JP6185023B2 - Tomographic image generating apparatus, method and program - Google Patents

Description

  The present invention relates to a tomographic image generation apparatus, method, and program for capturing a subject at each of a plurality of radiation source positions to acquire a plurality of projection images and generating a tomographic image from the plurality of projection images.

  In recent years, in a radiographic imaging apparatus using radiation such as X-rays and gamma rays, in order to observe the affected area in more detail, the radiation source is moved and the subject is irradiated with radiation from a plurality of radiation source positions, and imaging is performed. Tomosynthesis imaging has been proposed in which a plurality of projection images acquired in this way are added to generate a tomographic image in which a desired tomographic plane is emphasized. In tomosynthesis imaging, depending on the characteristics of the imaging device and the required tomographic image, the radiation source is moved in parallel with the radiation detector or moved so as to draw a circle or ellipse arc. A plurality of projection images are acquired by photographing a subject in step S1, and a tomographic image is generated by reconstructing these projection images using a back projection method such as a simple back projection method or a filtered back projection method.

  However, in tomosynthesis imaging, the angle at which a subject is irradiated with radiation is limited. For example, when a tomographic image is reconstructed by superimposing projection images by back projection, the structure in the subject is originally Artifacts that are virtual images of structures may appear in areas that do not exist. More specifically, artifacts may appear in a tomographic image of a tomographic plane that is different from a tomographic image of a tomographic plane in which a structure exists, due to back projection. If the artifact is too conspicuous, it is difficult to confirm the structure necessary for diagnosing a lesion or the like.

  Further, in tomosynthesis imaging, since the subject is irradiated with radiation a plurality of times, the lowest possible dose is used for one imaging in order to reduce exposure to the subject. However, if the dose is low, a lot of radiation quantum noise is included in the projection image acquired by photographing, and noise in the tomographic image becomes conspicuous.

  Various methods for reducing such artifacts or noise have been proposed. For example, in Patent Document 1, a single projection image is used as a reference projection image, and a pixel between the reference projection image and a corresponding pixel on each projection image that is cumulatively added to the same position on the tomographic image. For each pixel of the plurality of projection images, a weighting coefficient that increases as the degree of similarity increases, and a multiplication value of the pixel value of the pixel of the plurality of projection images and the weighting factor corresponding thereto is calculated. A technique for reconstructing a tomographic image by cumulative addition has been proposed.

  In addition, another method called an algebraic reconstruction method or a successive approximation reconstruction method has been proposed. These methods are methods for obtaining a tomographic image so that an image obtained by projecting the reconstructed tomographic image matches an actually captured projection image. Since these methods can incorporate various mathematical models into the reconstruction, artifact correction and noise reduction can be considered. Therefore, a tomographic image in which artifacts are suppressed and noise is reduced can be generated.

  However, the method described in Patent Document 1 and reconstruction processing such as the successive approximation reconstruction method have a problem that the computation time is enormous. For this reason, a plurality of band-limited images having different frequency response characteristics are generated from the projection image, and the band-limited image is nonlinearly transformed so that a portion where the band-limited image exceeds a predetermined value becomes small, and the plurality of non-linearly transformed band-limited images There has been proposed a technique for reducing artifacts while reducing computation time by accumulating images to generate a plurality of converted images and reconstructing a tomographic image from the plurality of converted images (see Patent Document 2). ).

  In Patent Documents 3 and 4, the calculation speed when reconstructing a tomographic image is improved by matching the pixel size of the tomographic image with the pixel size of the radiation detector regardless of the height of the tomographic plane. A technique to make it has been proposed.

JP 2013-000261 A JP 2013-031641 A JP 2005-013736 A JP 11-339050 A

  However, in the method described in Patent Document 2, although artifacts can be reduced, reduction of noise is not considered. Moreover, when it is going to shorten calculation time as described in patent documents 2-4, there exists a possibility that an artifact cannot fully be reduced. In the methods described in Patent Documents 1 to 4, the pixel value of the target pixel position in the tomographic image is calculated by adding only the pixel values of the projection image corresponding to the pixel position of the target pixel position. . For this reason, even if weighting or nonlinear conversion is performed, the influence of noise or artifacts cannot be sufficiently reduced.

  The present invention has been made in view of the above circumstances. When generating a tomographic image from a plurality of projection images acquired by performing imaging at a plurality of radiation source positions, such as tomosynthesis imaging, the tomography while reducing the calculation time. The object is to further improve the image quality.

A tomographic image generation apparatus according to the present invention includes a plurality of radiation sources captured by moving a radiation source relative to a detection unit and irradiating a subject with radiation at a plurality of radiation source positions by the movement of the radiation source. Image acquisition means for acquiring a plurality of projection images corresponding to each of the positions;
While maintaining the pixel values of the plurality of projection images, the pixel values of the plurality of projection images are set as desired for the subject based on the positional relationship between the radiation source position and the detection means at the time of shooting for each of the plurality of projection images. Pixel value projection means for projecting to coordinate positions on the tomographic plane;
A tomographic image of the tomographic plane is generated by calculating a pixel value of the coordinate position of interest based on a plurality of pixel values of the projected image projected in a predetermined range based on the target coordinate position on the tomographic plane. And a pixel value calculating means.

  “Moving the radiation source relative to the detection means” means either moving only the radiation source, moving only the detection means, or moving both the radiation source and the detection means. Is also included.

  The projection image and the tomographic image of the tomographic plane are composed of a plurality of pixels that are discretely arranged two-dimensionally at a predetermined sampling interval, and the pixels are arranged at lattice points that have the predetermined sampling interval. The The “pixel position” in the present invention means a position that becomes a grid point where a pixel value is arranged as an image in a projection image and a tomographic image. On the other hand, the “coordinate position” includes a grid point where a pixel is arranged as an image, that is, a pixel position, but also includes a position between which the pixel value is not arranged as an image. Therefore, the “coordinate position” includes not only the pixel position but also a position between the pixel positions.

  Further, the plurality of projection images to be projected may be all of the acquired plurality of projection images, or may be two or more of the plurality of projection images.

  The “desired tomographic plane” means a tomographic plane of a subject that is a target for generating a tomographic image.

  “While holding pixel values of a plurality of projection images” means that the pixel values of the projection images are not changed. In the present invention, the pixel value at the pixel position of the projected image may not be projected onto the coordinate position on the tomographic plane. That is, depending on the positional relationship between the radiation source position and the detection means, the pixel value of the projection image corresponding to the coordinate position on the tomographic plane does not exist at the pixel position of the projection image, but exists at the coordinate position between the pixel positions. May be. In such a case, the pixel value of the coordinate position projected on the tomographic plane in the projection image may be calculated by interpolating pixel values of pixel positions around the coordinate position, for example. Even in such a case, since the pixel value calculated by the interpolation is the pixel value of the projection image, if the pixel value of the projection image calculated by the interpolation is retained and projected to the coordinate position on the tomographic plane, Good.

  The “target coordinate position on the tomographic plane” means a coordinate position that is a target for calculating a pixel value when generating a tomographic image of the tomographic plane. Accordingly, by calculating the pixel value at the target coordinate position by sequentially changing the target coordinate position on the tomographic plane, a tomographic image on the tomographic plane can be generated.

  The “predetermined range based on the target coordinate position” means a range including the target pixel position and a predetermined number of coordinate positions or pixel positions around the target pixel position. For example, a 3 × 3 number of coordinate positions or a range of pixel positions centered on the target pixel position, a 5 × 5 number of coordinate positions or a range of pixel positions, etc. are determined in advance with reference to the target coordinate position. It can be a range. Note that the size of the predetermined range may be a fixed value or may be changed to an arbitrary value by an operator input.

  In the tomographic image generation device according to the present invention, the pixel value projecting unit is configured to detect the coordinates of the corresponding projection image that intersects a straight line connecting each source position and the pixel position on the tomographic plane at each of a plurality of source positions. The pixel value at the position may be projected onto the pixel value at the pixel position on the tomographic plane located on the straight line.

  In the tomographic image generation device according to the present invention, the pixel value projecting means is arranged on a tomographic plane intersecting a straight line connecting each source position and the corresponding pixel position on the projected image at each of a plurality of source positions. It is good also as a means to project the pixel value in the pixel position of the projection image located on a straight line to a coordinate position.

  In this case, the interval between the coordinate positions on the tomographic plane may be smaller than the interval between the pixel positions on the tomographic plane.

  In the tomographic image generation device according to the present invention, the pixel value calculation means may be a means for performing a regression analysis on the pixel value of the projection image projected on the tomographic plane to calculate the pixel value of the target coordinate position. Good.

  “Regression analysis” is a statistical technique for analyzing multivariate relationships. Here, it is assumed that the observed value on the observation point is observed with noise included in the true value. Regression analysis is a technique for solving an inverse problem for obtaining a true value at every observation point by a least square method, a moving average method, regression using a kernel, or the like. In the present invention, the coordinate position where the pixel value of the projection image on the tomographic plane is projected is the observation point, the pixel value of the observation point is the observation value, the pixel value at the target coordinate position is the true value, and the target coordinate position is The pixel value is calculated.

  Further, in the tomographic image generation device according to the present invention, the pixel value calculating means performs regression analysis to generate a regression surface representing the tomographic image on the tomographic plane, and samples the regression curved surface at a desired sampling interval, Means for generating a tomographic image by calculating a pixel value at a pixel position on the tomographic plane may be used.

  In this case, the sampling interval may be different from the sampling interval of the projection image.

