Method for reconstructing three-dimensional computed tomography image of moving object

Abstract

A method for reconstructing three-dimensional computed tomography image of a moving object is disclosed. It comprises steps of: a) placing an object on a rotational table in rotation which is placed on a carrier in moving, wherein the carrier passes through a projection space between an X-ray point source and an X-ray detector; b) fetching an X-ray image of the object and determining a spatial location and a posture of the object; c) repeating the step b) to fetch N X-ray images and corresponding spatial locations and postures; d) performing filtering operation for the N X-ray images; e) calculating paths of X-rays forming pixels of the X-ray images back projected to a reconstruction space one by one by using the spatial locations and postures fetched from the step c); f) accumulating values of the pixels of the N X-ray images after filtering operation to voxels; and g) outputting the reconstruction space.

Description

METHOD FOR RECONSTRUCTING THREE-DIMENSIONAL COMPUTED TOMOGRAPHY IMAGE OF MOVING OBJECTFIELD OF THE INVENTION

[0001] The present invention relates to a method for reconstructing a three-dimensional computed tomography image. More particularly, the present invention relates to a method for reconstructing a three-dimensional computed tomography image of a moving object.BACKGROUND OF THE INVENTION

[0002] Computed tomography (CT) image reconstruction is a process that uses computers to process original CT data, e.g., X-ray scanned photos, to generate 2D or 3D images that can show internal structure of an object. CT image reconstruction technology has important applications in fields such as medical diagnosis, industrial non-destructive testing and quality control. In terms of quality control, CT image reconstruction is used in manufacturing for dimensional measurements and inspections to ensure products meet design specifications. For example, check the accuracy of automotive parts, plastic molds or tiny components and detect internal damage which cannot be seen directly. Conventionally, these products need to be rotated on a special rotating table placed between an X-ray source and detector to irradiate X-rays in multiple directions and obtain corresponding CT images. In order to save costs during quality inspection, it is best for these products to be moved simultaneously on the production line to pass through the X-ray source and detector. This eliminates the need for additional manpower to move products on and off the rotating table. However, the movement of the rotating tablet on the production line 1P1130098 USincreases variables and difficulty of 3D reconstruction from the original CT images. Existing reconstruction methods cannot be used directly based on this scenario.

[0003] In order to settle the problem of CT image reconstruction mentioned above, an innovative method for reconstructing a three- dimensional computed tomography image of a moving object is provided.SUMMARY OF THE INVENTION

[0004] This paragraph extracts and compiles some features of the present invention; other features will be disclosed in the follow-up paragraphs. It is intended to cover various modifications and similar arrangements included within the spirit and scope of the appended claims.

[0005] According to an aspect of the present invention, a method for reconstructing three-dimensional computed tomography image of a moving object is disclosed. It comprises steps of: a) placing an object on a rotational table in rotation which is placed on a carrier in moving, wherein the carrier passes through a projection space between an X-ray point source and an X-ray detector; b) fetching an X-ray image of the object from the X-ray detector and determining a spatial location and a posture of the object at this moment; c) repeating the step b) in a fixed time manner or a non-fixed time manner to fetch N X-ray images and corresponding spatial locations and postures; d) performing filtering operation column by column for the N X-ray images to enhance contrast and edge sharpness of these X-ray images; e) calculating paths of X-rays forming pixels of the X-ray images back projected to a reconstruction space one by one by using the spatial locations and postures fetched from the step c); f) accumulating values of the pixels of the N X-ray images after filtering operation to voxels along respective paths 2P1130098 USin the reconstruction space by a distribution method; and g) outputting the reconstruction space containing a three-dimensional computed tomography image of the object. N is a positive integer greater than 1.

[0006] According to the present invention, the spatial location may contain spatial coordinates of points on the surface of the object relati ve to a luminous point of the X-ray point source.

[0007] According to the present invention, the posture may be a three- dimensional vector of a rotation axis of the object when rotating.

[0008] According to the present invention, the distribution method may be bilinear interpolation method or bicubic interpolation method.

