JP6485278B2 - Image processing method and X-ray fluoroscopic apparatus - Google Patents

Description

  The present invention relates to an X-ray fluoroscopic apparatus that acquires an X-ray image of a subject using X-rays, and more particularly to an image processing technique that removes artifacts generated in an X-ray image generated by X-ray fluoroscopy.

  In the medical field, radiation therapy is widely performed in which radiation such as X-rays or electron beams is irradiated to an affected part such as cancer. In such radiotherapy, it is necessary to accurately irradiate the affected area with radiation. However, when the affected part is located in the heart, lungs, etc., the affected part itself moves greatly due to the body movement of the subject due to breathing or pulsation, etc., so that the radiation treatment efficiency is reduced in the configuration in which radiation is constantly irradiated to a fixed point. . As a system for performing radiation therapy on an affected part that is moved by the body movement of the subject, a radiation therapy apparatus using a moving body tracking irradiation method has been proposed (for example, see Patent Documents 1 and 2).

  The conventional radiotherapy apparatus using the moving body tracking irradiation method includes a therapeutic radiation irradiation apparatus that irradiates a relatively strong radiation (treatment radiation) for treating an affected part of a subject, and an X that intermittently irradiates a relatively weak X-ray. And an X-ray fluoroscopic apparatus that intermittently acquires X-ray images of the subject through fluoroscopy. Using such an X-ray fluoroscopic imaging apparatus, X-ray fluoroscopy is performed from a plurality of directions, and an X-ray image showing a marker placed in the vicinity of the affected part is intermittently acquired, and the position of the affected part is determined based on the position of the marker. Is identified. The therapeutic normal device is controlled to irradiate the subject with therapeutic radiation when the position of the affected part of the specified subject is within a predetermined range.

  The X-ray fluoroscopic apparatus is provided with a plurality of imaging systems composed of an X-ray tube and an X-ray detector. The X-ray tube irradiates the subject with X-rays, and the X-ray detector detects the X-rays and outputs an X-ray detection signal. In recent years, flat panel X-ray detectors (FPD) are widely used as X-ray detectors. By sequentially reading out each of the pixels arranged in a two-dimensional matrix in the FPD in a predetermined direction (reading direction), an X-ray detection signal in each pixel is output from the FPD. An X-ray image is generated based on an X-ray detection signal output from each FPD.

  In each imaging system, each X-ray tube emits fluoroscopic X-rays from different directions, so that X-ray images from different directions are acquired intermittently. Based on the information of each X-ray image, the three-dimensional position information of the marker is calculated in real time. Since the three-dimensional position of the marker can be specified in real time by such moving body tracking irradiation method, even when the affected part moves intermittently due to the body movement of the subject, high-accuracy radiotherapy according to the movement of the affected part Can be executed.

JP-A-2005-111151 Japanese Patent Application Laid-Open No. 2014-128412

However, the conventional example having such a configuration has the following problems.
That is, in the X-ray fluoroscopic apparatus, a strip-like or linear artifact AF may occur in the generated X-ray image C (see the left arrow in FIG. 13). Each of the artifacts AF extends in a direction orthogonal to the reading direction Re of the FPD (the right diagram in FIG. 13).

  As to the cause of the artifact AF, the following hypothesis can be considered. That is, X-rays irradiated from a therapeutic radiation apparatus or the like are scattered and enter each FPD provided in the X-ray fluoroscopic apparatus. A difference occurs in the time when the X-ray detection signal is read out depending on the part of the FPD in the reading direction Re. Since the amount of incident scattered X-rays differs according to the time difference at which the X-ray detection signal is read, a luminance difference (pixel value difference) occurs in each pixel along the reading direction Re. Accordingly, it is considered that a strip-shaped artifact AF is generated in a direction orthogonal to the readout direction Re in accordance with the difference in pixel value based on the amount of scattered X-ray incidence.

  The occurrence of such a strip-like artifact AF reduces the visibility of the X-ray image shown in the X-ray image C. As a conventional image processing method for removing the band-shaped artifact AF from the X-ray image, a pixel value difference is calculated between pixels adjacent to each other in a direction orthogonal to the direction in which the artifact AF extends (here, the reading direction Re). (Reference 1: Japanese Patent Laid-Open No. 2014-42559, Reference 2: International Publication WO2010 / 134295).

  That is, when the luminance difference between pixels adjacent in the readout direction Re is equal to or greater than a preset threshold value, it is determined that a particularly noticeable artifact has occurred between the adjacent pixels. Then, by correcting the pixel value in each pixel so as to smooth the difference between the pixel values that are equal to or greater than the threshold value, the artifact that hinders diagnosis is removed from the X-ray image.

  However, there is a concern that such a conventional image processing method has insufficient artifact removal. That is, the pixel value is not corrected unless the difference in pixel value between adjacent pixels is equal to or greater than the threshold value. Therefore, if the difference in pixel value between adjacent pixels is equal to or less than the threshold value, even if an artifact occurs, it is not removed by image processing. Therefore, when the threshold value is set to a high value, it is difficult to sufficiently improve the visibility of the X-ray image because a relatively slight artifact cannot be removed by the conventional image processing method.

  On the other hand, when the threshold value is set low to remove relatively mild artifacts, even a small luminance difference that is caused by an actual X-ray detection signal indicating an X-ray image of the subject is subject to correction. . As a result, in the X-ray image corrected by the conventional image processing, it is difficult to obtain accurate image information regarding the affected part and the marker of the subject.

  Further, in the conventional image processing method, when performing X-ray fluoroscopy on a moving affected part in particular, there is a concern that the artifact cannot be suitably removed from each of the X-ray images. When a moving affected part is an imaging target, the position, brightness, and the like of the X-ray image reflected in the X-ray image change due to the movement of the affected part. Accordingly, since the pattern of pixel values of the entire X-ray image is greatly different in each X-ray image, the conventional image processing method has an artifact for each of the X-ray images according to the pixel value pattern of the entire X-ray image. It is necessary to appropriately set a threshold value suitable for removal.

  However, skill is required for the surgeon to appropriately set an appropriate threshold value for removing an appropriate artifact according to the pixel value pattern of the entire X-ray image. Further, in X-ray fluoroscopy, an X-ray image is generated at a frame rate of about 15 to 30 FPS as an example, and it is difficult to determine and set appropriate threshold values for all of a large number of X-ray images. As a result, when X-ray fluoroscopy is performed on a moving affected part, it is very difficult to suitably remove artifacts from each X-ray image.

  The present invention has been made in view of such circumstances, and provides an image processing method and an X-ray fluoroscopic imaging apparatus that suitably remove artifacts from an X-ray image, particularly in X-ray fluoroscopy in a moving affected part. For the purpose.

In order to achieve such an object, the present invention has the following configuration.
That is, in the image processing method according to the present invention, an X-ray conversion step in which an X-ray detector detects X-rays irradiated to an object from an X-ray tube and converts them into an X-ray detection signal; An image generation step for generating an X-ray image using the X-ray detection signal converted in the conversion step, and a band shape for generating the pixel value of the X-ray image generated in the image generation step in the X-ray image An image averaging step for generating an average pixel value profile by averaging in a direction orthogonal to the readout direction of the X-ray detector, and the entire average pixel value profile generated in the image averaging step An approximate curve generating step for calculating an approximate curve that approximates the difference, and a difference between the value of the approximate curve generated in the approximate curve generating step and the value of the average pixel value profile as a difference value A difference value calculation step to be output; and an image correction step of correcting the X-ray image by subtracting the difference value calculated in the difference value calculation step from each pixel value in the X-ray image. It is characterized by.

  [Operation / Effect] According to the image processing method of the present invention, in the image averaging step, the pixel values of the X-ray image are averaged in the direction in which the strip-like artifacts generated in the X-ray image extend, and the average pixel value profile is obtained. Generate. In the approximate curve generation step, an approximate curve that approximates the entire average pixel value profile is calculated, and in the difference value calculation step, the difference between the approximate curve value and the average pixel value profile value is calculated as a difference value. In the image correction step, the X-ray image is corrected by subtracting the difference value from each pixel value in the X-ray image.

  Since the average pixel value profile reflects the tendency of the pixel values in the entire X-ray image, information on the tendency of the pixel values in the entire X-ray image can be acquired in a lump by the averaging process. By calculating a difference value from the approximate curve obtained by smoothing the entire average pixel value profile, information regarding the magnitude of the influence of incident scattered radiation can be obtained accurately and promptly for each pixel column. As a result, the time required for the image processing for removing the artifact from the X-ray image can be shortened.

  Further, the difference value reflects the amount of change in pixel value due to incident scattered radiation in each pixel column. Therefore, by subtracting the difference value from the X-ray image, the pixel value of each pixel can be appropriately corrected according to the amount of change in the pixel value caused by the incident scattered radiation. The difference value is automatically and promptly calculated for all pixel columns every time an X-ray image is generated. Since the correction of the X-ray image is completed simply by subtracting the difference value from the pixel value in the X-ray image, a threshold value serving as a reference for determining whether correction is appropriate is set appropriately for each X-ray image as in conventional image processing. There is no need. Therefore, no experience is required for the operator, and the burden on the operator can be greatly reduced.