  Here, the resolution of the tomographic image can be changed by changing the sampling interval of the regression surface. For example, the smaller the sampling interval, the higher the resolution of the tomographic image. The “desired sampling interval” means a pixel interval at which a tomographic image having a required resolution is obtained. The desired sampling interval may be a fixed value or may be changed to an arbitrary value by an operator input.

  “Different from the sampling interval of the projection image” includes both cases where the sampling interval is larger and smaller than the sampling interval of the projection image. When it is larger than the sampling interval of the projection image, the resolution of the tomographic image is lower than the resolution of the projection image. On the contrary, when it is smaller than the sampling interval of the projection image, the resolution of the tomographic image becomes higher than the resolution of the projection image.

  In the tomographic image generation device according to the present invention, when the pixel value calculation unit receives an instruction to change the size of the attention area in the displayed tomographic image, the sampling interval of the area corresponding to the attention area on the regression surface is changed. It may be changed in accordance with the instruction and may be a means for generating a tomographic image of the region of interest.

  In the tomographic image generation device according to the present invention, the pixel value calculation means may be means for changing the sharpness of the pixel value at the coordinate position of interest when performing regression analysis.

  “Change sharpness” means that sharpness is enhanced to enhance edges included in the generated tomographic image, and sharpness is reduced to smooth the generated tomographic image and reduce noise. Including both reducing.

  In this case, the pixel value calculation means may be a means for changing the degree of change in sharpness according to information on at least one of the shooting conditions at the time of shooting and the structure of the subject included in the projected image.

  When a projected image is captured, the noise in the projected image increases as the amount of radiation reaching the detector decreases. Also, the amount of noise contained in the projected image may vary depending on whether the radiation quality is high or low, the type of material that constitutes the detection means, or the presence or absence of a grid that removes scattered radiation during imaging. The sharpness of the obtained image may be different. “Photographing conditions” means various conditions that affect the amount of noise and sharpness of the projected image. For example, the amount of radiation reaching the detector at the time of photographing, the type of detection means, or the presence or absence of a grid Can be used. The “subject structure” is a structure such as an edge included in the subject.

  Note that the sharpness of the tomographic image also depends on the preference of a user such as a doctor who views the image. For this reason, you may enable it to change a sharpness according to a user's liking.

  Further, in the tomographic image generation device according to the present invention, the pixel value calculation means causes the pixel value to be an outlier among a plurality of pixel values of the projection image projected in a predetermined range with the target coordinate position as a reference. The pixel value of the target coordinate position may be calculated by excluding the above or by reducing the weight of the pixel value that is an outlier.

  The “outlier” means a pixel value that is significantly different from other pixel values among a plurality of pixel values of a projected image projected in a predetermined range with the target coordinate position as a reference. Here, “substantially different” means, for example, that an average value of a plurality of pixel values projected on a predetermined range with reference to the target coordinate position is calculated, and becomes a value different from the average value by a predetermined threshold or more. Means that.

  In the tomographic image generation device according to the present invention, the pixel value projecting means includes two-dimensional coordinates of the target coordinate position on the projection image corresponding to the specific source position, and a pixel value of the target coordinate position on the projection image. The target coordinate position on the projection image is based on the positional relationship between the specific source position and the target coordinate position on the projection image so that the two-dimensional coordinates of the coordinate position on the tomographic plane on which the image is projected match. It is also possible to correct the coordinate positions on the tomographic plane on which the pixel values are projected and project the pixel values of a plurality of projection images onto the corrected coordinate positions on the tomographic plane.

In this case, the image acquisition means is a means for acquiring a radiographic image of the subject imaged by irradiating the subject with radiation at a specific radiation source position,
The pixel value calculation means is a means for generating a tomographic image by calculating a pixel value of a coordinate position of interest on a tomographic plane in which the pixel value of the projection image is projected at the corrected coordinate position,
Display means for displaying a radiation image and a tomographic image may be further provided.

  “Radiation image of a subject” is included in a subject that is acquired by performing shooting under a shooting condition that acquires a transmission image of the subject while fixing the source position without moving the source position. It is an image including the image which the structure to be transmitted was transmitted.

Further, in the tomographic image generation device according to the present invention, the pixel value projection means and the pixel value calculation means are means for generating tomographic images on a plurality of tomographic planes of the subject,
A pseudo image generation unit that generates a pseudo image from a plurality of tomographic images may be further provided.

  The pseudo image is a pseudo representation of an image of a different type from the tomographic image. By simply adding the pixel values of the corresponding pixel positions in a plurality of tomographic images, a transmission image acquired by normal radiography is obtained. An addition tomographic image representing the image in a pseudo manner can be given as an example. In addition to the addition tomographic image, the maximum value projection image obtained by a MIP (Maximum Intensity Projection) method for extracting the maximum value of the corresponding pixel position in a plurality of tomographic images, and the minimum value It is possible to use, as a pseudo image, a minimum-value projected image obtained by a minIP (Minimum Intensity Projection) method for extracting the three-dimensional image in a pseudo manner.

  In the tomographic image generation device according to the present invention, the pixel value calculation means calculates a weighting coefficient based on a difference between the pixel value at the target coordinate position and the pixel value at the coordinate position of the projected image corresponding to the target coordinate position. Then, the pixel value at the target coordinate position may be calculated again based on the plurality of pixel values and the weighting coefficient of the projection image, and a new pixel value at the coordinate position may be calculated.

  In this case, the pixel value calculation means calculates a new weighting coefficient using the new pixel value of the target coordinate position, and newly calculates the target coordinate position based on the plurality of pixel values and the new weighting coefficient of the projection image. It may be a means for repeating the calculation of the pixel value again.

  As the “weighting coefficient”, for example, a coefficient whose value decreases as the difference from the pixel value at the coordinate position of the projected image corresponding to the target coordinate position increases.

  Number of times the calculation of a new weighting coefficient using a new pixel value at the coordinate position of interest and the recalculation of a new pixel value at the coordinate position of interest based on a plurality of pixel values of the projected image and the new weighting coefficient May be predetermined, and the difference between the new pixel value at the target coordinate position and the pixel value at the coordinate position of the projected image corresponding to the target coordinate position is smaller than a predetermined threshold value. It may be repeated until a specific convergence condition such as

A tomographic image generation method according to the present invention includes a plurality of radiation sources captured by moving a radiation source relative to a detection unit and irradiating a subject with radiation at a plurality of radiation source positions by the movement of the radiation source. Acquire multiple projected images corresponding to each of the positions,
While maintaining the pixel values of the plurality of projection images, the pixel values of the plurality of projection images are set as desired for the subject based on the positional relationship between the radiation source position and the detection means at the time of shooting for each of the plurality of projection images. Project to the coordinate position on the fault plane,
A tomographic image of the tomographic plane is generated by calculating a pixel value of the coordinate position of interest based on a plurality of pixel values of the projected image projected in a predetermined range based on the target coordinate position on the tomographic plane. It is characterized by this.

  In addition, you may provide as a program for making a computer perform the tomographic image generation method by this invention.

  According to the present invention, while maintaining the pixel values of the plurality of projection images, the pixel values of the plurality of projection images based on the positional relationship between the radiation source position and the detection unit at the time of shooting for each of the plurality of projection images. Is projected onto the desired coordinate position on the tomographic plane of the subject. Then, based on a plurality of pixel values of the projected image projected onto a predetermined range based on the target coordinate position on the tomographic plane, a pixel value at the target coordinate position is calculated, and a tomographic image is generated. For this reason, the influence of the pixel values around the target coordinate position is considered in comparison with the conventional method of calculating the pixel value of the target coordinate position using only the pixel value of the projected image projected on the target coordinate position. As a result, artifacts can be reduced and a higher-quality tomographic image can be generated. In addition, unlike the successive approximation reconstruction method, it is not necessary to repeat projection and reprojection, so that the calculation time can be greatly reduced.

1 is a schematic configuration diagram of a radiographic imaging apparatus to which a tomographic image generation apparatus according to a first embodiment of the present invention is applied. The figure which looked at the radiographic imaging apparatus from the arrow A direction of FIG. The figure which shows schematic structure of the tomographic image generation apparatus implement | achieved by installing a tomographic image generation program in a computer in 1st Embodiment. Diagram for explaining acquisition of projection image The figure for demonstrating the projection of the pixel value in 1st Embodiment. The figure for demonstrating the interpolation of the pixel value of a projection image The figure which shows the pixel value projected on the tomographic plane in 1st Embodiment. Diagram for explaining generation of regression curve (regression surface) including outliers Diagram for explaining generation of regression curve (regression surface) with outliers removed The flowchart which shows the process performed in 1st Embodiment. The figure for demonstrating the projection of the pixel value in 2nd Embodiment. The figure which shows the pixel value projected on the tomographic plane in 2nd Embodiment. The figure which shows the positional relationship of the position of the structure on a tomographic plane, and the structure projected on a radiation detector The figure which shows the difference in the position of the structure in a tomographic image and a radiographic image The figure which shows schematic structure of the tomographic image generation apparatus implement | achieved by installing a tomographic image generation program in a computer in 4th Embodiment. The figure which shows the state which displayed the enlarged attention area on the display part The figure which shows schematic structure of the tomographic image generation apparatus implement | achieved by installing a tomographic image generation program in a computer in 5th Embodiment. The figure which shows the some tomographic image projection image when there is no position shift The figure which shows the some tomographic image projection image when there is position shift

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a schematic configuration diagram of a radiographic imaging apparatus to which the tomographic image generation apparatus according to the first embodiment of the present invention is applied. FIG. 2 is a diagram of the radiographic imaging apparatus as viewed from the direction of arrow A in FIG. In order to perform tomosynthesis imaging of a breast and generate a tomographic image, the radiographic imaging apparatus 1 generates a breast M (hereinafter sometimes referred to as a subject M) from a plurality of radiation source positions in different imaging directions. This is a mammography imaging apparatus that captures a plurality of radiographic images, that is, projection images. As shown in FIG. 1, the radiographic imaging device 1 includes an imaging unit 10, a computer 2 connected to the imaging unit 10, and a display unit 3 and an input unit 4 connected to the computer 2.