[0009] According to the present invention, a coordinate origin of the reconstruction space may lie on a line connecting a point on a luminous point of the X-ray point source and a point on a geometric center of a photosensitive surface of the X-ray detector, respectively.

[0010] According to the present invention, the filtering operation may be high-pass filtering operation.

[0011] According to the present invention, a moving path of the rotational table may be a straight line.

[0012] The method according to claim 1, wherein the rotational table rotates at a constant rotation speed.

[0013] Another method for reconstructing three-dimensional computed tomography image of a moving object is also disclosed in the present invention. It comprises steps of: a) placing an object on a rotational table in rotation which is placed on a carrier in moving, wherein the carrier passes through a projection space between an X-ray point source and an X-ray detector; b) fetching an X-ray image of the object from the X-ray detector and determining a spatial location and a posture of the object at this moment;3P1130098 USc) repeating the step b) in a fixed time manner or a non-fixed time manner to fetch N X-ray images and corresponding spatial locations and postures; d) calculating projection values of a reconstruction space on the X-ray detector to obtain M projected images by using the spatial locations and the postures fetched from the step c), wherein each projected image is corresponding to at least one X-ray image; e) calculating image difference values between each projected image and corresponding at least one X-ray image; f) calculating paths of X-rays forming pixels of the X-ray images back projected to the reconstruction space one by one by using the spatial locations and postures fetched from the step c); g) accumulating image difference values obtained from the step e) to voxels along respective paths in the step f) in the reconstruction space by a distribution method; and h) repeating steps d) to g) if a termination condition is not met, otherwise outputting the reconstruction space containing a three-dimensional computed tomography image of the object. M and N are positive integers greater than 1 and M is less than or equal to N.

[0014] According to the present invention, the spatial location may contain spatial coordinates of points on the surface of the object relative to a luminous point of the X-ray point source.

[0015] According to the present invention, the posture may be a three- dimensional vector of a rotation axis of the object when rotating.

[0016] According to the present invention, the distribution method may be bilinear interpolation method or bicubic interpolation method.

[0017] According to the present invention, a coordinate origin of the reconstruction space may lie on a line connecting a point on a luminous point of the X-ray point source and a point on a geometric center of a photosensitive surface of the X-ray detector, respectively.4P1130098 US

[0018] According to the present invention, a moving path of the rotational table may be a straight line.

[0019] According to the present invention, the rotational table may rotate at a constant rotation speed.

[0020] According to the present invention, a distance between values of all voxels of the reconstruction space obtained in the round of calculation and that obtained in a previous round of calculation is computed, and the termination condition is that a ratio of the distance of a current round of calculation to that of the previous round of calculation is less than a given threshold. The given threshold should be less than 1 but greater than 0.

[0021] Compared with the prior arts, the present invention has the advantage of being able to detect rotating and moving objects and reconstruct a three-dimensional computed tomography image for them. This advantage allows the present invention to be widely used in multiple aspects of industrial manufacturing, especially for quality inspection of products on automated assembly lines. By comparing the reconstructed CT image of the inspected product with the reconstructed CT image of the standard product, the quality can be identified without additional manual disassembly of the product. It can reduce significant labor costs.BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Fig. 1 illustrates hardware architecture that an embodiment of the present invention is applied to.

[0023] Fig. 2 illustrates a general motion aspect of an object.

[0024] Fig. 3 is a flow chart of an embodiment of the method for reconstructing three-dimensional computed tomography image of an object in the present invention.5P1130098 US

[0025] Fig. 4 illustrates an X-ray image of the object and a path plane of X-rays through the object at a starting time point.

[0026] Fig. 5 illustrates another X-ray image of the object and another path plane of X-rays through the object at a second time point.

[0027] Fig. 6 shows another path plane calculated from the path plane in Fig. 4 projected to a reconstruction space.

[0028] Fig. 7 shows another path plane calculated from the path plane in Fig. 5 projected to a reconstruction space.

[0029] Fig. 8 shows how a reconstructing three-dimensional computed tomography image represents the structure of the object.

