JP5647659B2 - Radiation image correction apparatus, operation method, and program - Google Patents

Description

  The present invention relates to a radiological image correction apparatus, method, and program for correcting a pixel value of one radiographic image of two radiographic images acquired for the same subject to be subjected to comparative interpretation.

  Conventionally, in the medical field, two or more medical images with different shooting times for a patient are comparatively read, the difference between the images is examined, an abnormal shadow is detected based on the difference, and the progression of the disease The treatment policy is examined by grasping the situation and the healing situation.

  However, two time-series images with different shooting timings to be compared have different image gradation characteristics, frequency characteristics, resolution due to differences in the shooting devices used at the time of shooting and differences in the set shooting conditions. Since image characteristics such as density and luminance may be shifted, processing for matching the image characteristics of these images is usually performed when comparative interpretation is performed. As a method for matching image characteristics, gradation processing, frequency processing, pixel size correction processing, dynamic range compression processing, alignment processing, and the like are performed on at least one of two images of the same subject to be compared. A technique has been proposed (Patent Document 1).

JP 2003-190125 A

  In recent years, there has been a growing need for imaging of patients who are difficult to move from a hospital room and emergency imaging in an operating room in a medical field. (Referred to as a round-trip car) is being put into practical use. In particular, for the purpose of respiratory and circulatory management of critically ill patients, X-ray images of the chest of the subject (patient) sleeping in bed with this round-trip car are taken regularly, and multiple chest images with different shooting times are taken. A comparative interpretation is performed to observe the progress.

  However, when taking pictures with this round-trip car, the cassette is placed on the back of the subject sleeping in the bed, so the position and orientation of the cassette with respect to the radiation source, which changes irregularly each time it is taken, the subject In some cases, non-reproducible density unevenness occurs in the radiation image due to the posture of the image. In particular, if such non-reproducible density unevenness exists in two radiographic images to be compared, the performance of comparative interpretation deteriorates. Further, in the above method, no influence is exerted on the performance of comparative interpretation due to such non-reproducible density unevenness, and measures against it are not assumed.

  In view of the above circumstances, the present invention provides a radiological image correction apparatus, method, and program capable of appropriately performing comparative interpretation even when there is non-reproducible density unevenness in a radiographic image to be subjected to comparative interpretation. Is intended to provide.

  The radiological image correction apparatus according to the present invention uses, for each of two chest front radiation images acquired for the same subject to be subjected to comparative interpretation, a pixel value in a predetermined region of the chest front radiation image, and Approximate density unevenness acquired from each chest front radiation image by data acquisition means for acquiring data approximating density unevenness that is the gradient of density appearing in at least one of the horizontal direction and vertical direction of the radiographic image Based on the obtained data, the front of one of the two chest front radiographs is matched so that the data approximating the density unevenness acquired by the data acquisition means from the corrected chest front radiograph matches each other. Correction means for correcting the pixel value of the radiographic image, and the data acquisition means includes a left and right outside of the rib cage in the chest front radiographic image. It is characterized in that an area where the overlap of ribs existing in the outline is imaged or a blank area where radiation is directly incident is acquired as a predetermined area, and data is acquired using pixel values of the acquired area To do.

  In addition, selection information input means for inputting information for selecting which area is to be acquired as a predetermined area among the area where the overlap of the ribs is imaged and the unexposed area in the chest front radiation image May be further provided.

The operation method of the radiological image correction apparatus according to the present invention is any one of a data acquisition unit that acquires data approximating density unevenness of each of two chest front radiation images acquired for the same subject, and a chest front radiation image. A correction method for correcting the radiation image correction apparatus, and a pixel value of a predetermined region of the chest front radiation image for each of two chest front radiation images to be subjected to comparative interpretation. A data acquisition step of acquiring data approximating density unevenness that is a gradient of density appearing in at least one of the horizontal direction and vertical direction of the chest front radiation image, and density unevenness acquired from each chest front radiation image. Based on the approximated data, the data approximating the density unevenness acquired from each chest front radiation image after correction should match each other. A method of operating a radiographic image correction device having a correction step of correcting the pixel value of one of the front chest radiographic image of the two front chest radiation image,
The data acquisition step acquires, as a predetermined region, a region where the overlap of ribs existing in the left and right outer contour portions of the thorax is imaged in the chest front radiation image, or a blank region where radiation is directly incident, Data is acquired using the pixel value of the acquired area.