  The imaging unit 10 includes an arm unit 12 connected to a base (not shown) by a rotation shaft 11. The imaging unit 13 is attached to one end of the arm 12, and the radiation irradiation unit 14 is attached to the other end so as to face the imaging table 13. The arm unit 12 is configured to be able to rotate only the end portion to which the radiation irradiating unit 14 is attached, so that only the radiation irradiating unit 14 can be rotated while fixing the imaging table 13. It is possible. Note that the rotation of the arm unit 12 is controlled by the computer 2.

  Inside the imaging table 13, a radiation detector 15 (detection means) such as a flat panel detector is provided. The imaging table 13 includes a charge amplifier that converts the charge signal read from the radiation detector 15 into a voltage signal, a correlated double sampling circuit that samples the voltage signal output from the charge amplifier, and a voltage signal. A circuit board or the like provided with an AD conversion unit for converting into a digital signal is also installed.

  The radiation detector 15 can repeatedly perform recording and reading of a radiation image, and may use a so-called direct type radiation detector that directly receives radiation to generate charges, or radiation. May be used as a so-called indirect radiation detector that converts the light into visible light and converts the visible light into a charge signal. As a radiation image signal reading method, a radiation image signal is read by turning on / off a TFT (thin film transistor) switch, or a radiation image signal by irradiating reading light. Although it is desirable to use a so-called optical readout system in which is read out, the present invention is not limited to this, and other types may be used.

  An X-ray source 16 (radiation source) is housed inside the radiation irradiation unit 14. The timing at which X-rays are emitted from the X-ray source 16 and the X-ray generation conditions (tube current, time, tube current time product, etc.) in the X-ray source 16 are controlled by the computer 2.

  In addition, the arm portion 12 includes a compression plate 17 that is disposed above the imaging table 13 and presses against the breast M, a support portion 18 that supports the compression plate 17, and a support portion 18 in the vertical direction of FIGS. A moving mechanism 19 is provided for moving to.

  The display unit 3 is a display device such as a CRT or a liquid crystal monitor, and displays messages necessary for operations in addition to a projection image acquired and a generated tomographic image as described later. The display unit 3 may include a speaker that outputs sound.

  The input unit 4 includes a keyboard, a mouse, or a touch panel type input device, and receives an operation of the radiation image capturing apparatus 1 by an operator. It also accepts input of various information such as imaging conditions and information correction instructions necessary for performing tomosynthesis imaging. In the present embodiment, each unit of the radiographic image capturing apparatus 1 operates in accordance with information input from the input unit 4 by the operator.

  The computer 2 is installed with a tomographic image generation program. In the present embodiment, the computer may be a workstation or a personal computer directly operated by an operator, or a server computer connected to them via a network. The tomographic image generation program is recorded and distributed on a recording medium such as a DVD or CD-ROM, and is installed in the computer from the recording medium. Alternatively, it is stored in a storage device of a server computer connected to a network or a network storage in a state where it can be accessed from the outside, and is downloaded to a computer and installed on demand.

  FIG. 3 is a diagram showing a schematic configuration of a tomographic image generation apparatus realized by installing a tomographic image generation program in the computer 2. As shown in FIG. 3, the tomographic image generation apparatus includes a CPU 21, a memory 22, and a storage 23 as a standard computer configuration.

  The storage 23 includes a storage device such as a hard disk or an SSD, and stores various information including a program for driving each unit of the radiographic imaging apparatus 1 and a tomographic image generation program. In addition, a projection image acquired by tomosynthesis imaging and a tomographic image generated as described later are also stored.

  The memory 22 temporarily stores programs stored in the storage 23 in order to cause the CPU 21 to execute various processes. The tomographic image generation program, as a process to be executed by the CPU 21, causes the radiographic imaging apparatus 1 to perform tomosynthesis imaging to acquire a plurality of projection images of the breast M, and holds pixel values of the plurality of projection images. On the other hand, based on the positional relationship between the position of the X-ray source 16 and the radiation detector 15 at the time of imaging for each of the plurality of projection images, the tomographic image in which the pixel values of the plurality of projection images are desired of the breast M as the subject The target coordinate position based on the pixel value projection process for projecting to the coordinate position on the surface, and a plurality of pixel values of the projected image projected to a predetermined range based on the target coordinate position on the desired tomographic plane The pixel value calculation process for generating the tomographic image by calculating the pixel value is defined.

  Then, the CPU 21 executes these processes in accordance with the tomographic image generation program, so that the computer 2 functions as the image acquisition unit 31, the pixel value projection unit 32, and the pixel value calculation unit 33. The computer 2 may include a CPU that performs image acquisition processing, pixel value projection processing, and pixel value calculation processing.

  The image acquisition unit 31 moves the X-ray source 16 by rotating the arm unit 12 around the rotation axis 11, and emits X-rays to the breast M as a subject at a plurality of source positions by the movement of the X-ray source 16. The X-rays irradiated and transmitted through the breast M are detected by the radiation detector 15 to obtain a plurality of projection images Gi (i = 1 to n, where n is the number of the source positions) at a plurality of source positions. FIG. 4 is a diagram for explaining the acquisition of the projection image Gi. As shown in FIG. 4, the X-ray source 16 is moved to each of the S1, S2,..., Sn source positions, and the X-ray source 16 is driven at each source position to irradiate the breast M with X-rays. By detecting X-rays transmitted through the breast M by the radiation detector 15, projection images G1, G2,..., Gn are obtained corresponding to the respective radiation source positions S1 to Sn. The acquired plurality of projection images Gi are stored in the storage 23. A plurality of projection images Gi may be acquired and stored in the storage 23 by a program separate from the tomographic image generation program. In this case, the image acquisition unit 31 reads a plurality of projection images Gi stored in the storage 23 from the storage 23 for pixel value projection processing and pixel value calculation processing.

  In the present embodiment, all of the plurality of projection images Gi stored in the storage 23 may be read out and used for the pixel value projection process and the pixel value calculation process. Two or more obtained projection images Gi may be read out and used for pixel value projection processing and pixel value calculation processing.

  The pixel value projection unit 32 projects the pixel values of the projection image to the desired coordinate position on the tomographic plane of the breast M while holding the pixel values of the plurality of projection images acquired by the image acquisition unit 31. FIG. 5 is a diagram for explaining the projection of pixel values in the first embodiment. In FIG. 5, a case where the projection image Gi acquired at the radiation source position Si is projected onto a desired tomographic plane Tj (j = 1 to m: m is the number of tomographic planes) of the breast M will be described. .

  Here, the tomographic image generated as described later in the projection image Gi and the tomographic plane Tj is composed of a plurality of pixels that are discretely arranged two-dimensionally at a predetermined sampling interval. Pixels are arranged at lattice points that serve as intervals. In FIG. 5 and FIG. 11 to be described later, a short line segment orthogonal to the projection image Gi and the tomographic plane Tj indicates a pixel separation position. Therefore, in FIG. 5 and FIG. 11 described later, the center position of the pixel delimiter position is the pixel position that is the lattice point. In the first embodiment, as shown in FIG. 5, the pixel values of the projection image Gi that intersect with the straight line connecting the source position Si and the pixel position on the tomographic plane Tj are displayed on the corresponding tomographic line. Projecting to a pixel value at a pixel position on the surface Tj.

  Here, when the coordinates of the radiation source position Si are (sxi, syi, szi) and the coordinates of the pixel position on the tomographic plane Tj are Tj (tx, ty, tz), the coordinates (pxi) of the coordinate position Pi on the projection image Gi. , Pyi) is represented by the following equation (1). In the present embodiment, the z axis is perpendicular to the detection surface of the radiation detector 15, the y axis is parallel to the direction in which the X-ray source 16 moves on the detection surface of the radiation detector 15, and y Assume that the x-axis is set in a direction perpendicular to the axis.

pxi = (tx * szi-sxi * tz) / (szi-tz)
pyi = (ty × szi−syi × tz) / (szi−tz) (1)
Note that the coordinate position Pi on the projection image Gi may not be the pixel position of the projection image Gi. For example, as illustrated in FIG. 6, the coordinate position Pi on the projection image Gi may be located between the four pixel positions O1 to O4 on the projection image Gi. In this case, as shown in FIG. 6, an interpolation operation using pixel values at the four pixel positions O1 to O4 of the projection image Gi that is closest to the coordinate position Pi is performed to calculate a pixel value at the coordinate position Pi. The calculated pixel value may be projected onto the pixel position Tj (tx, ty, tz) on the tomographic plane Tj. In addition to the linear interpolation calculation that weights the pixel values of the four pixel positions according to the distance between the coordinate position Pi and the four pixel positions, the interpolation calculation includes more pixel positions around the coordinate position Pi. Any method such as non-linear bicubic interpolation using pixel values and B-spline interpolation can be used. In addition to the interpolation calculation, the pixel value at the pixel position closest to the coordinate position Pi may be used as the pixel value at the coordinate position Pi.

