JP5821695B2 - Image processing apparatus and radiation tomography apparatus including the same - Google Patents

Description

  The configuration of the present invention includes an image processing apparatus that acquires a tomographic image of a subject based on a series of fluoroscopic images captured while a radiation source and an FPD are synchronously moved in opposite directions, and a radiation tomography apparatus including the same About.

  A medical institution is provided with a radiation imaging apparatus 51 that acquires a tomographic image of a subject. In such a radiation imaging apparatus 51, a series of fluoroscopic images are continuously shot while a radiation source 53 for irradiating radiation and an FPD 54 for detecting radiation move synchronously, and the series of fluoroscopic images are overlapped to obtain a tomographic image. Some are configured to acquire images (see FIG. 22). In such a radiation imaging apparatus 51, during the imaging of a series of fluoroscopic images, the radiation source 53 and the FPD 54 move so as to approach each other in the body axis direction of the subject, and the radiation source 53 and the FPD 54 in the body axis direction. After the positions coincide with each other, the radiation source 53 and the FPD 54 move away from each other in the body axis direction. Such a radiation imaging apparatus is described in, for example, Patent Document 1 (see Patent Document 1).

  An operation when the radiation imaging apparatus 51 captures a tomographic image as described above will be described. First, the radiation source 53 emits radiation intermittently while moving. That is, every time one irradiation is completed, the radiation source 53 moves in the direction of the body axis of the subject and irradiates the radiation again. In this way, 74 perspective images are acquired, and these are superimposed. The completed image is a tomographic image in which a tomographic image obtained by cutting the subject with a certain cut surface is reflected.

International Publication Number WO2011 / 086604

However, the conventional configuration as described above has the following problems.
That is, in the radiation imaging apparatus 51 having the conventional configuration, the image is disturbed at the end of the tomographic image.

  Since the fluoroscopic images are taken while moving the X-ray tube 3 and the FPD 54 with respect to the subject, the positions at which the subject is reflected differ among the fluoroscopic images. That is, an image of the subject that appears at the end of one fluoroscopic image does not appear in another fluoroscopic image. However, such a situation is not considered when generating the tomographic image D. That is, the tomographic image D is generated on the assumption that all the subject images necessary for reconstruction are included in the fluoroscopic image.

  Then, at the end of the generated tomographic image D, the image of the subject that is originally necessary for generating the tomographic image D is insufficient. Such a lack of image appears as a false image in the tomographic image D. More specifically, as shown in FIG. 23, the image is significantly disturbed at both ends of the tomographic image D in the direction in which the FPD 54 and the radiation source 53 move (the vertical direction of the tomographic image D).

  Therefore, in the conventional configuration, the shortage of the tomographic image is improved by compensating for the lack of the image and generating the tomographic image. The specific method is to make up for the shortage of the image by repeatedly pasting the end pattern of the fluoroscopic image and stretching the fluoroscopic image. However, this method is not sufficient for extrapolating images. That is, in the above-described method, the image of the subject appearing in the image is not taken into consideration, and the end of the image is simply pasted, so the joined subject images are discontinuous as shown in FIG. It becomes. As a result, the edge of the image may become unnatural due to extrapolation processing. Due to such circumstances, the visibility of the tomographic image may be lower than that in the conventional method when the tomographic image is generated without extrapolation.

  The present invention has been made in view of such circumstances, and an object of the present invention is to obtain a tomographic image excellent in visibility in a radiation tomography apparatus that obtains a tomographic image from a plurality of fluoroscopic images. A processing apparatus and a radiation tomography apparatus are provided.

The present invention has the following configuration in order to solve the above-described problems.
That is, the image processing apparatus according to the present invention, an image replication of the region linear structures are crowded-through to the side portions of the images are located side edge portion so as to be arranged in the extending direction of the linear structure An extrapolation unit that generates an extrapolated image by adding to the side of the object, a radiation source that irradiates the subject with radiation, and a radiation detection unit that detects the radiation that has passed through the subject move synchronously and in opposite directions. And a tomographic image generating means for generating a tomographic image by superimposing extrapolated images generated based on a series of images continuously taken, and the extrapolating means includes a radiation source and a radiation detecting means in the image. Image processing is performed at both ends in the moving direction.

[Operation and Effect] The image processing apparatus according to the present invention, the area where the linear structures are crowded-through to the side portions of the images are located side edge portion so as to be arranged in the extending direction of the linear structure An extrapolated image is generated by adding the duplicate to the side of the image. In addition, this adding operation is performed at both ends of the moving direction of the radiation source and the radiation detection means in the image. In this way, both ends in the moving direction of the image are added so as to extend the linear structure, and an extrapolated image is generated. Since the portion outside the side portion of the image is a portion that could not be photographed, it is impossible to actually know how the fluoroscopic image of this portion was. However, it is considered that the image of the subject in the part deviated from the image continues for a while with a pattern extending at least the image of the side part of the image.

  Therefore, if the pattern of the side part is appropriately shifted in the direction in which the side part extends so as to extend the linear structure reflected on the side part of the image, it can be taken on the image. The missing part can be extrapolated more naturally. That is, the end of the generated extrapolated image is such that the subject image is continuously extended. If a tomographic image is generated using an extrapolated image subjected to such extrapolation, a tomographic image with excellent visibility can be generated.

  Further, in the above-described image processing apparatus, the extrapolation means duplicates a strip-like pixel array located on the side portion extending in the orthogonal direction orthogonal to the moving direction in the image, and adds it to the side portion. It is more desirable to operate according to.

  [Operation / Effect] The above-described configuration shows a specific configuration of the extrapolation means of the present invention. In other words, the extrapolation means can be more reliably generated by duplicating the striped pixel array located on the side extending in the orthogonal direction perpendicular to the moving direction. is there.

  Further, in the above-described image processing apparatus, it is more preferable that the extrapolation means performs a smoothing process in the orthogonal direction to the strip-like pixel arrangement when adding the pixels.

  [Operation / Effect] The above-described configuration shows a specific configuration of the extrapolation means of the present invention. That is, the extrapolation means can acquire a clearer tomographic image by performing smoothing processing in the orthogonal direction to the strip-like pixel arrangement when adding pixels. It is expected that the fluoroscopic image of the subject in the outer portion of the image will be substantially the same as the subject image reflected on the side of the image. However, in actuality, the fluoroscopic image of the subject outside the image and the subject image reflected on the side of the image are not exactly the same. Therefore, when pixels are added as they are, a pattern different from the actual one is added to the image, and the tomographic image becomes unnatural. According to the above-described configuration, since the pixels are added after smoothing once, the unnaturalness of the tomographic image is not noticeable and the visibility of the tomographic image is improved.

  In the above-described image processing apparatus, it is more desirable that the extrapolation means gradually increase the blurring strength of the smoothing process applied to the added pixel array as the addition operation of the plurality of pixel arrays is repeated.