[0030] Fig. 9 is a flow chart of another embodiment of the method for reconstructing three-dimensional computed tomography image of an object in the present invention.DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0031] The present invention will now be described more specifically with reference to the following embodiments.

[0032] See Fig. 1. It illustrates hardware architecture that an embodiment of the present invention is applied to. According to the present invention, a method for reconstructing three-dimensional computed tomography image of an object (hereinafter referred to as the method), The three-dimensional image reconstructed by the method is the surface and internal image of an object 30 moving with rotation. In order to effectively obtain the internal structure information of the object 30, an X-ray source is used as a detection method. The X-ray source is an X-ray point source 10. The X-ray point source 10 emits uniform X-rays and an X-ray detector 20 receive them. The X-ray detector 20 further converts the intensity of X-rays 6P1130098 USreceived by different small areas (detection units) on its photosensitive surface 21 within a sampling time into corresponding voltage values, and combines them into a frame of CT image through a computer (not shown). The computer is also the main body for implementing the method of the present invention. The X-ray point source 10 is different from a general X- ray surface source. The former allows the X-rays received by any two adjacent detection units to have different incident angles, although the difference therebetween is very small, while most of the X-rays from the latter are parallel. Therefore, as shown in Fig. 1, a projection space S where the X-rays pass through is formed between the X-ray point source 10 and the X-ray detector 20. Since the X-rays from the X-ray point source 10 in the present invention are emitted to a specific area in space rather than to the entire adjacent area, the appearance shape of the projection space S is ideally a cone.

[0033] Regarding the object 30, its appearance and internal structure are not limited by the present invention. For convenience of explanation, the object 30 in Fig. 1 is set as a circular hollow tank with a solid cuboid fixed above the interior. There are thin walls encompassing the solid cuboid. Material-wise, as long as the X-rays are not completely blocked, any suitable materials may be used in the object 30. In this embodiment, the object 30 is made of metal, such as aluminum alloy. Fig. 2 illustrates another motion of the object 30 in another embodiment.

[0034] The object 30 is moving while rotating during the process of CT image taking. To achieve this, the object 30 is placed on a set of mobile devices. There are two mobile devices: a rotational table 40 and a carrier 41. The rotational table 40 may be any rotating machine which is able to make a load on it rotating around a fixed rotation axis. The rotational table 407P1130098 USrotates at a constant rotation speed so that the position of any point of the object 30 can be obtained with time. The carrier 41 may be any linear transport tool which carries the rotational table 40 to move along a straight line in space. Namely, a moving path of the rotational table 40 is s straight line for the method. Taking the example of production line quality control, the carrier 41 may be a conveyor belt. In practice, after the object 30 has gone through the projection space S, it will not come back or repeat the same trip again. In a wider range of applications, the carrier 41 may not be limited to this, but needs to maintain a constant linear transport speed. Therefore, the position of any point of the object 30 in moving can be calculated with time. Since the rotation speed and moving speed of the object 30 are known, the position of any point can also be obtained by calculating the change from the beginning to any point in time through the linear transport speed and the rotation speed.

[0035] In order to effectively describe the position of each point of the object 30 in space, a virtual coordinate system is defined in Fig. 1. The virtual coordinate system is a reference indicator and does not exist in real space. It is created for use by the aforementioned computer and has three mutually perpendicular axes, X axis, Y axis and Z axis. And thus, a reconstruction space is also built for the reconstructed three-dimensional CT image of the object 30. According to the present invention, a coordinate origin O of the reconstruction space should lie on a line connecting a point on a luminous point L of the X-ray point source 10 and a point on a geometric center G of the photosensitive surface 21 of the X-ray detector 20, respectively. Namely, the coordinate origin O may lie on any point on the Z axis shown in Fig. 1. In some cases, the distance between the luminous point L and the geometric center G is the shortest distance between the X-ray 8P1130098 USpoint source 10 and the X-ray detector 20. In this embodiment, a direction of the linear movement of the object 30 is parallel to the X axis indicated by a hollow arrow while a direction of the rotation axis of the object 30 is parallel to the Z axis indicated by a dashed arrow. It is obvious that this form of motion (movement and rotation) is one of many in space. For a more general motion aspect of the object 30, see Fig. 2. Locations and directions of the X-ray point source 10 and the X-ray detector 20 keep the same as that in Fig. 1 but the direction of the linear movement and the direction of the rotation axis of the object 30 change. The two directions both have x direction component, y direction component and z direction component and are not parallel to any axis. For the computer, all operations regarding position are essentially performed for three-dimensional space. The coordinates of each point on the object 30 have three coordinate data. Therefore, the method is not limited to specific motions in space.