  In the X-ray image corrected in the image correction process, the average pixel value profile is equal to the approximate curve. That is, in the average pixel value profile of the X-ray image before correction, the protruding portion due to the incident scattered radiation is preferably smoothed, and the tendency of the pixel value in the entire X-ray image is maintained. Therefore, it is possible to favorably correct the influence of the incident scattered radiation regardless of the size while retaining the pixel value information based on the actual X-ray image of the subject. Therefore, the accuracy of image information in the X-ray image can be improved while more accurately removing artifacts from the X-ray image.

  In the above-described invention, the approximate curve generation step is omitted when the X-ray image of the second and subsequent frames is generated, and the average pixel value profile calculated for the X-ray image is omitted in the difference value calculation step. It is preferable to calculate a difference between the value of the approximate curve generated in the X-ray image of the first frame as the difference value.

  [Operation / Effect] According to the fluoroscopic imaging apparatus of the present invention, the approximate curve is generated only for the X-ray image of the first frame, and the approximate curve generation step is performed when the X-ray images of the second and subsequent frames are generated. Is omitted. In the difference value calculation step, the difference between the value of the average pixel value profile calculated for the X-ray image and the value of the approximate curve generated in the X-ray image of the first frame is calculated as the difference value.

  With such a configuration, when the X-ray image for the second frame and thereafter is generated, the approximate curve generation step is omitted, so that the time required for image processing can be greatly reduced. Therefore, even when the frame rate of the X-ray image is increased, artifacts can be quickly removed from the X-ray image, and an X-ray image with improved visibility can be efficiently generated by image processing. Therefore, it is possible to more appropriately proceed with the treatment using an X-ray image with improved frame rate and visibility.

  In the above-described invention, after the approximate curve generation step, a weighting process step of calculating a processed approximate curve by weighting the approximate curve generated most recently and the approximate curve generated in the past. It is preferable that the difference value calculation step calculates a difference between the average pixel value profile value and the post-processing approximate curve value calculated in the weighting processing step as the difference value.

  [Operation / Effect] According to the image processing method of the present invention, after the approximate curve generating step, the approximate curve generated most recently and the approximate curve generated in the past are weighted, thereby processing the approximate curve after processing. Is further included. In the difference value calculating step, a difference between the value of the average pixel value profile and the value of the post-processing approximate curve calculated in the weighting processing step is calculated as a difference value.

  In such a configuration, even if the pattern of the approximate curve generated for each frame of the X-ray image is greatly different, the difference in pattern between the approximate curves after processing calculated by the weighting process is similar to that between the approximate curves. Smaller than pattern difference. The image correction unit subtracts a difference value, which is a difference between the value of the average pixel value profile and the value of the processed approximate curve, from each pixel value of the X-ray image. That is, the average pixel value profile in the corrected X-ray image matches the post-processing approximate curve.

  As a result, the difference in pixel value pattern between the corrected X-ray images becomes smaller. Therefore, when a plurality of X-ray images are generated intermittently, the X-ray images suddenly change between the plurality of X-ray images. Can be avoided. Therefore, the visibility as the whole X-ray image group produced | generated by X-ray fluoroscopy can be improved more.

  In the above-described invention, it is preferable that the approximate curve generation step calculates an n-order polynomial that approximates the entire average pixel value profile as the approximate curve.

  [Operation / Effect] According to the image processing method of the present invention, an n-order polynomial is calculated as an approximate curve that approximates the entire average pixel value profile. When approximation by an nth order polynomial is performed, the plot data of the average pixel value A can be approximated more accurately and quickly. Therefore, in the X-ray image after being corrected by the image processing, the pixel value information based on the actual X-ray image of the subject is reflected more accurately. Furthermore, the time required for image processing can be further shortened, and the workflow of the X-ray fluoroscopic apparatus can be further improved.

In order to achieve such an object, the present invention may take the following configurations.
That is, an X-ray fluoroscopic apparatus according to the present invention includes an X-ray tube that irradiates a subject with X-rays, and an X-ray detection unit that detects X-rays transmitted through the subject and outputs an X-ray detection signal. An image generation unit that generates an X-ray image using an X-ray detection signal output from the X-ray detection unit, and a strip-like artifact generated in the X-ray image extends in a reading direction of the X-ray detection unit Image averaging means for averaging pixel values of the X-ray image in an orthogonal direction to generate an average pixel value profile, approximate curve generating means for calculating an approximate curve that approximates the entire average pixel value profile, and Difference value calculating means for calculating a difference between the value of the average pixel value profile and the value of the approximate curve as a difference value; and subtracting the difference value from each of the pixel values in the X-ray image, correction It is characterized in further comprising a that image correction means.

  [Operation / Effect] According to the X-ray fluoroscopic apparatus of the present invention, the image averaging means averages the pixel values of the X-ray image in the direction in which the strip-like artifacts generated in the X-ray image extend, and the average pixel value Generate a profile. The approximate curve generation means calculates an approximate curve that approximates the entire average pixel value profile, and the difference value calculation means calculates the difference between the approximate curve value and the average pixel value profile value as a difference value. The image correcting unit corrects the X-ray image by subtracting the difference value from each of the pixel values in the X-ray image.

  Since the average pixel value profile reflects the tendency of the pixel values in the entire X-ray image, information on the tendency of the pixel values in the entire X-ray image can be acquired in a lump by the averaging process. By calculating a difference value from the approximate curve obtained by smoothing the entire average pixel value profile, information regarding the magnitude of the influence of incident scattered radiation can be obtained accurately and promptly for each pixel column. As a result, the time required for the image processing for removing the artifact from the X-ray image can be shortened.

  Further, the difference value reflects the amount of change in pixel value due to incident scattered radiation in each pixel column. Therefore, by subtracting the difference value from the X-ray image, the pixel value of each pixel can be appropriately corrected according to the amount of change in the pixel value caused by the incident scattered radiation. The difference value is automatically and promptly calculated for all pixel columns every time an X-ray image is generated. Since the correction of the X-ray image is completed simply by subtracting the difference value from the pixel value in the X-ray image, a threshold value serving as a reference for determining whether correction is appropriate is set appropriately for each X-ray image as in conventional image processing. There is no need. Therefore, no experience is required for the operator, and the burden on the operator can be greatly reduced.

  In the X-ray image corrected by the image correcting unit, the average pixel value profile is equal to the approximate curve. That is, in the average pixel value profile of the X-ray image before correction, the protruding portion due to the incident scattered radiation is preferably smoothed, and the tendency of the pixel value in the entire X-ray image is maintained. Therefore, it is possible to favorably correct the influence of the incident scattered radiation regardless of the size while retaining the pixel value information based on the actual X-ray image of the subject. Therefore, the accuracy of image information in the X-ray image can be improved while more accurately removing artifacts from the X-ray image.

  In the above-described invention, the approximate curve generation unit calculates the approximate curve only for the X-ray image of the first frame, and the difference value calculation unit calculates the value of the average pixel value profile and the first frame. It is preferable that a difference from the approximate curve value calculated for the X-ray image is calculated as the difference value.

  [Operation and Effect] According to the X-ray fluoroscopic apparatus according to the present invention, the approximate curve generating means calculates an approximate curve only for the X-ray image of the first frame. The difference value calculating means calculates a difference between the value of the average pixel value profile and the value of the approximate curve calculated for the X-ray image of the first frame as the difference value.

  In such a configuration, the step of generating the approximate curve is omitted when the X-ray images of the second and subsequent frames are generated, so that the time required for image processing can be greatly reduced. Therefore, even when the frame rate of the X-ray image is increased, artifacts can be quickly removed from the X-ray image, and an X-ray image with improved visibility can be efficiently generated by image processing. Therefore, it is possible to more appropriately proceed with the treatment using an X-ray image with improved frame rate and visibility.

  In the above-described invention, the approximate curve generating means further includes weighting processing means for calculating a processed approximate curve by weighting each of the approximate curves calculated by the approximate curve generating means, and the difference value calculating means includes the average value calculating means. It is preferable that the difference between the value of the pixel value profile and the value of the post-process approximation curve is calculated as the difference value.

  [Operation / Effect] The X-ray fluoroscopic apparatus according to the present invention further includes weighting processing means for calculating a processed approximate curve by weighting each of the approximate curves calculated by the approximate curve generating means. Yes. The difference value calculation means calculates a difference between the value of the average pixel value profile and the value of the post-process approximation curve calculated in the weighting process step as a difference value.

  In such a configuration, even if the pattern of the approximate curve generated for each frame of the X-ray image is greatly different, the difference in pattern between the approximate curves after processing calculated by the weighting process is similar to that between the approximate curves. Smaller than pattern difference. The image correction unit subtracts a difference value, which is a difference between the value of the average pixel value profile and the value of the processed approximate curve, from each pixel value of the X-ray image. That is, the average pixel value profile in the corrected X-ray image matches the post-processing approximate curve.

  As a result, the difference in pixel value pattern between the corrected X-ray images becomes smaller. Therefore, when a plurality of X-ray images are generated intermittently, the X-ray images suddenly change between the plurality of X-ray images. Can be avoided. Therefore, the visibility as the whole X-ray image group produced | generated by X-ray fluoroscopy can be improved more.

  In the above-described invention, it is preferable that the approximate curve generation unit calculates an n-order polynomial that approximates the entire average pixel value profile as the approximate curve.

  [Operation / Effect] According to the X-ray fluoroscopic apparatus according to the present invention, an nth order polynomial is calculated as an approximate curve that approximates the entire average pixel value profile. When approximation by an nth order polynomial is performed, the plot data of the average pixel value A can be approximated more accurately and quickly. Therefore, in the X-ray image after being corrected by the image processing, the pixel value information based on the actual X-ray image of the subject is reflected more accurately. Furthermore, the time required for image processing can be further shortened, and the workflow of the X-ray fluoroscopic apparatus can be further improved.

