Image processing device, method, and recording medium

Abstract

A processor detects a structure of interest from a plurality of tomographic images representing a plurality of tomographic planes of a subject. The processor selects a tomographic image from the plurality of tomographic images based on a frequency band in a region where the structure of interest is detected. The processor generates a composite two-dimensional image using the selected tomographic image in the region where the structure of interest is detected, and generates a composite two-dimensional image using a predetermined tomographic image in a region where the structure of interest is not detected.

Description

Technical Field

[0001] The present invention relates to an image processing device, method and recording medium. Background Art

[0002] In recent years, diagnostic imaging using radiographic imaging devices that capture breast images (called mammography) has attracted attention to promote early detection of breast cancer. In mammographic X-ray imaging devices, tomography has been proposed, which involves moving a radiation source to irradiate the breast from multiple source positions, reconstructing the multiple projection images obtained to generate a tomographic image that emphasizes the desired cross-sectional plane. In tomographic imaging, the radiation source and radiation detector are moved parallel to each other or in a circular or elliptical arc, depending on the characteristics of the imaging device and the desired cross-sectional image. Multiple projection images are captured from multiple source positions, and these projection images are reconstructed using back-projection methods such as simple back-projection or filtered back-projection, or by successive reconstruction methods, to generate a cross-sectional image.

[0003] By generating these tomographic images across multiple slices within the breast, structures that overlap in the depth direction of the slices can be separated. This allows for the detection of abnormalities such as lesions, which are difficult to detect using conventional simple radiographic images (hereinafter referred to as simple two-dimensional images) obtained by irradiating the subject with radiation from a predetermined direction.

[0004] Furthermore, there is known a technique for synthesizing a plurality of tomographic images obtained by tomography and having different distances (positions in the height direction) from the detection surface of the radiation detector toward the radiation source side by an addition method, an averaging method, a maximum projection method, a minimum projection method, etc., thereby generating a virtual two-dimensional image (hereinafter referred to as a synthesized two-dimensional image) equivalent to a simple two-dimensional image (see Japanese Patent Gazette No. 2014-128716).

[0005] On the other hand, in the medical field, computer-aided image diagnosis systems (CAD) are known that automatically detect structures such as abnormal shadows in images and highlight the detected structures. For example, CAD is used to detect diagnostically important structures such as tumors, burrs, and calcifications from tomographic images obtained by tomography. In addition, a method has been proposed that, when generating a composite two-dimensional image from multiple tomographic images obtained by tomography of the breast, a region of interest containing a structure is detected by CAD, and the detected region of interest is synthesized onto a projected image or a two-dimensional image obtained by simple photography, for example, to generate a composite two-dimensional image (see U.S. Patent No. 8,983,156). In addition, a method has been proposed that generates a composite two-dimensional image by synthesizing tomographic images containing only structures detected by CAD by averaging (see U.S. Patent No. 9,792,703). Summary of the Invention

[0006] Technical issues to be solved by the invention

[0007] However, in the composite 2D image generated using the method described in U.S. Patent No. 8,983,156, the structure of interest synthesized into the 2D image is limited to the structure of interest acquired from a single slice image. Therefore, if the structure of interest spans multiple slice images, the composite 2D image cannot reflect the presence of the structure of interest in the depth direction of the slice images. Furthermore, the method described in U.S. Patent No. 9,792,703 averages the structure of interest included in multiple slice images. Consequently, for example, subtle structures of interest such as calcifications and linear structures such as burrs within the breast appear blurred and difficult to observe.

[0008] The present invention has been made in view of the above circumstances, and an object of the present invention is to facilitate observation of a structure of interest in the depth direction and a minute structure of interest included in a subject in a synthesized two-dimensional image.

[0009] Means for solving technical problems

[0010] The image processing device according to the present invention comprises at least one processor.

[0011] The processor is configured as follows:

[0012] detecting a region of interest from a plurality of tomographic images representing a subject, selecting a tomographic image from the plurality of tomographic images according to a frequency band in the region where the region of interest was detected,

[0013] A synthetic two-dimensional image is generated using the selected tomographic image in a region where the structure of interest is detected, and a synthetic two-dimensional image is generated using a preset tomographic image in a region where the structure of interest is not detected.

[0014] In the image processing apparatus according to the present invention, the processor may be configured to perform frequency decomposition on the plurality of tomographic images to derive a plurality of frequency band tomographic images for each of the plurality of frequency bands.

[0015] For each pixel corresponding to a pixel of the synthesized two-dimensional image among the plurality of band tomographic images, a band tomographic image corresponding to a tomographic image in which a structure of interest is detected is selected according to the band,

[0016] In the region where the structure of interest is detected, a synthetic two-dimensional image is generated using the selected band tomographic images.

[0017] Furthermore, in the image processing apparatus according to the present invention, the processor may be configured to select, from the plurality of band tomographic images, a different number of band tomographic images corresponding to tomographic images in which a structure of interest is detected, depending on the frequency band. Furthermore, the different number may be 0. In other words, no band tomographic image may be selected for a certain frequency band.

[0018] Furthermore, in the image processing device according to the present invention, the plurality of frequency bands may include a first frequency band and a second frequency band lower than the first frequency band.

[0019] The processor is configured to select, in the first frequency band, a smaller number of frequency band tomographic images than in the second frequency band.

[0020] Furthermore, in the image processing apparatus according to the present invention, the processor may be configured to select, in the second band, all band tomographic images including the structure of interest for each pixel corresponding to a pixel of the synthesized two-dimensional image among the plurality of band tomographic images.

[0021] Furthermore, in the image processing apparatus according to the present invention, the processor may be configured to select, in the second band, one band tomographic image that best represents the structure of interest for each pixel corresponding to a pixel of the synthesized two-dimensional image among the plurality of band tomographic images.

[0022] Furthermore, in the image processing apparatus according to the present invention, the processor may be configured to select, in the first band, one band tomographic image that best represents the structure of interest for each pixel corresponding to a pixel of the synthesized two-dimensional image among the plurality of band tomographic images.

[0023] Furthermore, in the image processing apparatus according to the present invention, one band tomographic image that best represents the structure of interest can be the band tomographic image in which the structure of interest is largest or the band tomographic image in which the likelihood of detecting the structure of interest is highest.

[0024] Furthermore, in the image processing apparatus according to the present invention, the processor may further select the band tomographic image according to the type of the structure of interest.

[0025] Furthermore, in the image processing apparatus according to the present invention, the structure of interest may be a tumor, a spur, or a calcification.

[0026] Furthermore, in the image processing device based on the present invention, the processor can be configured as follows: using the selected band tomographic image among the pixels of the band tomographic image corresponding to the structure of interest, a synthetic band two-dimensional image is generated for each band, and the synthetic band two-dimensional image is frequency synthesized to generate a synthetic two-dimensional image.

[0027] Furthermore, in the image processing device based on the present invention, the processor can be configured as follows: when multiple band tomographic images are selected, among pixels corresponding to pixels of the synthetic band two-dimensional image in the multiple band tomographic images, a synthetic band two-dimensional image having pixel values of the band tomographic images determined based on the priority of a pre-set structure of interest is generated.

[0028] Furthermore, in the image processing apparatus according to the present invention, the processor may be configured to generate a first synthesized two-dimensional image by synthesizing a plurality of tomographic images,

[0029] In the pixels of the band tomographic image corresponding to the structure of interest, a selected band tomographic image is used for each type of the structure of interest, a synthetic band two-dimensional image is generated for each band, the synthetic band two-dimensional image is subjected to frequency synthesis to generate a second synthetic two-dimensional image for each type of the structure of interest, and the second synthetic two-dimensional image generated for each type of the structure of interest is synthesized with the first synthetic two-dimensional image to thereby generate a synthetic two-dimensional image.

[0030] Furthermore, in an image processing device based on the present invention, the processor can be configured as follows: replacing the pixel value of the structure of interest in the first synthesized two-dimensional image with the pixel value of the structure of interest in the second synthesized two-dimensional image, thereby synthesizing the second synthesized two-dimensional image into the first synthesized two-dimensional image.

