissue adjacent to the voxel including hard tissue (both have structures) at the boundary between hard tissue and soft tissue.

[0049] The structures of this embodiment may be classified. For example, biological hard tissues such as teeth and jaw bones are biological hard tissue structures, biological soft tissues such as gums are biological soft tissue structures, artificial structures that supplement or reinforce hard tissues added to biological hard tissue structures as medical treatments such as prosthetic metals are artificial hard tissue structures, and structures that exist as false images in image data such as metal artifact A are false image structures. Both biological hard tissue structures and artificial hard tissue structures may be considered as biological constitutive hard tissue structures that constitute the hard tissues of a living body. Among false image structures, those that have values ​​equal to or greater than those of hard tissue structures may be considered as false image hard tissue structures. A false image hard tissue structure may be considered as a purely computationally derived hard tissue structure that does not exist in real space. Even if it is a purely computationally derived hard tissue structure, a gradient vector can be calculated because it is a hard tissue structure.

[0050] The structures to be processed in this embodiment may be limited. The structures to be processed in this embodiment may be hard tissue structures and structures having values ​​equal to or greater than those of hard tissue structures, and soft tissue structures and structures having values ​​equal to or less than those of soft tissue structures may be excluded.

[0051] 9 is a schematic diagram for explaining a method for deleting voxels of a metal artifact A in the image processing device 8 according to the first embodiment. First, the position of each element is represented by the number of columns and rows in the coordinate system, such as B(column, row). For example, as shown in FIG. 9(d), voxels of 8 columns×10 rows, voxels B(1,1) to B(8,10), are illustrated, and a voxel in the third column and second row is represented as B(3,2). Note that voxels B(6,2) and B(3,2) are voxels in the same row, and voxels B(3,5) and B(3,2) are voxels in the same column.

[0052] FIG. 9(a) illustrates an example of a structure to be processed by the image processing device 8. Specifically, the structure is: Voxel B(1,1)~Voxel B(6,5) [5 rows from column 1 to column 6], Voxel B(1,6)~Voxel B(3,6) [1 row from column 1 to column 3], Voxel B(7,1)~Voxel B(7,4)[7th column, 3 rows] In addition, in Fig. 9, for the sake of simplicity, the coordinates of voxel B are illustrated as two-dimensional, but the coordinates of voxel B are actually three-dimensional.

[0053] The value stored in each voxel is, for example, a value of "1" if there is a structure (tooth S1, jawbone S2) of the subject M1, and a value of "0" if there is no such structure. Of course, the value stored in each voxel may be a continuous value instead of only a value of "1" or "0". In addition, it is possible to make the value stored in each voxel a continuous value by using a Gaussian filter or the like after storing a value of "1" or "0" in each voxel. Since a value is stored in each voxel, the gradient vector can be calculated as the differential value of the function between each voxel. Note that it is desirable to normalize the differential value so that it is easy to perform numerical calculation. For example, as an example of adopting a process that handles differently depending on the presence or absence of a structure, since a structure exists in all adjacent voxels of voxel B(1,1), no gradient occurs and the gradient vector K(1,1) is "0".

[0054] On the other hand, voxel B(7,1) has no structure in the adjacent voxel (8,1), so a gradient occurs in the direction from voxel B(7,1) to voxel B(8,1), resulting in gradient vector K(7,1) shown in Figure 9(a). Also, voxel B(7,4) has no structure in the adjacent voxels (8,4), (8,5), and (7,5), so gradients occur in each of these directions, and the composite vector of these gradients is gradient vector K(7,4) shown in Figure 9(a).

[0055] Here, the direction whose angular difference with the XY directions (S direction) is within a first predetermined range is defined as the first direction. The direction whose angular difference with the direction perpendicular to the axial parallel direction (S direction) is within a first predetermined range may also be considered as the first direction. Moreover, the direction whose angular difference with the Z direction is within a second predetermined range is defined as the second direction. The angular differences within the first and second predetermined ranges may be considered as appropriate. For example, various possible processing methods are possible, such as a process in which the first direction is a direction that has an angular difference of less than 45 degrees from the XY directions (S direction) and the second direction is a direction that has an angular difference of less than 45 degrees from the Z direction, a process in which the first direction is a direction that has an angular difference of less than 45 degrees from the XY directions (S direction) and the second direction is a direction that has an angular difference of less than 45 degrees from the Z direction, a process in which the first direction is a direction that has an angular difference of less than 45 degrees from the XY directions (S direction) and the second direction is a direction that has an angular difference of less than 45 degrees from the Z direction, a process in which the angular difference between each is less than 30 degrees, a process in which the angular difference between each is less than 20 degrees, a process in which the angular difference between each is less than 10 degrees, or a process in which the angular difference between each is 0 degrees.