  Furthermore, the radiographic image correction program according to the present invention is for causing a computer to function as a radiographic image correction apparatus.

  In the radiographic image device, the operation method, and the program, the correction that the data that approximates the density unevenness acquired from each radiographic image after correction agrees with each other is performed by either one of the two radiographic images. What is necessary is just to be performed on the pixel value of the image, and it is not necessary to actually acquire data approximating density unevenness from each corrected radiation image.

  In addition, “each corrected radiographic image” does not have to be an image in which all radiographic images are corrected. When correction is performed only on one radiation, one is corrected and the other is corrected. Means a radiographic image that is not.

  According to the radiological image correction method, apparatus, and program of the present invention, the density of each of two radiographic images acquired for the same subject, which is the target of comparative interpretation, using the pixel value of a predetermined region of the radiographic image. Data that approximates the unevenness is acquired, and based on the data that approximates the density unevenness acquired from each radiographic image, the data that approximates the density unevenness acquired from each corrected radiographic image matches each other. By correcting the pixel value of one of the two radiographic images, it is possible to make the density unevenness present in the two radiographic images to be compared coincide with each other and make it easier to visually recognize the difference between the images. And the comparative interpretation performance can be improved.

  In the above method, apparatus, and program, either a region where the overlap of the ribs existing in the left and right outer contour portions of the rib cage is imaged or a blank region where radiation is directly incident is acquired as the predetermined region. Can do. When there is no density unevenness in the radiographic image, the region where the overlap of the ribs present on the left and right outer contours of this rib cage is photographed has a density distribution that is at least symmetrical to the body axis of the subject, Since it has a substantially uniform density distribution, it is possible to acquire density unevenness in the radiation image by examining the pixel values in any of these areas.

1 is a schematic configuration diagram showing an embodiment of a radiological image correction apparatus of the present invention. The figure for demonstrating the correction process of the radiographic image by the radiographic image correction apparatus of FIG. The figure which shows an example of the body axis of the subject extracted from the target image The figure which shows an example of the body axis of the to-be-extracted object extracted from the reference image The figure which shows an example of the angle between the body axes of the to-be-extracted object respectively extracted from the target image and the reference image The figure which shows an example of the blank region in a radiation image

  Embodiments of the radiation image correcting apparatus of the present invention will be described below with reference to the drawings. The radiological image correction apparatus 1 according to the embodiment of the present invention shown in FIG. 1 is realized by executing a radiological image correction program read into the auxiliary storage device on a computer (for example, a personal computer). At this time, the radiation image correction program is stored in an information storage medium such as a CD-ROM or distributed via a network such as the Internet and installed in a computer.

The radiological image correction apparatus 1 has a density trend of one of the two radiographic images acquired for the same subject, which is a target for comparative interpretation (hereinafter referred to as a target image I (x, y)). The pixel value of the target image I (x, y) is corrected so as to coincide with the density trend of the other reference image (hereinafter referred to as the reference image I ₀ (x, y)). As shown in FIG. 1, an area acquisition unit 10, a correction unit 20, an input unit 30, a display unit 40, a storage unit 50, and the like are provided.

  Here, due to the position and orientation of the cassette with respect to the radiation source, which changes irregularly every imaging, and the posture of the subject, the density unevenness existing in the radiographic image obtained by the imaging is collectively called, This is called concentration trend.

The area acquisition unit 10 acquires an area for extracting a density trend of each of the body image I (x, y) and the reference image I ₀ (x, y).