  According to the image processing method of the present invention, in the image averaging step, the average pixel value profile is generated by averaging in the extending direction of the band-like artifacts generated in the X-ray image. In the approximate curve generation step, an approximate curve that approximates the entire average pixel value profile is calculated, and in the difference value calculation step, the difference between the approximate curve value and the average pixel value profile value is calculated as a difference value. In the image correction step, the X-ray image is corrected by subtracting the difference value from each pixel value in the X-ray image.

  Since the average pixel value profile reflects the tendency of the pixel values in the entire X-ray image, information on the tendency of the pixel values in the entire X-ray image can be acquired in a lump by the averaging process. By calculating a difference value from the approximate curve obtained by smoothing the entire average pixel value profile, information regarding the magnitude of the influence of incident scattered radiation can be obtained accurately and promptly for each pixel column. As a result, the time required for the image processing for removing the artifact from the X-ray image can be shortened.

  Further, the difference value reflects the amount of change in pixel value due to incident scattered radiation in each pixel column. Therefore, by subtracting the difference value from the X-ray image, the pixel value of each pixel can be appropriately corrected according to the amount of change in the pixel value caused by the incident scattered radiation. The difference value is automatically and promptly calculated for all pixel columns every time an X-ray image is generated. Since the correction of the X-ray image is completed simply by subtracting the difference value from the pixel value in the X-ray image, a threshold value serving as a reference for determining whether correction is appropriate is set appropriately for each X-ray image as in conventional image processing. There is no need. Therefore, no experience is required for the operator, and the burden on the operator can be greatly reduced.

  In the X-ray image corrected in the image correction process, the average pixel value profile is equal to the approximate curve. That is, in the average pixel value profile of the X-ray image before correction, the protruding portion due to the incident scattered radiation is preferably smoothed, and the tendency of the pixel value in the entire X-ray image is maintained. Therefore, it is possible to favorably correct the influence of the incident scattered radiation regardless of the size while retaining the pixel value information based on the actual X-ray image of the subject. Therefore, the accuracy of image information in the X-ray image can be improved while more accurately removing artifacts from the X-ray image.

It is the schematic explaining the whole structure of the radiotherapy apparatus to which the X-ray fluoroscopic apparatus which concerns on Example 1 is applied. (A) is a front view of a radiotherapy apparatus, (b) is a right view of a radiotherapy apparatus. It is a functional block diagram explaining the structure of the radiotherapy apparatus to which the X-ray fluoroscopic apparatus which concerns on Example 1 is applied. It is a flowchart explaining the process of operation | movement of the X-ray fluoroscopic apparatus which concerns on each Example. (A) is a flowchart according to the first embodiment, (b) is a flowchart according to the second embodiment, and (c) is a flowchart according to the third embodiment. It is a figure which shows the image before correction | amendment produced | generated by step S1 which concerns on Example 1. FIG. The figure on the left shows the actual image before correction, and the figure on the right shows the image before correction and the artifacts schematically. FIG. 10 is a diagram for explaining a process of generating an average pixel value profile in step S2 according to the first embodiment. (A) is a figure which shows an average pixel value profile, (b) is a figure which shows the approximated curve which approximates an average pixel value profile. In Example 1, it is a figure explaining the positional relationship of an average pixel value profile and an artifact. It is a figure which shows the approximated curve produced | generated by step S3 which concerns on Example 1. FIG. It is a figure which shows the image after correction | amendment produced | generated by step S5 which concerns on Example 1. FIG. (A) is a figure which shows the average pixel value profile which concerns on the image after correction | amendment, (b) is a figure which shows an actual image after correction | amendment, (c) is the figure which showed the image after correction | amendment typically. . It is a figure explaining the operation | movement in step S7 which concerns on Example 1. FIG. (A) is a figure which shows the state determined to be irradiated with therapeutic radiation, (b) is a figure which shows the state determined to stop irradiation of therapeutic radiation. It is a figure explaining the effect by the composition of Example 1. (A) is a figure which shows the state which the wide strip | belt-shaped artifact has generate | occur | produced, (b) is a figure which shows the pixel value of each pixel in the m-th line in the conventional image processing, (c) FIG. 6 is a diagram illustrating an average pixel value profile and an approximate curve generated in image processing according to the first embodiment. 6 is a functional block diagram illustrating a configuration of an X-ray fluoroscopic apparatus according to Embodiment 3. FIG. FIG. 10 is a functional block diagram for explaining the effect of the configuration of the third embodiment. The upper row is a diagram showing an image before correction, the middle row is a diagram showing an approximate curve, and the lower row is a diagram showing an approximate curve after processing generated by weighting processing of the approximate curve. It is a figure for demonstrating the artifact which generate | occur | produces in a X-ray image in the X-ray fluoroscopic imaging apparatus which concerns on a prior art example. The left figure is a diagram showing an actual X-ray image, and the right figure is a diagram schematically showing the X-ray image and artifacts.

  Embodiment 1 of the present invention will be described below with reference to the drawings. Here, a case will be described as an example in which radiotherapy by the moving body pursuit irradiation method is performed using the radiotherapy apparatus to which the X-ray fluoroscopic apparatus according to the first embodiment is applied. FIG. 1A is a front view for explaining the overall configuration of a radiotherapy apparatus 1 using a moving body pursuit irradiation method to which the X-ray fluoroscopic imaging apparatus according to the first embodiment is applied, and FIG. 1B is a radiotherapy apparatus 1. It is a right view explaining the whole structure.

  As shown in FIG. 1A, the radiotherapy apparatus 1 includes a top 3 on which a subject M is placed, a therapeutic radiation irradiation apparatus 5, and an X-ray fluoroscopic imaging apparatus 7. The therapeutic radiation irradiation apparatus 5 includes a therapeutic radiation source 5a. The therapeutic radiation source 5a irradiates the affected part B of the subject M with relatively strong radiation G as therapeutic radiation. Examples of the radiation G used for therapeutic radiation include X-rays and electron beams.

  The X-ray fluoroscopic apparatus 7 includes a first imaging system composed of an X-ray tube 9a and an FPD 11a, and a second imaging system composed of an X-ray tube 9b and an FPD 11b. The X-ray tube 9a irradiates the affected part B of the subject M with X-ray Ha from the first direction, and the X-ray tube 9b applies the second direction different from the first direction to the affected part B of the subject M. To X-rays Hb.

  In each of the FPD 11a and the FPD 11b, pixels that detect X-rays and convert them into X-ray detection signals that are electrical signals are arranged in a two-dimensional matrix. Although the number of pixels may be changed as appropriate, the number of pixels in the first embodiment is assumed to be 1,024 vertical × 1,024 horizontal. The FPD 11a detects the X-ray Ha and outputs an X-ray detection signal, and the FPD 11b detects the X-ray Hb and outputs an X-ray detection signal.

  Each of the FPD 11a and the FPD 11b outputs an X-ray detection signal for all the pixels by reading the X-ray detection signal converted by each pixel in a predetermined direction (reading direction Re). In each of FIGS. 1A and 1B, the reading direction Re in the FPD 11a is illustrated by a symbol Re1, and the reading direction Re in the FPD 11b is illustrated by a symbol Re2.

  The X-ray tube 9a is supported by the pedestal 13a, and the X-ray tube 9b is supported by the pedestal 13b. On the floor W of the diagnosis room, a rail 15 having a substantially U shape in plan view is disposed so as to surround the top plate 3. The pedestal 13a and the pedestal 13b are guided by the rail 15 and configured to be movable in the x direction (longitudinal direction of the top plate 3) and the y direction (short direction of the top plate 3).

  The FPD 11a is supported by the pedestal 17a, and the FPD 11b is supported by the pedestal 17b. On the ceiling surface Y of the diagnosis room, a rail 19 having a substantially U shape in plan view is disposed so as to surround the top plate 3. The pedestal 17a and the pedestal 17b are configured to be guided by the rail 19 so as to be movable in the x direction and the y direction. Each of the X-ray tube 9a and the X-ray tube 9b is provided with a collimator 21 for adjusting the X-ray irradiation field to a pyramid shape or the like. The FPD 11a and the FPD 11b correspond to an X-ray detector and X-ray detection means.

  The X-ray tube 9a and the FPD 11a, and the X-ray tube 9b and the FPD 11b are arranged so as to always face each other with the top plate 3 interposed therebetween. That is, when the X-ray tube 9 a moves along the rail 15, the FPD 11 a moves along the rail 19 synchronously. Further, the FPD 11 b moves along the rail 19 in synchronization with the movement of the X-ray tube 9 b along the rail 15. As a result, the first imaging system and the second imaging system can change the X-ray irradiation position and irradiation direction, respectively. FIG. 1A shows a state where the irradiation direction of X-ray Ha and the irradiation direction of X-ray Hb intersect on the xz plane, and the irradiation direction of X-ray Ha and the irradiation direction of X-ray Hb are on the yz plane. FIG. 1 (b) shows a state of crossing at. In the first embodiment, the z direction indicates the vertical direction.