[0056] The gradient vector of each voxel corresponds to the normal direction of the surface of the structure. Therefore, the gradient vector of each voxel of the structure of the subject M1 is in the first direction (mainly XY and S directions), whereas the gradient vector of each voxel of the metal artifact A is in the second direction (mainly Z direction). Strictly speaking, the gradient vector of each voxel of the metal artifact A is a gradient vector in the normal direction of the X-ray irradiation rotation plane XSF formed by the trajectory of the X-ray irradiation axis Xax, and the second direction includes at least the normal direction of the X-ray irradiation rotation plane XSF. Therefore, as a configuration example in which deletion is performed as a subtraction process, the metal artifact A can be deleted from the 3D shape data by deleting voxels whose gradient vectors are mainly in the second direction (mainly Z direction).

[0057] Returning to FIG. 8, the calculation unit 84 judges whether the angle difference of the direction of the gradient vector of the voxel calculated in step S104 with respect to the axis parallel direction (Z direction) is within a predetermined range (step S105). If the angle difference of the direction of the gradient vector of the voxel with respect to the axis parallel direction is within the predetermined range (YES in step S105), the calculation unit 84 judges that the direction of the gradient vector of the voxel is the second direction and performs subtraction processing on the voxel. As the subtraction processing, for example, the voxel is deleted from the three-dimensional shape data (step S106). The value may be set to a value that is low enough not to cause a problem as a subtraction processing. When there is a level difference in the structure, subtraction processing according to the level difference may be performed. For example, it is assumed that the value of a standard biological hard tissue is "1" and the value of a metal is "1.2". When there is such a level difference, subtraction of "-1" may be performed for a value of "1" and subtraction of "-1.2" for a value of "1.2". The attenuation coefficient for multiplication may be changed for each level. If the value stored in each voxel is a continuous value, where a value of "1" indicates the presence of a structure and a value of "0" indicates the absence of a structure, subtraction processing is performed by adding a value of, for example, "-1" to a specific value stored in a voxel. Of course, in the subtraction processing, the value to be added may be a value such as "-0.8".

[0058] Specifically, a description will be given with reference to FIG. 9(b). Now, let us look at the gradient vector K(7,1) of voxel B(7,1) and the gradient vector K(3,6) of voxel B(3,6). The gradient vector K(7,1) is oriented in the left-right direction (XY direction) in the figure, and is therefore a gradient vector in the first direction. In contrast, the gradient vector K(3,6) is oriented in the up-down direction (Z direction) in the figure, and is therefore a gradient vector in the second direction. As described above, the second direction is a direction in which the angle difference with respect to the axial parallel direction (Z direction) is within a predetermined range. For example, if the predetermined range is 45 degrees, the gradient vector K(7,4) of voxel B(7,4) is oriented at 45 degrees with respect to the axial parallel direction, and is therefore a gradient vector in the second direction. The predetermined range can be set in advance by the operator, and may be set based on the processing results after the processing is repeated multiple times.

[0059] In Figure 9(b), voxels with gradient vectors in the second direction are Voxel(1,6)~Voxel(3,6), Voxel(3,5)~Voxel(6,5), Voxel(6,4)~Voxel(7,4). On the other hand, voxels with gradient vectors in the first direction are Voxel(7,1)~Voxel(7,3). Therefore, as shown in Fig. 9(c), Voxel(1,6)~Voxel(3,6), Voxel(3,5)~Voxel(6,5), Voxel(6,4) through voxel(7,4) will be deleted.

[0060] It has been described that the calculation unit 84 determines that a voxel is to be deleted when the angle difference between the direction of the gradient vector of the voxel and the parallel direction (Z direction) is within a predetermined range, but this determination is merely an example and is not limited thereto. For example, when a certain voxel B(m,n) is focused on, a vector extending from this focused voxel B(m,n) in the second direction is set as a second direction vector VH. A certain vector amount is set as the second direction vector VH, which is set as a calculation target vector VC. The calculation unit 84 may determine that a voxel to be deleted from the three-dimensional shape data is a voxel whose gradient vector has a value of an inner product with the second direction vector VH (specifically, the calculation target vector VC) that is equal to or greater than a predetermined value. Note that, when the value stored in each voxel is a continuous value, the calculation unit 84 may change the value (negative value) to be added depending on the value of the inner product. The calculation unit 84 may linearly change the value to be added, for example, to "-1" if the value of the inner product is "1", and to "0" if the value of the inner product is "cos(π / 4)".

[0061] Returning to Fig. 8, after step S106, the calculation unit 84 returns the process to step S104. That is, in one voxel deletion, the voxels on the surface of the three-dimensional shape data are simply deleted as shown in Fig. 9(c), so the metal artifact A only becomes thinner in the Z direction as shown in Fig. 7(b). Therefore, the calculation unit 84 further calculates gradient vectors for each voxel of the three-dimensional shape data from which the voxels have been deleted, and repeats the process of deleting voxels having gradient vectors in the second direction from the three-dimensional shape data multiple times.