[0036] As the object 30 moves into the projection space S, some X-rays pass through the object 30 and make a projection on the photosensitive surface 21 of the X-ray detector 20. X-rays on different straight paths will pass through different parts of the object 30. Depending on the portion and material of the object 30 that the X-rays passed through, the energy of the X- rays will be absorbed (decreased) to varying degrees. A projection image P on the photosensitive surface 21 shows the result of the X-rays passing through the object 30 at the same point in time. The projection image P has two areas with different shades of gray. The darker area indicates that the X- rays have passed through the solid cuboid and thin walls. It absorbs more energy, so the displayed image is dimmer. In contrast, the brighter area indicates that X-rays have only passed through thin walls. Less energy is absorbed so the displayed image is slightly brighter. However, the brightness 9P1130098 USof this part is far less than that of the X-ray projection that X-rays have not passed through the object 30 and is shown with a white background around the projection image P.

[0037] For convenience of explanation, a reconstructed three-dimensional computed tomography image R is shown in Fig. 1. The location of the reconstructed three-dimensional CT image R may not be on the path of the object 30 in move and the size of the reconstructed three-dimensional CT image R is a little larger than that of the object 30, but orientation of the reconstructed three-dimensional CT image R keeps the same as the object 30. It is illustrated by dashed lines.

[0038] See Fig. 3. It is a flow chart of an embodiment of the method in the present invention. A first step of the method in this embodiment is placing an object on a rotational table in rotation which is placed on a carrier in moving, wherein the carrier passes through a projection space between an X-ray point source and an X-ray detect (SOI). The purpose of the step SOI is to build up an application scenario as mentioned above. It is not to be elaborated here. Illustration of steps after the step SOI will use the component names and symbols disclosed above.

[0039] A second step of the method is fetching an X-ray image of the object from the X-ray detector and determining a spatial location and a posture of the object at this moment (S02). Assume that at the starting time point tl, the object 30 is illuminated by X-rays for the first time and the X-ray detector obtains a first frame of X-ray image of the object 30. Almost at the same time, the computer calculates locations of important points, the spatial location, that make up the object 30 on the virtual coordinate system. According to the present invention, the spatial location should at least contain spatial coordinates of points on the surface of the object relative to 10P1130098 USthe luminous point L of the X-ray point source 10. Here, the “point” is defined by a set of coordinate data, x-axis value, y-axis value and z-axis value. A point actually represents the space within a certain range of the location, and this range is half the distance between two adjacent points. For example, if the distance between any two adjacent points on any axis is 1, for the point (0,0,0), its spatial range is defined by (0.5, 0.5, 0.5), (0.5, 0.5,-0.5), (0.5, -0.5, 0.5), (0.5, -0.5,-0.5), (-0.5, 0.5,0.5), (-0.5, 0.5, -0.5), (-0.5, -0.5, 0.5), and (-0.5, -0.5, -0.5). Unit volume is 1. In practice, the present invention does not limit the numerical value of the minimum distance between points on the coordinate axis. Attribute values calculated in each spatial range will be stored in association with the coordinates of the corresponding point. This is the smallest unit of three-dimensional space segmentation of digital data used in three-dimensional imaging, scientific data, medical imaging and other fields, voxel. The present invention uses voxels to construct and describe points of the reconstructed three-dimensional CT image R. Although voxels are convenient to use, it is still necessary to further determine the distance between adjacent voxels so that the size of the reconstructed three-dimensional CT image R calculated can be the same as the real size of the object 30. In order to achieve this goal, a relative position of each point in the virtual coordinate system relative to a reference point must be obtained. The distance between the point and the reference point can be obtained by multiplying the coordinate values of the relative position by the unit length. The space enclosed by the relative coordinates of all points on the surface of the reconstructed three-dimensional CT image R is the same size as the object 30. Therefore, the reference point is the luminous point L in the method. On the other hand, since the object 30 also rotates when it moves, because the direction of the rotation axis and the rotation 11P1130098 USspeed of the object 30 remain unchanged, the computer can calculate the rotated positions of all points of the object 30 at different times. By superimposing the moved position, the actual position of the object 30 at any time can be known. Hence, the posture is a three-dimensional vector of a rotation axis of the object 30 when rotating.