  As shown in FIG. 2, the X-ray fluoroscopic apparatus 7 includes an X-ray irradiation control unit 23 and an image processing unit 25. The X-ray irradiation control unit 23 is connected to each of the X-ray tube 9a and the X-ray tube 9b, and determines the dose of X-rays irradiated from each of the X-ray tubes 9a and 9b, the timing of irradiating X-rays, and the like. Control. In the first embodiment, the X-ray irradiation control unit 23 controls each X-ray tube so as to perform X-ray fluoroscopy with a relatively low dose of irradiated X-rays. Each of the X-ray tubes 9 a and 9 b irradiates with a low dose of X-rays intermittently under the control of the X-ray irradiation control unit 23.

  As a characteristic configuration of the present invention, the image processing unit 25 further includes an image generation unit 27, an average value calculation unit 29, an approximate expression calculation unit 31, a difference value calculation unit 33, and an image correction unit 35. Yes. The image generation unit 27 is provided in the subsequent stage of each of the FPD 11a and the FPD 11b. Based on the X-ray detection signals output from the FPDs 11a and 11b, the image generation unit 27 generates an X-ray image that displays the X-ray image of the affected area B of the subject and the marker R placed in the vicinity of the affected area B. Generate intermittently. Here, each of the X-ray images generated by the image generation unit 27 by X-ray fluoroscopy is referred to as “pre-correction image”.

  The average value calculation unit 29 generates an average pixel value profile in the reading direction Re by performing an operation for averaging the pixel values in a direction orthogonal to the reading direction Re for each of the uncorrected images. The direction orthogonal to the reading direction Re corresponds to the direction in which the artifact generated in the pre-correction image extends. The approximate expression calculation unit 31 performs n-order polynomial approximation by the least square method on the average pixel value profile generated by the average value calculation unit 29, and calculates an approximate curve that approximates the average pixel value profile. The average value calculator 29 corresponds to the image averaging means in the present invention. The approximate expression calculation unit 31 corresponds to the approximate curve generation means in the present invention.

  The difference value calculation unit 33 calculates, as a difference value, the difference between the value of the average pixel value profile and the value on the approximate curve that approximates the average pixel value profile for each pixel row arranged in the readout direction Re. The image correcting unit 35 performs image processing for correcting each of the uncorrected images by subtracting the difference value calculated by the difference value calculating unit 33 from the pixel value of each pixel constituting the uncorrected image. The pre-correction image after the image processing by the image correction unit 35 is executed is referred to as “post-correction image”. The difference value calculation unit 33 corresponds to the difference value calculation means in the present invention.

  A marker position calculation unit 37 is provided following the image correction unit 35. The marker position calculation unit 37 calculates the three-dimensional position information of the marker R on the subject M as needed based on the position of the marker R shown in the corrected image generated by the image correction unit 35. Based on the three-dimensional position information of the marker R, the marker position calculation unit 37 determines whether or not the current position of the affected part B is located in a predetermined range defined in advance. The image correction unit 35 corresponds to the image correction unit in the present invention.

  The therapeutic radiation irradiation apparatus 5 includes a radiation irradiation control unit 39 in addition to the therapeutic radiation source 5a. The radiation irradiation control unit 39 is connected to the therapeutic radiation source 5a, and controls the dose of the radiation G irradiated from the therapeutic radiation source 5a, the timing at which the radiation G is irradiated, and the like. The radiation irradiation control unit 39 outputs the radiation G from the therapeutic radiation source 5a when the marker position calculation unit 37 determines that the current position of the affected part B is located in a predetermined range (the therapeutic radiation irradiation range) defined in advance. Irradiate as therapeutic radiation.

  The X-ray fluoroscopic apparatus 7 further includes an input unit 41, a monitor 43, a storage unit 45, and a main control unit 47. The input unit 41 inputs an operator's instruction, and examples thereof include a keyboard input type, mouse input type, or touch input type panel. The monitor 43 displays a corrected image generated by the image correction unit 35, and the marker position calculation unit 37 displays the three-dimensional position information of the marker R to be calculated.

  The storage unit 45 stores information such as each of the corrected images generated by the image correction unit 35, an average pixel value profile, an approximate curve that approximates the average pixel value profile, and a difference value calculated by the difference value calculation unit 33. The main control unit 47 comprehensively controls each component such as the X-ray irradiation control unit 23, the image processing unit 25, the radiation irradiation control unit 39, the monitor 43, and the storage unit 45. The main controller 47 controls the movement of the X-ray tube 9 including the X-ray tubes 9a and 9b and the FPD 11 including the FPDs 11a and 11b via an imaging system moving mechanism (not shown).

<Description of operation>
Next, the operation of the radiotherapy apparatus 1 to which the X-ray fluoroscopic apparatus 7 according to the first embodiment is applied will be described. FIG. 3A is a flowchart for explaining an operation process of the radiotherapy apparatus 1 to which the X-ray fluoroscopic apparatus 7 according to the first embodiment is applied. In the first embodiment, a case will be described as an example in which radiation therapy is performed by the moving body tracking irradiation method using the radiation therapy apparatus 1. Further, the irradiation position at which the X-ray tubes 9a and 9b irradiate X-rays is the position shown in FIG.

  Note that the affected part B of the subject M shown in FIG. 1A is a part that periodically moves due to the body movement of the subject M, for example, a heart or a lung. In addition, it is assumed that a marker R is placed in the vicinity of the affected part B in advance. Examples of the configuration of the marker R include small pieces and granular materials made of a material having high radiopacity such as gold.

Step S1 (Generation of X-ray image)
In performing radiotherapy, first, a subject M that takes a supine posture is placed on the top 3. Then, in order to obtain position information of the affected part B that moves due to the breathing and pulsation of the subject M, that is, the body movement of the subject M, an X-ray image is generated by fluoroscopy. That is, the surgeon operates the input unit 41 to intermittently irradiate the X-ray tube 9a and the X-ray tube 9b with X-rays from different oblique directions. The content of the instruction input to the input unit 41 is transmitted to the X-ray irradiation control unit 23 via the main control unit 47. Under the control of the X-ray irradiation control unit 23, the X-ray tube 9a irradiates the subject M with X-ray Ha, and the X-ray tube 9b irradiates X-ray Hb.

  The X-ray Ha transmitted through the subject M is detected by each of the pixels arranged in the FPD 11a and converted into an electric signal. The X-ray Hb that has passed through the subject M is detected by each of the pixels disposed in the FPD 11b and converted into an electrical signal. Each of the FPDs 11a and 11b outputs an X-ray detection signal by sequentially reading the electrical signals converted by all the pixels in a predetermined direction (reading direction Re).

  Based on the X-ray detection signals output from the FPDs 11a and 11b, the image generation unit 27 intermittently forms X-ray images on which the X-ray images of the affected part B and the marker R are projected. In the first embodiment, generation of an X-ray image by X-ray fluoroscopy is performed at a frame rate of about 15 to 30 FPS, for example. Each of the X-ray images generated by the image generation unit 27 is hereinafter referred to as “pre-correction image C”. An uncorrected image C generated based on the X-ray detection signal output from the FPD 11a is referred to as an uncorrected image Ca, and an uncorrected image C generated based on the X-ray detection signal output from the FPD 11b is referred to as an uncorrected image Cb. To do. The generation of the pre-correction image C ends the step S1.

  Due to a difference in luminance value (pixel value) caused by a difference in the amount of incident scattered radiation, an artifact AF extending in a direction orthogonal to the readout direction Re may occur in each of the uncorrected images C (FIG. 4 ( a), (b)). FIG. 4A is an actual pre-correction image C, and the position where the artifact AF occurs is indicated by an arrow. FIG. 4B is a diagram schematically showing a strip-like or linear artifact AF generated in the pre-correction image C. The direction orthogonal to the reading direction Re, that is, the direction in which each of the artifacts AF extends is hereinafter referred to as the S direction.

  As shown in FIG. 4B, the artifact AF extending in the S direction is generated, so that the visibility of the affected part B and the marker R reflected in the pre-correction image C is significantly lowered. Therefore, image processing for removing the artifact AF from the pre-correction image C is performed by the steps S2 to S5 characteristic of the present invention.

Step S2 (Generation of average pixel value profile)
When performing the image processing of the pre-correction image C, first, an average pixel value profile is generated based on the pixel value information of the pre-correction image C. Information on pixel values of all the pixels constituting the uncorrected image P is transmitted from the image generation unit 27 to the average value calculation unit 29. The average value calculation unit 29 generates an average pixel value profile for the reading direction Re by performing a calculation process that averages the pixel values of the pre-correction image C in the direction in which the artifact AF extends, that is, the S direction.

  Here, the process of generating the average pixel value profile will be described in detail. As shown in FIG. 5, for a pixel 1024 (vertical 1024 × horizontal 1024) constituting the pre-correction image C, a pixel located in m rows and n columns is defined as a pixel D (m, n). A pixel value E (m, n) is assumed to be a pixel value E in the pixel D (m, n). In the first embodiment, the S direction coincides with the vertical column direction.

  The average value calculation unit 29 averages the pixel values E from the pixel value E (1, n) to the pixel value E (1024, n) for the 1024 pixels D constituting the n column, thereby obtaining the n-th column. An average pixel value A (n) for is calculated. In the first embodiment, the average value calculator 29 averages the pixel values E by averaging the pixel values E for the pixel columns from the pixel value (1, n) to the pixel value E (1024, n). The average pixel value A (n) is calculated. For the calculation for averaging, a known method such as a geometric average in addition to the addition average may be appropriately used.