[0062] The process is repeated multiple times, and when voxels having gradient vectors in the second direction are deleted from the three-dimensional shape data and the direction of the gradient vector of each voxel becomes less than a predetermined angle from the first direction (YE in step S105), the calculation unit 84 ends the process. This makes it possible to delete the metal artifact A from the three-dimensional shape data as shown in Fig. 7(c). The number of times the process is repeated may be set by the operator, or may be set while observing the degree of deletion of the metal artifact A.

[0063] As described above, the image processing device 8 according to the first embodiment is an image processing device that reconstructs X-ray projection data acquired by rotating the X-ray detector 20 around the axis of the rotation axis in the X-ray CT imaging device 1 with the maxillofacial region of the subject M1 as the imaging region CA. The image processing device 8 includes a communication interface 86 that receives input of X-ray projection data from the X-ray CT imaging device 1, a calculation unit 84 that reconstructs the X-ray projection data received by the communication interface 86, and a display unit 81 that outputs the data reconstructed by the calculation unit 84. When the X-ray projection data is reconstructed and the three-dimensional CT image data contains a metal artifact A, the calculation unit 84 calculates a vector of the direction of the change as a gradient vector when there is a change in the continuity of the structure between each voxel of the three-dimensional shape data extracted from the three-dimensional CT image data and an adjacent voxel, and performs subtraction processing on voxels having gradient vectors whose angular difference with respect to the axis parallel direction is within a predetermined range, with the direction parallel to the axis of the rotation axis being set as the axis parallel direction. The calculation unit 84 preferably further calculates a gradient vector for each voxel of the three-dimensional shape data after the subtraction process is performed on the voxel, and performs the subtraction process multiple times on voxels having gradient vectors whose angular difference with respect to the axis-parallel direction is within a predetermined range. The subtraction process is, for example, a process of deleting voxels having gradient vectors whose angular difference with respect to the axis-parallel direction is within a predetermined range from the three-dimensional shape data.

[0064] As a result, in the image processing device 8 according to embodiment 1, voxels having gradient vectors in the second direction whose angular difference with respect to the axial parallel direction is within a predetermined range are deleted from the three-dimensional shape data, so that it is possible to delete the metal artifact A from the three-dimensional shape data with high accuracy. (Embodiment 2) As described in the first embodiment, metal artifact A may contain not only parts whose surface normal direction is the Z direction, but also parts with random normal directions as noise. In the second embodiment, a method for removing noise parts in order to remove metal artifact A with greater accuracy will be described. Furthermore, in the second embodiment, a method for interpolating a part where a part of a structure of subject M1 has been removed together with metal artifact A will also be described.

[0065] Fig. 10 is a flowchart showing a method for removing and interpolating a metal artifact A in the image processing device according to the second embodiment. Note that, in the flowchart shown in Fig. 10, steps S101 to S106 are the same as those in the flowchart shown in Fig. 8, so the same step numbers are given to the same steps and detailed descriptions will not be repeated. In the flowchart shown in Fig. 10, after removing voxels whose gradient vectors are in the second direction, the calculation unit 84 finds the variation in the calculated gradient vectors and removes voxels from the three-dimensional shape data based on the found variation (step S107).

[0066] Fig. 11 is a diagram showing three-dimensional shape data including a noise portion. Fig. 11(a) is the three-dimensional shape data before deleting voxels whose gradient vectors are in the second direction, and Fig. 11(b) is the three-dimensional shape data after deleting voxels whose gradient vectors are in the second direction. As shown in Fig. 11(b), simply deleting voxels whose gradient vectors are in the second direction leaves metal artifact A portions having gradient vectors in random directions that cannot be deleted. Therefore, in step S107, the calculation unit 84 deletes voxels from the three-dimensional shape data based on the variation in the gradient vectors, thereby deleting the remaining metal artifact A portions.

[0067] FIG. 12 is a schematic diagram for explaining a method for deleting voxels of a metal artifact A in the image processing device 8 according to the second embodiment. FIG. 12(a) shows an example of a structure to be processed by the image processing device 8. Specifically, the structure is Voxel B(2,2)~Voxel B(2,3) [2 rows of 2nd column], Voxel B(4,2)~Voxel B(6,3) [2 rows from column 4 to column 6], Voxel B(2,6)~Voxel B(7,8)+Voxel B(5,9)~Voxel B(7,9)[3 rows from column 2 to column 7]+[1 row from column 5 to column 7] In other words, three clusters exist in Fig. 12(a). Note that, for the sake of simplicity, the coordinates of voxel B are shown in two dimensions in Fig. 12, but the coordinates of voxel B are actually three-dimensional.