[0040] A third step of the method is repeating the step S02 in a fixed time manner or a non-fixed time manner to fetch N X-ray images and corresponding spatial locations and postures (S03). In the method, N is a positive integer greater than 1, may be 2, 3, and more. For a better understanding of this step, please refer to Fig. 4 and Fig. 5. Fig. 4 illustrates an X-ray image XI of the object 30 and a path plane of X-rays through the object 30 at the starting time point tl. Fig. 5 illustrates another X-ray image X2 of the object 30 and another path plane of X-rays through the object 30 at a second time point t2. “Fixed time manner” means the time difference between any two adjacent time points is the same. Namely, the X-ray point source 10 emits X-rays at a fixed frequency to take photos. “Non-fixed time manner” means the time difference between any two adjacent time points may not be the same. The X-ray point source 10 may randomly emit X-rays. Any of the above manners are applicable to the present invention. It can be seen from Figure 4 and Figure 5 that the positions and image sizes of the circular hollow’ tank and the solid cuboid will change in the X-ray images obtained at different time points. There are N X-ray images taken during the time that the object 30 moves through the projection space S. Corresponding spatial locations and postures at all N time points tl to tN can be calculated and recorded by the computer. Thus, the step S02 actually be carried out N times. Theoretically, the larger the N value, the closer the reconstructed image will be to reality. Here, in this embodiment, N is set to be 18.12P1130098 US

[0041] A fourth step of the method is performing filtering operation column by column for the N X-ray images to enhance contrast and edge sharpness of these X-ray images (S04). After the X-rays pass through the object 30, the edge portions of the X-ray images will be blurred due to the movement of the object 30, In order to solve this problem, the N X-ray images fetched must be processed to enhance contrast and edge sharpness. A better way to achieve this is using filtering operation. Preferably, the filtering operation is high-pass filtering operation. During the filtering operation, it is done column by column. As shown in Fig. 4, there are j columns of pixels in the X-ray image XI. The filtering operation is conducted from the first column to the j th column.

[0042] A fifth step of the method is calculating paths of X-rays forming pixels of the X-ray images back projected to a reconstruction space one by one by using the spatial locations and postures fetched from the step S03 (S05). Because of the spatial locations and postures of the object 30 at all time points, each point on the reconstructed three-dimensional CT image in the virtual coordinate system can be found by back projection at corresponding time point. However, back projection requires knowing the path of each X-ray that generates corresponding pixel. This step describes how to make it. The reconstruction space has been defined in the preceding text. All points on the surface of the reconstructed three-dimensional CT image are known. The spatial locations and postures also be fetched in the step S03. The paths of X-rays forming pixels of the X-ray images can be obtained. A pixel P on the X-ray image XI is shown in Fig. 4. The path of X-rays forming a pixel P is calculated and shown by dashed line segments. Due to straightness of light beams, for one column of pixels, a collection of the paths of X-rays may form a plane or an arc surface. A gray path plane is 13P1130098 USa collection of paths of X-rays forming a column of pixels and cuts through the solid cuboid of the object 30. Another path plane calculated from the path plane in Fig. 4 projected to the reconstruction space is shown in Fig. 6. The path plane in Fig. 6 passes through corresponding voxels in the reconstructed three-dimensional CT image R. Similar situation happens in Fig. 5. Another gray plane cutting through the solid cuboid of the object 30 in different direction represents another collection of paths of X-rays at time points tl. A corresponding path plane calculated from the path plane in Fig.5 projected to the reconstruction space is shown in Fig. 7. The path plane in Fig. 7 passes through corresponding voxels in the reconstructed three- dimensional CT image R, A difference between the path plane in Fig. 4 and that in Fig. 5 is that the area the former cuts the solid cuboid is smaller than that of the latter. Since the X-ray image XI and X-ray image X2 are obtained from different spatial locations and postures of the object 30, the numbers of columns are different in this embodiment.