  The pixel value projection unit 32 projects the pixel values of the projection image Gi on the tomographic plane Tj for all the radiation source positions Si. As a result, as shown in FIG. 7, n pixel values corresponding to the number of projection images are projected at each pixel position on the tomographic plane Tj. FIG. 7 shows a state in which pixel values of five projected images are projected at each pixel position. In FIG. 7 and FIGS. 8, 9, 12, 18, and 19 to be described later, a short line segment orthogonal to the tomographic plane Tj indicates a pixel separation position, and the center position of the pixel separation position is a lattice point. A certain pixel position is shown.

  The pixel value calculation unit 33 generates a tomographic image on the tomographic plane Tj by calculating each pixel value on the tomographic plane Tj. Specifically, the pixel value of the target coordinate position is calculated based on a plurality of pixel values of the projected image projected on a predetermined range based on the target coordinate position that is the target of calculation of the pixel value. Note that the target coordinate position may be a pixel position on the tomographic plane Tj. In the first embodiment, the pixel value of the projection image Gi is projected at the pixel position on the tomographic plane Tj. When calculating the pixel value at the target coordinate position, the pixel value projected at the target coordinate position is calculated. It may or may not be used. Hereinafter, calculation of the pixel value of the target coordinate position will be described.

  The pixel values of the projection image Gi projected on the tomographic plane Tj by the pixel value projection unit 32 tend to be the same value as they are closer to each other. For this reason, the pixel value calculation unit 33 performs a process of changing the sharpness so that the pixel values projected on the tomographic plane Tj are smoothly continuous. In the present embodiment, a filtering process using a smoothing filter is performed on the pixel values projected onto the tomographic plane Tj. Specifically, for example, a filtering process using a Gaussian filter is performed on pixel values in a pixel range in a predetermined range such as 3 × 3 or 5 × 5 centered on the target coordinate position. As a result, since the pixel values smoothly continue at the target coordinate position and the surrounding pixels, noise such as quantum noise included in the projected image Gi is suppressed in the pixel values projected onto the tomographic plane Tj. Can do.

  The size of the predetermined range may be stored in the storage 23 as a fixed value. Further, it may be changed to an arbitrary value by an input from the input unit 4 by the operator. In this case, the value of the predetermined range stored in the storage 23 is rewritten by the input from the input unit 4 by the operator, and the predetermined range is changed. .

  Further, the degree of smoothing, that is, the degree of noise suppression can be changed by changing the filter size of the Gaussian filter. Specifically, noise can be further suppressed as the filter size is increased to increase the filtering range centered on the target coordinate position. Here, when the projection image Gi is photographed, the noise included in the projection image Gi increases as the X-ray dose reaching the radiation detector 15 decreases, and as a result, the noise of the pixel values projected onto the tomographic plane Tj increases. Become. The amount of noise included in the projection image Gi also changes depending on the quality of the X-rays, that is, whether the X-rays are high pressure or low pressure. Further, the amount of noise included in the projection image Gi also changes depending on the type of the radiation detector 15 used during imaging. Furthermore, a scattered radiation removal grid may be arranged between the radiation detector 15 and the subject M in order to prevent the influence of X-ray scattered rays on the subject M when photographing the projection image Gi. The amount of noise included in the projection image Gi also changes depending on the type of the scattered radiation removal grid or the presence or absence of the scattered radiation removal grid.

  Therefore, in the present embodiment, the imaging conditions such as the X-ray dose reaching the radiation detector 15, the X-ray quality, the type of the radiation detector 15, the type of the scattered radiation removal grid, and the presence or absence of the scattered radiation removal grid are used. Based on this, the characteristics of the smoothing filter are changed. For example, in the case of shooting conditions in which noise included in the projection image Gi increases, the filter size is increased so that the noise is further suppressed.

  Further, when a Gaussian filter is used when performing filtering, there is a possibility that an edge which is a structure of the subject M included in a tomographic image generated as described later may be blurred. For this reason, the pixel adjacent to the target coordinate position is filtered using a bilateral filter that weights according to the distance between the pixels and a normal distribution in which the weighting decreases as the difference between the pixel values increases. You may do it. Also, filtering may be performed using a non-local mean filter (weighting) based on the similarity between the target coordinate position on the tomographic plane Tj and the neighborhood of each arbitrary pixel. Good. As a result, the edge can be preserved while suppressing noise, so that it is possible to prevent the sharpness from being lowered in the tomographic image generated as described later.

  Further, for example, by filtering the pixel value projected on the tomographic plane Tj using a differential filter, an edge that is a structure of a subject in which the pixel value rapidly changes beyond a predetermined threshold is detected. The degree of change in sharpness may be changed by changing the filter characteristics so as to perform the filtering process along the existing direction. Further, filtering processing may be performed on the pixel values at the edge boundary so as not to use the pixel values existing at positions beyond the edge. Thereby, since the edge is not smoothed, it is possible to prevent the sharpness from being lowered in the distribution of the pixel values projected onto the tomographic plane Tj while suppressing noise.

  Instead of smoothing or in addition to smoothing, the edge may be emphasized by performing processing for enhancing sharpness. In this case, it is preferable to perform a process of enhancing the sharpness along the direction in which the edge exists.

  After the filtering is performed in this manner, the pixel value calculation unit 33 performs a regression analysis on the pixel value of the projection image projected on the tomographic plane Tj, thereby obtaining a curved surface representing the tomographic image on the tomographic plane Tj. Generate. Here, in order to facilitate explanation, the regression surface is considered as a regression curve. Regression analysis is a statistical technique for analyzing multivariate relationships. Here, it is assumed that the observed value on the observation point is observed with noise included in the true value. Regression analysis is a technique for solving an inverse problem for obtaining a true value at every observation point by a least square method, a moving average method, regression using a kernel, or the like. In the first embodiment, the coordinate position where the pixel value of the projection image Gi on the tomographic plane Tj is projected is the observation point uk, the pixel value of the projection image projected on the observation point uk is the observation value qk, and the target coordinate position. The pixel value calculated in um is set to the true value rm, and the pixel value rm at the target coordinate position um is calculated by regression analysis.

  Here, when the least squares method is used, it is assumed that the true value follows a function whose distribution is defined by γ parameters a. That is, it is assumed that r = f (u | a1, a2,... Aγ). Parameters a1, a2,... That minimize the square error between the true value and the observed value. The function f can be determined by obtaining aγ. Specifically, the pixel value rm at the target coordinate position is calculated by determining the parameters of the function f so that the sum of the errors of the observation values at the observation point is minimized by the following equation (3), and the regression is performed. Generate a curve (curved surface). As shown in equation (4), a weight wk is set for each observation point uk, and a regression curve (curved surface) is generated by calculating the pixel value rm at the coordinate position of interest um by the weighted least square method. May be.

Further, when using the moving average, the pixel value at the target coordinate position is calculated by the moving average to generate a regression surface. Specifically, in order to simplify the explanation, when a regression surface is considered as a regression curve, three pixel positions adjacent to the target coordinate position um, that is, for example, the coordinate position uk, for the pixel value of the target coordinate position um. The average value {(qk−1) + qk + (qk + 1)} / 3, which is the average value of the pixel values of the projected image Gi projected onto −1, uk, uk + 1, is calculated, and the calculated average value is used as the pixel value of the target coordinate position um. And A weight may be set for each pixel value. For example, the weight may be set so as to decrease as the distance from the target coordinate position um increases.

  When a regression method using a kernel is used, the regression is determined by determining the kernel according to the following equation (5) for the observation point uk and the target coordinate position um on the tomographic plane Tj on which the pixel value of the projection image is projected. A curve (curved surface) is calculated. In equation (5), argmin represents calculating the value of r (um) that minimizes the right side.

Note that the above-described filtering process for smoothing can be incorporated into the regression analysis. When the least square method is used, the degree of smoothing of the generated regression surface can be increased by using a low-dimensional function as the function f. On the contrary, the sharpness of the generated regression surface can be increased by using a high-dimensional function. When the moving average is used, the degree of smoothing can be changed by changing the number of pixels for which the average is calculated or changing the degree of weighting. That is, the degree of smoothing can be increased by increasing the number of pixels for which the average is calculated. Further, as described above, the degree of weighting may be changed according to the shooting conditions to change the degree of smoothing.

  In the regression method using the kernel, the degree of smoothing can be changed according to the design of the kernel. In particular, in the regression method using the kernel, the same effect as the weighted least square method can be obtained or the edge can be emphasized according to the design of the kernel. For example, when the kernel shown in the following equation (6) is used, a pixel value that is closer to the observation value qk of the observation point uk is calculated as the distance between the target coordinate position um and the observation point uk is shorter. A regression surface will be generated. For this reason, the same effect as smoothing using a Gaussian filter can be obtained. In Expression (6), hx is a bandwidth parameter.

By the way, in FIG. 7, among the five pixel values projected on the rightmost pixel position on the tomographic plane Tj, two pixel values are greatly different from those of adjacent pixel positions. When there is a pixel value that is significantly different from the pixel value of the adjacent pixel as described above, when a regression surface is generated, the pixel position including the outlier is significantly different from the adjacent pixel position as shown in FIG. It becomes. For this reason, when a tomographic image is generated from the calculated regression surface as described later, artifacts occur at pixel positions that are outliers.