  [Operation / Effect] The above-described configuration shows a specific configuration of the extrapolation means of the present invention. Prediction of a subject image that is not shown in the image becomes more difficult as the portion where the subject image is predicted is farther from the side of the image. For example, in a region outside the image, it is expected that the fluoroscopic image of the subject adjacent to the side portion will be substantially the same as the subject image reflected on the side portion. However, in the outer area of the image, the image of the subject should be farther away from the subject image reflected on the side portion as the distance from the side portion increases. According to the present invention, as the adding operation of a plurality of pixel arrays is repeated, the blurring strength of the smoothing process applied to the added pixel array is gradually increased. In this way, it is possible to prevent a situation in which a pattern that is far from the original subject image is gradually added as the adding operation is repeated.

  In the radiation tomography apparatus equipped with the above-described image processing apparatus, a radiation source that irradiates a subject with radiation, a radiation detection means that detects radiation transmitted from the subject, a radiation source and a radiation detection means A moving unit that moves the subject synchronously in the moving direction and in opposite directions; a movement control unit that controls the moving unit; and an image generating unit that generates an image based on the output of the radiation detecting unit. More desirable.

  [Operation / Effect] The configuration described above shows the configuration of a radiation tomography apparatus equipped with the image processing apparatus of the present invention. With the radiation tomography apparatus according to the present invention, a tomographic image with improved visibility at both ends of the tomographic image can be acquired.

  The image processing apparatus according to the present invention first generates a gradient vector map indicating the extending direction of the linear structure reflected in the image. Then, a copy of the region located on the side is copied so that the linear structures reflected on the side of the image with reference to the gradient vector map are arranged in the extending direction of the linear structure. An extrapolated image is generated by adding to the section. Then, if the patterns of the side portions are joined together while shifting the pattern of the side portions so as to extend the linear structure reflected in the side portions of the image, the subject image is more naturally supplemented. If a tomographic image is generated after this operation, a tomographic image with excellent visibility can be generated.

1 is a functional block diagram illustrating a configuration of an image processing apparatus according to a first embodiment. 1 is a functional block diagram illustrating a configuration of an X-ray tomography apparatus according to Embodiment 1. FIG. FIG. 3 is a schematic diagram for explaining the principle of tomography according to the first embodiment. In the X-ray tomography apparatus which concerns on Example 1, it is a schematic diagram explaining a test subject appearing in an original image. In the X-ray tomography apparatus which concerns on Example 1, it is a schematic diagram explaining a test subject appearing in an original image. In the X-ray tomography apparatus which concerns on Example 1, it is a schematic diagram explaining a test subject appearing in an original image. In the X-ray tomography apparatus which concerns on Example 1, it is a schematic diagram explaining a test subject appearing in an original image. In the X-ray tomography apparatus which concerns on Example 1, it is a schematic diagram explaining a test subject appearing in an original image. It is a schematic diagram explaining operation | movement of the map production | generation part which concerns on Example 1. FIG. It is a schematic diagram explaining operation | movement of the map production | generation part which concerns on Example 1. FIG. It is a schematic diagram explaining operation | movement of the map production | generation part which concerns on Example 1. FIG. It is a schematic diagram explaining operation | movement of the map production | generation part which concerns on Example 1. FIG. FIG. 6 is a schematic diagram for explaining the operation of the extrapolation unit according to the first embodiment. It is a schematic diagram explaining a mode that the total vector which concerns on Example 1 is calculated. FIG. 6 is a schematic diagram for explaining the operation of the extrapolation unit according to the first embodiment. FIG. 6 is a schematic diagram for explaining the operation of the extrapolation unit according to the first embodiment. FIG. 6 is a schematic diagram for explaining the operation of the extrapolation unit according to the first embodiment. FIG. 6 is a schematic diagram for explaining the operation of the extrapolation unit according to the first embodiment. FIG. 6 is a schematic diagram for explaining the operation of the extrapolation unit according to the first embodiment. FIG. 6 is a schematic diagram for explaining the operation of the extrapolation unit according to the first embodiment. It is a schematic diagram explaining the operation | movement which concerns on 1 modification of this invention. It is a schematic diagram explaining the structure of the conventional X-ray tomography apparatus. It is a schematic diagram explaining the structure of the conventional X-ray tomography apparatus. It is a schematic diagram explaining the problem of the conventional X-ray tomography apparatus.

  As shown in FIG. 1, the image processing apparatus according to the first embodiment inputs an image (referred to as an original image P0) acquired by taking an image of the subject with X-rays. The tomographic image D when cut is output. The original image P0 corresponds to the image of the present invention.

<Overall configuration of image processing apparatus>
As shown in FIG. 1, the image processing apparatus 12 according to the first embodiment calculates a gradient vector indicating the extending direction of the linear structure reflected in the original image P0, arranges the gradient vector, and arranges the gradient vector map (simply simply A map generation unit 12a that generates a map m) and a region located on the side so as to extend a linear structure reflected on the side of the original image P0 with reference to the map m. An extrapolation unit 12b that generates an extrapolated image P1 by adding to the side part of the original image P0 while shifting in the direction in which the side part extends, an X-ray tube 3 that irradiates the subject with X-rays, and an X after the subject is transmitted A tomogram that generates a tomographic image D by superimposing an extrapolated image P1 generated based on a series of original images P0 that are continuously shot while the FPD 4 that detects a line is moved in a direction opposite to each other synchronously with the moving direction. And an image generation unit 12c. The operations of these units will be described later. The X-ray tube 3 corresponds to the radiation source of the present invention, and the FPD 4 corresponds to the radiation detection means of the present invention. The map generator 12a corresponds to the map generator of the present invention, and the extrapolator 12b corresponds to the extrapolator of the present invention. The tomographic image generation unit 12c corresponds to the tomographic image generation means of the present invention, and the map m corresponds to the gradient vector map of the present invention.

<Overall configuration of X-ray tomography apparatus>
FIG. 2 is a functional block diagram illustrating the configuration of the radiation tomography apparatus according to the first embodiment. As illustrated in FIG. 2, the X-ray imaging apparatus 1 according to the first embodiment includes a top plate 2 on which a subject M that is a target of X-ray tomography is placed, and an upper portion of the top plate 2 (1 of the top plate 2). X-ray tube 3 that irradiates the subject M provided on the surface side) with a cone-shaped X-ray beam, and the lower part of the top plate 2 (on the other side of the top plate) and transmits the subject M. X-ray in a state where the sheet-like flat panel type X-ray detector (hereinafter abbreviated as FPD) 4 that detects the X-ray and the central axis of the cone-shaped X-ray beam and the center point of the FPD 4 always coincide. A synchronous movement mechanism 7 that synchronously moves each of the tube 3 and the FPD 4 in opposite directions across the region of interest of the subject M, a synchronous movement control unit 8 that controls the synchronous movement mechanism 7, and an X-ray that detects the X-rays of the FPD 4 And an X-ray grid 5 for absorbing scattered X-rays provided so as to cover the detection surface. In this way, the top plate 2 is disposed at a position sandwiched between the X-ray tube 3 and the FPD 4. The X-ray tube 3 corresponds to the radiation source of the present invention, and the FPD 4 corresponds to the radiation detection means of the present invention. The synchronous movement control unit 8 corresponds to the movement control means of the present invention, and the synchronous movement mechanism 7 corresponds to the movement means of the present invention. The synchronous movement mechanism 7 corresponds to the movement means of the present invention, and the synchronous movement control unit 8 corresponds to the movement control means of the present invention.