[0031] Furthermore, in an image processing device based on the present invention, the processor can be configured as follows: when a plurality of focus structures are included in corresponding pixels between a plurality of second synthetic two-dimensional images, a synthetic two-dimensional image having pixel values of the second synthetic two-dimensional image determined according to the priority of a pre-set focus structure is generated.

[0032] Furthermore, in the image processing apparatus according to the present invention, the processor may be configured to generate a first synthesized two-dimensional image by synthesizing a plurality of tomographic images,

[0033] extracting a region of a predetermined specific type of structure of interest from the first synthesized two-dimensional image,

[0034] For other structures of interest other than the specific type of structure of interest, the selected band tomographic image is used in the pixels of the band tomographic images corresponding to the other structures of interest, a synthetic band two-dimensional image is generated for each type of the other structures of interest, and the synthetic band two-dimensional image is frequency-synthesized to generate a second synthetic two-dimensional image for each type of the other structures of interest. The second synthetic two-dimensional image for the other structures of interest is synthesized into the first synthetic two-dimensional image, and the area of the specific type of structure of interest is synthesized on the first synthetic two-dimensional image synthesized with the second synthetic two-dimensional image, thereby generating a synthetic two-dimensional image.

[0035] Furthermore, in the image processing apparatus according to the present invention, the specific structure of interest may be calcification, and the other structures of interest may be tumors and spurs.

[0036] Furthermore, in an image processing device based on the present invention, the processor can be configured as follows: replacing the pixel value of the structure of interest in the first synthesized two-dimensional image with the pixel value of the structure of interest in the second synthesized two-dimensional image, thereby synthesizing the second synthesized two-dimensional image into the first synthesized two-dimensional image.

[0037] Furthermore, in an image processing device based on the present invention, the processor can be configured as follows: when a plurality of other structures of interest are included in corresponding pixels between a plurality of second synthetic two-dimensional images, a synthetic two-dimensional image having pixel values of the second synthetic two-dimensional image determined according to the priority of a pre-set structure of interest is generated.

[0038] Furthermore, in an image processing device based on the present invention, the processor can be configured as follows: replacing the pixel value of the structure of interest in the first synthetic two-dimensional image synthesized with the second synthetic two-dimensional image with the pixel value of an area of a specific type of structure of interest, thereby synthesizing an area of a specific type of structure of interest on the first synthetic two-dimensional image synthesized with the second synthetic two-dimensional image.

[0039] In the image processing method according to the present invention, a structure of interest is detected from a plurality of tomographic images representing a plurality of tomographic planes of an object.

[0040] In the region where the structure of interest is detected, a tomographic image is selected from a plurality of tomographic images according to a frequency band,

[0041] A synthetic two-dimensional image is generated using the selected tomographic image in a region where the structure of interest is detected, and a synthetic two-dimensional image is generated using a preset tomographic image in a region where the structure of interest is not detected.

[0042] Furthermore, the image processing method according to the present invention can be provided as a program for execution on a computer.

[0043] Effects of the Invention

[0044] According to the present invention, it is possible to easily observe a structure of interest in the depth direction and a minute structure of interest included in a subject in a synthesized two-dimensional image. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a schematic diagram of the configuration of a radiographic imaging system to which an image processing device according to an embodiment of the present invention is applied.

[0046] Figure 2 It is from Figure 1 FIG. 1 is a diagram of the radiographic imaging device as viewed in the direction of arrow A.

[0047] Figure 3 This is a diagram showing a schematic configuration of an image processing device according to the first embodiment.

[0048] Figure 4 This is a diagram showing the functional configuration of the image processing device according to the first embodiment.

[0049] Figure 5 It is a diagram for explaining acquisition of a projection image.

[0050] Figure 6 A diagram for explaining the generation of a tomographic image.

[0051] Figure 7 3 is a diagram for explaining detection of a structure of interest.

[0052] Figure 8 is a diagram showing the detection results of the attention structure.

[0053] Figure 9 A diagram for explaining a band tomographic image.

[0054] Figure 10 This is a diagram for explaining selection of a band tomographic image of a tumor in the middle and low frequency bands.

[0055] Figure 11 This is a diagram for explaining selection of a band tomographic image regarding a tumor in a high-frequency band.

[0056] Figure 12 This is a diagram for explaining selection of a band tomographic image regarding spurs in a high-frequency band.

[0057] Figure 13 This is a diagram for explaining selection of a band tomographic image regarding calcification in a high-frequency band.

[0058] Figure 14 This figure shows glitches that span multiple tomographic images.

[0059] Figure 15 This is a diagram for explaining the generation of a composite band two-dimensional image.

[0060] Figure 16 This is a diagram for explaining the generation of a synthesized band two-dimensional image in the middle and low frequency bands.

[0061] Figure 17 This is a diagram for explaining the generation of a synthesized band two-dimensional image in the high-frequency band.

[0062] Figure 18 A diagram showing a display screen of a synthesized two-dimensional image.

[0063] Figure 19 This is a flowchart showing the processing performed in the first embodiment.

[0064] Figure 20 This is a diagram for explaining the generation of the second synthesized frequency band two-dimensional image in the middle and low frequency bands related to the tumor.

[0065] Figure 21 This is a diagram for explaining the generation of a second synthesized frequency band two-dimensional image in a high-frequency band related to a tumor.

[0066] Figure 22 This is a diagram for explaining the generation of the second synthesized frequency band two-dimensional image in the high-frequency band related to spurs.

[0067] Figure 23 This is a diagram for explaining the generation of a second synthesized frequency band two-dimensional image in a high-frequency band related to calcification.

[0068] Figure 24 This is a diagram for explaining the generation of the composite two-dimensional image CG0 in the second embodiment.

[0069] Figure 25 This is a flowchart showing the processing performed in the second embodiment.

[0070] Figure 26 This is a diagram for explaining the extraction of calcified regions.

[0071] Figure 27 This is a diagram for explaining the generation of a synthesized two-dimensional image in the third embodiment.

[0072] Figure 28This is a flowchart showing the processing performed in the third embodiment.

[0073] Figure 29 This is a diagram for explaining the generation of a synthesized band two-dimensional image in the high-frequency band according to the fourth embodiment.

[0074] Figure 30 This is a diagram for explaining the generation of a second synthesized frequency band two-dimensional image in a high-frequency band related to a tumor according to the fourth embodiment.

[0075] Figure 31 This is a diagram for explaining the generation of a synthesized band two-dimensional image in the high-frequency band according to the fifth embodiment.

[0076] Figure 32 This is a diagram for explaining the generation of a second synthesized frequency band two-dimensional image in a high-frequency band related to a tumor according to the fifth embodiment. DETAILED DESCRIPTION

[0077] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. Figure 1 is a schematic structural diagram of a radiographic imaging system to which an image processing device according to an embodiment of the present invention is applied. Figure 2 It is from Figure 1 A diagram showing a mammography apparatus in a radiographic imaging system viewed in the direction of arrow A. Figure 1 As shown, in the radiographic imaging system 100 according to this embodiment, in order to perform breast tomosynthesis and generate tomographic images, the breast M, serving as the imaging subject, is imaged from multiple radiation source positions, thereby acquiring multiple radiographic images, i.e., multiple projection images. The radiographic imaging system 100 according to this embodiment includes a mammographic X-ray imaging apparatus 1, a console 2, an image storage system 3, and an image processing apparatus 4.

[0078] The mammography apparatus 1 includes an arm 12 connected to a base (not shown) via a rotation axis 11. An imaging table 13 is mounted on one end of the arm 12, and a radiation irradiation unit 14 is mounted on the other end so as to face the imaging table 13. The arm 12 is configured so that only the end on which the radiation irradiation unit 14 is mounted can be rotated, thereby fixing the imaging table 13 and rotating only the radiation irradiation unit 14.

[0079] A radiation detector 15, such as a flat-panel detector, is provided within the imaging table 13. The radiation detector 15 has a radiation detection surface 15A. Furthermore, a circuit board is provided within the imaging table 13. This circuit board 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 an analog-to-digital (AD) converter that converts the voltage signal into a digital signal.