[0043] A sixth step of the method is accumulating values of the pixels of the N X-ray images after filtering operation to voxels along respective paths in the reconstruction space by a distribution method (S06). This step is to conduct back projection with average distribution of attribute values. See Fig. 8. It shows how the reconstructing three-dimensional computed tomography image represents the structure of the object 30. The object 30 can be seen as a collection of small parts and each small part is represented by a corresponding voxel in the reconstructed three-dimensional CT image R. A portion of the reconstructed three-dimensional CT image R is enlarged and the image is shown in a round frame. There are many cubes of voxels stacked to form the surface portion of the reconstructed three-dimensional CT image R. Values of the pixels are used to describe how dark or how 14P1130098 USbright each pixel is shown after the X-rays fall on the detection unit on the photosensitive surface 21. For example, 0 is for the brightest and 255 for the darkest. The back projection is to add all values of the pixels to the voxels the paths of X-rays pass. One voxel may accumulate a lot of values of the pixels since it is on the intersection of many paths of X-rays. A final accumulated value of the voxel from the pixels of the N X-ray images represents how much energy of the X-rays has been absorbed by the corresponding small part of the object 30. The darkness of voxels shown in the round frame in Fig. 8 is converted from the final accumulated values. Since the paths of X-rays forming pixels of the X-ray images may not exactly pass through all centers of the voxels, how to distribute the value when one path goes between two adjacent voxels should be considered. According to this embodiment, the distribution method is the solution. According to the present invention, the distribution method may be, but not limited to bilinear interpolation method and bicubic interpolation method.

[0044] A last step of the method in this embodiment is outputting the reconstruction space containing a three-dimensional computed tomography image of the object (S07). This is to output the reconstructed three-dimensional CT image R closest to the object 30.

[0045] The present invention also provide another method for reconstructing three-dimensional computed tomography image of a moving object. The method is described in the following embodiment.

[0046] See Fig. 9. It is a flow chart of another embodiment of a method for reconstructing three-dimensional computed tomography image of an object (hereinafter referred to as this method) in the present invention. A first step of this method is placing an object on a rotational table in rotation which is placed on a carrier in moving, wherein the carrier passes through a 15P1130098 USprojection space between an X-ray point source and an X-ray detector (S11). Like the step SOI in the previous embodiment, the step S11 is to build up an application scenario as mentioned above. A second step of this method is fetching an X-ray image of the object from the X-ray detector and determining a spatial location and a posture of the object at this moment (S12). Obviously, the step S12 is the same as the step S02. Details of the step S12 is not repeated here. It should be noticed that some restrictions remain unchanged: the spatial location should contain spatial coordinates of points on the surface of the object relative to a luminous point of the X-ray point source; the posture is defined as a three-dimensional vector of a rotation axis of the object when rotating; a coordinate origin of the reconstruction space should lie on a line connecting a point on a luminous point of the X-ray point source and a point on a geometric center of a photosensitive surface of the X-ray detector respectively; a moving path of the rotational table is a straight line; and the rotational table should rotate at a constant rotation speed.

[0047] A third step of this method is repeating the step S12 in a fixed time manner or a non-fixed time manner to fetch N X-ray images and corresponding spatial locations and postures (S13). This step is the same as the step S03. The purpose of this step is to fetch N X-ray images that are created by X-rays passing through the object 30 in different paths. N is also any positive greater than 1.