[0068] First, a local average of gradient vectors is calculated for each cluster. The calculation unit 84 uses, for example, a Gaussian filter for gradient vectors as the local average of gradient vectors. Specifically, the cluster (cluster CL1) of voxels B(2,2) to B(2,3) has two gradient vectors K(2,2) and K(2,3) whose directions are opposite to each other. Therefore, when the local average is calculated for the cluster, the gradient vectors K(2,2) and K(2,3) cancel each other out, and the values ​​(vector lengths) of the gradient vectors Ka(2,2) and Ka(2,3) after the local average are smaller as shown in FIG. 12(b). Similarly, the values ​​(vector lengths) of the gradient vectors Ka(4,2) to Ka(6,3) after the local average are smaller than the gradient vectors K(4,2) to K(6,3) before the local average in the cluster (cluster CL2) of voxels B(4,2) to B(6,3) due to the cancellation of the gradient vectors.

[0069] On the other hand, the cluster (cluster CL3) of voxels B(2,6) to B(7,9) has gradient vectors K(2,6) to K(7,9) in various directions. Therefore, when the local average is calculated for the cluster, the gradient vectors K(2,6) to K(7,9) cannot cancel each other out, and the values ​​(vector lengths) of the gradient vectors Ka(2,6) to Ka(7,9) after the local average become large, as shown in FIG. 12(b).

[0070] Next, the calculation unit 84 calculates the difference between the gradient vector Ka after the local averaging and the original gradient vector K. Specifically, in the cluster (cluster CL1) of voxels B(2,2) to B(2,3), the values ​​(vector lengths) of the gradient vectors Ka(2,2) and Ka(2,3) after the local averaging are small, so the differences between the original gradient vectors K(2,2) and K(2,3) are large. FIG. 12(c) illustrates the gradient vectors Kb(2,2) and Kb(2,3). Similarly, in the cluster (cluster CL2) of voxels B(4,2) to B(6,3), the values ​​(vector lengths) of the gradient vectors Ka(4,2) to Ka(6,3) after the local averaging are small, so the differences between the original gradient vectors K(4,2) to K(6,3) are large. FIG. 12(c) illustrates the gradient vectors Kb(4,2) to Kb(6,3).

[0071] On the other hand, in the cluster (cluster CL3) of voxels B(2,6) to B(7,9), the values ​​(vector lengths) of the gradient vectors Ka(2,6) to Ka(7,9) after local averaging are large, so the differences from the original gradient vectors K(2,6) to K(7,9) are small. Figure 12(c) illustrates the differential gradient vectors Kb(2,6) to Kb(7,9). Note that the differential gradient vectors Kb(2,6) to Kb(7,9) have smaller values ​​(vector lengths) than the differential gradient vectors Kb(2,2), Kb(2,3) and the differential gradient vectors Kb(4,2) to Kb(6,3).

[0072] Next, for each cluster, the local average of the value (vector length) of the differential gradient vector Kb is calculated. For example, the calculation unit 84 uses a Gaussian filter on the value of the gradient vector Kb as the local average of the values ​​of the gradient vector Kb. Specifically, the cluster (cluster CL1) of voxels B(2,2) to B(2,3) has large values ​​of the differential gradient vectors Kb(2,2) and Kb(2,3), so the local average of the values ​​of the gradient vector Kb is also large. Similarly, the cluster (cluster CL2) of voxels B(4,2) to B(6,3) has large values ​​of the differential gradient vectors Kb(4,2) to Kb(6,3), so the local average of the values ​​of the gradient vector Kb is also large.

[0073] On the other hand, the cluster (cluster CL3) of voxels B(2,6) to B(7,9) has small values ​​of differential gradient vectors Kb(2,6) to Kb(7,9), so the local average of the gradient vector Kb is also small. A predetermined threshold is then set for the local average of the gradient vector Kb, and for example, the cluster (cluster CL3) of voxels B(2,6) to B(7,9) is not subject to deletion because the local average of the gradient vector Kb is smaller than the predetermined threshold. However, the cluster (cluster CL1) of voxels B(2,2) to B(2,3) and the cluster (cluster CL2) of voxels B(4,2) to B(6,3) are subject to deletion because the local average of the gradient vector Kb is equal to or greater than the predetermined threshold.