  The X-ray tube 3 is configured to repeatedly irradiate the subject M with a cone-shaped and pulsed X-ray beam according to the control of the X-ray tube control unit 6. The X-ray tube 3 is provided with a collimator that collimates the X-ray beam into a cone shape that is a pyramid. The X-ray tube 3 and the FPD 4 generate imaging systems 3 and 4 that capture X-ray fluoroscopic images.

  The synchronous movement mechanism 7 is configured to move the X-ray tube 3 and the FPD 4 in synchronization. The synchronous movement mechanism 7 linearly moves the X-ray tube 3 along a linear trajectory (longitudinal direction of the top 2) parallel to the body axis direction A of the subject M according to the control of the synchronous movement control unit 8. The moving direction of the X-ray tube 3 and the FPD 4 coincides with the longitudinal direction of the top 2. Moreover, during the examination, the cone-shaped X-ray beam irradiated by the X-ray tube 3 is always irradiated toward the region of interest of the subject M. The X-ray irradiation angle is determined by the X-ray tube 3. Is changed from, for example, an initial angle of −20 ° to a final angle of 20 °. Such an X-ray irradiation angle change is performed by the X-ray tube tilting mechanism 9.

  Further, the synchronous movement mechanism 7 synchronizes with the linear movement of the X-ray tube 3 described above, and causes the FPD 4 provided at the lower part of the top 2 to move in the body axis direction A (the longitudinal direction of the top 2) of the subject M. Move straight ahead. The moving direction is opposite to the moving direction of the X-ray tube 3. That is, the cone-shaped X-ray beam whose irradiation source position and irradiation direction change as the X-ray tube 3 moves is always received by the entire surface of the X-ray detection surface of the FPD 4. In this way, in one inspection, the FPD 4 acquires, for example, 74 original images P0 while moving in synchronization with the X-ray tube 3 in the opposite directions. Specifically, the imaging systems 3 and 4 are opposed to the position indicated by the one-dot chain line shown in FIG. 2 via the position indicated by the broken line, with the position of the solid line as the initial position. That is, a plurality of fluoroscopic images are taken while changing the positions of the X-ray tube 3 and the FPD 4. By the way, since the cone-shaped X-ray beam is always received by the entire surface of the X-ray detection surface of the FPD 4, the central axis of the cone-shaped X-ray beam during imaging always coincides with the center point of the FPD 4. During imaging, the center of the FPD 4 moves straight, but this movement is in the direction opposite to the movement of the X-ray tube 3. That is, the X-ray tube 3 and the FPD 4 are moved in the body axis direction A synchronously and in directions opposite to each other.

  A collimator provided in the X-ray imaging apparatus 1 will be described. The collimator is attached to the X-ray tube 3, and collimates X-rays emitted from the X-ray tube 3 to form a quadrangular pyramid (cone-shaped) X-ray beam 19.

  Next, the principle of obtaining a tomographic image of the X-ray imaging apparatus 1 according to the first embodiment will be described. FIG. 3 is a diagram for explaining a tomographic image acquisition method of the X-ray imaging apparatus according to the first embodiment. For example, the reference cutting surface MA parallel to the top plate 2 (horizontal with respect to the vertical direction) will be described. As shown in FIG. 3, the points P and Q positioned on the reference cutting surface MA are always X-rays of the FPD 4. A series of FPDs 4 are synchronously moved in the opposite direction of the X-ray tube 3 in accordance with the irradiation direction of the cone-shaped X-ray beam 19 by the X-ray tube 3 so as to be projected onto the fixed points p and q of the detection surface. Is generated by the image generation unit 11. In the series of original images P0, the projected image of the subject M is reflected while changing the position. Then, if this series of original images P0 (exactly, an extrapolated image P1 subjected to extrapolation processing described later) is superimposed by the image processing unit (image processing device) 12, an image positioned on the reference cut surface MA. (For example, fixed points p and q) are accumulated and imaged as an X-ray tomographic image. On the other hand, the point I that is not located on the reference cut surface MA is reflected as a point i in a series of subject images while changing the projection position on the FPD 4. Unlike the fixed points p and q, such a point i is blurred without forming an image when the image processing unit 12 superimposes the fluoroscopic images. In this way, by superimposing a series of original images P0 (extrapolated image P1), an X-ray tomographic image in which only an image located on the reference cutting plane MA of the subject M is reflected is obtained. In this way, when the X-ray fluoroscopic images are simply superimposed, an X-ray tomographic image at the reference cut surface MA is obtained. The position in the vertical direction of the reference cutting section MA is the reference cutting position of the present invention. The image generation unit 11 corresponds to an image generation unit of the present invention.

  Further, by changing the setting of the image processing unit 12, a similar tomographic image can be obtained at any cutting position horizontal to the reference cut surface MA. During shooting, the projection position of the point i moves in the FPD 4, but this moving speed increases as the separation distance between the point I before projection and the reference cut surface MA increases. By using this to superimpose a series of acquired subject images while shifting in the body axis direction A at a predetermined pitch, an X-ray tomographic image at a cutting position parallel to the reference cut surface MA is obtained. The image processing unit 12 superimposes a series of such subject images. A method for acquiring a tomographic image in this way is called filter back projection.

  When comparing the original images P0 captured while moving the imaging systems 3 and 4, the image of the subject M is reflected in the original image P0 while shifting in the moving direction of the imaging systems 3 and 4. This is because the projection shape of the subject M projected on the FPD 4 changes as the positions of the imaging systems 3 and 4 change in the movement direction. On the other hand, the shadow of the leaf reflected in the series of original images P0 does not follow the displacement of the image of the subject M, and appears in the same manner in any original image P0.

  If the tomographic image D is to be acquired using the original image P0 as it is, the tomographic image of the subject M is obtained particularly at both ends in the longitudinal direction of the tomographic image D (direction corresponding to the moving direction of the imaging systems 3 and 4). The image is disturbed. The principle that the tomographic image is disturbed will be described. Assume that a tomographic image of the subject M is generated at the cut surface MB including the point p inside the subject M shown in FIG. Referring to FIG. 4, it can be seen that the position of the point p reflected in the FPD 4 gradually moves to the end of the FPD 4 due to the movement of the imaging systems 3 and 4. That is, when the imaging systems 3 and 4 are at the position of the solid line, the point p is positioned at the center of the X-ray beam, and when the imaging systems 3 and 4 are at the position of the chain line, the point p is the X-ray beam. It is designed to be located at the end.