  A group A (1) of 1024 average pixel values A obtained for each pixel column in the readout direction Re by performing such an arithmetic operation on each pixel column from the first column to the 1024th column. -A (1024) is calculated. The average value calculation unit 29 generates an average pixel value profile P by plotting the group A (1) to A (1024) of average pixel values (see FIG. 6). The vertical direction of the average pixel value profile P indicates the average pixel value A, and the horizontal direction indicates the position (nth column) of the pixel D corresponding to the average pixel value A in the readout direction Re.

  Since the reading direction Re is a predetermined direction according to the standard of the X-ray fluoroscopic apparatus 7, the S direction is also determined in advance. Therefore, the direction in which the average value calculation unit 29 performs the addition averaging is also determined in advance, so that the average value calculation unit 29 can generate the average pixel value profile P automatically and quickly by generating the pre-correction image C. By averaging the pixel values of all the pixels of the correction image C in the S direction and generating an average pixel value profile P, the process of step S2 ends. Step S2 corresponds to the image averaging step in the present invention.

  Since the artifact AF extends in the S direction, when the artifact AF occurs in the pixel D in the nth column, the pixel values E in the nth column are all compared to the pixel value E of the pixel D in which no artifact AF occurs. It changes by the same value. Therefore, in the average pixel value profile P obtained by averaging in the S direction, it is considered that a particularly prominent artifact AF occurs in the portion where the average pixel value A protrudes (see the symbol L in FIG. 6).

Step S3 (calculation of approximate curve)
Information on the average pixel value profile P generated for the pre-correction image C is transmitted from the average value calculation unit 29 to the approximate expression calculation unit 31. The approximate expression calculation unit 31 calculates an approximate curve Q that approximates the average pixel value profile P (see FIG. 7). In the first embodiment, the approximate expression calculation unit 31 uses the least square method for the data of the average pixel value profile P, thereby calculating the n-order polynomial F (x) as the approximate curve Q that approximates the average pixel value profile P. To do. The order n of the polynomial F (x) that is the approximate curve Q may be appropriately selected according to the conditions. The value of the approximate curve Q relating to the pixel D in the n-th column is defined as Q (n). The value of Q (n) corresponds to F (n) that is a value obtained by substituting n into the nth order polynomial F (x).

  The n-th order polynomial F (x) is data obtained by smoothly reflecting the tendency of the average pixel value profile P and smoothing the average pixel value profile P. Therefore, in the average pixel value profile P, the portion L where the average pixel value A protrudes due to the artifact AF is suitably smoothed in the nth-order polynomial F (x) that is the approximate curve Q. By calculating the approximate curve Q that approximates the data of the average pixel value profile P for all the pixels in the reading direction Re, the process of step S3 is completed. Step S3 corresponds to the approximate curve generation step in the present invention.

Step S4 (calculation of difference value)
Information on the approximate curve Q, that is, information on the nth order polynomial F (x) is transmitted from the approximate expression calculation unit 31 to the difference value calculation unit 33 together with information on the average pixel value profile P. The difference value calculation unit 33 calculates the difference value J by taking the difference between the value of the average pixel value profile P and the value of the approximate curve Q. As an example, the value of the difference value J (n) in the pixel D in the n-th column is the value of the average pixel value profile P in the n-th pixel column, that is, the average pixel value An, and the approximate curve in the n-th pixel column It is calculated by the difference between the value of Q, that is, Q (n). Step S4 corresponds to the difference value calculation step in the present invention.

  Thus, the difference value calculation unit 33 calculates the difference values J (1) to J (1024) for each of the first column to the 1024th column. Each of the calculated difference values J (1) to J (1024) information is transmitted from the difference value calculation unit 33 to the image correction unit 35. Further, in synchronization with the transmission of the information on the difference value J, the information on the pre-correction image C is transmitted from the image generation unit 27 to the image correction unit 35. When the difference value calculation unit 33 calculates the difference value J for each of the pixel columns aligned in the readout direction Re and extending in the S direction, the process of step S4 ends.

Step S5 (X-ray image correction)
The image correction unit 35 corrects the pre-correction image C by subtracting the difference value J from the pixel values of all the pixels constituting the pre-correction image C. Specifically, the difference value J (1) calculated by the difference value calculation unit 33 for the pixel D in the first column from each of the pixel values E (1, 1) to E (1024, 1) in the pixel D in the first column. ) Is subtracted. Then, the difference value J (n) calculated by the difference value calculation unit 33 for the pixel D in the n-th column is subtracted from each of the pixel values E (1, n) to E (1024, n) in the pixel D in the n-th column. To do.

  In this way, for each of the first to 1024th columns, the image correction unit 35 subtracts the difference value J calculated for each pixel column from the pixel value E for each pixel D, so that the image correction unit 35 can determine from the pre-correction image C. Correction for subtracting the difference value J is performed. Hereinafter, the image C before correction after being corrected by the image correction unit 35 is referred to as “image T after correction”. By the correction process executed by the image correction unit 35, the average pixel value profile Pt in the corrected image T becomes equal to the approximate curve Q, that is, the nth order polynomial F (x) (see FIG. 8A).

  That is, the protruding portion in the average pixel value profile P in the pre-correction image C is smoothed in the average pixel value profile Pt in the post-correction image T (see FIG. 8A, arrow L). As a result, the artifact AF generated in the pre-correction image C is removed in the post-correction image T (see FIGS. 8B and 8C). FIG. 8B is a diagram illustrating the actual corrected image T, and FIG. 8C is a diagram schematically illustrating the corrected image T. The corrected image T is displayed on the monitor 43. By generating the corrected image T, the process of step S5 ends. Step S5 corresponds to an image correction step in the present invention.

  The approximate curve Q is obtained by further smoothing the data of the average pixel value profile P obtained by averaging the pixel values of the pre-correction image C in the S direction. As the dose of scattered radiation incident on the pixel D increases, the difference between the pixel value E output from the pixel D and the pixel value based on the actual X-ray image of the subject M increases. As a result, a more prominent artifact AF occurs in a pixel that is more affected by scattered radiation. Therefore, the difference between the difference value, that is, the value of the average pixel value profile P and the value of the approximate curve Q becomes large in the pixel where the artifact AF is conspicuous.

  In addition, the dose of scattered radiation incident on the pixel D changes according to the time difference at which the FPD reads the pixel. Therefore, a difference in pixel values (luminance difference) based on the incident dose of scattered radiation occurs between adjacent pixels in the readout direction Re. That is, the amount of change in the pixel value based on the amount of incident scattered radiation in each of the pixels D (m, 1) to (m, 1024) in the m-th row is different. On the other hand, the change amount of the pixel value based on the amount of incident scattered radiation in each of the pixels D (1, n) to (1024, n) in the n-th column may be considered constant.

  Therefore, by subtracting the difference value J (n) calculated for the n-th pixel column from each of the pixel values E of the pixels D in the n-th column, the pixels D constituting the n-th pixel column The correction for removing the artifact AF can be performed for all of the above. Then, by subtracting the difference value J calculated for each pixel column (first column to 1024 column) in the readout direction Re, image processing that suitably removes the artifact AF is performed on the entire uncorrected image C. Can be executed.

  As a result of the image processing constituted by the steps S2 to S5 as described above, the artifact AF generated in the pre-correction image C is preferably removed, and the post-correction image in which the visibility of the affected part B and the marker R is improved. T is generated. The image processing steps according to steps S2 to S5 are performed on each of the pre-correction image Ca based on the output signal of the FPD 11a and the pre-correction image Cb based on the output signal of the FPD 11b. An image corrected by the image correction unit 35 with respect to the pre-correction image Ca is referred to as a post-correction image Ta, and an image corrected by the image correction unit 35 with respect to the pre-correction image Cb is referred to as a post-correction image Tb.

Step S6 (calculates the marker position)
After image processing for removing artifacts is performed, the position of the marker is calculated using the corrected image T. That is, the image information of the corrected image Ta and the corrected image Tb is transmitted from the image correction unit 35 to the marker position calculation unit 37, respectively. The marker position calculation unit 37 calculates the three-dimensional position information of the marker R in the subject M based on the position of the marker R displayed in each of the corrected image Ta and the corrected image Tb.

  Since the X-ray tube 9a and the X-ray tube 9b emit X-rays from different oblique directions, each of the corrected image Ta and the corrected image Tb is an image in which the marker R is projected from different directions. Therefore, the marker position calculation unit 37 can calculate the three-dimensional position information of the marker R using the corrected image Ta and the corrected image Tb. Further, since the artifact AF is preferably removed from each of the corrected image Ta and the corrected image Tb, the visibility of the marker R is greatly improved. Therefore, the marker position calculation unit 37 can calculate the three-dimensional position information of the marker R more accurately by the image processing steps from step S2 to S5. By calculating the three-dimensional position information of the marker R, the process according to step S6 ends.

Step S7 (determination of therapeutic radiation irradiation)
The marker position calculation unit 37 further determines whether or not the radiation G, which is therapeutic radiation, should be irradiated based on the three-dimensional position information of the marker R. The position of the affected part B in the subject M periodically moves according to the body movement of the subject M. Since the relative position between the marker R placed in the vicinity of the affected part B and the affected part B is unchanged, the three-dimensional position of the marker R is periodically displaced according to the movement of the affected part B caused by the body movement of the subject M. .