[0074] In FIG. 12(d), voxel B(2,2) to voxel B(2,3) of cluster CL1 and voxel B(4,2) to voxel B(6,3) of cluster CL2, which are to be deleted, are hatched differently from the other voxels. This hatching indicates that the voxels are equal to or greater than a predetermined threshold. FIG. 12(e) is a diagram after voxel B(2,2) to voxel B(2,3) of cluster CL1 and voxel B(4,2) to voxel B(6,3) of cluster CL2, which are to be deleted, have been deleted. In this way, the calculation unit 84 deletes voxels from the three-dimensional shape data based on the variation in gradient vectors, thereby making it possible to delete the portion of the metal artifact A having gradient vectors in random directions shown in FIG. 11(b). Note that the method of deleting voxels from the three-dimensional shape data based on the variation in gradient vectors shown in FIG. 12 is just one example, and the calculation unit 84 may delete voxels from the three-dimensional shape data based on the variation in gradient vectors using other methods (e.g., a method using the variance of gradient vectors).

[0075] If the size (volume) of a cluster becomes equal to or smaller than a predetermined size, the cluster may be uniformly subtracted. For example, consider an example in which the cluster is set to be deleted uniformly when it becomes 2 columns x 2 rows or less. Assume that there is a cluster CLA as shown in FIG. 12(f) and it is composed of voxel B(a,a) to voxel B(b,d). If the result of subtracting (deleting) the voxel whose gradient vector gradient is in the second direction becomes a cluster CLB as shown in FIG. 12(g) composed of voxel B(a,b) to voxel B(b,c), the cluster CLB is a cluster of 2 columns x 2 rows or less, and therefore is a target for subtraction, for example, deletion. The predetermined size (volume) may be called the size of the cluster to be subtracted. In addition, when the size (volume) of the cluster to be subtracted is limited to a predetermined size or less, a step (step S107(a)) of judging whether the cluster of the structure shown by the dashed line is a predetermined size or not may be added before step S107 as shown in FIG. 10.

[0076] Returning to FIG. 10, the calculation unit 84 deletes voxels from the three-dimensional shape data based on the variation in the gradient vectors in step S107, and then interpolates the portion of the structure of the subject M1 that was deleted in step S106 (step S108). FIG. 13 is a diagram displaying the three-dimensional shape data from which a portion of the structure of the subject M1 has been deleted. FIG. 13(a) shows the three-dimensional shape data before the metal artifact A is deleted. The calculation unit 84 deletes the metal artifact A from the three-dimensional shape data in step S106, but a portion Sa2 of the structure of the subject M1 is also deleted as shown in FIG. 13(b).

[0077] Fig. 14 is a diagram showing a schematic diagram of a state in which a part of a structure of a subject M1 is removed in an image processing device 8 according to the second embodiment. Fig. 14(a) shows a cross-sectional view of the structure of the subject M1 on which a metal artifact A is formed. The structure of the subject M1 includes teeth S1 and a jawbone S2. The cross-sectional view shown in Fig. 14(a) is a cross-sectional view of the three-dimensional shape data of the subject M1.

[0078] The calculation unit 84 can reduce the thickness of the metal artifact A in the Z direction as shown in FIG. 14(b) by deleting voxels having gradient vectors in the second direction in step S106. However, since there are voxels having gradient vectors in the second direction at the upper and lower ends of the structure of the subject M1 in the figure, they are also deleted in the processing in step S106. Therefore, as shown in FIG. 14(b), voxels of a part Sa1 of the tooth S1 of the subject M1 and a part Sa2 of the jawbone S2 are also deleted. In FIG. 14(c), the metal artifact A is deleted from the three-dimensional shape data of the subject M1, but the voxels of a part Sa1 of the tooth S1 and a part Sa2 of the jawbone S2 are further deleted.

[0079] The calculation unit 84 interpolates the voxels of the part Sa1 of the tooth S1 and the part Sa2 of the jawbone S2 that were deleted in step S108. FIG. 15 is a schematic diagram for explaining a method for interpolating the voxels of the structure of the subject M1 that was deleted in the image processing device 8 according to the second embodiment. First, in FIG. 15(a), voxels B(1,1) to B(8,10) are illustrated, and a structure exists in the hatched voxel B. Note that in FIG. 15, the coordinates of voxel B are explained in two dimensions for the sake of simplicity, but the coordinates of voxel B are actually three dimensional.

[0080] The calculation unit 84 deletes voxels having a gradient vector in the second direction from the structure of the subject M1 in which the metal artifact A is formed as shown in Fig. 15(a), thereby deleting voxels such as voxel B(6,1) in addition to voxel B(1,4). As a result, voxels are deleted from the structure of the subject M1 up to voxel B(5,6) (an example of a portion of the deleted portion) as shown in Fig. 15(b), leaving voxel B(3,3) (an example of a portion of the remaining portion) on the side portion of the structure of the subject M1.