[0080] A radiation source 16 is housed within the radiation irradiation unit 14. The radiation source 16 emits X-rays as radiation, and the timing of irradiation from the radiation source 16 and the radiation generation conditions in the radiation source 16, namely, the selection of target and filter materials, the tube voltage, and the irradiation time, are controlled by the console 2.

[0081] Furthermore, the arm 12 is provided with: a compression plate 17, which is arranged above the imaging platform 13 and presses and compresses the breast M; a support 18, which supports the compression plate 17; and a moving mechanism 19, which moves the support 18 along the imaging platform 13. Figure 1 and Figure 2 In addition, the interval between the compression plate 17 and the photographic table 13, ie, the compression thickness, is input to the console 2.

[0082] The console 2 has the function of controlling the mammography apparatus 1 via a network such as a wireless communication LAN (Local Area Network), using imaging orders and various information received from a radiology information system (RIS) (not shown), as well as direct instructions from a technician, etc. Specifically, the console 2 causes the mammography apparatus 1 to perform tomosynthesis of the breast M, acquires multiple projection images, and reconstructs these projection images to generate multiple tomographic images, as described below. In this embodiment, a server computer is used as the console 2, for example.

[0083] The image archiving system 3 stores image data such as radiographic images and tomographic images captured by the mammography apparatus 1. From the stored images, the image archiving system 3 extracts images corresponding to requests from the console 2, the image processing apparatus 4, and the like, and transmits them to the requesting device. A specific example of the image archiving system 3 is a PACS (Picture Archiving and Communication Systems).

[0084] Next, the image processing device according to the first embodiment will be described. Figure 3, the hardware structure of the image processing device involved in the first embodiment is described. Figure 3 As shown, the image processing device 4 is a computer such as a workstation, server, or personal computer, and includes a CPU (Central Processing Unit) 21, a nonvolatile storage device 23, and a memory 26 serving as a temporary storage area. Furthermore, the image processing device 4 includes a display 24 such as a liquid crystal display, input devices 25 such as a keyboard and a mouse, and a network interface (Interface) 27 for connection to a network (not shown). The CPU 21, storage device 23, display 24, input device 25, memory 26, and network interface 27 are connected to a bus 28. The CPU 21 is an example of a processor according to the present invention.

[0085] The storage device 23 is implemented by an HDD (Hard Disk Drive), an SSD (Solid State Drive), a flash memory, or the like. The storage device 23, serving as a storage medium, stores the image processing program 22 installed in the image processing device 4. The CPU 21 reads the image processing program 22 from the storage device 23, expands it into the memory 26, and executes the expanded image processing program 22.

[0086] The image processing program 22 is stored in a storage device of a server computer connected to a network or a network storage device in a state that allows external access, and is downloaded upon request and installed in the computer constituting the image processing device 4. Alternatively, the image processing program 22 may be distributed by recording it on a recording medium such as a DVD (Digital Versatile Disc) or a CD-ROM (Compact Disc Read Only Memory), and then installed in the computer constituting the image processing device 4 from the recording medium.

[0087] Next, the functional configuration of the image processing device according to the first embodiment will be described. Figure 4 1 is a diagram showing the functional structure of the image processing device according to the first embodiment. Figure 4 As shown, the image processing device 4 includes an image acquisition unit 30, a structure of interest detection unit 31, a frequency decomposition unit 32, a selection unit 33, a synthesis unit 34, and a display control unit 35. The CPU 21 executes the image processing program 22 to function as the image acquisition unit 30, the structure of interest detection unit 31, the frequency decomposition unit 32, the selection unit 33, the synthesis unit 34, and the display control unit 35.

[0088] The image acquisition unit 30 acquires a tomographic image acquired by the mammography apparatus 1 performing imaging via the console 2. The image acquisition unit 30 acquires the tomographic image from the console 2 or the image storage system 3 via the network I / F 27.

[0089] Here, we will describe tomographic imaging and the generation of tomographic images in the console 2. When performing tomographic imaging for generating tomographic images, the console 2 rotates the arm 12 about the rotation axis 11 to move the radiation source 16. Based on the movement of the radiation source 16, the breast M, serving as the imaging subject, is irradiated with radiation according to pre-set imaging conditions for tomographic imaging at multiple radiation source positions. The radiation detector 15 detects the radiation that has passed through the breast M, thereby acquiring multiple projection images Gi (i = 1 to n, where n is the number of radiation source positions, e.g., n = 15) at the multiple radiation source positions.

[0090] Figure 5 : is a diagram for explaining the acquisition of the projection image Gi. Figure 5 As shown, the radiation source 16 is moved to each radiation source position S1, S2, ..., Sn. At each radiation source position, the radiation source 16 is driven to irradiate the breast M with radiation. The radiation detector 15 detects the X-rays that have passed through the breast M, thereby acquiring projection images G1, G2, ..., Gn corresponding to each radiation source position S1 to Sn. Furthermore, the same dose of radiation is irradiated to the breast M at each radiation source position S1 to Sn.

[0091] In addition, Figure 5 In FIG, the radiation source position Sc is a radiation source position where the optical axis X0 of the radiation emitted from the radiation source 16 is perpendicular to the detection surface 15A of the radiation detector 15. The radiation source position Sc is referred to as a reference radiation source position Sc.

[0092] Then, the console 2 reconstructs the plurality of projection images Gi to generate a tomographic image that emphasizes a desired tomographic plane of the breast M. Specifically, the console 2 reconstructs the plurality of projection images Gi using a well-known back projection method such as a simple back projection method or a filter back projection method. Figure 6 As shown, multiple slice images Dj (j = 1 to m) are generated for each of the multiple slices of the breast M. At this time, a three-dimensional coordinate position in a three-dimensional space including the breast M is set, and the pixel values of the corresponding pixels of the multiple projection images Gi are reconstructed for the set three-dimensional coordinate position, thereby calculating the pixel value at the coordinate position.

[0093] The console 2 transmits the generated tomographic image Dj directly to the image processing device 4 or to the image archiving system 3 .

[0094] The structure-of-interest detection unit 31 detects the structure of interest from the plurality of tomographic images Dj. In this embodiment, a tumor, spurs, and calcification included in the breast M are detected as the structure of interest. Figure 7 : This is a diagram for explaining the detection of the structure of interest. Here, the detection of the structure of interest from the six tomographic images D1 to D6 is explained. Figure 7 As shown, tomographic image D1 contains calcification K13. Tomographic image D2 contains a tumor K21. Tomographic image D3 contains the tumor K21 from tomographic image D2, a tumor K31 that is continuous within breast M, and a spur K32. Tomographic image D4 contains the tumor K21 from tomographic image D2 and the tumor K31 from tomographic image D3, a tumor K41a that is continuous within breast M, a tumor K41b that is only present in tomographic image D4, and a spur K42. Tomographic image D5 contains a spur K52. Tomographic image D6 contains calcification K63.

[0095] The structure of interest detection unit 31 uses a well-known computer-aided diagnostic imaging (CAD) algorithm to detect structures of interest from the tomographic image Dj. CAD-based algorithms derive the probability (likelihood) that a pixel in the tomographic image Dj is a structure of interest, and detect pixels whose probability exceeds a predetermined threshold as structures of interest. Furthermore, a CAD-based algorithm is prepared for each type of structure of interest. In this embodiment, CAD algorithms are prepared for tumor detection, burr detection, and calcification detection.

[0096] The detection of the structure of interest is not limited to detection using CAD, and the structure of interest can be detected from the tomographic image Dj by using filtering processing using a filter for detecting the structure of interest or a detection model that has undergone machine learning such as deep learning for detecting the structure of interest.

[0097] The attention structure detection unit 31 Figure 7 The tomographic images D1 to D6 shown in FIG. 1 are used to detect tumors, burrs, and calcifications as structures of interest. Figure 8 As shown, tumor detection result R1, spur detection result R2, and calcification detection result R3 are derived. In tumor detection result R1, tumors are detected in tomographic images D2 to D4. In spur detection result R2, spurs are detected in tomographic images D3 to D5. In calcification detection result R3, calcification is detected in tomographic images D1 and D6.