  When the three-dimensional position of the marker R is within the range of the therapeutic radiation irradiation range V, which is a position suitable for therapeutic radiation irradiation, the marker position calculation unit 37 determines that the radiation G should be irradiated. On the other hand, when the three-dimensional position of the marker R is outside the therapeutic radiation irradiation range V, the marker position calculation unit 37 determines that the irradiation of the radiation G should be stopped.

  That is, as shown in FIG. 9A, when the marker R shown in the corrected image T is within the range of the therapeutic radiation irradiation range V defined in advance, the marker position calculation unit 37 is connected via the main control unit 47. The control signal of the content which irradiates the radiation G to the radiation irradiation control part 39 is transmitted. The radiation irradiation control unit 39 irradiates the subject M with the radiation G from the treatment radiation source 5a according to the control signal. In this case, since the affected part B has moved to a position where the radiation G is suitably irradiated, radiotherapy with the radiation G can be performed effectively.

  As shown in FIG. 9B, when the marker R shown in the corrected image T is outside the range of the therapeutic radiation irradiation range V defined in advance, the marker position calculation unit 37 receives the radiation via the main control unit 47. A control signal for stopping the radiation G is transmitted to the irradiation control unit 39. The radiation irradiation control unit 39 controls the therapeutic radiation source 5a according to the control signal so as to stop the irradiation of the radiation G. In this case, since the position of the affected part B is out of the range in which the radiation G is suitably irradiated, exposure by unnecessary radiation G can be avoided.

  Determination of therapeutic radiation irradiation is performed based on the three-dimensional position information of the marker R calculated using each of the corrected images T. Since the visibility of the corrected image T is improved by image processing that suitably removes the artifact AF, the marker position calculation unit 37 can more accurately execute the treatment radiation irradiation determination. By determining whether or not therapeutic radiation can be irradiated, the process of step S7 ends.

  After completion of step S7, it is determined whether or not to continue fluoroscopy and the process branches. Further, when X-ray fluoroscopy is continued to track the marker R, the process returns to step S1 and steps S1 to S7 are repeated. That is, the image processing according to steps S2 to S5 is performed on each of the uncorrected images C generated at the frame rate of 15 to 30 FPS in step S1, and the corrected image T is generated. In steps S6 to S7, the position of the marker R is calculated on the basis of the corrected image T, and whether or not the radiation G can be irradiated is determined. If no further radiotherapy is performed, fluoroscopy is stopped and all steps are terminated.

<Effects of Configuration of Example 1>
The X-ray fluoroscopic apparatus 7 according to the first embodiment includes an average value calculation unit 29, an approximate expression calculation unit 31, a difference value calculation unit 33, and an image correction unit 35. The average value calculator 29 generates an average pixel value profile P by averaging pixel values in a predetermined direction S in which the artifact AF extends with respect to the pre-correction image. The approximate expression calculation unit 31 calculates an approximate curve Q that approximates the entire average pixel value profile P by performing polynomial approximation. The difference value calculation unit 33 extends in the extension direction S of the artifact AF, and calculates the difference between the value of the average pixel value profile P and the value of the approximate curve Q as a difference value for each of the pixel columns arranged in the readout direction Re.

  The average pixel value profile P reflects the tendency of the pixel values of the entire image before correction in the reading direction Re. The approximate curve Q is obtained by appropriately smoothing the average pixel value profile P while reflecting the tendency of the entire average pixel value profile P. The difference value calculated by the difference value calculation unit 33 for each pixel row extending in the S direction reflects the magnitude of the influence of the scattered radiation that is the cause of the artifact on the pixel value. The image correcting unit 35 subtracts the difference value from each pixel value in the pre-correction image, so that each pixel value in the pre-correction image is preferably smoothed. As a result, the artifact can be suitably removed from the pre-correction image by the image correction performed by the image correction unit 35.

  When removing the artifact extending in the predetermined direction S, the conventional image processing method determines whether or not the difference between the pixel values adjacent to the readout direction Re that is orthogonal to the S direction exceeds a predetermined threshold value. To do. Then, when the difference between the pixel values exceeds the threshold value, correction is performed to smooth the pixel values of the adjacent pixels. However, in such a conventional method, it is necessary to perform an arithmetic processing for taking a difference in all the combinations of adjacent pixels in the readout direction Re and comparing it with a threshold value, so that the time required for image processing becomes long. In addition, when the threshold value to be set is high, only very conspicuous artifacts can be removed, and image correction becomes insufficient. On the other hand, if the threshold value to be set is low, the brightness difference is smoothed more than necessary, and as a result, the brightness difference based on the actual X-ray image of the subject M is smoothly corrected, so the accuracy of the X-ray image information is significantly reduced. .

  Therefore, in the configuration according to the first embodiment, correction is performed by subtracting the difference value J between the average pixel value profile P and the approximate curve Q from the pre-correction image. By generating the average pixel value profile P, information about the tendency of the pixel values in the entire pre-correction image can be acquired collectively. Then, by calculating a difference value J from the approximate curve Q obtained by smoothing the entire average pixel value profile P, information regarding the magnitude of the influence of incident scattered radiation can be accurately obtained for each of the pixel columns arranged in the readout direction Re. Can be acquired promptly. As a result, the time required for image processing for removing artifacts from the pre-correction image can be shortened. By subtracting the difference value J from the pre-correction image, it is possible to correct the change amount of the pixel value caused by the incident scattered radiation more accurately regardless of the size.

  In the post-correction image generated by subtracting the difference value from the pre-correction image, the average pixel value profile Pt obtained by averaging in the S direction is equal to the approximate curve Q. That is, in the average pixel value profile P of the pre-correction image, the protruding portion due to the incident scattered radiation is preferably smoothed, and the tendency of the entire average pixel value profile P is maintained. Therefore, it is possible to suitably exclude the influence of the incident scattered radiation regardless of the size while maintaining the information of the pixel value based on the actual X-ray image of the subject M. Therefore, it is possible to improve the accuracy of the image information in the post-correction image while more accurately removing artifacts from the pre-correction image.

  In the configuration according to the first embodiment, an effect of suitably removing the entire wide band-shaped artifact can be expected. Such an effect will be described below with reference to FIG. Here, as shown in FIG. 10A, the band-shaped artifact AF indicated by the oblique lines is generated in a wide range from the pixel D in the n-th column to the pixel D in the (n + 5) -th column in the pre-correction image C. Shall.

  In the conventional image processing method, whether or not correction is possible is determined by comparing the difference between the pixel values E of pixels adjacent to the direction S in which the artifact AF extends, that is, the direction Re, with a predetermined threshold value. As an example, when determining each pixel D in the m-th row, the pixel value E in the m-th row is calculated. FIG. 10B schematically graphs the pixel value E of the pixel D in the m-th row.

  As shown in FIG. 10B, the difference in pixel value E between adjacent pixels is large at the end of the artifact AF. That is, the difference between the pixel value E (m, n) and the pixel value E (m, n-1) in the pixel D (m, n), the pixel value E (m, n) and the pixel value (m, n + 1) The difference between is generally greater than the threshold. Therefore, in the pixel D (m, n) located at the end of the artifact AF, the pixel value E is corrected also in the conventional image processing method.

  However, in the central part of the strip-shaped artifact AF, the difference in pixel value E between adjacent pixels is very small. That is, the difference between the pixel value E (m, n + 1) and the pixel value E (m, n + 2) and the difference between the pixel value E (m, n + 2) and the pixel value (m, n + 3) are generally smaller than the threshold value. . Therefore, in the conventional method, the pixel value E is corrected for the pixel D (m, n + 2), the pixel D (m, n + 3), and the like located in the center regardless of the occurrence of the artifact AF. Absent. As a result, the wide belt-like artifact AF cannot be sufficiently removed.

  As shown in FIG. 10A, the range in which the artifact AF occurs in the pixel D in the (n + 5) th column is relatively narrow. Therefore, depending on the size of the threshold value to be set, the difference between the pixel value E (m, n + 5) and the pixel value E (m, n + 6) is below the threshold value. As a result, the correction of the pixel value E is not executed in the pixel D (m, n + 5), despite the occurrence of the artifact AF. As described above, it is difficult to appropriately remove the artifact AF by the conventional image processing method. In order to remove the artifact AF, it is necessary to appropriately set an appropriate threshold value for each frame of the X-ray image. Therefore, the surgeon receives a great burden to set the threshold appropriately, and the operator is required to have experience to set an appropriate threshold.

  On the other hand, in the image processing according to the first embodiment, the average pixel value profile P and the approximate curve Q of the pre-correction image C are calculated. Then, a difference value J between the value of the average pixel value profile P and the value of the approximate curve Q is calculated as a value reflecting the magnitude of the influence of the artifact AF based on the scattered radiation. FIG. 10C is a graph schematically showing the average pixel value profile P and the approximate curve Q of the image C before correction.

  As shown in FIG. 10C, the influence of the artifact AF is large in the pixels D from the (n + 1) th column to the (n + 4) th column. Therefore, the difference value J between the value of the average pixel value profile P and the value of the approximate curve Q is large in any pixel column from the (n + 1) th column to the (n + 4) th column. Therefore, since the pixel value E of the pixel D is greatly corrected not only at the end portion of the artifact AF but also at the center portion, the entire wide artifact AF can be suitably removed from the pre-correction image C.

  Further, in the pixel D in the nth column or the (n + 5) th column, the difference value J decreases as the influence of the artifact AF decreases. Then, the difference value J is almost 0 in the pixel D in which the artifact AF does not occur as in the (n−1) th column. Accordingly, by subtracting the difference value J from the pixel value E in the pre-correction image C for each pixel row in the direction Re, the pixel value E of each pixel D can be appropriately corrected according to the magnitude of the influence of the artifact AF. .