For example, when the images I (x, y) and I ₀ (x, y) are chest radiographic images, for example, as shown in FIG. 2, an overlap of ribs existing in the left and right outer contour portions of the rib cage was photographed. Regions A and A ₀ (regions surrounded by solid lines in the drawing) are acquired as regions for extracting the density trend of the image. The area where the overlap of ribs present on the left and right outer contours of this rib cage is photographed is located outside the lung field area, which is usually the observation target of pathological changes, and is not dependent on the pathological condition. If there is no density unevenness, the density distribution is symmetric at least with respect to the body axis of the subject. Therefore, the density trend in the radiographic image can be acquired by examining the pixel values in this area.

Specifically, the region acquisition unit 10 uses the technique described in, for example, Japanese Patent Application Laid-Open No. 2003-6661 for each image I (x, y) and I ₀ (x, y). Determine the area surrounded by the contour, right inner lung contour, right lower lung contour, left lung outer contour, left lung inner contour, left lung lower contour, and outward from the left and right contours of the determined region. By extracting a region having a width that is generally included in the region where the overlap is photographed, a region where the overlap of the ribs existing in the left and right outer contour portions of the rib cage is photographed is acquired.

In addition, in the target image I (x, y) and the reference image I ₀ (x, y), for example, a blank region where radiation is directly incident as shown in FIG. 6 shows the overall density trend of the radiation image. If the approximate data is wide enough to be found, specifically, the area of the continuous missing area or the figure surrounding the whole of the plurality of missing areas that are discontinuously spread When the area is sufficiently large, this blank region (region B surrounded by a dashed line in the drawing) can be acquired as a region for extracting the density trend of the image. This non-exposed area has a substantially uniform density distribution when there is no density unevenness in the radiographic image. Therefore, by examining the pixel values in this area, the position of the cassette, in particular, relative to the radiation source of the cassette, It is possible to acquire a density trend that occurs in the radiation image due to the arrangement direction.

In addition, when the images I (x, y) and I ₀ (x, y) are chest front radiation images, the region where the overlap of the ribs present on the left and right outer contour portions of the rib cage is photographed, and the radiation For example, the user can determine which of the directly incident elementary regions is to be acquired as a region for extracting the density trend of each image I (x, y), I ₀ (x, y). Information for selecting which of these areas is to be acquired may be input to the input unit 30, and one of the areas selected based on the input information may be acquired.

  Further, the area acquisition unit 10 automatically detects and acquires an area where the overlapping of ribs existing in the left and right outer contour parts of the rib cage is imaged, or a blank area where radiation is directly incident. Alternatively, the user may be allowed to input information specifying these areas to the input unit 30 and obtain the information based on the input information. Here, the information for specifying a region broadly means information that can specify the range of the region, for example, some points specified to surround the region to be specified on the radiographic image. Position information, position information of each pixel in an area designated as an area to be specified by painting or the like.

The correction unit 20 uses the pixel value of a predetermined area in the image acquired by the area acquisition unit 10 for each of the images I (x, y) and I ₀ (x, y), and the density trend of the image. The approximate data acquisition unit 21 that acquires data approximating the data, the correction data generation unit 22 that generates correction data for correcting the target image I (x, y) using the acquired approximate data, and the generated correction data Is used to correct the pixel value of the radiation image.

Approximate data acquisition unit 21, for example by using the area A, the pixel value of A ₀ obtained in the region acquisition unit 10, the target image I (x, y) and the reference image I _{0 (x,} y) each concentration trends to get the data _T, T ₀ approximating the.

Specifically, for the target image I (x, y), a linear model Z (x, y) that approximates the image signal of the region A is created as in the following formula (1). Here, the constants a, b, and c are parameters selected so as to constitute the most probable linear model Z (x, y) by, for example, the least square method. Specifically, a parameter that minimizes the sum of squares of the difference between the pixel values A (x, y) and Z (x, y) of each pixel in the region A is obtained by searching.