[0081] The calculation unit 84 uses morphological processing to interpolate the deleted structure of the subject M1. Specifically, the calculation unit 84 fills in the deleted voxels as shown in FIG. 15(c) using a closing process in the morphological processing. FIG. 15(c) illustrates three-dimensional shape data in which voxels B(5,3) and voxels B(5,6) are filled in by the closing process. Furthermore, the calculation unit 84 uses a dilation process in the morphological processing to expand the periphery of the three-dimensional shape data in which the voxels are filled as shown in FIG. 15(d). FIG. 15(d) illustrates three-dimensional shape data expanded to voxels B(2,2) and voxels B(8,10) around the three-dimensional shape data in which the voxels are filled in by the dilation process.

[0082] The calculation unit 84 uses Boolean operations to obtain the union (AND) of the 3D shape data expanded by the dilation process and the original 3D shape data (FIG. 15(a)). The result of the union is shown in FIG. 15(e). FIG. 15(e) shows 3D shape data that has been interpolated up to voxel B(2,4) and voxel B(6,1), etc.

[0083] 15(d) has been described as an example of dilation processing in which the periphery of the three-dimensional shape data is expanded by one voxel, but the degree of expansion can be set arbitrarily. However, if the degree of expansion is increased, the part of the deleted metal artifact A is also interpolated, whereas if the degree of expansion is decreased, the deleted structure of the subject M1 cannot be sufficiently interpolated, resulting in a trade-off relationship. Therefore, the degree of expansion in the dilation processing can be changed for each region of the structure of the subject M1 to reduce the trade-off relationship.

[0084] Specifically, Fig. 16 is a diagram showing three-dimensional shape data for explaining the degree of expansion in the dilation process. Fig. 17 is a diagram showing other three-dimensional shape data for explaining the degree of expansion in the dilation process. Fig. 16(a) shows three-dimensional shape data of the structure of the subject M1 before the metal artifact A is deleted. Fig. 16(b) shows three-dimensional shape data when the degree of expansion in the dilation process is reduced and interpolation is performed on the structure of the subject M1 after the metal artifact A is deleted. In Fig. 16(b), a part Sa2 of the deleted structure of the subject M1 remains without being interpolated.

[0085] On the other hand, Fig. 17(a) shows 3D shape data obtained by performing interpolation with a larger degree of expansion in the dilation process on the structure of the subject M1 after deleting the metal artifact A. In Fig. 17(a), a part Sa2 of the deleted structure of the subject M1 has been interpolated, but up to the metal artifact A has been interpolated.

[0086] Metal artifact A is often formed mainly in the tooth S1 region because it is caused by a metal dental crown or the like. Therefore, the trade-off relationship is reduced by decreasing the degree of expansion in the dilation process for the tooth S1 region and increasing the degree of expansion in the dilation process for the jawbone S2 region. FIG. 17(b) shows 3D shape data when interpolating the structure of the subject M1 after deleting the metal artifact A by decreasing the degree of expansion in the dilation process for the tooth S1 region and increasing the degree of expansion in the dilation process for the jawbone S2 region. In FIG. 17(b), a part Sa2 of the deleted structure of the subject M1 can be interpolated, and the metal artifact A is prevented from being interpolated.

[0087] In this way, the calculation unit 84 can interpolate the three-dimensional shape data from which a part of the structure of the subject M1 has been deleted. However, the method of interpolation is not limited to a method that combines the closing process and the dilation process of the morphological process, and may be, for example, a method of replacing the part of the three-dimensional shape data from which a part of the structure of the subject M1 has been deleted with the original three-dimensional shape data.

[0088] Furthermore, the calculation unit 84 may set in advance conditions for excluding voxels from the voxels to be deleted from the three-dimensional shape data when deleting voxels in step S106. That is, the calculation unit 84 excludes the ends of the tooth S1 and the jawbone S2 shown in Fig. 14(a) from the voxels to be deleted in step S106. This prevents the voxels of the tooth S1 and the jawbone S2 from being deleted in the process in step S106, making it possible to eliminate the need for a process of interpolating voxels.

[0089] The above is an example of a process for compensating the value of a hard tissue structure for a voxel that is the target of subtraction process when a subtraction process is performed on a voxel of a hard tissue structure having a gradient vector whose angular difference with respect to the axial parallel direction is within a predetermined range, with the direction parallel to the axial direction of the rotation axis 31 being taken as the axial parallel direction, for the voxel of the three-dimensional shape data. As a result, when subtraction process is performed on a voxel of a bioconstitutive hard tissue structure, this process is a process for compensating the value of a hard tissue structure for a voxel that is the target of subtraction process. When considered from the direction of subtraction process, this process is a process for compensating the value of a hard tissue structure for a voxel in the opposite direction to the direction in which the subtraction process was carried out.