[0048] A fourth step of this method is calculating projection values of a reconstruction space on the X-ray detector to obtain M projected images by using the spatial locations and the postures fetched from the step S13, wherein each projected image is corresponding to at least one X-ray image (S14). The purpose of this step is to use the N sets of spatial locations and 16P1130098 USpostures obtained in the previous step to create multiple spatial virtual images in the reconstruction space. The spatial positions of these virtual images are the same as the spatial positions of the object 30 at some time points in the virtual coordinate system. These virtual images will also form the projection image P on the X-ray detector 20. Using the calculated projection values is equivalent to establishing background values as the initial basis for subsequent iterative calculations. These projection values can also be set to 0, and iterative calculations can be performed on this basis. The number of projected images can be the same as the number of X-ray images, so that each frame of projected image can be applied to the corresponding frame of X-ray image. The number of projected images can also be less than the number of X-ray images, so that two or more X-ray images can be applied to one projected image. Therefore, N is a positive integer greater than 1 and M is less than or equal to N.

[0049] A fifth step of this method is calculating image difference values between each projected image and corresponding at least one X-ray image (S15). This step is to find out a usable value which has no effect of the background value for each pixel.

[0050] A sixth step of this method is calculating paths of X-rays forming pixels of the X-ray images back projected to the reconstruction space one by one by using the spatial locations and postures fetched from the step S13 (S16). This step is the same as the step S05. Its purpose is to find out paths for back projection.

[0051] A seventh step of this method is accumulating image difference values obtained from the step S15 to voxels along respective paths in the step S16 in the reconstruction space by a distribution method (S17). This step is similar to the step S06 but has an essential difference. Background 17P1130098 USvalues are subtracted from the values for back projection. It is the same that bilinear interpolation method or bicubic interpolation method can be chosen as the distribution method.

[0052] A last step of this method is repeating steps SI 4 to S17 if a termination condition is not met, otherwise outputting the reconstruction space containing a three-dimensional computed tomography image of the object (S18). It is clear that this step conducts an iterative operation to calculate different three-dimensional computed tomography images until one of them can be deemed as the three-dimensional CT image R to output. The termination condition is a judging point. According to the present invention, for each round of calculation, a distance between values of all voxels of the reconstruction space obtained in the round of calculation and that obtained in a previous round of calculation is computed, and the termination condition is that a ratio of the distance of a current round of calculation to that of the previous round of calculation is less than a given threshold. The given threshold should be less than 1 but greater than 0. Like the definition in mathematics, the “distance” means the length of two points in a high dimensional space. In practice, the distance may be Euclidean distance, between two points in Euclidean space. A formulaic description of the distance is as followswhere Dt is the distance of a t-th round of calculation, Vt refers to values of all voxels of the reconstruction space obtained in the t-th round of calculation, Vt-i refers to values of all voxels of the reconstruction space obtained in the previous round ((t-l)-th round) of the t-th round of18P1130098 UScalculation,means a value of an i-th voxels of the reconstruction space obtained in the t-th round of calculation, andmeans a value of an i-th voxel of the reconstruction space obtained in the (t-1)-th round of calculation There are N voxels in the reconstruction space. The given threshold can be described as follows0 < — < 1Dt_iwhere Dt-1 is the distance of the (t-1 )-th round of calculation.

[0053] Since the projection values of the projected images in step S14 are related to their positions, the start setting of calculation can start from any value or an extreme value, so the result of each iteration operation, the sum of values of all voxels of the reconstruction space, may increase or decrease. Based on the different M projected images used for calculation, the total difference between adjacent rounds will also change. This step is to find out the total difference that starts to converge and is less than the preset threshold value after convergence. After finding it, the three-dimensional computed tomography image in the computer at this time can be defined as the final acceptable reconstructed three-dimensional CT image R, which is output together with the overall reconstruction space.