Next, using the parameters a and b and the coordinates (xc, yc) of the center of the target image I (x, y), the target image I (x, y) ), Approximate data T (x, y) is obtained by approximating the concentration trend in the horizontal direction x and the vertical direction y to a linear model.

For the reference image I ₀ (x, y), the density trend of the reference image I (x, y) ₀ using the pixel values in the region A _{0 is performed} in the same manner as in the case of the target image I (x, y). Data T ₀ (x, y) = a ₀ (x-xc) + b ₀ (y-yc) is obtained by approximating to the linear model. FIG. 2 shows an example of data T (x, y) and T ₀ (x, y) acquired by approximating the density trends of the target image I (x, y) and the reference image I ₀ (x, y).

The correction data creation unit 22 uses the approximate data T (x, y) and T ₀ (x, y) acquired by the approximate data acquisition unit 21 to correct the correction data C for correcting the target image I (x, y). (x, y) is generated. For example, as shown in FIG. 2, the target image I (x, y) is obtained from data T ₀ (x, y) approximating the density trend of the reference image I ₀ (x, y). Correction data C (x, y) = T ₀ (x, y) −T (x, y) is created by subtracting data T (x, y) approximating the density trend of y).

It should be noted that the region for extracting the density trend of each image I (x, y), I ₀ (x, y) acquired by the region acquisition unit 10 overlaps the ribs present in the left and right outer contour portions of the rib cage. Is the imaged area and the body axes of the subjects in the images I (x, y) and I ₀ (x, y) are significantly different from each other, data T _{0 (x,} y) data T ₀ after correction that is corrected according to the deviation φ _'(x, y) to get the, the T _0' (x, y) data T _{0 (x,} y It is more preferable to perform the subtraction instead of).

Hereinafter, with reference to FIG. 3 to FIG. 5, processing for obtaining the corrected data T ₀ ′ (x, y) will be described. First, as shown in FIGS. 3 and 4, the correction data creation unit 22 extracts the body axes V and V ₀ of the subject from the target image I (x, y) and the reference image I ₀ (x, y). Specifically, for example, as described in Japanese Patent Application No. 2008-037145, first, in each pixel of the chest image using a Gabor filter, an edge direction value corresponding to the thickness of the vertebral body and the vertebral body Extract edge component values including edge strength values corresponding to the thickness of the image, and then in the region of interest that is set as an area that does not include the left and right edge measurement areas, which are areas where at least the clavicle and ribs overlap in the chest image The direction obtained by weighting and averaging the edge direction value corresponding to the highest edge intensity value in each pixel by the highest edge intensity value in that pixel is estimated as the vertebral body direction. Next, the chest image is scanned substantially perpendicularly to the estimated vertebral body direction, each pixel having a predetermined pixel value or less is extracted as a vertebral body region, and the median line of the extracted vertebral body region is extracted from the subject. Detect as body axis.

Next, as shown in FIG. 5, the angle φ between the body axes V and V ₀ extracted from the images I (x, y) and I ₀ (x, y) is obtained, and the following equation (3) is obtained. , Calculate the values of parameters a ₀ ′ and b ₀ ′ from the values of parameters a ₀ and b ₀ in data T ₀ (x, y) = a ₀ (x−xc) + b ₀ (y−yc) Thus, the corrected data T ₀ ′ (x, y) = a ₀ ′ (x−xc) + b ₀ ′ (y−yc) is acquired.

  The parameters A and B in the correction data C (x, y) = A (x−xc) + B (y−yc) created as described above are output to the correction processing unit 22. Prior to this output, the parameters A and B once created are displayed on the display screen of the display unit 40, the correction value by the user is received from the input unit 30, and the input correction value is used to change them. It is also possible to correct the parameters and output the corrected data to the correction processing unit 22.