[0090] In addition, voxel B(4,3), voxel B(5,2), voxel B(6,1), etc. are surfaces connected to voxel B(3,3) to voxel B(3,9) and voxel B(7,2) to voxel B(7,9), which are ZS plane surfaces of voxel B(3,3) to voxel (7,9) with small variation, and for voxels of such surfaces, even if subtraction processing is performed once as described above, interpolation processing may be performed, or processing may be performed to exclude them from subtraction targets from the beginning. The ZS plane surface is an example of a surface having continuity in a direction in which the angle difference with respect to the axial parallel direction is within a predetermined range. The determination that the target surface is a surface having continuity in a direction in which the angle difference with respect to the axial parallel direction is within a predetermined range may be made, for example, by the presence of continuity in a direction in which the angle difference with respect to the axial parallel direction is within a predetermined range, and further, by the presence of continuity of hard tissue structures in a direction perpendicular to the axial parallel direction. A threshold value may be set for the number of voxels required to determine continuity, and calculations may be performed so that a certain number of voxels or more is sufficient. This process results in the voxels of bioconstitutive hard tissue structures being excluded from the subtraction process.

[0091] In this way, when a subtraction process is performed on voxels of bioconstitutive hard tissue structures contained in a structure, a process of supplementing voxels with values ​​of hard tissue structures in the direction opposite to the direction in which the subtraction process was performed, or a process of excluding voxels of hard tissue structures contained in a structure from the target of the subtraction process, may be referred to as hard tissue structure preservation processing, and in particular, hard tissue structure preservation processing for bioconstitutive hard tissue structures may be referred to as bioconstitutive hard tissue structure preservation processing.

[0092] In addition, although the flowchart shown in FIG. 10 illustrates a configuration for performing both the processes of step S107 and step S108, the image processing device 8 according to embodiment 2 may be configured to perform either one of the processes of step S107 and step S108.

[0093] In the above embodiment, a method for deleting metal artifact A from three-dimensional shape data has been described. However, the calculation unit 84 can apply the three-dimensional shape data from which the voxels of the metal artifact A have been deleted to the three-dimensional CT image data to obtain three-dimensional CT image data from which the metal artifact A has been deleted. FIG. 18 is a diagram showing three-dimensional CT image data from which the metal artifact A has been deleted. FIG. 18(a) shows three-dimensional CT image data of a part of the dental arch of the subject M1 viewed from the Z direction before the metal artifact A has been deleted. The metal artifact A occurs because the tooth S1 has a metal crown or the like.

[0094] Metal artifact A is deleted from the three-dimensional shape data extracted from the three-dimensional CT image data by the method described in the above embodiment. By applying the three-dimensional shape data from which the metal artifact A has been deleted to the original three-dimensional CT image data, it is possible to delete the metal artifact A located in the same position from the original three-dimensional CT image data. Fig. 18(b) shows the three-dimensional CT image data of a part of the dental arch of subject M1 viewed from the Z direction after the metal artifact A has been deleted.

[0095] The embodiments disclosed herein should be considered to be illustrative and not restrictive in all respects. The scope of the present disclosure is defined by the claims, not the above description, and is intended to include all modifications within the meaning and scope of the claims. [Explanation of symbols]

[0096] 1 imaging device, 3 imaging mechanism, 8 image processing device, 10 X-ray generator, 13 beam shaping mechanism, 20 X-ray detector, 30 swivel arm, 30D X-ray detection unit, 30G X-ray generation unit, 30K swivel arm drive unit, 30R swivel unit, 31 swivel axis, 34 swivel axis moving mechanism, 35 swivel drive mechanism, 40 swivel arm lift unit, 41 upper frame, 42 lower frame, 50 support, 60 control unit, 61, 81 display unit, 62, 82 operation unit, 63, 86 communication interface, 64, 85 memory unit, 65 switch unit, 70 X-ray room, 80 image processing main body unit, 84 calculation unit, 90 notification unit, 100 imaging system, 422 subject holding arm, 423 subject fixing unit.

Claims

1. An image processing device that reconstructs X-ray projection data acquired by rotating an X-ray detector about a rotation axis of an X-ray CT imaging device having a maxillofacial region of a subject as an imaging region, comprising: an input interface that receives input of the X-ray projection data from the X-ray CT imaging device; a processor that reconstructs the X-ray projection data received through the input interface; an output interface for outputting the data reconstructed by the arithmetic processing device; The arithmetic processing device includes: When a metal artifact is included in the three-dimensional shape data extracted from the three-dimensional CT image data reconstructed from the X-ray projection data, a false image structure existing as a false image is included in the structure of the three-dimensional shape data, and when there is a change between each voxel of the three-dimensional shape data and an adjacent voxel as to whether or not the structure is included, or a change between hard tissue and soft tissue in the structure, a vector indicating a direction to a voxel having the change is calculated as a gradient vector; An image processing device that performs subtraction processing on voxels having gradient vectors whose angular difference with respect to a direction parallel to the axial direction of the rotation axis is within a predetermined range, the direction being parallel to the axial direction.