  For this reason, the pixel value calculation unit 33 determines pixel values greatly different from adjacent pixel values as pixel outliers in the pixel values projected on the tomographic plane Tj, excludes pixel values that are outliers, and coordinates of interest The pixel value of the position is calculated. For example, on the tomographic plane Tj, the difference between the average value of the pixel values of the pixel positions adjacent to the target coordinate position and each of the plurality of pixel values projected on the target coordinate position is calculated, and the difference is set to a predetermined threshold value. What is necessary is just to determine the pixel value which exceeded as an outlier and exclude the pixel value which becomes an outlier in the regression analysis. Instead of excluding outliers, the weighting of pixel values that are outliers may be reduced.

  When the regression curve (curved surface) is calculated by excluding outliers or reducing outlier weighting in this way, as shown in FIG. 9, even at pixel positions including outliers, adjacent pixel positions and values are calculated. Will not change significantly. Thereby, it is possible to prevent an artifact from being included in the tomographic image.

  It is also possible to incorporate a process for removing outliers in the regression analysis. In the case of using the least square method, the weighted least square method shown in the above equation (4) is used, and the weighting of pixel values that are outliers may be set to 0 or smaller. When the moving average is used, a weighted average is obtained, and the weighting of pixel values that are outliers may be set to 0 or small.

  When the regression surface is generated, the pixel value calculation unit 33 samples the regression surface at a desired sampling interval to generate a tomographic image. Note that the sampling interval may be stored in the storage 23 as a fixed value. Further, it may be changed to an arbitrary value by an instruction from the input unit 4. For example, if the sampling interval is the same as the projection image, the tomographic image has the same resolution as the projection image Gi, and if the sampling interval is smaller than the projection image G1, the tomographic image has a higher resolution than the projection image Gi. Can do. Conversely, if the sampling interval is made larger than that of the projection image Gi, the tomographic image can have a lower resolution than that of the projection image Gi. In this case, by the input from the input unit 4 by the operator, the value of the sampling interval stored in the storage 23 is rewritten, and the sampling interval is changed. Further, the sampling interval may be adjusted according to the resolution of the display unit 3.

  Next, processing performed in the first embodiment will be described. FIG. 10 is a flowchart showing processing performed in the first embodiment. When the input unit 4 accepts an instruction to start processing by the operator, tomosynthesis imaging is performed, and the image acquisition unit 31 acquires a plurality of projection images Gi (step ST1). Next, the pixel value projection unit 32 holds the pixel values of the plurality of projection images acquired by the image acquisition unit 31 and sets the pixel values of the projection image Gi at the coordinate position on the desired tomographic plane Tj of the breast M. Projection is performed (step ST2).

  Then, the pixel value calculation unit 33 performs regression analysis on the pixel value of the projection image Gi projected on the tomographic plane Tj (step ST3), and generates a regression surface representing the tomographic image on the tomographic plane Tj (step ST4). ). Further, the pixel value calculation unit 33 samples the regression curved surface at a predetermined sampling interval to generate a tomographic image (step ST5), and ends the process. Note that when generating a tomographic image on another tomographic plane, the position of the tomographic plane may be changed and the processing of step ST1 to step ST5 may be performed.

  As described above, in the first embodiment, the positions of the X-ray source 16 and the radiation detector 15 at the time of imaging for each of the plurality of projection images Gi are held while maintaining the pixel values of the plurality of projection images Gi. Based on the positional relationship, the pixel values of the plurality of projected images Gi are projected onto the desired coordinate position on the tomographic plane Tj of the breast M as the subject, and predetermined based on the target coordinate position on the tomographic plane Tj. Based on a plurality of pixel values of the projected image Gi projected onto the range, for example, a regression surface is generated by regression analysis, thereby calculating a pixel value of the coordinate position of interest and generating a tomographic image. . For this reason, the influence of the pixel values around the target coordinate position is considered as compared with the conventional method of calculating the pixel value of the target coordinate position using only the pixel value of the projection image Gi projected on the target coordinate position. As a result, artifacts can be reduced and a higher-quality tomographic image can be generated. In addition, unlike the successive approximation reconstruction method, it is not necessary to repeat projection and reprojection, so that the calculation time can be greatly reduced.

  In addition, a tomographic image having a desired resolution can be generated by sampling a regression surface at a desired sampling interval and calculating a pixel value at a target coordinate position.

  Next, a second embodiment of the present invention will be described. The configuration of the tomographic image generation apparatus according to the second embodiment is the same as the configuration of the tomographic image generation apparatus according to the first embodiment, and only the processing to be performed is different. Therefore, detailed description of the apparatus is omitted here. To do. In the first embodiment, as shown in FIG. 5, the pixel value of the projection image Gi that intersects with the straight line connecting the source position and the pixel position on the tomographic plane Tj is taken as the tomographic plane located on this straight line. Projecting to the pixel value at the pixel position on Tj. In the second embodiment, in each of the plurality of radiation source positions Si, the coordinate position on the tomographic plane Tj intersecting the straight line connecting each radiation source position Si and the pixel position on the projection image Gi is on this straight line. The point which projected the pixel value in the pixel position of the projection image Gi located differs from 1st Embodiment.

  FIG. 11 is a diagram for explaining the projection of pixel values in the second embodiment. In FIG. 11, the projection image Gi acquired at the radiation source position Si is projected onto a desired tomographic plane Tj (j = 1 to m: m is the number of tomographic planes) among the tomographic planes in the breast M. The case where it does is demonstrated. In the second embodiment, as shown in FIG. 11, the line connecting the line source position Si and the pixel position on the projected image Gi and the coordinate position where the tomographic plane Tj intersects are located on this line. The pixel value of the projection image Gi is projected. Here, the coordinates (sxi, syi, szi) of the radiation source position at the radiation source position Si, the coordinates Tj (tx, ty, tz) of the pixel position on the tomographic plane Tj, and the coordinates of the coordinate position Pi on the projection image Gi ( The relationship of pxi, pyi) is as shown in the above equation (1). Therefore, pxi and pyi in the expression (1) are set as pixel positions of the projection image Gi, and by solving the expression (1) for tx and ty, the pixel value at the pixel position of the projection image Gi is projected on the tomographic plane Tj. The coordinate position can be calculated. Therefore, the pixel value of the projection image Gi may be projected onto the calculated coordinate position on the tomographic plane Tj.

  In this case, as shown in FIG. 11, the intersection Tjx between the line connecting the radiation source position Si and the pixel position on the projection image Gi and the tomographic plane Tj may not be the pixel position on the tomographic plane Tj. In the second embodiment, even in such a case, the pixel value of the projection image Gi is projected onto the coordinate position between the pixel positions. For this reason, in the second embodiment, the pixel value of the projection image is projected at a coordinate position other than the pixel position on the tomographic plane Tj.

  In the second embodiment, the pixel value projection unit 32 projects the pixel values of all the projection images G1 to Gn on the tomographic plane Tj for all the radiation source positions Si. As a result, as shown in FIG. 12, pixel values corresponding to the number of pixels of all the projection images Gi are projected at coordinate positions including pixel positions on the tomographic plane Tj.

  Also in the second embodiment, the pixel value calculation unit 33 performs a filtering process on the pixel values of the projection image Gi projected on the tomographic plane Tj, and performs a regression analysis, as in the first embodiment. By doing so, a regression surface is generated. Note that when performing the filtering process, the filtering may be performed around the coordinate position where the pixel value is projected on the tomographic plane.

  Here, in the second embodiment, since the pixel value of the projection image Gi is projected onto the coordinate position on the tomographic plane Tj as described above, the pixel value of the projection image Gi as in the first embodiment. Need not be interpolated. Therefore, compared with the first embodiment, the image quality of the projected image is not deteriorated during projection. Thereby, the image quality of the generated tomographic image can be made higher.

  In the second embodiment, the interval between the coordinate positions at which the pixel values are projected on the tomographic plane Tj is smaller than the interval between the pixel positions on the tomographic plane Tj, and as a result, the pixels on the tomographic plane Tj. The pixel value is at a coordinate position that is not a position. Therefore, a highly accurate regression surface can be generated on the basis of the pixel value of the actually acquired projection image even in a frequency band higher than the Nyquist frequency determined by the interval between the pixel positions on the tomographic plane Tj. Accordingly, in the second embodiment, even when the sampling interval of the regression surface is reduced, image quality deterioration can be suppressed, so that a higher quality tomographic image can be generated.

  Next, a third embodiment of the present invention will be described. The configuration of the tomographic image generation apparatus according to the third embodiment is the same as the configuration of the tomographic image generation apparatus according to the first embodiment, and only the processing to be performed is different. Therefore, detailed description of the apparatus is omitted here. To do. In the third embodiment, the image acquisition unit 31 further acquires a radiation image by normal X-ray imaging, and the pixel value projection unit 32 corrects the projection position of the pixel value of the projection image on the tomographic plane Tj. This is different from the first embodiment. Note that normal X-ray imaging is performed by irradiating a subject with X-rays according to imaging conditions for acquiring a transmission image of the subject while the X-ray source 16 is fixed at a specific source position without moving. Shooting.

  Here, as shown in FIG. 13, the X-rays emitted from the X-ray source 16 become a cone beam that spreads away from the X-ray source 16. Further, the surface of the radiation detector 15 from which the projection image Gi is acquired is located farther from the X-ray source 16 than the tomographic plane Tj. Therefore, when the breast M is imaged at a specific source position among the plurality of source positions Si, the positions of the structures B0 such as the mammary gland and calcification included in the breast M are as shown in FIG. The projection image Gi detected by the radiation detector 15 is different from the tomographic image TGj on the tomographic plane Tj.