[0098] The frequency decomposition unit 32 performs frequency decomposition on each of the plurality of tomographic images Dj, deriving a plurality of band tomographic images representing the frequency components of each of the plurality of frequency bands for each of the plurality of tomographic images Dj. In addition to performing multi-resolution transformation on radiographic images, any known method such as wavelet transform and Fourier transform can be used as a frequency decomposition method. Furthermore, the number of frequency bands used for frequency decomposition can be two or more. Furthermore, in this embodiment, the frequency bands are categorized into low, medium, and high bands for explanation. However, the frequency components contained in the band tomographic images increase in order of low, medium, and high bands. Furthermore, when frequency decomposition is performed using four or more frequency bands, the low, medium, and high bands can be arbitrarily set. Furthermore, when the number of frequency decompositions is two, the low frequency band is referred to as the low-medium frequency band, and the high frequency band is referred to as the high frequency band. Furthermore, when the number of frequency decompositions is four or more, the low and medium frequency bands may also be collectively referred to as the low-medium frequency band.

[0099] Figure 9 is a diagram for explaining a band tomographic image. Figure 9 For simplicity of explanation, only the mid-low frequency band MLf and the high frequency band Hf are shown among the multiple frequency bands. The high frequency band Hf corresponds to the first frequency band of the present invention, while the mid-low frequency band MLf corresponds to the second frequency band of the present invention. Furthermore, the band tomographic images of the mid-low frequency band MLf are referred to as DML1 to DML6, while the band tomographic images of the high frequency band Hf are referred to as DH1 to DH6. Band tomographic images DML1 to DML6 of the mid-low frequency band MLf only include relatively large tumor structures among the tumors, spurs, and calcifications included in tomographic images D1 to D6. Band tomographic images DH1 to DH6 of the high frequency band include fine structures such as spurs and calcifications, as well as the fine structure of the tumor.

[0100] The selection unit 33 selects a tomographic image from the multiple tomographic images Dj based on the type and frequency band of the structure of interest in the region where the structure of interest is detected. Specifically, in the first embodiment, the selection unit 33 selects a band tomographic image corresponding to the tomographic image in which the structure of interest was detected from the multiple band tomographic images, based on the type and frequency band of the structure of interest, for each pixel in the multiple band tomographic images corresponding to a pixel of the synthesized two-dimensional image CG0 (described later). Furthermore, when selecting the band tomographic images, the selection unit 33 associates the position of the structure of interest in the tomographic image Dj with the positions of the band tomographic images DMLj and DHj for each type of structure of interest detected by the structure of interest detection unit 31.

[0101] Figure 10 is a diagram for explaining selection of a band tomographic image of a tumor in the middle and low frequency bands. Figure 11 is a diagram for explaining selection of a band tomographic image of a tumor in a high-frequency band. Figure 12 is a diagram for explaining selection of a band tomographic image regarding burrs in a high-frequency band. Figure 13 This is used to explain the selection of a band tomographic image of calcification in the high frequency band. Figures 10-13 In FIG, a tomographic image is schematically shown in one dimension. Figures 10-13 In FIG, an index 40 for indicating the correspondence between the pixels of the band tomographic image and the pixels of the synthesized two-dimensional image CG0 is shown in a one-dimensional manner. In addition, in a band tomographic image including a structure of interest, the pixels of the detected structure of interest are displayed thicker than the pixels other than the structure of interest. Figure 10 、 11 The pixels of the tumor are painted black in Figure 12 Set the pixels of the burr to blank in Figure 13 Calcified pixels are marked with vertical hatching. In addition, in the index 40, 15 pixels P1 to P15 corresponding to the pixels of the synthesized two-dimensional image CG0 are shown. In addition, in the index 40, only the pixels P1, P5, P10, and P15 are marked with reference symbols. Figures 10 to 13 The same diagram is shown in the same way Figures 10 to 13 same.

[0102] First, the selection of band tomographic images related to the tumor will be described. The selection unit 33 selects all band tomographic images containing the tumor for each pixel corresponding to the pixel of the synthesized two-dimensional image CG0 in the middle and low frequency bands MLf. Figure 10 As shown, no tumor was detected in any of the band tomographic images DMLj for pixels P1, P4-P6, and P11-P15 of the band tomographic image DMLj in the mid-low frequency band MLf. Therefore, the selection unit 33 selects no band tomographic images for pixels P1, P4-P6, and P11-P15. Furthermore, a tumor was detected only in the band tomographic image DML4 for pixels P2 and P3. Therefore, the selection unit 33 selects the band tomographic image DML4 for pixels P2 and P3. A tumor was detected only in the band tomographic image DML3 for pixels P7 and P10. Therefore, the selection unit 33 selects the band tomographic image DML3 for pixels P7 and P10. Furthermore, a tumor was detected in the band tomographic images DML2-DML4 for pixels P8 and P9. Therefore, the selection unit 33 selects all of the band tomographic images DML2-DML4 in which a tumor was detected for pixels P8 and P9.

[0103] On the other hand, the selection unit 33 selects one band tomographic image that best represents the tumor for each pixel corresponding to the pixel of the synthesized two-dimensional image CG0 among the plurality of band tomographic images in the high frequency band Hf. Figure 11 As shown, no tumor was detected in any of the band tomographic images DHj for pixels P1, P4-P6, and P11-P15 of the band tomographic image DHj in the high-frequency band Hf. Therefore, the selection unit 33 selects no band tomographic images for pixels P1, P4-P6, and P11-P15. Furthermore, a tumor was detected only in the band tomographic image DH4 for pixels P2 and P3. Therefore, the selection unit 33 selects the band tomographic image DH4 for pixels P2 and P3. A tumor was detected only in the band tomographic image DH3 for pixels P7 and P10. Therefore, the selection unit 33 selects the band tomographic image DH3 for pixels P7 and P10. Furthermore, a tumor was detected in the band tomographic images DH2-DH4 for pixels P8 and P9. Here, among the tumors detected in band tomographic images DH2 to DH4, the largest tumor is detected in band tomographic image DH3, and band tomographic image DH3 best represents the tumor among the band tomographic images DH2 to DH4. Therefore, the selection unit 33 selects band tomographic image DH3 for pixels P8 and P9. Alternatively, instead of the largest tumor, the band tomographic image containing the tumor with the highest probability (likelihood) derived by the structure of interest detection unit 31 during detection may be selected.

[0104] Next, the selection of the band tomographic image for the burr will be described. The structure of the burr is contained only in the band tomographic image DHj of the high-frequency band Hf. Therefore, the selection unit 33 selects one band tomographic image that best represents the burr for each pixel corresponding to the pixel of the synthesized two-dimensional image CG0 among the multiple band tomographic images only in the high-frequency band Hf. Figure 12 As shown, no glitches were detected in any of the band tomographic images DHj for pixels P1, P2, P7, and P12-P15. Therefore, regarding glitches, the selection unit 33 selects no band tomographic images DHj for pixels P1, P2, P7, and P12-P15. Furthermore, glitches were detected only in the band tomographic image DH4 for pixels P3, P4, and P11. Therefore, the selection unit 33 selects the band tomographic image DH4 for pixels P3, P4, and P11. Furthermore, glitches were detected in the band tomographic images DH4 and DH5 for pixels P5 and P10. Of the glitches detected in the band tomographic images DH4 and DH5, the glitches detected in the band tomographic image DH4 are the largest, and of the band tomographic images DH4 and DH5, the band tomographic image DH4 best represents the glitches. Therefore, the selection unit 33 selects the band tomographic image DH4 for pixels P5 and P10. Alternatively, instead of the largest spur, a tomographic image including a spur with the highest probability (likelihood) derived by the structure-of-interest detection unit 31 during detection may be selected.

[0105] For pixels P6 and P9, glitches are detected only in the band tomographic image DH5. Therefore, the selection unit 33 selects the band tomographic image DH5 for pixels P6 and P9. Furthermore, for pixel P8, glitches are detected in the band tomographic image DH3. Therefore, the selection unit 33 selects the band tomographic image DH3 for pixel P8.