  FIG. 5 schematically shows how the point p is reflected in the 74 original images P0. The original image P01 in the figure shows the first image taken, and represents an image acquired when the imaging systems 3 and 4 in FIG. 4 are at the positions of the solid lines, and the original image P037 in the figure. Indicates the 37th photographed image and represents an image obtained when the imaging systems 3 and 4 in FIG. Also, the original image P074 in the figure shows the 74th image taken, and represents an image acquired when the imaging systems 3 and 4 in FIG. 4 are at the position of the alternate long and short dash line. As can be seen by comparing each image, when the original image P0 is compared in the order of acquisition, the point p moves toward the lower end of the original image P0. The point p is reflected in all the original images P0 while moving.

  The description using FIG. 5 represents an ideal original image P0. However, in reality, the point p is not reflected in any of the original images P0 as shown in FIG. That is, a part of the image of the subject M that should appear in the peripheral portion of each original image P0 is out of the field of view while the X-ray tube 3 and the FPD 4 are moved. In the case of the point p, as the original image P0 is continuously shot, the point p moves to the lower end side of the original image P0 and is finally out of the field of view. Thus, the point p is not reflected in all the original images P0.

  If an attempt is made to form the point p on the tomographic image D using the original image P0 as shown in FIG. 6, the added image becomes insufficient, and the pixel value of the point p becomes extremely dark. Such disturbance of the tomographic image D is particularly noticeable at the upper end and the lower end when the direction corresponding to the moving direction of the imaging systems 3 and 4 in the tomographic image D is the vertical direction. This is because the image of the subject that is reflected at the end of the tomographic image D is likely to be out of the field of view as the imaging systems 3 and 4 move.

  FIG. 7 shows a comparison of the positions at which the subject M is captured between the first image and the last image of the original image P0 obtained by continuous shooting. In the drawing, the position of the X-ray tube 3 in the first imaging is represented by a solid line, and the position of the X-ray tube 3 in the last imaging is represented by a broken line. Then, the subject image on the reference cut surface MA indicated by the arrow in FIG. 7 is reflected in the last imaged original image P0, although it is reflected in the first imaged original image P0. Not. Such a shortage of the subject image causes disturbance of the tomographic image D. Here, of the end portions of the X-ray beam emitted from the X-ray tube 3 when the X-ray tube 3 is at the continuous shooting start position, the reference cutting of the end portion on the side opposite to the moving direction of the X-ray tube 3 is performed. A position on the surface MA is set as a start side reference position k1.

  FIG. 8 also shows a comparison of the positions at which the subject M appears between the original image P0 photographed first and the original image P0 photographed last. In the drawing, the position of the X-ray tube 3 in the first imaging is represented by a solid line, and the position of the X-ray tube 3 in the last imaging is represented by a broken line. Then, the subject image on the reference cut surface MA indicated by the arrow in FIG. 8 is reflected in the original image P0 captured first, although it is reflected in the last captured original image P0. Not. Such a lack of the subject image also causes disturbance of the tomographic image D. Here, out of the end portions of the X-ray beam emitted from the X-ray tube 3 when the X-ray tube 3 is at the continuous shooting end position, the end portion in the moving direction of the X-ray tube 3 is on the reference cutting plane MA. The position is set as an end side reference position k2.

  Therefore, according to the configuration of the first embodiment, extrapolation processing is performed on the original image P0 in a region indicated by hatching in FIG. The hatched portion in FIG. 7 is a region projected by the X-ray on the plane to which the detection surface of the FPD 4 belongs in the range of the arrow in FIG. Also in FIG. 8, extrapolation processing is performed for the hatched portion in the drawing. 8 is a region projected by X-rays on the plane to which the detection surface of the FPD 4 belongs in the range of the arrow in FIG.

  The image processing unit 12 according to the first embodiment is provided for the purpose of generating a tomographic image D in which image distortion is suppressed. That is, a series of original images P0 generated by the image generation unit 11 is sent to the image processing unit 12, and is subjected to various image processing by the image processing unit 12 and converted into a tomographic image D. Hereinafter, operations of the map generation unit 12a, the extrapolation unit 12b, and the tomographic image generation unit 12c constituting the image processing unit 12 will be described.

<Operation of map generator>
The map generation unit 12a performs gradient analysis on each of the original images P0, and calculates gradient vectors corresponding to the pixels constituting the original image P0. Then, the map m is generated by two-dimensionally arranging each gradient vector following the corresponding pixel. Therefore, the map m is generated for each of the original images P0.

  FIG. 9 shows a general gradient vector calculation method. FIG. 9 shows how the gradient vector is obtained for the pixel a belonging to the original image P0. In order to obtain the gradient vector for the pixel a, first, a difference in pixel value between the pixel a and each of the surrounding pixels surrounding the pixel a is obtained. The peripheral pixels surrounding the pixel a are adjacent to the left side, the right side, the upper side, the lower side, the upper right diagonal side, the upper left diagonal side, the lower right diagonal side, and the lower left diagonal side as viewed from the pixel a, as shown in FIG. There are 8 to do. For each of the eight peripheral pixels, a pixel value difference is obtained from the pixel a. Then, vectors are generated for the eight peripheral pixels, with the pixel value as the vector length and the direction passing through the pixel a and the peripheral pixels as the vector direction. Then, as shown in FIG. 9, eight vectors extending in the direction of the left side, the right side, the upper side, the lower side, the upper right side, the upper left side, the lower right side, and the lower left side are obtained from the pixel a. become. Then, these eight vectors are added to obtain one vector. This obtained vector is the gradient vector for the pixel a.

  The gradient vector represents the direction in which the image reflected in the original image P0 extends. That is, assume that there is an image as shown on the left side of FIG. In this image, an image extending in the vertical direction is reflected in the center of the image, and the pixel value is darker than the surrounding area. When gradient analysis is performed on such an image and the obtained gradient vectors are arranged following the pixel positions, a vector map as shown on the right side of FIG. 10 is obtained. That is, a vector map is obtained in which gradient vectors extending in the horizontal direction are arranged in the vertical direction at positions indicated by shading.

  The gradient vector represents the direction in which the difference in pixel values between a certain pixel and its adjacent pixels is significant and the magnitude of the difference in pixel values. Therefore, the following can be said from the gradient vector. That is, it can be said that an image extending in a direction orthogonal to the gradient vector exists in the original image, and the position where the image appears in the original image coincides with the position of the gradient vector in the vector map. The gradient vectors arranged in the vector map in this way represent the extending direction and the position on the image of the streak structure reflected in the original image. The length of the gradient vector represents how clearly the streak-like structure is reflected.

  The actual map generation unit 12a performs gradient analysis on the end portion of the original image P0. Therefore, as shown in FIG. 11, the map generation unit 12a obtains the gradient vector of the pixel a using the five peripheral pixels surrounding the pixel a at the end of the original image P0. Then, the map generation unit 12a obtains a gradient vector for one column of pixel arrays located at the end of the original image P0, and further, another column of pixels located at the end opposite to the side where the gradient vector is obtained. A gradient vector is also obtained for the array. Then, the map generation unit 12a generates a map m by arranging each of the obtained gradient vectors following the arrangement of pixels in the original image P0. FIG. 12 shows an outline of the map m. As shown in FIG. 10, on the map m, gradient vectors corresponding to the pixels are mapped for each of the pixels located at both ends of the original image P0.