  When acquiring a radiographic image by normal X-ray imaging, the position of the X-ray source 16 is fixed to a specific source position, and imaging of the breast M is performed under imaging conditions for acquiring a transmission image of the breast M. For this reason, the geometric positional relationship between the projection image acquired at the specific radiation source position and the radiation image is the same. Therefore, when acquiring a radiographic image and making a diagnosis by arranging the radiographic image and the tomographic image, the position of the corresponding structure B0 differs between the radiographic image and the tomographic image. For this reason, in the third embodiment, the pixel value projection unit 32 matches the coordinate position of interest on the projection image Gi with the coordinate position on the tomographic plane Tj on which the pixel value of the coordinate position of interest is projected. As described above, on the tomographic plane Tj on which the pixel value of the target coordinate position on the projection image Gi is projected based on the positional relationship between the source position Si at the time of shooting the projection image Gi and the target coordinate position on the projection image Gi. The coordinate position is corrected.

  Hereinafter, the correction of the projection position will be described. As shown in FIG. 13, assuming that the coordinates of the radiation source position at the radiation source position Si are (sxi, syi, szi) and the coordinates of the structure B0 on the tomographic plane Tj are Tj (tx, ty, tz), the radiation detector. The coordinates Pi (pxi, pyi) of the projection position of the structure B0 on 15 are expressed by the above formula (1).

  Here, when Pi is the coordinate position of interest, when the coordinate position is not corrected, the coordinate position Pi of interest is projected onto the coordinate position Tj on the tomographic plane Tj, and by solving Equation (1) for tx and ty, The coordinate position on the tomographic plane Tj on which the pixel value of the target coordinate position Pi is projected can be calculated.

  On the other hand, as shown in FIG. 13, when the X-ray source 16 exists on a line that passes through the coordinate position Pi (pxi, pyi) and is orthogonal to the detection surface of the radiation detector 15, in the tomographic image of the tomographic plane Tj, The target coordinate position Pi is projected at the intersection of the tomographic plane Tj and the straight line connecting the radiation source position Sc and the coordinate position Pi. As a result, in the tomographic image of the tomographic plane Tj, the structure B0 exists at the same two-dimensional coordinates as the projection image Gi. As a result, the position of the structure B0 is the projection taken at the radiation source position Sc. The image Gi matches the tomographic image on the tomographic plane Tj. Therefore, the reference source position when the pixel value of the projection image is projected onto the tomographic plane Tj is changed to the source position Sc on a straight line orthogonal to the coordinate position on the projection image as the projection pixel value. Thus, the coordinate position on the tomographic plane Tj on which the pixel value at the target coordinate position is projected is corrected. Here, if the coordinate position of the source position Sc that is the reference for correction is (sxc, syc, szc), the relationship between the corrected coordinate position Pi and the coordinate position (tx, ty) on the tomographic plane Tj is It is represented by the following formula (7).

In Expression (7), (szc−tz) / szc in the first term of the expression representing each of pxi and pyi is the enlargement ratio of the coordinate position, and the third term is the amount of movement of the coordinate position in the x and y directions. It represents. Therefore, by correcting Equation (7) for tx and ty, the corrected coordinate position on the tomographic plane Tj on which the pixel value of the target coordinate position Pi is projected can be calculated.

  As described above, by correcting the coordinate position on the tomographic plane Tj on which the pixel value of the projection image Gi is projected, the position of the structure such as a tumor included in the tomographic image and the correspondence included in the projection image, that is, the radiographic image. The positions of the structures to be matched can be matched. Therefore, the diagnosis using the radiographic image and the tomographic image can be performed more accurately.

  Next, a fourth embodiment of the present invention will be described. FIG. 15 is a diagram illustrating a schematic configuration of a tomographic image generation device according to the fourth embodiment. In FIG. 15, the same components as those in FIG. 3 are denoted by the same reference numerals, and detailed description thereof is omitted. The tomographic image generation apparatus according to the fourth embodiment is different from the first embodiment in that it includes a pseudo image generation unit 34 that generates a pseudo image from a plurality of tomographic images on a plurality of tomographic planes of the breast M.

  The pseudo image generating unit 34 generates an added tomographic image as a pseudo image by adding a plurality of tomographic images TGj generated on each of the plurality of tomographic planes Tj at corresponding pixel positions. The added tomographic image generated in this way is a pseudo-representation of a transmission image of the subject that is the same as the radiographic image acquired by normal X-ray imaging. Here, in the present embodiment, the tomographic image has a high image quality with reduced noise. Therefore, by adding a plurality of such tomographic images, a high-quality transmission image, that is, a pseudo radiation image can be generated.

  Here, in the fourth embodiment, when the pixel value of the projection image is projected onto the coordinate position of the tomographic plane, the coordinate position on the tomographic plane Tj is corrected as in the third embodiment. The position of the corresponding structure can be matched between the tomographic images. Therefore, it becomes easy to associate the pixel values at the time of addition, so that a higher quality radiation image can be generated.

  Note that the pseudo image generation unit 34 may generate, as a pseudo image, a maximum value projection image obtained by the MIP method for extracting the maximum value of the corresponding pixel position in a plurality of tomographic images instead of the addition tomographic image. Further, a minimum value projection image obtained by the minIP method for extracting the minimum value of a plurality of tomographic images may be generated as a pseudo image.

  Further, in each of the above embodiments, the generated tomographic image is displayed on the display unit 3, an instruction to change the size of the attention area on the tomographic image is received, and the attention area for which the change instruction has been received is enlarged and displayed. Good. In this case, on the regression surface, a region corresponding to the region of interest for which the size change instruction has been received is extracted, and sampling is performed with the sampling interval of the extracted region being smaller than the sampling interval of the tomographic image. Thereby, as shown in FIG. 16, an enlarged image A2 of the attention area A1 designated in the tomographic image TGj can be displayed. Here, the sampling interval is preferably a sampling interval determined by the resolution of the display unit 3. In FIG. 16, the enlarged image A2 is superimposed on the tomographic image TGj, but the enlarged image A2 may be displayed instead of the tomographic image TGj, or the tomographic image TGj and the enlarged image A2 may be displayed side by side. Good. Conversely, when an instruction to reduce the attention area is given, the sampling interval of the area corresponding to the attention area extracted on the regression surface may be set to be larger than the sampling interval of the tomographic image.

  Next, a fifth embodiment of the present invention will be described. FIG. 17 is a diagram showing a schematic configuration of a tomographic image generation device according to the fifth embodiment. In FIG. 17, the same components as those in FIG. 3 are denoted by the same reference numerals, and detailed description thereof is omitted. In the tomographic image generation device according to the fifth embodiment, an image composed of pixel values of a plurality of projection images Gi projected on the tomographic plane Tj is defined as a tomographic plane projection image TGi, and between the plurality of tomographic plane projection images TGi. This is different from the first embodiment in that a misalignment correction unit 35 that corrects the misalignment is provided.

  Here, when tomosynthesis imaging is performed, it is difficult to accurately move the radiation source to the calculated radiation source position due to the influence of apparatus errors such as vibration or mechanical shift during imaging. For this reason, the source position at the time of imaging | photography will shift | deviate from the calculated source position. Due to the effect of this shift, the projection position of the object cannot be accurately aligned, and as a result, the image quality of the tomographic image deteriorates.

  Also, when performing such tomosynthesis shooting, it takes several seconds as the shooting time from the start of shooting to the end of shooting since a plurality of images of the subject are shot based on one shooting start instruction. The subject may move inside. When such a body motion, which is the movement of the subject, occurs, the projection position of the object cannot be accurately aligned at the time of reconstruction, and as a result, the image quality of the tomographic image deteriorates.

  Moreover, since the effects of apparatus errors and body movements occur three-dimensionally and nonlinearly in the subject, a tomographic image can be obtained by simply translating, rotating, and enlarging / reducing a plurality of projection images Gi by affine transformation or the like. It is difficult to remove the effects of device errors and body movements from

  In the fifth embodiment, the misalignment correction unit 35 corrects misalignment between a plurality of tomographic projection images TGi. Here, when an apparatus error or body movement of the breast M does not occur during imaging with the X-ray source 16 moved to a plurality of source positions Si, the pixel values of the plurality of tomographic projection images TGi are: As shown in FIG. However, when an apparatus error or body movement occurs, the tomographic plane projection image corresponding to the projection image taken when the apparatus error or body movement occurs corresponds to another tomographic plane projection image as shown in FIG. The pixel position to be shifted is shifted. The position deviation correction unit 35 corrects the position deviation between the plurality of tomographic plane projection images TGi so that the positions of the plurality of tomographic plane projection images TGi match. Specifically, the misregistration correction unit 35 uses an algorithm such as SIFT (Scale-Invariant Feature Transform) or SURF (Speeded Up Robust Features), and includes an edge and an intersection of edges included in each tomographic plane projection image TGi. Then, feature points such as corners of edges are detected, and the positional deviation is corrected by converting each tomographic plane projection image TGi so as to match the detected feature points. As a result, the positional deviation of the plurality of tomographic plane projection images TGi is corrected as shown in FIG. 18, and the distribution of pixel values in the plurality of tomographic plane projection images TGi substantially matches.

  Here, SIFT is a technique for describing a feature quantity that is invariant to image rotation and scale change at a feature point, and aligning a plurality of images based on the described feature quantity. SURF is a technique for performing alignment at higher speed by replacing the processing performed in SIFT with approximation processing. Note that the misalignment correction processing performed in the present embodiment is not limited to SIFT or SURF, and any method can be used.