[0106] Next, the selection of band tomographic images related to calcification will be described. The structure of calcification is contained only in the band tomographic images DHj of the high-frequency band Hf. Therefore, the selection unit 33 selects one band tomographic image that best represents calcification for each pixel corresponding to the pixel of the synthesized two-dimensional image CG0 among the multiple band tomographic images only in the high-frequency band Hf. Figure 13 As shown, no calcification was detected in any of the band tomographic images DHj for pixels P1 to P11, P13, and P15. Therefore, regarding calcification, the selection unit 33 selects no band tomographic images for pixels P1 to P11, P13, and P15. Furthermore, calcification was detected only in the band tomographic image DH1 for pixel P12. Therefore, the selection unit 33 selects the band tomographic image DH1 for pixel P12. Calcification was detected only in the band tomographic image DH6 for pixel P14. Therefore, the selection unit 33 selects the band tomographic image DH6 for pixel P14.

[0107] In addition, regarding burrs, such as Figure 14 As shown, sometimes a glitch K7 spreads two-dimensionally in a direction perpendicular to the tomographic plane and exists across multiple band tomographic images DHk-1, DHk, and DHk+1. In this case, the glitch is detected in pixel P100 of band tomographic image DHk-1, in pixel P101 of band tomographic image DHk, and in pixel P102 of band tomographic image DHk+1. Therefore, as Figure 14 As shown, when there is a burr K7 in the breast M, if you select Figure 14 For the band tomographic image DHk shown in FIG. 1 , a plurality of band tomographic images DHk-1 and DHk+1 located above and below the band tomographic image DHk, to which the burr K7 included in the band tomographic image DHk is connected, are further selected.

[0108] The synthesizing unit 34 generates a synthesized two-dimensional image using the band tomographic images selected by the selection unit 33 for each type of the structure of interest according to the band. Specifically, the synthesizing unit 34 generates a synthesized band two-dimensional image for each band using the selected band tomographic images in the pixels of the band tomographic images corresponding to the structure of interest, and performs frequency synthesis on the synthesized band two-dimensional images to generate a synthesized two-dimensional image. The synthesized two-dimensional image is a simulated two-dimensional image equivalent to a simple two-dimensional image taken by irradiating the breast M with radiation from the reference radiation source position Sc. In this embodiment, as Figure 15As shown, the synthesis unit 34 stacks a plurality of band tomographic images ( Figure 15 Only the state of DHj) is shown in FIG. 1 , and the direction from the reference ray source position Sc toward the viewpoint of the radiation detector 15 is Figure 5 The pixel values of the corresponding pixels in each band tomographic image DHj are synthesized along the optical axis X0 shown, thereby generating a composite band two-dimensional image CGH0. The generation of a composite band two-dimensional image or a composite two-dimensional image will be described below.

[0109] Figure 16 : is a diagram for explaining the generation of a two-dimensional image of a composite band in the middle and low frequency band MLf. Figure 16 As shown, for pixels P1, P4-P6, and P11-P15 where no tumor was detected, the synthesizing unit 34 calculates the arithmetic mean of the pixel values of all band tomographic images DML1-DML6 and uses this arithmetic mean as the pixel values of pixels P1, P4-P6, and P11-P15 of the synthesized two-dimensional band image CGML0 for the mid-low frequency band MLf. At this point, all band tomographic images become the predetermined tomographic images of the present invention. For pixels P2 and P3, band tomographic image DML4, where a tumor was detected, has been selected. Therefore, the synthesizing unit 34 uses the pixel values of pixels P2 and P3 in band tomographic image DML4 as the pixel values of pixels P2 and P3 of the synthesized two-dimensional band image CGML0. Since band tomographic image DML3 is selected from pixels P7 and P10 of band tomographic image DMLj, the synthesizing unit 34 uses the pixel values of pixels P7 and P10 of band tomographic image DML3 as the pixel values of pixels P7 and P10 of the synthesized two-dimensional band image CGML0. Since band tomographic images DML2 to DML4 are selected from pixels P8 and P9 of band tomographic image DMLj, the synthesizing unit 34 uses the sum of the pixel values of pixels P8 and P9 of band tomographic images DML2 to DML4 as the pixel values of pixels P8 and P9 of the synthesized two-dimensional band image CGML0. Alternatively, a weighted sum or weighted average value may be used instead of the sum. In this case, the weight of band tomographic image DML3 may be greater than that of band tomographic images DML2 and DML4.

[0110] Figure 17 This is a diagram for explaining the generation of a composite band two-dimensional image in the high frequency band Hf. In addition, the band tomographic images DH3 and DH4 contain both tumors and burrs. Figure 17 In FIG, the band tomographic images DH3-1 and DH4-1 containing only the detection result of the tumor and the band tomographic images DH3-2 and DH4-2 containing only the detection result of the burr are virtually arranged. Figure 17 In the same figure, the diagram is the same as Figure 17 same.

[0111] like Figure 17 As shown, for pixels P1, P13, and P15 of the structure of interest, where no tumor, spur, or calcification was detected, the synthesizing unit 34 calculates the arithmetic mean of the pixel values of all band tomographic images DH1 to DH6 and uses this arithmetic mean as the pixel value of pixels P1, P13, and P15 in the synthesized band two-dimensional image CGH0 of the high-frequency band Hf. With respect to pixel P2, since the band tomographic image DH4 in which a tumor was detected was selected, the synthesizing unit 34 uses the pixel value of pixel P2 in the band tomographic image DH4 as the pixel value of pixel P2 in the synthesized band two-dimensional image CGH0. With respect to pixel P3, since the band tomographic image DH4 in which both a tumor and a spur were detected was selected, the synthesizing unit 34 uses the pixel value of pixel P3 in the band tomographic image DH4 as the pixel value of pixel P3 in the synthesized band two-dimensional image CGH0. Since the band tomographic image DH4 in which the burr is detected is selected for pixels P4 and P5 , the synthesizing unit 34 uses the pixel values of pixels P4 and P5 in the band tomographic image DH4 as the pixel values of pixels P4 and P5 in the synthesized band two-dimensional image CGH0 .

[0112] With respect to pixel P6, the band tomographic image DH5, in which a spur was detected, was selected. Therefore, the synthesizing unit 34 uses the pixel value of pixel P6 in the band tomographic image DH5 as the pixel value of pixel P6 in the synthesized band two-dimensional image CGH0. With respect to pixel P7, the band tomographic image DH3, in which a tumor was detected, was selected. Therefore, the synthesizing unit 34 uses the pixel value of pixel P7 in the band tomographic image DH3 as the pixel value of pixel P7 in the synthesized band two-dimensional image CGH0. With respect to pixel P8, the band tomographic image DH3, in which both a tumor and a spur were detected, was selected. Therefore, the synthesizing unit 34 uses the pixel value of pixel P8 in the band tomographic image DH3 as the pixel value of pixel P8 in the synthesized band two-dimensional image CGH0.

[0113] For pixel P9, the band tomographic image DH3, in which a tumor was detected, and the band tomographic image DH5, in which a spur was detected, were selected. In this embodiment, when different band tomographic images are selected for the same pixel in the band tomographic image DHj, respectively, for a tumor, a spur, and a calcification, the pixel value of the band tomographic image is assigned based on the order of priority, with the priority increasing in the order of tumor, spur, and calcification. Therefore, the synthesis unit 34 uses the pixel value of pixel P9 in the band tomographic image DH5, in which a spur was detected, as the pixel value for pixel P9 in the synthesized two-dimensional band image CGH0.

[0114] For pixel P10, the band tomographic image DH3 where a tumor was detected and the band tomographic image DH4 where a spur was detected were selected. Therefore, the synthesizing unit 34 uses the pixel value of pixel P10 of the band tomographic image DH4 where a spur was detected as the pixel value of pixel P10 in the synthesized band two-dimensional image CGH0.

[0115] With respect to pixel P11, the band tomographic image DH4, in which burrs were detected, was selected. Therefore, the synthesizing unit 34 uses the pixel value of pixel P11 in the band tomographic image DH4 as the pixel value of pixel P11 in the synthesized band two-dimensional image CGH0. With respect to pixel P12, the band tomographic image DH1, in which calcification was detected, was selected. Therefore, the synthesizing unit 34 uses the pixel value of pixel P12 in the band tomographic image DH1 as the pixel value of pixel P12 in the synthesized band two-dimensional image CGH0. With respect to pixel P14, the band tomographic image DH6, in which calcification was detected, was selected. Therefore, the synthesizing unit 34 uses the pixel value of pixel P14 in the band tomographic image DH6 as the pixel value of pixel P14 in the synthesized band two-dimensional image CGH0.