  Both ends of the original image P0 for which the map generation unit 12a obtains the gradient vector coincide with both ends of the original image P0 in the moving direction of the X-ray tube 3 and the FPD 4.

<Operation of extrapolation unit>
The map m and the corresponding original image P0 are sent to the extrapolation unit 12b. The extrapolation unit 12b pays attention to one side of the map m where the gradient vector is obtained, and knows how the linear structure reflected in the original image P0 extends at the end of the original image P0. . As illustrated in FIG. 13, the extrapolation unit 12 b adds up the gradient vectors arranged on one side of the map m to generate a single total vector. Further, the extrapolation unit 12b generates a single total vector by adding the gradient vectors for the other side of the map m for which the gradient vector is obtained.

  The meaning of the total vector will be described. First, as shown on the left side of FIG. 14, it is assumed that a structure on the line is reflected at the end of the original image P0. When the map m is generated for the original image P0, the map m obtained at this time is as shown on the right side of FIG. That is, a plurality of short gradient vectors facing various directions and a small number of long gradient vectors facing one direction are obtained. The reason for this will be described. In the original image P0, similar pixels are arranged in a portion where the structure is not reflected, so that the length of the gradient vector is shortened. Since noise is superimposed on this portion, the pixel value variation is irregular, and the direction of the gradient vector varies. On the other hand, in the map m, the gradient vector at a position where the structure of the original image P0 is clearly reflected is located in a direction perpendicular to the extending direction of the structure. In addition, since this structure is clearly reflected in the original image P0, the pixel values of the pixels belonging to the structure in the original image P0 are greatly different from the pixel values of the surrounding pixels. Therefore, the length of this gradient vector is longer than the surrounding gradient vectors.

  The right side of FIG. 14 also represents a total vector obtained by adding these gradient vectors. According to this, short vectors that are directed in various directions are canceled out, and a long gradient vector representing the direction in which the structure extends is strongly expressed in the total vector. That is, the total vector represents the direction in which the linear structure extends in the original image P0 (more precisely, the direction orthogonal thereto).

  The left side of FIG. 15 is an example of an original image P0 that is actually captured, and includes an image of a knee joint. Consider a pixel array R1 located on one side of the original image P0. In the pixel array R1, pixels are arranged in one column. The right side of FIG. 15 shows a map m corresponding to the original image P0 superimposed on the original image P0. As can be seen from the right side of FIG. 15, in the pixel array R1, a long gradient vector appears at the contour of the femur. Note that a short gradient vector is omitted on the right side of FIG. The extrapolation unit 12b adds these gradient vectors to obtain a total vector. Then, the extrapolation unit 12b recognizes the direction orthogonal to the total vector (the direction indicated by the thick arrow on the right side of FIG. 15) as the extending direction of the linear structure reflected in the end of the original image P0.

  FIG. 16 shows how the extrapolation unit 12b extends the original image P0 by extrapolation. The extrapolation unit 12b duplicates the elongated region R1a related to one side for which the total vector is calculated, and extends the original image P0 by joining the duplicated image to one side of the original image P0 while appropriately shifting the duplicate in the extending direction of the elongated region R1a. To do. This shift amount is determined so that the replicas are stacked in a direction orthogonal to the total vector. For example, when the total vector is orthogonal to the direction in which the elongated region R1a extends, the extrapolation unit 12b recognizes this direction as the direction in which the linear structure reflected in the elongated region R1a extends. Then, the extrapolation unit 12b adds the copy as it is to the original image P0 without shifting in the direction in which the elongated region R1a extends.

  Further, when the total vector is oriented in the diagonally right direction as shown in the right side of FIG. 15, the extrapolation portion 12b extends in a linear structure that is reflected in the elongated region R1a in the diagonally left direction orthogonal to this direction. Recognize. Then, the extrapolation part 12b adds to the original image P0 while shifting the copy so that the linear structure extends in this direction (see FIG. 16). That is, the extrapolation unit 12b has the subject image reflected in the end portion of the original image P0 and the subject image reflected in the duplicate to be added in the diagonally left direction (the orthogonal direction of the total vector). The position of the replica is adjusted with respect to the direction in which the elongated region R1a extends so as to be shifted. In this way, the original image P0 is extrapolated while the linear structure of the subject reflected in the elongated region R1a is extended in a continuous state without interruption. As described above, the duplicates are added while being shifted from the original image P0, so that the linear structures reflected in the elongated region R1a of the original image P0 are arranged in the extending direction while maintaining continuity.

  The left side of FIG. 17 shows how one side of the original image P0 is extended. At this time, the elongated regions R1a are successively arranged so as to be further joined to the elongated regions R1a joined together. In this way, the extrapolation of the original image P0 is performed while extending the linear structure reflected in the original image P0. As a result, the subject image of the original image P0 is continuous and extrapolated more naturally without interruption, rather than simply joining the elongated regions R1a without shifting.

  The extent to which the original image P0 is extended by extrapolation will be described. When the copy of the elongated region R1a is added, the original image P0 becomes longer in the vertical direction (moving direction of the imaging systems 3 and 4) by the width in the short direction in the elongated region R1a. Since the duplication of the copy is repeated, the original image P0 extends by the width in the short direction of the elongated region R1a each time the duplication is added.

  The number of times of duplication of the elongated region R1a will be described. The number of times of addition is determined based on the image overlay operation when the tomographic image D is generated. That is, until the subject image reflected in one end of the 74 original images P0 is extrapolated to all the other 73 original images P0 (full extrapolation state). The extrapolation unit 12b continues the arrangement of the elongated regions R1a of the original images P0. Thereafter, the extrapolation unit 12b continues the extrapolation operation until all the 74 original images P0 are in the extrapolation state.

  For example, when 74 original images P0 are continuously shot, it is assumed that the field of view is moved one pixel at a time in the original image P0. At this time, an area of 1 pixel width arranged on the upper side of the original image P0 photographed first does not appear in the other 73 original images P0. This is because the field of view has moved during continuous shooting. The extrapolation unit 12b generates an extrapolated image P1 by repeating the extrapolation operation on the 74 original images P0 in order to compensate for the loss of this region. That is, the extrapolation unit 12b continues the extrapolation operation until an area corresponding to this area is extrapolated to the other 73 original images P0.

  For example, the second acquired original image P0 is extrapolated only once on the upper side. Thereafter, when the extrapolation operation is performed in the order of acquisition, the number of times the extrapolation is repeated on the original image P0 increases one by one, and the last 74th acquired original image P0 is 73 on the upper side. Extrapolations are made.

  The same applies to the lower side of the original image P0. That is, the area of 1 pixel width arranged below the last photographed original image P0 is not reflected in the other 73 original images P0. The extrapolation unit 12b continues the extrapolation operation until an area corresponding to this area is extrapolated to the other 73 original images P0. For example, the original image P0 acquired first is extrapolated 73 times on the lower side. Similarly, the 73rd original image P0 is extrapolated only once on the lower side.