2. 2. The image processing device according to claim 1, wherein the arithmetic processing device further calculates a gradient vector for each voxel of the three-dimensional shape data on which the subtraction process has been performed, and repeats the process of performing the subtraction process multiple times on voxels having gradient vectors whose angular difference with respect to the axial parallel direction is within the predetermined range.

3. 3. The image processing apparatus according to claim 1, wherein the subtraction process is a process of deleting from the three-dimensional shape data voxels having gradient vectors whose angular difference with respect to the axial parallel direction is within the predetermined range.

4. The image processing device according to claim 1 , wherein the arithmetic processing device determines whether or not continuity of the structure changes based on whether or not the structure is included in the three-dimensional shape data.

5. The arithmetic processing device includes: A direction in which an angular difference with respect to a direction perpendicular to the axial parallel direction is within a first predetermined range is defined as a first direction, and a direction in which an angular difference with respect to the axial parallel direction is within a second predetermined range is defined as a second direction, 3 . The image processing device according to claim 1 , wherein the voxels to be subjected to the subtraction process are voxels having a gradient vector whose inner product with the vector in the second direction is equal to or greater than a predetermined value.

6. 3. The image processing device according to claim 1, wherein the arithmetic processing device is capable of setting in advance a condition for excluding voxels from those to be subjected to the subtraction process.

7. 3. The image processing device according to claim 1, wherein the arithmetic processing device determines a variation in the calculated gradient vector, and selects voxels as targets for the subtraction process based on the determined variation.

8. The structure of the subject includes a bioconstitutive hard tissue structure that constitutes hard tissue of a living body, The arithmetic processing device includes: The image processing device according to claim 1 or claim 2, which performs a bioconstitutive hard tissue structure preservation process, which is a process of supplementing voxels of the bioconstitutive hard tissue structure with values ​​of hard tissue structures in a direction opposite to the direction in which the subtraction process was performed when the subtraction process was performed on the voxels of the bioconstitutive hard tissue structure, or a process of excluding voxels of the bioconstitutive hard tissue structure from the target of the subtraction process.

9. The image processing apparatus of claim 8 , wherein the processor is configured to interpolate removed voxels of the bioconstitutive hard tissue structures.

10. The image processing device according to claim 9 , wherein the arithmetic processing device interpolates the deleted voxels by a combination of a closing process and a dilation process of morphological processing.

11. 3. The image processing device according to claim 1, wherein the arithmetic processing device applies the three-dimensional shape data from which the voxels of the metal artifact have been deleted to the three-dimensional CT image data to obtain the three-dimensional CT image data from which the metal artifact has been deleted.

12. an X-ray CT imaging device that acquires the X-ray projection data using an X-ray image sensor that faces an X-ray generator across the subject; 3. An X-ray CT imaging system comprising: the image processing device according to claim 1.

13. An image processing method for performing subtraction processing on metal artifacts from 3D CT image data reconstructed from X-ray projection data acquired by rotating an X-ray detector about a rotation axis in an X-ray CT imaging device having an imaging area of ​​a maxillofacial region of a subject, the method comprising: a step of calculating, when the three-dimensional shape data extracted from the three-dimensional CT image data includes a metal artifact, a false image structure that exists as a false image is included in a structure of the three-dimensional shape data, and when there is a change between each voxel of the three-dimensional shape data and an adjacent voxel as to whether or not the structure is included, or a change between hard tissue and soft tissue in the structure, as a gradient vector, a vector indicating a direction to a voxel having the change; performing a subtraction process on voxels having gradient vectors whose angular difference with respect to a direction parallel to the axial direction of the rotation axis is within a predetermined range, the direction being parallel to the axial direction of the rotation axis.

14. 14. The image processing method according to claim 13, further comprising the step of: calculating a gradient vector for each voxel of the three-dimensional shape data for which the subtraction process has been performed; and repeating the subtraction process a plurality of times for voxels having gradient vectors whose angular difference with respect to the axial parallel direction is within the predetermined range.

15. 15. The image processing method according to claim 13, further comprising the step of: determining a variation in the calculated gradient vector; and deleting voxels from the three-dimensional shape data based on the determined variation.

16. The image processing method according to claim 13 or 14, wherein the structures of the subject include bioconstitutive hard tissue structures that constitute hard tissues of a living organism, and further comprises a step of performing a bioconstitutive hard tissue structure preservation process, which is a process of supplementing voxels of the bioconstitutive hard tissue structures with values ​​of the hard tissue structures in a direction opposite to the direction in which the subtraction process was performed when the subtraction process was performed on the voxels of the bioconstitutive hard tissue structures, or a process of excluding voxels of the bioconstitutive hard tissue structures from the target of the subtraction process.

Images