[0116] The synthesis unit 34 then performs frequency synthesis on the synthesized two-dimensional band image CGML0 of the mid-low frequency band MLf and the synthesized two-dimensional band image CGH0 of the high frequency band Hf, thereby generating a synthesized two-dimensional band image CG. The frequency synthesis method may be a method corresponding to the frequency decomposition performed by the frequency decomposition unit 32. For example, if frequency decomposition is performed using a wavelet transform, frequency synthesis may be performed using an inverse wavelet transform.

[0117] The display control unit 35 displays the synthesized two-dimensional image CG0 generated by the synthesizing unit 34 on the display 24 . Figure 18 is a diagram showing a display screen of a synthesized two-dimensional image. Figure 18 As shown, the synthesized two-dimensional image CG0 is displayed on the display screen 50 of the display 24. Figure 18 The synthetic two-dimensional image CG0 shown is based on Figure 7 The synthesized two-dimensional image is generated by the tomographic images D1 to D6 shown in FIG. Figure 18 The synthesized two-dimensional image CG0 shown clearly includes calcification K13 in tomographic image D1, tumor K31 in tomographic image D3, tumor K41b in tomographic image D4, calcification K63 in tomographic image D6, and burrs K32, K42, and K52 in tomographic images D3 to D5. Burrs K32, K42, and K52 are omitted from illustration. Burrs K32, K42, and K52 partially overlap with tumor K31, but the pixel values of tumor K31 have been replaced with the pixel values of burrs K42 and K52 in tomographic images D4 and D5.

[0118] Next, the processing performed in the first embodiment will be described. Figure 19This is a flowchart illustrating the processing performed in the first embodiment. It is assumed that multiple tomographic images Dj have been previously acquired and stored in the storage device 23. When the input device 25 receives a processing start instruction from the operator, processing begins, and the structure of interest detector 31 detects a structure of interest from the multiple tomographic images Dj (step ST11). Next, the frequency decomposition unit 32 performs frequency decomposition on each of the multiple tomographic images Dj, thereby deriving multiple frequency band tomographic images representing frequency components of each of the multiple frequency bands for each of the multiple tomographic images Dj (step ST12).

[0119] Next, the selection unit 33 selects a band tomographic image corresponding to the tomographic image in which the structure of interest is detected from among the plurality of band tomographic images, for each corresponding pixel in the plurality of band tomographic images, according to the type and frequency band of the structure of interest (step ST13 ).

[0120] The synthesis unit 34 then generates synthesized two-dimensional band images CGML0 and CGH0 using the selected band tomographic images (step ST14). It then frequency synthesizes the synthesized two-dimensional band images CGML0 and CGH0 to generate a synthesized two-dimensional image CG0 (step ST15). The display control unit 35 then displays the synthesized two-dimensional image CG0 on the display 24 (step ST16), and the process ends.

[0121] In this manner, in the first embodiment, a tomographic image is band-decomposed, and a band tomographic image containing the structure of interest is selected from multiple band tomographic images DMLj and DHj based on the type and frequency band of the structure of interest. A composite two-dimensional image CG0 is generated using the selected band tomographic image in the region where the structure of interest is detected. Therefore, compared to the method described in U.S. Patent No. 9,792,703, which generates a composite two-dimensional image by weighted averaging all tomographic images, the composite two-dimensional image CG0 is generated using fewer tomographic images for the region of the structure of interest. Consequently, fine structures of interest are not blurred in the composite two-dimensional image CG0. In particular, in the first embodiment, a single band tomographic image that best represents the structure of interest is selected for each corresponding pixel in the multiple band tomographic images, thereby reducing blurring of fine structures of interest in the composite two-dimensional image CG0.

[0122] Furthermore, in the first embodiment, all band tomographic images containing the structure of interest are selected for each pixel corresponding to a pixel of the synthesized two-dimensional image CG0 from among the multiple band tomographic images in the low- to medium-frequency band MLf. Therefore, even if a single structure of interest extends in the direction in which the band tomographic images are arranged, i.e., in the depth direction of the breast M, the state of the structure of interest in the depth direction can be reflected in the synthesized two-dimensional image CG0 by using the multiple selected band tomographic images to generate the synthesized two-dimensional image CG0.

[0123] Furthermore, in the first embodiment, in the high-frequency band Hf, a single band tomographic image that best represents the structure of interest is selected for each pixel corresponding to a pixel of the synthesized two-dimensional image CG0 from among the multiple band tomographic images. Therefore, when a single structure of interest expands two-dimensionally in a direction perpendicular to the optical axis X0 of the radiation and in the depth direction of the breast M, in which the band tomographic images are arranged, multiple band tomographic images are also selected for that structure of interest. Therefore, by generating the synthesized two-dimensional image CG0 using these multiple selected band tomographic images, the state of the structure of interest, expanding two-dimensionally and in the depth direction, can be reflected in the synthesized two-dimensional image CG0.

[0124] Furthermore, when different band tomographic images are selected for the same pixel in band tomographic images DMLj and DHj, for each of a tumor, a spur, and a calcification, the pixel value of the band tomographic image is assigned according to a priority order, with the tumor, spur, and calcification increasing in that order. In breast M, the degree of malignancy increases in the order of tumor, spur, and calcification. Therefore, by selecting band tomographic images according to this priority order, a synthetic two-dimensional image CG0 can be generated that makes the more malignant structures of interest stand out.

[0125] Next, the second embodiment of the present invention will be described. The structure of the image processing device according to the second embodiment is the same as that of the image processing device according to the first embodiment, except for the processing performed. Therefore, a detailed description of the device will be omitted here. In the second embodiment, the synthesis unit 34 generates a first synthesized two-dimensional image CG1 by synthesizing multiple tomographic images Dj. Then, for each structure of interest, the synthesis unit 34 generates a synthesized band two-dimensional image for each band using the selected band tomographic image in the pixels of the band tomographic image corresponding to the structure of interest, and performs frequency synthesis on the synthesized band two-dimensional images to generate second synthesized two-dimensional images CG21, CG22, and CG23 for each structure of interest. Furthermore, the synthesis unit 34 generates a synthesized two-dimensional image CG0 by synthesizing the second synthesized two-dimensional images CG21, CG22, and CG23 for each structure of interest with the first synthesized two-dimensional image CG1.

[0126] In the second embodiment, the synthesis unit 34 first synthesizes a plurality of tomographic images Dj to generate a first synthesized two-dimensional image CG1. Specifically, the synthesis unit 34 generates the first synthesized two-dimensional image CG1 by, for example, averaging the pixel values of corresponding pixels in the plurality of tomographic images Dj.

[0127] Furthermore, in the second embodiment, the synthesis unit 34 generates second synthesized two-dimensional images CG21, CG22, and CG23 according to the type and frequency band of the structure of interest. Specifically, a second synthesized two-dimensional image CG21 for the tumor, a second synthesized two-dimensional image CG22 for the spur, and a second synthesized two-dimensional image CG23 for the calcification are generated. First, the generation of the second synthesized two-dimensional image CG21 for the tumor will be described. Furthermore, the selection of the frequency band tomographic images for each frequency band, for the tumor, the spur, and the calcification, is performed by the selection unit 33, similar to the first embodiment.

[0128] In the second embodiment, the synthesis unit 34 generates a second synthesized band two-dimensional image CGML21 using only the selected band tomographic images for pixels where a tumor is detected. First, the generation of the second synthesized band two-dimensional image CGML21 in the low- to medium-frequency band MLf will be described. Furthermore, band tomographic images DML2 to DML4 are selected for the tumor in the low- to medium-frequency band MLf. Figure 20 This is a diagram for explaining the generation of the second synthesized frequency band two-dimensional image in the middle and low frequency bands related to the tumor.