  The full extrapolation state in this example will be specifically described. The first extrapolated state of the original image P0 photographed is a state where 73 extrapolations have been completed on the lower side, and the second extrapolated state of the original image P0 acquired second is once on the upper side. This is a state where the extrapolation is performed 72 times on the lower side. In addition, the entire extrapolation state of the original image P0 acquired 73rd is a state in which the extrapolation is performed 72 times on the upper side and once on the lower side, and the entire extrapolation of the last captured original image P0 is performed. The state is a state where 73 extrapolations have been completed on the upper side. By the way, when the extrapolation unit 12b performs an extrapolation operation once, the extrapolation is performed by 1 pixel width in the vertical direction, and the image extends by 1 pixel in the vertical direction. This means that the original image P0 is stretched by 73 pixels in the vertical direction (moving direction) and the extrapolated image P1 is generated.

  Generally speaking, the extrapolation operation is performed m-1 times in total, and accordingly, the original image P0 is extended by n × (m−1) to generate the extrapolated image P1. . In this case, m is the number of times of continuous shooting, and n is a movement amount indicating how much the field of view to be moved on the original image P0 per shooting in the continuous shooting of the original image P0.

  The range of extrapolation can also be described using the start side reference position k1 and the end side reference position k2 described above with reference to FIGS. That is, the extrapolation unit 12b acquires the start side reference position k1 and the end side reference position k2 on the standard cutting plane MA based on the positions of the X-ray tube 3 and FPD4 and the movement mode of the X-ray tube 3 and FPD4. . Similarly, the extrapolation unit 12b calculates where the both ends (the upper end and the lower end in FIG. 15) of the original image P0 in the moving direction of the X-ray tube 3 are located on the reference cutting plane MA. And extrapolation part 12b repeats extrapolation operation from these calculation results. That is, the extrapolation unit 12b extrapolates the original image P0 up to the projection position of the start side reference position k1 for the end portion (for example, the upper end in FIG. 15) in the direction opposite to the moving direction of the X-ray tube 3 in the original image P0. To do. Similarly, the extrapolation unit 12b extrapolates the original image P0 to the projection position of the end-side reference position k2 for the end portion (for example, the lower end in FIG. 15) in the movement direction of the X-ray tube 3 in the original image P0.

  Further, if the elongated regions R1a are simply joined together, the elongated regions R1a protrude in a direction orthogonal to the direction in which the image is extended as shown on the right side of FIG. In order to prevent this from happening, the extrapolation part 12b trims the part that protrudes from the width of the original image P0. That is, the original image P0 is shaped so as to be rectangular after being extended.

  The elongated region R1a may be the same region as the pixel array R1 used for searching the linear structure. The elongated region R1a may be a pixel array in which pixels are arranged in a row (in a direction orthogonal to the moving direction), or the elongated region R1a may have a thickness in the vertical direction (moving direction). .

<Extrapolation of missing parts at image vertices>
Next, an operation in which the extrapolation unit 12b compensates for an image shortage will be described. The portion indicated by shading at the apex portion of the image on the left side of FIG. 17 is not extrapolated by the above-described arrangement operation. This is because this portion does not have the original elongated region R1a. Therefore, the extrapolation unit 12b arranges the elongated regions R1a in the movement direction, and then extrapolates the portion indicated by the hatching using the elongated regions R1a again.

  FIG. 18 shows a state in which the vertex of the image is extrapolated. In this description, as an example, a state of extrapolating a shortage of an image with respect to the elongated region R1a1 on the left side of FIG. 18 is shown. This elongated region R1a1 is one of the replicas of the elongated region R1a arranged by the extrapolation operation of the extrapolation portion 12b. The extrapolation portion 12b adds a region R1b, which is a region that is deficient in the elongated region R1a1, to the elongated region R1a1 in a direction perpendicular to the moving direction. This region R1b belongs to the elongated region R1a. The right side of FIG. 18 shows a state in which a copy of the region R1b is arranged adjacent to the elongated region R1a1. The extrapolation unit 12b compensates for the pixel shortage that occurs at the vertex of the image by repeating the same operation, and generates a rectangular image.

<About the side where the extrapolation unit performs extrapolation on the original image>
Next, the sides of the original image P0 where the extrapolation unit 12b performs the extrapolation operation will be described. The extrapolation unit 12b performs an extrapolation operation on two sides that are parallel to each other among the four sides of the original image P0. These two sides are sides extending in a direction orthogonal to the moving direction of the X-ray tube 3 and the FPD 4 indicated by arrows in FIG. In other words, the extrapolation unit 12b reproduces the striped pixel array located on the side part extending in the orthogonal direction that is the direction orthogonal to the moving direction in the original image P0, and adds it to the side part. Operate. That is, the extrapolation unit 12b is not limited to the one side part on the pixel array R1 side illustrated in FIG. 15 in the original image P0, but the side part on the pixel array R2 side located on the side opposite to the pixel array R1. Does the same. Pixels are arranged in one column in the pixel array R2. In this way, the extrapolation unit 12b performs image processing on both ends in the moving direction of the imaging systems 3 and 4 in the original image P0.

<About the smoothing process performed by the extrapolation unit>
The extrapolation portion 12b does not simply duplicate and arrange the elongated region Ra1. The extrapolation part 12b performs an addition operation after smoothing the elongated region Ra1 in the direction in which the region extends (orthogonal direction orthogonal to the moving direction). Moreover, the smoothing process becomes stronger as the addition of the elongated region Ra1 is repeated. Therefore, when the replica of the elongated region Ra1 added to the extrapolated portion 12b is viewed in order from the center of the image to the end portion, the blur gradually becomes intense. That is, as the extrapolation unit 12b repeats the adding operation of the plurality of elongated regions Ra1, the extrapolation unit 12b gradually increases the blurring strength of the smoothing process applied to the added elongated region Ra1.

  The reason why such a blurring operation is added will be described. There is no way to know exactly what the subject image looks like outside the original image P0. However, at least a portion of the outer region adjacent to the end portion of the original image P0 should have a pattern similar to that of the end portion of the original image P0. In other words, the farther from the end of the original image P0 in the outer region, the less likely the subject image will be.

  Even if the elongated region R1a is simply arranged without blurring, there is no problem in the portion adjacent to the end portion of the original image P0. However, after that, when the elongated regions R1a are successively arranged, the pattern of the image reflected in the elongated region R1a gradually differs from the actual subject image. This is because the outer portion of the image of the subject image does not always have a pattern similar to the elongated region R1a. That is, the extrapolated image P1 generated by the extrapolation unit 12b has a pattern that does not resemble the original subject image at the extreme end where the extrapolation process has been performed. When the tomographic image D is generated using such an extrapolated image P1, a false image appears in the tomographic image D. This false image appears strongly as it goes to the end of the tomographic image D. At the extreme end of the tomographic image D, the discrepancy between the actual subject image and the extrapolated image P1 is severe, and this discrepancy appears in the tomographic image D as a false image.