  In the fifth embodiment, the pixel value calculation unit 33 calculates each pixel value on the tomographic plane Tj after correcting the positional deviation of the plurality of tomographic plane projection images TGi, as in the above embodiments. Then, a tomographic image on the tomographic plane Tj is generated.

  As described above, in the fifth embodiment, since the positional deviation of the plurality of tomographic plane projection images TGi projected on the tomographic plane Tj is corrected, the influence of the apparatus error and the body movement that occur three-dimensionally. Is considered only as a two-dimensional displacement on the tomographic plane Tj. Therefore, it is possible to effectively remove the position error caused by the apparatus error and the body movement in the tomographic plane Tj, and as a result, the image quality of the tomographic image can be further improved.

  Note that in the tomographic images of adjacent tomographic planes, the pixel values at corresponding pixel positions tend to be the same value. Therefore, in each of the above embodiments, a regression surface may be generated on a plurality of tomographic planes in the breast M, and the regression surface of the tomographic plane of interest may be smoothed using the regression surface on the adjacent tomographic plane. . Specifically, for the regression surface of the tomographic plane Tj, the correlation between the pixel values of the corresponding pixel positions on the regression surfaces of the three adjacent tomographic planes Tj−1, Tj, and Tj + 1 is calculated. Then, only when the correlation exceeds a predetermined threshold, smoothing is performed by calculating the average of the pixel values of the three tomographic planes Tj−1, Tj, and Tj + 1 for the pixel position. For the correlation, for example, an absolute value of a difference value between pixel values at corresponding pixel positions can be used. Thereby, since the influence of the pixel value of the adjacent tomographic plane can be taken into consideration, a high-quality tomographic image in which noise is further reduced can be generated.

  In each of the above embodiments, after generating the regression surface, the absolute value of the difference value at the corresponding coordinate position between the pixel value projected on the tomographic plane Tj and the regression surface is calculated, and the absolute value of the difference value is Pixel values exceeding a predetermined threshold value may be determined as outliers, and outliers may be excluded, or the weight of outliers may be reduced to correct the regression surface. For example, as shown in FIG. 8, when a regression curve is generated, the difference value between the five pixel values projected on the rightmost pixel position on the tomographic plane Tj and the value of the regression curve at this pixel position is calculated. Calculate the absolute value. In this case, the absolute value of the difference value is larger than the threshold value for two pixel values having a large value. For this reason, the regression curve is corrected by excluding two pixels having large values as outliers or by reducing the weighting of pixel values that are outliers. As a result, as shown in FIG. 9, a regression curve can be generated so that the value does not vary greatly from the adjacent pixel position even at a pixel position including an outlier.

  In each of the above embodiments, the regression surface is generated after the pixel values of all the projection images Gi are projected onto the tomographic plane Tj. However, the regression surface may be generated for each projection image Gi. In this case, the correlation at the corresponding coordinate position is calculated between the regression surfaces calculated for each projection image Gi. Note that the absolute value of the difference value can be used as the correlation. If there is a regression surface having a small correlation, that is, a pixel having a pixel value whose absolute value of the difference value exceeds a predetermined threshold, a pixel value at a coordinate position having a small correlation on the regression surface is determined as an outlier. Then, the regression surface may be generated based on the pixel values of all the projection images Gi projected on the coordinate positions on the tomographic plane Tj by excluding the outliers or reducing the weighting of the outliers.

  Further, when tomosynthesis imaging is performed, the X-ray dose reaching the radiation detector 15 differs depending on whether the X-rays are incident so as to be orthogonal to the radiation detector 15 or obliquely incident. For this reason, in the projection image Gi, the image quality differs for each radiation source position at the time of shooting. Specifically, in the projection image acquired when the X-rays are incident so as to be orthogonal to the radiation detector 15 and the projection image acquired when the X-rays are incident obliquely, the latter is more radiation detection. Since the radiation dose to the instrument 15 is reduced, the density of the projected image is lowered and the image quality is whitish. Further, density unevenness may occur on the projected image due to the influence of the characteristics of the radiation image capturing apparatus. For this reason, it is preferable to project the pixel value of the projection image Gi on the coordinate position on the tomographic plane Tj after correcting the difference in image quality and density unevenness between the projection images Gi.

  Here, the density difference and density unevenness based on the arrival dose are included in the low-frequency component of the projection image Gi. The low frequency is a frequency that is appropriately set at 50% or less of the Nyquist frequency determined by the interval between the pixel positions of the projection image Gi. For this reason, it is preferable to remove the low frequency component from each projection image Gi and project the pixel value of the projection image Gi from which the low frequency component is removed onto the coordinate position of the tomographic plane Tj. Thereby, since the difference in image quality between the projection images Gi can be corrected, a tomographic image with higher image quality can be generated. The removal of the low frequency component may be performed by, for example, calculating the low frequency component by performing frequency decomposition on the projection image and subtracting the calculated low frequency component from the projection image. Alternatively, a blurred image of the projected image may be generated and the blurred image may be subtracted from the projected image.

  In the above embodiment, the generation of the regression surface may be repeated sequentially. For example, in the method of calculating the regression surface using the weighted least square method described above, after generating the regression surface, the weight wk of each observation point uk is calculated by the following equation (8). F (uk) in Equation (8) is a value at the observation point uk in the calculated regression curve.

According to Equation (8), the greater the difference between the observed value qk and the value of the observation point uk on the regression curve, the smaller the value of the weight k. Then, when the regression curve is generated by the above formula (4) using the calculated weight wk, the newly generated regression curve is generated so that the influence of the value deviating from the previously generated regression curve is smaller. Is done. In particular, when there is an outlier, the outlier is weighted less, so that when a tomographic image is generated, artifacts that occur at pixel positions that are outliers can be reduced. Note that the number of times the regression surface generation is repeated may be set in advance, or may be repeated until a predetermined convergence condition is satisfied. As the convergence condition, for example, a condition in which a difference by repeated calculation, that is, a value of qk−f (uk) is equal to or less than a predetermined threshold value can be used.

  In addition, when generating a regression curve using a moving average or a kernel, the generation of the regression surface may be repeated sequentially. When using the moving average, the weight for the observation point uk may be calculated in the same manner as the above equation (8), and the calculation of the weighted moving average may be repeated. In the case of the method using the kernel, a term for calculating the weight wk as in the above equation (8) is incorporated in the equation for determining the kernel, and the regression method is generated by repeating the method using the kernel.

  In each of the above embodiments, the pixel value calculation unit 33 generates a regression surface to generate a tomographic image. However, without generating a regression surface, the pixel value at the pixel position on the tomographic plane Tj is A tomographic image may be generated by calculation by regression analysis. In this case, the sampling interval of the generated tomographic image is set in advance, the coordinate position that becomes the sampling interval is set as the target pixel position on the tomographic plane Tj, and the pixel value of this target pixel position is calculated by regression analysis. You can do it.

  In each of the embodiments described above, tomosynthesis imaging is performed using the breast M as a subject, but the present invention can of course be applied to cases where tomosynthesis imaging is performed using subjects other than the breast.

  In each of the above embodiments, only the X-ray source 16 is moved. However, depending on the imaging apparatus, the X-ray source 16 and the radiation detector 15 can be moved in synchronization. In such a case, the X-ray source 16 and the radiation detector 15 may be moved in synchronization. Alternatively, the X-ray source 16 may be fixed and only the radiation detector 15 may be moved.

  In each of the above embodiments, the present invention is applied to an imaging apparatus that performs tomosynthesis imaging. However, the present invention is applicable to an arbitrary imaging apparatus that acquires a plurality of projection images by imaging a subject at a plurality of radiation source positions. The invention can be applied. For example, a radiation source and a radiation detector are arranged so as to face each other with the subject as the center, and these sets are rotated around the subject, and radiation is irradiated from various angles to obtain a plurality of projection images. The present invention can also be applied to a CT imaging apparatus.

  In the above embodiments, the trajectory of the X-ray source 16 is an arc, but may be a straight line.

  Hereinafter, the effect of the embodiment of the present invention will be described.

  At each of the plurality of radiation source positions, the pixel value at the pixel position of the projection image located on the straight line is set to the pixel value at the coordinate position where the straight line connecting the radiation source and the pixel position on the projection image intersects the tomographic plane. By projecting, interpolation processing is not performed when projecting pixel values. For this reason, a higher-quality tomographic image can be generated. In particular, since the pixel value is projected on the tomographic plane at a position different from the coordinate position to be the pixel position of the tomographic image, a high-resolution and high-quality tomographic image can be generated.

  In addition, by performing regression analysis, generating a regression surface representing a tomographic image on the tomographic plane, sampling the regression curved surface at a desired sampling interval, and calculating the pixel value of the pixel position on the tomographic plane, A tomographic image having a desired resolution can be generated.

  Also, when performing regression analysis, by changing the sharpness of the pixel value at the target coordinate position, by enhancing the sharpness in the pixel value at the target coordinate position, or by smoothing by reducing the sharpness Noise can be suppressed. For this reason, a tomographic image having a desired image quality can be generated.

  In addition, by changing the degree of change in sharpness according to the shooting conditions at the time of shooting and at least one information of the structure of the subject included in the projection image, a desired value corresponding to at least one of the shooting conditions and the structure of the subject is obtained. A tomographic image having the following image quality can be generated.