  The difference value J that reflects the magnitude of the influence of the artifact AF is automatically and promptly calculated for all pixel columns in the readout direction Re every time the pre-correction image C is generated. Since the image processing in the first embodiment is completed simply by subtracting the difference value J from the pixel value E in the pre-correction image C, it is not necessary to appropriately set a threshold value for each image frame as in the conventional image processing. Therefore, no experience is required for the operator, and the burden on the operator can be greatly reduced.

  Next, Embodiment 2 of the present invention will be described with reference to the drawings. The overall configuration of the X-ray fluoroscopic apparatus according to Embodiment 2 is the same as the overall configuration of Embodiment 1 shown in FIGS. 1 and 2. However, Example 1 and Example 2 are partially different from each other in the operation process of the fluoroscopic imaging apparatus. FIG. 3B is a flowchart for explaining an operation process of the X-ray fluoroscopic apparatus according to the second embodiment. Here, the average pixel value profile P calculated by averaging the pre-correction images C in the kth frame is referred to as “average pixel value profile Pk”. An approximate curve that approximates the entire average pixel value profile Pk is referred to as an “approximate curve Qk”.

  In the first embodiment, for each of the pre-correction images C generated in step S1, the average value calculator 29 generates an average pixel value profile P, and the approximate expression calculator calculates an approximate curve Q. That is, the average pixel value profile Pk and the approximate curve Qk corresponding to the k-th image before correction C are calculated. Then, the difference value calculation unit 33 calculates the difference value J by taking the difference between the value of the average pixel value profile Pk and the value of the approximate curve Qk.

  On the other hand, in the second embodiment, the average value calculation unit 29 generates an average pixel value profile P for each of the uncorrected images C. However, even when image processing is performed on the pre-correction image C in the second and subsequent frames, the approximate expression calculation unit 31 applies the approximate curve Q1 calculated based on the pre-correction image C1 in the first frame. That is, the difference value calculation unit 33 calculates the difference value J by taking the difference between the value of the average pixel value profile Pk and the value of the approximate curve C1 in the image processing for the pre-correction image C in the k-th frame.

  As shown in FIG. 3A, in the first embodiment, every time the pre-correction image C is generated intermittently by X-ray fluoroscopy, the steps S1 to S7 are repeated. On the other hand, in the second embodiment, as shown in FIG. 3B, when the pre-correction image C of the first frame is generated, steps S1 to S7 are executed. When the pre-correction image C for the second and subsequent frames is generated, the step S3 is omitted. That is, after the average pixel value profile P is generated in step S2, the process proceeds to step S4 to calculate a difference value.

<Effects of Configuration of Example 2>
In the X-ray fluoroscopic apparatus according to the second embodiment, the approximate curve Q1 calculated for the pre-correction image of the first frame is applied to the pre-correction images of the second and subsequent frames. In this case, the process of calculating the approximate curve Q by performing polynomial approximation that approximates the entire average pixel value profile can be omitted in the second and subsequent frames. As a result, the time required for image processing for removing artifacts from the pre-correction image can be greatly reduced.

  When radiotherapy is performed by the moving body tracking irradiation method, X-ray fluoroscopy is performed in a state where the position of the imaging system is fixed at a predetermined position, and a plurality of pre-correction images are generated. Therefore, in the X-ray fluoroscopic apparatus, when the incident pattern of scattered rays with respect to the FPD is constant, the regions where the artifact AF is likely to occur in each of the pre-correction images C are substantially the same. Therefore, even when the approximate curve Q1 calculated for the pre-correction image of the first frame is used for image processing of the pre-correction images of the second and subsequent frames, the strip-like artifact AF can be suitably removed from the pre-correction image. .

  In the configuration according to the second embodiment, the time required for image processing of the pre-correction image can be reduced, so even if the X-ray fluoroscopy frame rate is increased, the strip-like artifact AF is quickly removed from the pre-correction image. A corrected image T with high visibility can be generated efficiently. Therefore, by using the corrected image T having a higher frame rate, the position of the moving marker can be tracked with higher accuracy, so that the therapeutic effect of the radiation irradiation apparatus 1 can be further improved.

  Next, Embodiment 3 of the present invention will be described with reference to the drawings. As shown in FIG. 11, the X-ray fluoroscopic apparatus 7A according to the third embodiment is different from the configuration of the X-ray fluoroscopic apparatus 7 according to the first embodiment in that it further includes a weighting processing unit 49.

<Configuration Characteristic of Example 3>
The weighting processing unit 49 is provided after the approximate expression calculation unit 31 and is provided before the difference value calculation unit 33. The weighting processing unit 49 newly generates an approximate curve used for correcting the most recently generated pre-correction image by performing an arithmetic processing of appropriately weighting and adding each of the approximate curves generated by the approximate expression calculating unit 31. . A configuration in which the weighting processing unit 49 performs weighting calculation processing includes a recursive filter. The weighting processing unit 49 corresponds to the weighting processing means in the present invention.

  FIG. 3C is a flowchart for explaining an operation process of the X-ray fluoroscopic apparatus according to the third embodiment. The third embodiment is common to the first embodiment with respect to steps S1 to S3. That is, for each pre-correction image C generated in the k-th frame in step S1, the average value calculation unit 29 generates an average pixel value profile Pk by averaging the pixel values in the S direction (step S2). Then, the approximate expression calculation unit 31 calculates a high-order polynomial that approximates the entire average pixel value profile Pk as an approximate curve Qk by polynomial approximation using the mean square method (step S3).

  Here, the process branches according to the number of frames of the pre-correction image C. That is, when the image generation unit 27 generates the pre-correction image C of the first frame, the process proceeds from step S3 to step S4, and the processes from step S4 to S7 are executed similarly to the first embodiment. On the other hand, when the image generation unit 27 generates the pre-correction image C for the second and subsequent frames, the process proceeds to step S3-2 after step S3 ends. Here, the step S3-2 characteristic of the third embodiment will be described.

Step S3-2 (weighting process)
After the approximate curve Qk related to the image C before correction in the k-th frame is calculated in step S3, information about the approximate curve Qk is transmitted from the approximate expression calculation unit 31 to the weighting processing unit 49. In addition, the weighting processing unit 49 transmits information about approximate curves Q1 to Q (k-1) related to the respective pre-correction images C from the first frame to the (k-1) th frame.

  The weighting processing unit 49 newly calculates an approximate curve that is actually used to correct the pre-correction image C of the kth frame by appropriately performing a weighting process based on the approximate curves Q1 to Qk. Hereinafter, the approximate curve calculated by the weighting process performed by the weighting processing unit 49 is referred to as a “post-process approximate curve Z”. The post-processing approximate curve Z used for correcting the pre-correction image C of the k-th frame is hereinafter referred to as “post-processing approximate curve Zk”.

An arithmetic processing method related to the weighting process for calculating the post-processing approximate curve Zn may be appropriately selected. As an example, the post-processing approximate curve Zk is calculated by using the constant α satisfying 0 <α <1, the approximate curve Q (k−1), and the approximate curve Qk, using the following equation (1). The For example, the value of α is preferably about 0.2 to 0.3.
Zk = (1−α) · Q (k−1) + α · Qk (1)

  Note that the mathematical formula used to calculate the post-processing approximate curve Zk is not limited to the approximate curve Q, and has been applied to past image frames before correction, such as Z (k-1) instead of Q (k-1). The formula of the approximate curve Z after processing may be used. A plurality of mathematical expressions from Q1 to Q (k-1) and Z1 to Z (k-1) may be selected and used to calculate the post-processing approximate curve Zk. Information on the processed approximate curve Zn is transmitted from the weighting processing unit 49 to the difference value calculation unit 33. When the weighting processing unit 49 calculates the processed approximate curve Zn, the process of step S3-2 ends, and the process proceeds to step S4.

Step S4 (calculation of difference value)
In step S4 according to the first embodiment, the difference value calculation unit 33 calculates a difference between the value of the average pixel value profile Pk and the value of the approximate curve Qk, thereby subtracting and correcting the pre-correction image C in the k-th frame. The difference value J is calculated. On the other hand, in step S4 according to the third embodiment, the difference value calculation unit 33 subtracts the value of the average pixel value profile Pk and the value of the processed approximate curve Zk from the pre-correction image C in the k-th frame. A difference value J for correction is calculated.

  Example 3 and Example 1 are common about the process after step S5. That is, the image correction unit 35 generates the corrected image T by subtracting the difference value J calculated in step S4 from each pixel value of the pre-correction image C (step S5). The marker position calculation unit 37 calculates the three-dimensional position information of the marker R using the corrected image T based on each of the FPDs 11a and 11b (step S6). Furthermore, the marker position calculation unit 37 determines whether or not the radiation G, which is therapeutic radiation, can be irradiated based on whether the marker R is within the therapeutic radiation irradiation range V (step S7).

  The X-ray fluoroscopic apparatus 7A according to the third embodiment includes a weighting processing unit 49, and generates a processed approximate curve by appropriately performing a weighting process on each generated approximate curve. Then, the difference value calculation unit 33 calculates the difference value J by taking the difference between the value of the average pixel value profile Pk and the value of the processed approximate curve Zk, and the image correction unit 35 calculates the pixel value of the pre-correction image C. The corrected image T is generated by subtracting the difference values J from each other.