  In order to solve such a problem, the extrapolation unit 12b adds a copy of the elongated region R1a to the original image P0 while performing a smoothing process. Specifically, the extrapolation process is performed while increasing the strength of the smoothing process for each addition. By doing so, the replica of the elongated region R1a that is added as it deviates from the original image P0 becomes gradually unclear. As a result, the image that appears in the elongated region R1a that causes the false image gradually becomes unclear, and the false image that appears in the tomographic image D also loses clarity. By doing so, it is possible to prevent a false image from appearing in the tomographic image D.

<Extrapolation processing for images with multiple linear structures>
The operation of the extrapolation unit 12b described with reference to FIGS. 16 and 17 is described for the sake of convenience of description with respect to the extension direction of the femoral contour in the region on the map m corresponding to the pixel array R1. Yes. In the original image P0, various linear structures are reflected in addition to the femoral contour. The extending direction of the linear structure does not always coincide with the extending direction of the contour portion of the femur. Therefore, if the extrapolation part 12b adds an image based on the extending direction of the femoral contour part, the other linear structures reflected in the original image P0 are not necessarily extended.

  FIG. 19 illustrates this problem. When attention is paid to one of the plurality of linear structures shown in the original image on the left side of FIG. 19 and the images are added in the direction indicated by the arrow, the noticeable linear structure is obtained as shown on the right side of FIG. Is extended. However, other linear structures appearing in the image are discontinuous and discontinuous.

  Therefore, the extrapolation unit 12b divides the elongated region R1a into a plurality of parts and performs extrapolation processing. That is, the direction in which the image is added to the original image P0 is changed depending on the portion of the image. FIG. 20 shows how the divided elongated regions R1a are individually joined together. In FIG. 20, the elongated region R1a extending in the horizontal direction is divided at a predetermined interval in the horizontal direction. This divided fragment is represented by a symbol R1p. The extrapolation part 12b calculates | requires a total vector independently about each of fragment | piece R1p, and calculates the extending | stretching direction of a linear structure. Then, the extrapolation unit 12b extends the image by joining the fragments R1p in the vertical direction of FIG. 17 based on the image stretching direction calculated for each fragment R1p. That is, the images are joined together while shifting the copy of the fragment R1p in different directions depending on each fragment R1p as shown by the arrows in FIG.

  At this time, a gap may be formed between the joined replicas. In such a case, the extrapolation part 12b extrapolates this gap by duplicating and fitting a part of the fragment R1p adjacent to the fragment R1p into this gap. At this time, a part of the fragment R1p to be duplicated has the same size and shape as the gap.

  The extrapolation unit 12b performs the above-described operation on the pixel array R1 in the original image P0, and then performs the same operation on the pixel array R2 in the original image P0 to extend both sides of the original image P0 (see FIG. 15). The extrapolation unit 12b performs the same extrapolation operation on all of the continuously shot original images P0 to generate a series of extrapolated images P1.

<Operation of the tomographic image generator>
The extrapolated image P1 is sent to the tomographic image generation unit 12c. The tomographic image generation unit 12c superimposes a series of extrapolated images P1 while shifting in the moving direction of the X-ray tube 3 and the FPD 4, and generates a tomographic image D in which a tomographic image of the subject at a predetermined cutting plane is reflected.

  Further, the X-ray imaging apparatus 1 according to the first embodiment includes a main control unit 25 that comprehensively controls the control units 6 and 8 and a display unit 27 that displays a tomographic image. The main control unit 25 is constituted by a CPU, and realizes the control units 6 and 8 and the later-described units 11 and 12 by executing various programs. The storage unit 23 stores all parameters necessary for control of the X-ray imaging apparatus 1 and image processing. The storage unit 23 temporarily stores an intermediate image generated during image processing.

<Operation of X-ray imaging apparatus>
Next, the operation of the X-ray imaging apparatus 1 will be described. In order to acquire the tomographic image D using the X-ray imaging apparatus 1 according to the first embodiment, first, the subject M is placed on the top 2. When the surgeon gives an instruction to acquire the original image P0 through the console 26, the synchronous movement control unit 8 moves the X-ray tube 3 and the FPD 4 to a predetermined initial position. The imaging systems 3 and 4 at this time are arranged as shown by a solid line in FIG. That is, the X-ray tube 3 in the initial position is located in the front stage in the body axis direction A (longitudinal direction of the top plate 2), and the FPD 4 is located in the rear stage in the body axis direction A. At this time, the X-ray tube 3 is inclined to an initial angle of −20 degrees.

  The X-ray tube control unit 6 controls the X-ray tube 3, and the X-ray tube 3 irradiates the X-ray beam toward the FPD 4 with a predetermined pulse width, tube voltage, and tube current. The X-ray beam passes through the top plate 2 and then enters the FPD 4. The image generation unit 11 assembles the detection signal output from the FPD 4 into the original image P0.

  Thereafter, the synchronous movement control unit 8 moves the X-ray tube 3 and the FPD 4 synchronously and in directions opposite to each other. The X-ray tube controller 6 intermittently irradiates the X-ray beam during the movement, and the image generator 11 generates the original image P0 each time. In this way, a series of original images P0 is generated. At this time, the synchronous movement control unit 8 moves the X-ray tube 3 to the rear stage side in the body axis direction A, and moves the FPD 4 to the front stage side in the body axis direction A.

  Then, the synchronous movement control unit 8 moves the X-ray tube 3 and the FPD 4 to a predetermined final position. The imaging systems 3 and 4 at this time are arranged as shown by a one-dot chain line in FIG. That is, the X-ray tube 3 at the final position is located at the rear stage of the body axis direction A (longitudinal direction of the top plate 2), and the FPD 4 is located at the front stage of the body axis direction A. At this time, the X-ray tube 3 is inclined to a final angle of 20 degrees. In this state, the last original image P0 is acquired, and acquisition of a series of original images P0 ends. In the first embodiment, 74 original images P0 are acquired.

  The image processing unit 12 creates a map m using the map generation unit 12a every time the original image P0 is generated. Then, every time the map m is created, the image processing unit 12 extrapolates the original image P0 using the extrapolation unit 12b. The extrapolated image P1 thus generated is an image obtained by extending the original image P0 in the moving direction of the imaging systems 3 and 4.

  A series of extrapolated images P1 generated in this way generates a tomographic image D by the tomographic image generating unit 12c of the image processing unit 12. This tomographic image D includes a tomographic image of the subject M suitable for diagnosis. The tomographic image D is displayed on the display unit 27 and the operation ends.

  As described above, the image processing apparatus according to the present invention first generates a map m indicating the extending direction of the linear structure reflected in the original image P0. Then, with reference to the map m, a copy of the region located on the side portion is arranged so that the linear structures reflected on the side portion of the original image P0 are arranged in the extending direction of the linear structure P0. An extrapolated image P1 is generated by adding to the side of the image. In addition, this adding operation is performed at both ends of the moving direction of the X-ray tube 3 and the FPD 4 in the original image P0. In this way, both ends of the original image P0 in the moving direction are added so as to extend the linear structure, and an extrapolated image P1 is generated. Since the portion outside the side portion of the original image P0 is a portion that could not be photographed, it is impossible to actually know how the original image P0 of this portion was. However, it is considered that the image of the subject in the portion deviated from the original image P0 is at least a pattern in which the image of the side portion of the original image P0 is extended for a while.