  Further, out of the pixel values of the projected image projected in a predetermined range based on the target coordinate position, pixel values that are outliers are excluded, or weighting of pixel values that are outliers is reduced. By calculating the pixel value of the target coordinate position, it is possible to reduce the influence of the pixel value that is highly likely to be an artifact, thereby generating a higher-quality tomographic image.

  Note that low-frequency unevenness may occur in a plurality of projected images due to differences in radiation source positions. For this reason, by removing the low-frequency component of the projection image and projecting the pixel value of the projection image from which the low-frequency component has been removed onto the coordinate position on the tomographic plane, it is possible to reduce the influence of low-frequency unevenness. As a result, a tomographic image with higher image quality can be generated.

  Further, the radiation emitted from the radiation source is a cone beam that spreads away from the radiation source. Here, since the surface of the detection means from which the projection image is acquired is located farther from the radiation source than the tomographic plane, the position on the two-dimensional coordinate of the structure included in the subject is the tomographic plane including this structure. The tomographic image and the projected image are different.

  For this reason, the two-dimensional coordinates of the target coordinate position on the projection image projected from the specific radiation source position and the two-dimensional coordinates of the coordinate position on the tomographic plane onto which the pixel value of the target coordinate position is projected match. In addition, based on the positional relationship between the specific source position and the target coordinate position, the coordinate position on the tomographic plane where the pixel value of the target coordinate position is projected is corrected, and the pixel values of a plurality of projected images are By projecting to the corrected coordinate position above, the position of the two-dimensional coordinate of the structure included in the subject can be matched between the tomographic image and the projection image acquired at the specific source position.

  In this case, a tomographic image is obtained by photographing a subject at a specific source position, obtaining a radiation image, and calculating a pixel value at a coordinate position of interest on a tomographic plane in which the pixel value of the projection image is projected at the corrected coordinate position. Is generated and the radiographic image and the tomographic image are displayed, so that the positions of the two-dimensional coordinates of the structures included in the subject coincide in the displayed radiographic image and the tomographic image. Therefore, diagnosis can be performed more appropriately.

  By generating tomographic images of a plurality of tomographic planes of a subject and adding the plurality of tomographic images to generate an added tomographic image, a transmission image with higher image quality can be generated in a pseudo manner.

DESCRIPTION OF SYMBOLS 1 Radiographic imaging apparatus 2 Computer 3 Display part 4 Input part 10 Imaging part 15 Radiation detector 16 X-ray source 17 Compression board 21 CPU
22 Memory 23 Storage 31 Image acquisition unit 32 Pixel value projection unit 33 Pixel value calculation unit 34 Pseudo image generation unit 35 Position shift correction unit

Claims (19)

A plurality of radiation sources corresponding to each of the plurality of radiation source positions, which are photographed by moving the radiation source relative to the detection unit and irradiating the subject with radiation at the plurality of radiation source positions by the movement of the radiation source. Image acquisition means for acquiring a projected image of
While holding the pixel values of the plurality of projection images, the pixel values of the plurality of projection images are determined based on the positional relationship between the radiation source position and the detection unit at the time of photographing for each of the plurality of projection images. Pixel value projecting means for projecting to a desired coordinate position on the tomographic plane of the subject;
By calculating the pixel value of the coordinate position of interest based on a plurality of pixel values of the projected image projected onto a predetermined range based on the coordinate position of interest on the tomographic plane, the tomographic plane of the tomographic plane is calculated. A tomographic image generation apparatus comprising a pixel value calculation means for generating an image.   The pixel value projection means, at each of the plurality of radiation source positions, calculates a pixel value at a coordinate position of the corresponding projection image that intersects a straight line connecting the radiation source position and the pixel position on the tomographic plane. The tomographic image generation apparatus according to claim 1, wherein the tomographic image generation device is a unit that projects the pixel value at a pixel position on the tomographic plane located on a straight line.   In each of the plurality of source positions, the pixel value projecting means is arranged on the straight line at a coordinate position on the tomographic plane that intersects a straight line connecting each source position and the corresponding pixel position on the projection image. The tomographic image generation device according to claim 1, wherein the tomographic image generation device is a unit that projects a pixel value at a pixel position of the projection image located at a position.   The tomographic image generation apparatus according to claim 3, wherein an interval between coordinate positions on the tomographic plane is smaller than an interval between pixel positions on the tomographic plane.   The pixel value calculation means is means for calculating a pixel value at the target coordinate position by performing regression analysis on the pixel value of the projection image projected onto the tomographic plane. The tomographic image generation apparatus according to item 1.   The pixel value calculation means performs the regression analysis, generates a regression surface representing a tomographic image on the tomographic plane, samples the regression surface at a desired sampling interval, and pixels at pixel positions on the tomographic plane The tomographic image generation apparatus according to claim 5, which is means for generating the tomographic image by calculating a value.   The tomographic image generation device according to claim 6, wherein the sampling interval is different from a sampling interval of the projection image.   When the pixel value calculation unit receives an instruction to change the size of the region of interest in the displayed tomographic image, the pixel value calculation unit changes the sampling interval of the region corresponding to the region of interest on the regression surface according to the change instruction. The tomographic image generation apparatus according to claim 6 or 7, which is means for generating a tomographic image of the region of interest.   The tomographic image generation apparatus according to claim 5, wherein the pixel value calculation unit is a unit that changes the sharpness of the pixel value at the target coordinate position when performing the regression analysis.   10. The pixel value calculating means is means for changing the degree of change of the sharpness according to information on at least one of the photographing condition at the time of photographing and the structure of the subject included in the projected image. Tomographic image generator.   The pixel value calculation means excludes a pixel value that is an outlier from a plurality of pixel values of the projection image projected on a predetermined range with the target coordinate position as a reference, or the outlier 11. The tomographic image generation apparatus according to claim 1, wherein the tomographic image generation apparatus is a unit that calculates a pixel value of the target coordinate position by reducing a weight of a pixel value that becomes.   2. The pixel value projecting unit is a unit that removes a low-frequency component of the projection image and projects the pixel value of the projection image from which the low-frequency component has been removed onto a coordinate position on the tomographic plane. The tomographic image generation device according to any one of 1 to 11.   The pixel value projection means includes a two-dimensional coordinate of a target coordinate position on the projection image corresponding to a specific source position, and a coordinate on the tomographic plane onto which a pixel value of the target coordinate position on the projection image is projected. The pixel value of the target coordinate position on the projection image is projected based on the positional relationship between the specific source position and the target coordinate position on the projection image so that the two-dimensional coordinates of the position coincide with each other. 13. The tomography according to claim 1, wherein the tomography unit is a unit that corrects the coordinate position on the tomographic plane and projects pixel values of the plurality of projection images onto the corrected coordinate position on the tomographic plane. Image generation device. The image acquisition means is means for acquiring a radiographic image of the subject imaged by irradiating the subject with the radiation at the specific radiation source position,
The pixel value calculation means is means for generating the tomographic image by calculating a pixel value of the target coordinate position on the tomographic plane in which the pixel value of the projection image is projected on the corrected coordinate position,
The tomographic image generation apparatus according to claim 13, further comprising display means for displaying the radiation image and the tomographic image. The pixel value projection means and the pixel value calculation means are means for generating the tomographic images on a plurality of tomographic planes of the subject,
The tomographic image generation apparatus according to claim 1, further comprising a pseudo image generation unit configured to generate a pseudo image from the plurality of tomographic images.   The pixel value calculation means calculates a weighting coefficient based on a difference between a pixel value at the coordinate position of interest and a pixel value at the coordinate position of the projection image corresponding to the coordinate position of interest, and the plurality of projection images The pixel value of the target coordinate position is calculated again based on the pixel value and the weighting coefficient, and a new pixel value of the target coordinate position is calculated. The tomographic image generation device according to item.   The pixel value calculation means calculates a new weighting coefficient using the new pixel value at the target coordinate position, and the target coordinates based on the plurality of pixel values and the new weighting coefficient of the projection image. The tomographic image generation apparatus according to claim 16, which is means for repeatedly calculating a new pixel value of a position again. A plurality of radiation sources corresponding to each of the plurality of radiation source positions, which are photographed by moving the radiation source relative to the detection unit and irradiating the subject with radiation at the plurality of radiation source positions by the movement of the radiation source. The projection image of
While holding the pixel values of the plurality of projection images, the pixel values of the plurality of projection images are determined based on the positional relationship between the radiation source position and the detection unit at the time of photographing for each of the plurality of projection images. Project to the desired coordinate position on the tomographic plane of the subject,
By calculating the pixel value of the coordinate position of interest based on a plurality of pixel values of the projected image projected onto a predetermined range based on the coordinate position of interest on the tomographic plane, the tomographic plane of the tomographic plane is calculated. A tomographic image generation method characterized by generating an image. A plurality of radiation sources corresponding to each of the plurality of radiation source positions, which are photographed by moving the radiation source relative to the detection unit and irradiating the subject with radiation at the plurality of radiation source positions by the movement of the radiation source. To obtain a projected image of
While holding the pixel values of the plurality of projection images, the pixel values of the plurality of projection images are determined based on the positional relationship between the radiation source position and the detection unit at the time of photographing for each of the plurality of projection images. Projecting to a desired coordinate position on the tomographic plane of the subject;
By calculating the pixel value of the coordinate position of interest based on a plurality of pixel values of the projected image projected onto a predetermined range based on the coordinate position of interest on the tomographic plane, the tomographic plane of the tomographic plane is calculated. A tomographic image generation program that causes a computer to execute a procedure for generating an image.