The correction processing unit 23 corrects the pixel value of the target image I (x, y) using the correction data generated by the correction data generation unit 22, and as shown in FIG. , y) is added with the correction data C (x, y) created by the correction data creation unit 22, so that the density trend of the corrected target image I (x, y) is changed to the reference image I ₀ (x, y). A corrected image I ′ (x, y) is created by correcting the pixel value of the target image I (x, y) so as to match the density trend of y).

  The corrected image I ′ (x, y) created as described above is displayed on the display screen of the display unit 40. At this time, information such as parameters of the correction data C (x, y) used for correction by the correction processing unit 23 is also displayed on the screen so that the user can easily understand what correction processing has been performed. can do.

  When the user inputs a correction value to the parameter displayed on the display screen together with the corrected image I ′ (x, y) by the input unit 30, the correction data creation unit 21 sets the correction value by the user. The parameter is corrected using the input correction value received from the input unit 30, the corrected data is output to the correction processing unit 22, and the correction processing unit 22 uses the input correction data after correction. Thus, the pixel value of the target image I (x, y) can be corrected again to create the unevenness corrected image I ′ (x, y).

  The recording unit 50 stores the created corrected image I ′ (x, y) and the correction data C (x, y) used for the correction in a recording medium in association with each other.

With the above configuration, when performing comparative interpretation using two radiographic images (target image I (x, y) and reference image I ₀ (x, y)) acquired for the same subject, first, an area acquisition unit 10 shows, for each of the target image I (x, y) and the reference image I ₀ (x, y), an area where the overlap of ribs existing in the left and right outer contours of the rib cage is photographed, or a blank area. Acquired as an area for extracting the density trend of the image. Next, the approximate data acquisition unit 21 uses the pixel values of the acquired areas to approximate the density trends of the target image I (x, y) and the reference image I ₀ (x, y), respectively. (x, y) and T ₀ (x, y) are acquired. Next, the correction data creation unit 22 creates correction data C (x, y) by subtracting the acquired approximate data T (x, y) and T ₀ (x, y), and the correction processing unit 23. However, the correction data created by the correction data creation unit 22 is used to correct the pixel value of the radiation image I (x, y) to create a corrected image I ′ (x, y). Next, the display unit 40 displays the corrected image I ′ (x, y) created in the correction processing unit 23, the correction data C (x, y) used for the correction, and the like on the monitor screen. .

  When the user inputs a correction value to the parameter of the correction data C (x, y) displayed on the screen by the input unit 30, the correction data creation unit 22 receives the correction value by the user from the input unit 30, The parameter is corrected using the input correction value, and the correction processing unit 23 corrects the pixel value of the radiation image again using a function defined by the corrected parameter, and the corrected image I ′ (x, Create y). Then, the recording unit 50 stores the corrected image I ′ (x, y) finally created and the correction data used for the correction in association with each other and stored in the recording medium.

  Here, after the corrected data I ′ (x, y) is generated by performing correction using the correction data C (x, y) generated by the correction data generating unit 22 as it is, the result is sent to the user. Presented and explained the case where correction data is corrected as necessary and correction is performed again using the corrected correction data. However, the correction data created by the correction data creation unit 22 before the first correction is performed. The data C (x, y) or the like may be temporarily displayed on the monitor screen, and correction by the user may be received as necessary in the correction data, and correction may be performed using the corrected correction data.

  According to the above embodiment, by correcting the pixel value of one of the two radiographic images acquired for the same subject that is the target of comparative interpretation, the radiographic image exists in those radiographic images. It is possible to make the density unevenness coincide with each other, make it easy to visually recognize the difference between the images, and improve the comparative interpretation performance.

  Note that it is possible to arbitrarily select / determine which of the two radiographic images to be subjected to comparative interpretation is used as the reference image for the correction processing according to the present invention. However, the positioning is relatively good and the density unevenness is relatively high. It is more preferable to select the radiographic image without the reference image as the reference image.