  Therefore, if the pattern of the side portion is appropriately shifted in the extending direction of the side portion so as to extend the linear structure reflected on the side portion of the original image P0, the image on the image is captured. The part that could not be done can be extrapolated more naturally. That is, the end of the generated extrapolated image P1 is such that the subject image is continuously extended. If the tomographic image D is generated using the extrapolated image P1 subjected to such extrapolation, the tomographic image D having excellent visibility can be generated.

  Moreover, the above-mentioned structure shows the specific structure of the extrapolation part 12b of this invention. That is, if the extrapolation part 12b adds, the extrapolated image P1 is more reliably generated by duplicating the strip-like pixel array located in the side part extending in the orthogonal direction orthogonal to the moving direction. Is possible.

  The above-described configuration shows a specific configuration of the extrapolation portion 12b of the present invention. That is, the extrapolation unit 12b can acquire a clearer tomographic image D by performing a smoothing process in the orthogonal direction on the strip-like pixel array when adding pixels. It is expected that the fluoroscopic image of the subject in the outer portion of the original image P0 will be substantially the same as the subject image reflected on the side portion of the original image P0. However, in actuality, the fluoroscopic image of the subject on the outer side of the original image P0 and the subject image reflected on the side portion of the original image P0 are not exactly the same. Therefore, when pixels are added as they are, a pattern different from the actual pattern is added to the original image P0, and the tomographic image D becomes unnatural. According to the above configuration, since the pixels are added after smoothing once, the unnaturalness of the tomographic image D is not noticeable, and the visibility of the tomographic image D is improved.

  The above-described configuration shows a specific configuration of the extrapolation portion 12b of the present invention. Prediction of a subject image that is not shown in the original image P0 becomes more difficult as the portion where the subject image is predicted is farther from the side portion of the original image P0. For example, in a region outside the original image P0, it is expected that the perspective image of the subject in the portion adjacent to the side portion will be substantially the same as the subject image reflected in the side portion. However, in the region outside the original image P0, the image of the subject should be farther away from the subject image reflected on the side portion as it is farther from the side portion. According to the present invention, as the adding operation of a plurality of pixel arrays is repeated, the blurring strength of the smoothing process applied to the added pixel array is gradually increased. In this way, it is possible to prevent a situation in which a pattern that is far from the original subject image is gradually added as the adding operation is repeated.

  The above configuration shows the configuration of a radiation tomography apparatus equipped with the image processing apparatus of the present invention. According to the radiation tomography apparatus according to the present invention, a tomographic image D with improved visibility at both ends of the tomographic image D can be acquired.

  The present invention is not limited to the above-described configuration, and can be modified as follows.

  (1) In the above configuration, the total vector is obtained from the pixel array R1 in which the pixels located at the end of the original image P0 are arranged in one column, but the present invention is not limited to this configuration. That is, the total vector may be obtained using a region having a wider width in the short direction instead of the pixel array R1.

  (2) In the above-described configuration, the pixel array R1 is divided into a plurality of fragments, and a total vector is obtained for each of them to perform extrapolation processing of the image. However, the present invention is not limited to this configuration. That is, the operation may be performed based on a profile regarding the length of the gradient vector.

  This modification will be specifically described. The left side of FIG. 21 represents an example in which a plurality of linear structures are reflected at the end of the original image P0. The map generator 12a obtains a gradient vector for the original image P0 and arranges it to generate a map m. The extrapolation unit 12b generates a profile in which the position on the image and the magnitude of the gradient vector are related based on the map m. In this profile, three peaks appear as shown on the right side of FIG. The extrapolation unit 12b recognizes these three peaks and knows in which part the linear structure is located at the end of the original image P0. Whether it is a peak or not is determined by whether or not the maximum value in the profile exceeds a predetermined threshold. When the value of the maximum portion is equal to or greater than the threshold value, the extrapolation unit 12b operates by determining the maximum portion as a peak.

  The extrapolation unit 12b operates by dividing the pixel array R1 according to the position where the peak appears. That is, the extrapolation unit 12b obtains an intermediate point located between two adjacent peaks on the profile, and divides the pixel array R1 at the position indicated by the intermediate point. According to the example of FIG. 21, since three peaks appear, there are two peaks between the first peak and the second peak adjacent to each other and between the second peak and the third peak adjacent to each other. Intermediate points ca and cb are provided at the locations. Then, the extrapolation unit 12b has three sections, a section from the left end to the intermediate point ca, a section from the intermediate point ca to the intermediate point cb, and a section from the intermediate point cb to the right end in the pixel array R1 on the original image P0. The total vector is obtained and the original image P0 is stretched for each section. Thereby, each of the linear structures on the original image P0 is stretched according to the extending direction, and the original image P0 is extrapolated. Thereby, the tomographic image D with less image disturbance can be acquired. As a method for determining the intermediate point, a position between two adjacent peaks may be used as the intermediate point, and it is more preferable that the intermediate point is determined as a position having the same distance from both of the two peaks.

m Map (gradient vector map)
P0 Original image (image)
3 X-ray tube (radiation source)
4 FPD (radiation detection means)
7 Synchronous movement mechanism (moving means)
8 Synchronous movement control unit (movement control means)
11 Image generation unit (image generation means)
12a Map generator (map generator)
12b Extrapolation part (extrapolation means)
12c tomographic image generating unit (tomographic image generating means)

Claims (5)

Extrapolation is performed by adding a duplicate of the region located on the side portion to the side portion of the image so that the linear structure reflected on the side portion of the image is arranged in the extending direction of the linear structure. Extrapolation means for generating an image;
Extrapolation generated based on a series of images continuously shot while the radiation source for irradiating the subject with radiation and the radiation detecting means for detecting the radiation after passing through the subject are moved synchronously and in opposite directions. A tomographic image generating means for generating a tomographic image by superimposing images,
The image processing apparatus, wherein the extrapolation unit performs image processing on both ends of the image in a direction in which the radiation source and the radiation detection unit move. The image processing apparatus according to claim 1.
The extrapolation means operates by duplicating a strip-like pixel array located in a side portion extending in an orthogonal direction orthogonal to the moving direction in the image and adding the copied pixel to the side portion. An image processing apparatus. The image processing apparatus according to claim 2,
The extrapolation unit performs smoothing processing in the orthogonal direction on a strip-like pixel arrangement when adding pixels. The image processing apparatus according to claim 3.
The extrapolation means gradually increases the blurring strength of the smoothing process applied to the added pixel array as the addition operation of the plurality of pixel arrays is repeated. A radiation tomography apparatus equipped with the image processing apparatus according to claim 1,
A radiation source for irradiating the subject with radiation;
Radiation detecting means for detecting radiation transmitted from the subject;
Moving means for moving the radiation source and the radiation detecting means in a direction opposite to each other in a moving direction with respect to the subject;
Movement control means for controlling the movement means;
A radiation tomography apparatus comprising: an image generation unit configured to generate the image based on an output of the radiation detection unit.

Images