[0129] First, for pixels P1, P4-P6, and P11-P15 where no tumor was detected in any of the band tomographic images DMLj, the synthesizing unit 34 calculates the arithmetic mean of the pixel values of the band tomographic images DML1-DML6 and uses this arithmetic mean as the pixel values of pixels P1, P4-P6, and P11-P15 in the second synthesized band two-dimensional image CGML21 of the low-mid-frequency band MLf. With respect to pixels P2 and P3, since the band tomographic image DML4 has been selected, the synthesizing unit 34 uses the pixel values of pixels P2 and P3 in the band tomographic image DML4 as the pixel values of pixels P2 and P3 in the second synthesized band two-dimensional image CGML21. Since the band tomographic image DML3 is selected from pixels P7 and P10 of the band tomographic image DMLj, the synthesizing unit 34 uses the pixel values of pixels P7 and P10 of the band tomographic image DML3 as the pixel values of pixels P7 and P10 of the second synthesized band two-dimensional image CGML21. Since the band tomographic images DML2 to DML4 are selected from pixels P8 and P9 of the band tomographic image DMLj, the synthesizing unit 34 uses the sum of the pixel values of pixels P8 and P9 of the band tomographic images DML2 to DML4 as the pixel values of pixels P8 and P9 of the second synthesized band two-dimensional image CGML21. Alternatively, a weighted sum or weighted average value may be used instead of the sum. In this case, the band tomographic image DML3 may be given a greater weight than the band tomographic images DML2 and DML4.

[0130] Next, generation of the second composite band two-dimensional image CGH21 in the high-frequency band Hf of the tumor will be described. Figure 21 : is a diagram for explaining the generation of the second synthetic frequency band two-dimensional image in the high frequency band of the tumor. Figure 21 As shown, for pixels P1, P4-P6, and P11-P15 where no tumor was detected in any of the band tomographic images DHj, the synthesizing unit 34 calculates the arithmetic mean of the pixel values of the band tomographic images DH1-DH6 and uses this arithmetic mean as the pixel values of pixels P1, P4-P6, and P11-P15 in the second synthesized band two-dimensional image CGH21 of the high frequency band Hf. With respect to pixels P2 and P3, the band tomographic image DH4 was selected, so the synthesizing unit 34 uses the pixel values of pixels P2 and P3 in the band tomographic image DH4 as the pixel values of pixels P2 and P3 in the second synthesized band two-dimensional image CGH21. With respect to pixels P7-P10 of the band tomographic image DHj, the band tomographic image DH3 was selected, so the synthesizing unit 34 uses the pixel values of pixels P7-P10 in the band tomographic image DH3 as the pixel values of pixels P7-P10 in the second synthesized band two-dimensional image CGH21.

[0131] Then, the synthesizing unit 34 generates a second synthesized two-dimensional image CGML21 of the low-medium frequency band MLf and the second synthesized two-dimensional image CGH2 of the high-frequency band Hf regarding the tumor by frequency synthesizing the image.

[0132] Next, the generation of the second synthesized two-dimensional image CG22 for burrs will be described. In the second embodiment, the synthesizing unit 34 also generates the second synthesized band two-dimensional image CG22 using only the selected band tomographic images for pixels where burrs are detected. Furthermore, the structure of the burr is contained only in the band tomographic images DHj of the high-frequency band Hf. Therefore, the synthesizing unit 34 calculates the arithmetic mean of the pixel values of all pixels P1 to P15 in the band tomographic images DMLj of the mid- and low-frequency band MLf as the pixel values of pixels P1 to P15 in the second synthesized band two-dimensional image CGML22 for the mid- and low-frequency band MLf.

[0133] Figure 22 : is a diagram for explaining the generation of the second synthetic band two-dimensional image in the high frequency band of burrs. Figure 22 As shown, for pixels P1, P2, P7, and P12-P15 where no glitch was detected in any of the band tomographic images DHj, the synthesizing unit 34 derives the arithmetic mean of the pixel values of the band tomographic images DH1-DH6 and uses this arithmetic mean as the pixel values of pixels P1, P2, P7, P12-P15 of the second synthesized band two-dimensional image CGH22 for the high frequency band Hf. Since the band tomographic image DH4 is selected among the pixels P3-P5, P10, and P11, the synthesizing unit 34 uses the pixel values of pixels P3-P5, P10, and P11 in the band tomographic image DH4 as the pixel values of pixels P3-P5, P10, and P11 of the second synthesized band two-dimensional image CGH22. Since the band tomographic image DH5 is selected for pixels P6 and P9, the synthesizing unit 34 uses the pixel values of pixels P6 and P9 in the band tomographic image DH5 as the pixel values of pixels P6 and P9 in the second synthesized band two-dimensional image CGH22. Since the band tomographic image DH3 is selected for pixel P8, the synthesizing unit 34 uses the pixel value of pixel P8 in the band tomographic image DH3 as the pixel value of pixel P8 in the second synthesized band two-dimensional image CGH22.

[0134] Then, the synthesizing unit 34 generates a second synthesized two-dimensional image CG22 regarding the glitch by frequency synthesizing the second synthesized band two-dimensional image CGML22 regarding the mid-low frequency band MLf and the second synthesized band two-dimensional image CGH22 regarding the high frequency band Hf.

[0135] Next, the generation of the second synthesized two-dimensional image CG23 regarding calcification will be described. In the second embodiment, regarding calcification, the synthesizing unit 34 also generates the second synthesized band two-dimensional image CGML23 using only the selected band tomographic images for pixels where calcification is detected. Furthermore, the calcified structure is contained only in the band tomographic images DHj of the high-frequency band Hf. Therefore, the synthesizing unit 34 calculates the arithmetic mean of the pixel values of all pixels P1 to P15 in the band tomographic images DMLj of the mid- and low-frequency band MLf as the pixel values of pixels P1 to P15 in the second synthesized band two-dimensional image CGML23 of the mid- and low-frequency band MLf.

[0136] Figure 23 1 is a diagram for explaining the generation of a second synthesized frequency band two-dimensional image in a high frequency band related to calcification. Figure 23 As shown, for pixels P1 to P11, P13, and P15 where no calcification was detected in any of the band tomographic images DHj, the synthesizing unit 34 calculates the arithmetic mean of the pixel values of the band tomographic images DH1 to DH6 and uses this arithmetic mean as the pixel value of pixels P1 to P11, P13, and P15 in the second synthesized band two-dimensional image CGH23 for the high-frequency band Hf. For pixel P12, the band tomographic image DH1 was selected, so the synthesizing unit 34 uses the pixel value of pixel P12 in the band tomographic image DH1 for the high-frequency band Hf as the pixel value of pixel P12 in the second synthesized band two-dimensional image CGH23. For pixel P14, the band tomographic image DH6 was selected, so the synthesizing unit 34 uses the pixel value of pixel P14 in the band tomographic image DH6 as the pixel value of pixel P14 in the second synthesized band two-dimensional image CGH23.

[0137] Then, the synthesizing unit 34 generates a second synthesized two-dimensional image CG23 of the calcification by frequency synthesizing the second synthesized band two-dimensional image CGML23 of the low-medium frequency band MLf and the second synthesized band two-dimensional image CGH23 of the high-frequency band Hf.

[0138] The synthesis unit 34 sequentially synthesizes the second synthesized two-dimensional image CG21 regarding the tumor, the second synthesized two-dimensional image CG22 regarding the spur, and the second synthesized two-dimensional image CG23 regarding the calcification generated as described above with the first synthesized two-dimensional image CG1 to generate a synthesized two-dimensional image CG0 . Figure 24 : is a diagram for explaining the generation of the synthesized two-dimensional image CG0 in the second embodiment. Figure 24 As shown, the synthesis unit 34 first replaces the tumor region in the first synthesized two-dimensional image CG1 with the tumor region in the second synthesized two-dimensional image CG21 related to the tumor, thereby synthesizing the second synthesized two-dimensional image CG21 related to the tumor with the first synthesized two-dimensional image CG1. This generates an intermediate synthesized two-dimensional image CG11.

[0139] Next, the synthesis unit 34 replaces the burr region in the intermediate synthesized two-dimensional image CG11 with the burr region in the second synthesized two-dimensional image CG22 regarding burrs, thereby synthesizing the second synthesized two-dimensional image CG22 regarding burrs with the intermediate synthesized two-dimensional image CG11. Thus, the intermediate synthesized two-dimensional image CG12 is generated.