<Effects of Configuration of Example 3>
In the configuration according to the third embodiment, the visibility of the entire X-ray image group obtained intermittently by X-ray fluoroscopy, that is, the entire moving image, even when the luminance pattern changes greatly between image frames. Can be avoided. In X-ray fluoroscopy, for example, a large number of X-ray images are generated at a high frame rate of about 15 to 30 FPS, and the generated X-ray image groups are displayed intermittently. The surgeon performs various treatments with reference to a group of X-ray images displayed as a so-called moving image.

  In radiotherapy using the moving body tracking irradiation method, X-ray fluoroscopy is performed using the affected area B that periodically moves due to body movement of the subject as a region of interest. Therefore, as shown in the upper part of FIG. 12, the position of the affected part B shown in the k-th pre-correction image C (upper left figure) and the affected part B shown in the (k + 1) -th pre-correction image C (upper right figure). May be different in the reading direction Re. In this case, as shown in the middle stage of FIG. 12, the pattern of the approximate curve Qk based on the k-th pre-correction image C (middle stage left figure) and the approximate curve Q (based on the pre-correction image C of the (k + 1) th frame This is significantly different from the pattern of k + 1) (middle stage right diagram).

  In the image processing according to the first embodiment, since the difference value J is calculated using the approximate curve Q calculated for each frame of the pre-correction image C as it is, the average pixel value profile Pt related to the post-correction image T is the approximate curve Q. (Fig. 8 (a)). Therefore, when the position of the affected part B moves at high speed in the direction Re, in the image processing according to the first embodiment, the pixel value pattern in the corrected image T in the k frame and the pixel in the corrected image T in the (k + 1) frame. The value pattern is very different. As a result, a situation occurs in which the X-ray image suddenly changes between image frames in the entire moving image.

  On the other hand, in the image processing according to the third embodiment, the post-processing approximate curve Z is calculated by appropriately performing weighting processing on each of the approximate curves Q calculated for each frame of the pre-correction image C. In such a configuration, even if the pattern of the approximate curve Qk and the pattern of the approximate curve Q (k + 1) are greatly different, the pattern of the processed approximate curve Z (k + 1) calculated by the weighting process is the approximate curve. The difference from the Qk pattern becomes smaller (lower line in FIG. 12, dotted line).

  The difference value calculation unit 33 calculates a difference value J by taking the difference between the value of the average pixel value profile Pk and the value of the processed approximate curve Zk, and the image correction unit 35 calculates a difference from each pixel value of the pre-correction image C. The corrected image T is generated by subtracting the value J. Accordingly, in the third embodiment, the average pixel value profile Pt (k + 1) based on the corrected image T in the (k + 1) th frame matches the post-processing approximate curve Z (k + 1).

  As a result, the difference between the pixel value pattern in the corrected image T in the k-th frame and the pixel value pattern in the corrected image T in the (k + 1) -th frame is smaller. Can be avoided. Therefore, in the third embodiment, the visibility of the entire moving image can be further improved.

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

  (1) In each of the above-described embodiments, the method for calculating the n-th order polynomial as the approximate curve Q is not limited to the least square method, and a known calculation method such as Legendre polynomial approximation or Hermite polynomial approximation may be used. The configuration of the approximate curve Q is not limited to an nth order polynomial as long as it approximates the average pixel value profile P, and may be another approximate curve such as a moving average.

  In the case of approximation using the moving average method, a section F that takes a moving average is set so as to favorably reflect the tendency of the average pixel value profile P. Based on the F consecutive average pixel values A, the moving average is calculated as the approximate curve Q. However, the method of approximating with the n-order polynomial is particularly preferable in that the plot data of all the average pixel values A in the readout direction Re can be approximated more accurately and quickly.

  (2) In each of the above-described embodiments, the radiation therapy apparatus 1 using the moving body tracking imaging method including the therapeutic radiation irradiation apparatus 5 has been described as an example. However, the configuration of the image processing according to each embodiment is arbitrary X-ray fluoroscopy. It can be applied to a photographing apparatus. That is, you may apply to the general X-ray fluoroscopic imaging apparatus etc. which abbreviate | omitted structures, such as the therapeutic radiation irradiation apparatus 5 and the rail 15, from the radiotherapy apparatus 1. FIG. More specifically, the present invention can be applied to an X-ray fluoroscopic apparatus having a set of imaging systems in which an X-ray tube 9 is provided at one end of a C-type arm and an FPD 11 is provided at the other end of the C-type arm.

  (3) In each of the above-described embodiments, two sets of imaging systems including the X-ray tube 9 and the FPD 11 are provided. However, the number of imaging systems may be increased or decreased as appropriate. In addition to a configuration for acquiring an X-ray image by X-ray fluoroscopy in which X-rays with relatively low doses are intermittently irradiated, a configuration for acquiring an X-ray image by X-ray imaging with X-rays with relatively high doses. The image processing according to each embodiment can be applied to the above.

  (4) In each of the above-described embodiments, the case where the artifact AF extends in the direction orthogonal to the reading direction Re is described as an example. However, the direction in which the artifact AF generated in the pre-correction image C extends is the reading direction Re. The direction does not have to be orthogonal to. In this case, the average value calculation unit 29 can generate the average pixel value profile P by averaging the pixel values of the pre-correction image C in the direction in which the artifact AF extends.

DESCRIPTION OF SYMBOLS 1 ... Radiation therapy apparatus 3 ... Top plate 5 ... Treatment radiation irradiation apparatus 7 ... X-ray fluoroscopic imaging apparatus 9 ... X-ray tube 11 ... FPD (X-ray detection means)
DESCRIPTION OF SYMBOLS 21 ... Collimator 23 ... X-ray irradiation control part 25 ... Image processing part 27 ... Image generation part 29 ... Average value calculating part 31 ... Approximation formula calculating part (approximate curve calculating means)
33 ... Difference value calculation unit (difference value calculation means)
35 ... Image correction unit (image correction means)
37 ... Marker position calculation unit 39 ... Radiation irradiation control unit 47 ... Main control unit 49 ... Weighting processing unit (weighting processing means)

Claims (8)

An X-ray conversion step in which an X-ray detector detects X-rays irradiated to the subject from the X-ray tube and converts them into an X-ray detection signal;
An image generation step of generating an X-ray image using the X-ray detection signal converted in the X-ray conversion step;
The pixel value of the X-ray image generated in the image generation step is averaged in a direction orthogonal to the readout direction of the X-ray detector, in which a band-shaped artifact generated in the X-ray image extends, and an average pixel value An image averaging process to generate a profile;
An approximate curve generating step for calculating an approximate curve that approximates the entire average pixel value profile generated in the image averaging step;
A difference value calculating step of calculating a difference between the value of the approximate curve generated in the approximate curve generating step and the value of the average pixel value profile as a difference value;
An image processing method comprising: an image correction step of correcting the X-ray image by subtracting the difference value calculated in the difference value calculation step from each pixel value in the X-ray image. The image processing method according to claim 1,
In the case of generating the X-ray image of the second and subsequent frames, the approximate curve generation step is omitted,
In the difference value calculating step, a difference between the value of the average pixel value profile calculated for the X-ray image and the value of the approximate curve generated in the X-ray image of the first frame is calculated as the difference value. Image processing method. The image processing method according to claim 1,
A weighting process step of calculating a processed approximate curve by performing a weighting process on the approximate curve generated most recently and the approximate curve generated in the past after the approximate curve generating step;
In the difference value calculation step, an image processing method for calculating a difference between the value of the average pixel value profile and the value of the post-process approximation curve calculated in the weighting processing step as the difference value. The image processing method according to any one of claims 1 to 3,
The approximate curve generation step is an image processing method for calculating, as the approximate curve, an nth order polynomial that approximates the entire average pixel value profile. An X-ray tube that irradiates the subject with X-rays;
X-ray detection means for detecting X-rays transmitted through the subject and outputting an X-ray detection signal;
An image generation unit that generates an X-ray image using an X-ray detection signal output by the X-ray detection unit;
Image averaging means for averaging the pixel values of the X-ray image in a direction orthogonal to the readout direction of the X-ray detection means, and generating an average pixel value profile, in which strip-like artifacts generated in the X-ray image are stretched ; ,
An approximate curve generating means for calculating an approximate curve that approximates the entire average pixel value profile;
A difference value calculation means for calculating a difference between the value of the average pixel value profile and the value of the approximate curve as a difference value;
An X-ray fluoroscopic imaging apparatus comprising: an image correcting unit that corrects the X-ray image by subtracting the difference value from each pixel value in the X-ray image. The X-ray fluoroscopic apparatus according to claim 5,
The approximate curve generation means calculates the approximate curve only for the X-ray image of the first frame,
The X-ray fluoroscopic apparatus, wherein the difference value calculation means calculates a difference between the value of the average pixel value profile and the value of the approximate curve calculated for the X-ray image of the first frame as the difference value. The X-ray fluoroscopic apparatus according to claim 5,
Weighting processing means for calculating a processed approximate curve by weighting each of the approximate curves calculated by the approximate curve generating means;
The difference value calculation means is an X-ray fluoroscopic apparatus that calculates a difference between a value of the average pixel value profile and a value of the processed approximate curve as the difference value. The image processing method according to any one of claims 5 to 7,
The approximate curve generation unit is an X-ray fluoroscopic apparatus that calculates an n-th order polynomial that approximates the entire average pixel value profile as the approximate curve.