DESCRIPTION OF SYMBOLS 1 Radiation image correction apparatus 10 Area | region acquisition part 20 Correction | amendment part 21 Trend approximation data acquisition part 22 Correction | amendment data creation part 23 Correction | amendment process part 30 Input part 40 Display part 50 Storage part I Target image I ₀ reference _| standard image C Correction data

Claims (4)

For each of the two chest front radiographs acquired for the same subject that is the subject of comparative interpretation, the horizontal and vertical directions of the chest front radiograph are obtained using the pixel values of a predetermined region of the chest front radiograph. Data acquisition means for acquiring data approximating density unevenness that is a gradient of density appearing in at least one direction of
Based on the data approximating the density unevenness acquired from each chest front radiation image by the data acquisition means, the density unevenness acquired by the data acquisition means from the corrected chest front radiation image was approximated Correction means for correcting the pixel value of either one of the two chest front radiation images so that the data match each other,
The data acquisition means uses, as the predetermined area, an area in which an overlap of ribs existing in the left and right outer contours of the thorax is imaged in the chest frontal radiographic image, or a blank area where radiation is directly incident acquired state, and are not to acquire the data by using the pixel values of the acquired region,
For the data approximating the density unevenness, the pixel value A (x, y) of the predetermined area is expressed by the following linear model
Z (x, y) = ax + by + c (a, b, c are constants)
In approximated Z (x, y) der Rukoto radiographic image correcting apparatus according to claim.   Selection information for inputting information for selecting which area is to be acquired as the predetermined area from among the area where the overlap of the ribs is imaged and the unexposed area in the chest front radiation image The radiation image correction apparatus according to claim 1, further comprising an input unit. A radiographic image comprising data acquisition means for acquiring data approximating the density unevenness of each of two chest frontal radiographic images acquired for the same subject, and correction means for correcting one of the chest frontal radiographic images A method of operating the correction device,
For each of the two chest front radiation images to be subjected to comparative interpretation by the data acquisition means, using the pixel value of a predetermined region of the chest front radiation image, the horizontal and vertical directions of the chest front radiation image A data acquisition step of acquiring data approximating density unevenness, which is a density gradient appearing in at least one direction of the direction;
Based on the data approximating the density unevenness acquired from the chest frontal radiation images by the correction means, the data approximating the density unevenness acquired from the corrected chest frontal radiographic images match each other. A correction step of correcting a pixel value of one of the two chest front radiation images among the two chest front radiation images,
In the data acquisition step, the region where the overlap of ribs existing in the left and right outer contour portions of the chest is imaged in the chest front radiation image, or the blank region where radiation is directly incident is set as the predetermined region. acquired state, and are not to acquire the data by using the pixel values of the acquired region,
For the data approximating the density unevenness, the pixel value A (x, y) of the predetermined area is expressed by the following linear model
Z (x, y) = ax + by + c (a, b, c are constants)
In approximated Z (x, y) der operation method Rukoto radiographic image correcting apparatus according to claim. Computer
For each of the two chest front radiographs acquired for the same subject that is the subject of comparative interpretation, the horizontal and vertical directions of the chest front radiograph are obtained using the pixel values of a predetermined region of the chest front radiograph. Data acquisition means for acquiring data approximating density unevenness that is a gradient of density appearing in at least one direction of
Based on the data approximating the density unevenness acquired from each chest front radiation image by the data acquisition means, the density unevenness acquired by the data acquisition means from the corrected chest front radiation image was approximated A radiological image correction program for functioning as a correction means for correcting a pixel value of any one of the two chest front radiation images among the two chest front radiation images so that the data match each other,
The data acquisition means uses, as the predetermined area, an area in which an overlap of ribs existing in the left and right outer contours of the thorax is imaged in the chest frontal radiographic image, or a blank area where radiation is directly incident Obtaining the data using the pixel value of the obtained area ,
For the data approximating the density unevenness, the pixel value A (x, y) of the predetermined area is expressed by the following linear model
Z (x, y) = ax + by + c (a, b, c are constants)
Radiation image correction program that functions as Z (x, y) approximated by .