[0001] While the invention has been described in terms of what is presently considered to be the most practical and preferred embodiments, it is to be understood that the invention needs not be limited to the disclosed embodiments. On the contrary, it is intended to cover various modifications and similar arrangements included within the spirit and scope of the appended claims, which are to be accorded with the broadest interpretation so as to encompass all such modifications and similar structures.19P1130098 US

Claims

WHAT IS CLAIMED IS:

1. A method for reconstructing three-dimensional computed tomography image of a moving object, comprising steps of:a) placing an object on a rotational table in rotation which is placed on a carrier in moving, wherein the carrier passes through a projection space between an X-ray point source and an X-ray detector;b) fetching an X-ray image of the object from the X-ray detector and determining a spatial location and a posture of the object at this moment; c) repeating the step b) in a fixed time manner or a non-fixed time manner to fetch N X-ray images and corresponding spatial locations and postures; d) performing filtering operation column by column for the N X-ray images to enhance contrast and edge sharpness of these X-ray images;e) calculating paths of X-rays forming pixels of the X-ray images back projected to a reconstruction space one by one by using the spatial locations and postures fetched from the step c);f) accumulating values of the pixels of the N X-ray images after filtering operation to voxels along respective paths in the reconstruction space by a distribution method; andg) outputting the reconstruction space containing a three-dimensional computed tomography image of the object,wherein N is a positive integer greater than 1.

2. The method according to claim 1, wherein the spatial location contains spatial coordinates of points on the surface of the object relative to a luminous point of the X-ray point source.

3. The method according to claim 1, wherein the posture is a three- dimensional vector of a rotation axis of the object when rotating.

4. The method according to claim 1, wherein the distribution method is 20P1130098 USbilinear interpolation method or bicubic interpolation method.

5. The method according to claim 1, wherein a coordinate origin of the reconstruction space lies on a line connecting a point on a luminous point of the X-ray point source and a point on a geometric center of a photosensitive surface of the X-ray detector respectively.

6. The method according to claim 1, wherein the filtering operation is high-pass filtering operation.

7. The method according to claim 1, wherein a moving path of the rotational table is a straight line.

8. The method according to claim 1, wherein the rotational table rotates at a constant rotation speed.

9. A method for reconstructing three-dimensional computed tomography image of a moving object, comprising steps of:a) placing an object on a rotational table in rotation which is placed on a carrier in moving, wherein the carrier passes through a projection space between an X-ray point source and an X-ray detector;b) fetching an X-ray image of the object from the X-ray detector and determining a spatial location and a posture of the object at this moment; c) repeating the step b) in a fixed time manner or a non-fixed time manner to fetch N X-ray images and corresponding spatial locations and postures; d) calculating projection values of a reconstruction space on the X-ray detector to obtain M projected images by using the spatial locations and the postures fetched from the step c), wherein each projected image is corresponding to at least one X-ray image;e) calculating image difference values between each projected image and corresponding at least one X-ray image;f) calculating paths of X-rays forming pixels of the X-ray images back 21P1130098 USprojected to the reconstruction space one by one by using the spatial locations and postures fetched from the step c);g) accumulating image difference values obtained from the step e) to voxels along respective paths in the step f) in the reconstruction space by a distribution method; andh) repeating steps d) to g) if a termination condition is not met, otherwise outputting the reconstruction space containing a three-dimensional computed tomography image of the object,wherein M and N are positive integers greater than 1 and M is less than or equal to N.

10. The method according to claim 9, wherein the spatial location contains spatial coordinates of points on the surface of the object relative to a luminous point of the X-ray point source.

11. The method according to claim 9, wherein the posture is a three- dimensional vector of a rotation axis of the object when rotating.

12. The method according to claim 9, wherein the distribution method is bilinear interpolation method or bicubic interpolation method.

13. The method according to claim 9, wherein a coordinate origin of the reconstruction space lies on a line connecting a point on a luminous point of the X-ray point source and a point on a geometric center of a photosensitive surface of the X-ray detector, respectively.

14. The method according to claim 9, wherein a moving path of the rotational table is a straight line.

15. The method according to claim 9, wherein the rotational table rotates at a constant rotation speed.

16. The method according to claim 9, wherein for each round of calculation, a distance between values of all voxels of the reconstruction space obtained 22P1130098 USin the round of calculation and that obtained in a previous round of calculation is computed, and the termination condition is that a ratio of the distance of a current round of calculation to that of the previous round of calculation is less than a given threshold.

17. The method according to claim 16, wherein the given threshold is less than 1 but greater than 0.?P1130098 US

Images