[0140] The synthesis unit 34 then replaces the calcified areas in the intermediate synthesized two-dimensional image CG12 with the calcified areas in the second synthesized two-dimensional image CG23 related to calcification, thereby synthesizing the second synthesized two-dimensional image CG23 related to calcification with the intermediate synthesized two-dimensional image CG12. This generates a synthesized two-dimensional image CG0 according to the second embodiment.

[0141] Next, the processing performed in the second embodiment will be described. Figure 25 This is a flowchart illustrating the processing performed in the second embodiment. It is assumed that a plurality of tomographic images Dj have been previously acquired and stored in the storage device 23. When the input device 25 receives a processing start instruction from the operator, processing begins, and the structure of interest detector 31 detects a structure of interest from the plurality of tomographic images Dj (step ST21). Next, the frequency decomposition unit 32 performs frequency decomposition on each of the plurality of tomographic images Dj, thereby deriving a plurality of frequency band tomographic images representing frequency components of each of a plurality of frequency bands for each of the plurality of tomographic images Dj (step ST22).

[0142] Next, the selection unit 33 selects a band tomographic image corresponding to the tomographic image in which the structure of interest is detected from the plurality of band tomographic images, for each corresponding pixel in the plurality of band tomographic images, according to the type and frequency band of the structure of interest (step ST23 ).

[0143] Next, the synthesis unit 34 generates a first synthesized two-dimensional image CG1 from the multiple tomographic images Dj (step ST24). In addition, the processing of step ST24 can be performed before each processing of steps ST21 to ST23, or can be performed in parallel with them. Then, the synthesis unit 34 generates second synthesized two-dimensional images CG21, CG22, and CG23 regarding the tumor, burr, and calcification, respectively (step ST25). Moreover, the synthesis unit 34 generates a synthesized two-dimensional image CG0 by sequentially synthesizing the second synthesized two-dimensional images CG21, CG22, and CG23 regarding the tumor, burr, and calcification, respectively, with the first synthesized two-dimensional image CG1 (step ST26). Next, the display control unit 35 displays the synthesized two-dimensional image CG0 on the display 24 (step ST27) and ends the processing.

[0144] Next, the third embodiment of the present invention will be described. The structure of the image processing device according to the third embodiment is the same as that of the image processing device according to the second embodiment, differing only in the processing performed. Therefore, a detailed description of the device will be omitted here. In the third embodiment, the synthesis unit 34 synthesizes multiple tomographic images Dj to generate a first synthesized two-dimensional image CG1. The synthesis unit 34 then uses the selected band tomographic images to generate a synthesized band two-dimensional image for each frequency band in the pixels of the synthesized two-dimensional image CG0 corresponding to the tumor, spur, and calcification. The synthesized band two-dimensional images are then frequency synthesized to generate second synthesized two-dimensional images CG21 and CG22. Meanwhile, the synthesis unit 34 extracts calcified regions from the first synthesized two-dimensional image CG1 as calcified regions. Furthermore, the synthesis unit 34 synthesizes the second synthesized two-dimensional images CG21 and CG22, respectively, for the tumor and spur, with the first synthesized two-dimensional image CG1, and also synthesizes the calcified regions, thereby generating a synthesized two-dimensional image CG0.

[0145] In the third embodiment, the synthesis unit 34 generates the first synthesized two-dimensional image CG1 , the second synthesized frequency band two-dimensional image CG21 of the tumor, and the second synthesized two-dimensional image CG22 of the spur, similarly to the second embodiment.

[0146] Figure 26 The diagram is used to explain the extraction of calcified regions from the first synthesized two-dimensional image CG1. The synthesizing unit 34 extracts regions corresponding to the calcified regions detected by the attention structure detecting unit 31 from the tomographic images Dj as calcified regions 42A and 42B from the first synthesized two-dimensional image CG1.

[0147] Figure 27 : is a diagram for explaining the generation of the synthesized two-dimensional image CG0 in the third embodiment. Figure 27 As shown, the synthesis unit 34 first replaces the tumor region in the first synthesized two-dimensional image CG1 with the tumor region in the second synthesized two-dimensional image CG21 related to the tumor, thereby synthesizing the second synthesized two-dimensional image CG21 related to the tumor with the first synthesized two-dimensional image CG1. This generates an intermediate synthesized two-dimensional image CG11.

[0148] Next, the synthesis unit 34 replaces the burr region in the intermediate synthesized two-dimensional image CG11 with the burr region in the second synthesized two-dimensional image CG22 regarding burrs, thereby synthesizing the second synthesized two-dimensional image CG22 regarding burrs with the intermediate synthesized two-dimensional image CG11. Thus, the intermediate synthesized two-dimensional image CG12 is generated.

[0149] Furthermore, in the third embodiment, the synthesis unit 34 replaces the calcified regions in the intermediate synthesized two-dimensional image CG12 with the calcified regions 42A and 42B, thereby synthesizing the calcified regions 42A and 42B with the intermediate synthesized two-dimensional image CG12. Thus, a synthesized two-dimensional image CG0 according to the third embodiment is generated.

[0150] Next, the processing performed in the third embodiment will be described. Figure 28 This is a flowchart illustrating the processing performed in the third embodiment. It is assumed that a plurality of tomographic images Dj have been previously acquired and stored in the storage device 23. When the input device 25 receives a processing start instruction from the operator, processing begins, and the structure of interest detector 31 detects a structure of interest from the plurality of tomographic images Dj (step ST31). Next, the frequency decomposition unit 32 performs frequency decomposition on each of the plurality of tomographic images Dj, thereby deriving a plurality of frequency band tomographic images representing frequency components of each of a plurality of frequency bands for each of the plurality of tomographic images Dj (step ST32).

[0151] Next, the selection unit 33 selects a band tomographic image corresponding to the tomographic image in which the structure of interest is detected from the plurality of band tomographic images, for each corresponding pixel in the plurality of band tomographic images, according to the type and frequency band of the structure of interest (step ST33 ).

[0152] Next, the synthesis unit 34 generates a first synthesized two-dimensional image CG1 from the plurality of tomographic images Dj (step ST34). The processing of step ST34 may be performed before or in parallel with the processing of steps ST31 to ST33. The synthesis unit 34 then generates second synthesized two-dimensional images CG21 and CG22 for the tumor and burr, respectively (step ST35). Furthermore, the synthesis unit 34 extracts the calcified regions 42A and 42B from the first synthesized two-dimensional image CG1 (step ST36). The processing of step ST36 may be performed before or in parallel with any processing, as long as it is performed after the first synthesized two-dimensional image CG1 is generated.

[0153] Next, the synthesis unit 34 sequentially synthesizes the second synthesized two-dimensional images CG21 and CG22, respectively, for the tumor and spur, with the first synthesized two-dimensional image CG1, thereby generating an intermediate synthesized two-dimensional image CG12 (step ST37). The synthesis unit 34 then synthesizes the calcified regions 42A and 42B onto the intermediate synthesized two-dimensional image CG12 to generate a synthesized two-dimensional image CG0 (step ST38). The display control unit 35 then displays the synthesized two-dimensional image CG0 on the display 24 (step ST39), and the process ends.

[0154] In each of the above-described embodiments, with respect to the tumor, for each pixel in the multiple band tomographic images corresponding to a pixel in the synthesized two-dimensional image CG0 in the mid- and low-frequency bands MLf, all band tomographic images containing the tumor are selected. Furthermore, for each pixel in the multiple band tomographic images corresponding to a pixel in the synthesized two-dimensional image CG0 in the high-frequency band Hf, a single band tomographic image that best represents the tumor is selected. However, the selection of band tomographic images is not limited to this. With respect to the tumor, for each pixel in the multiple band tomographic images corresponding to a pixel in the synthesized two-dimensional image CG0 in the mid- and low-frequency bands MLf, all band tomographic images containing the tumor may be selected. This will be described below as a fourth embodiment.

[0155] When the band tomographic images are selected as in the fourth embodiment and the processing of the first embodiment is performed, the synthesizing unit 34 generates a synthesized band two-dimensional image CGML0 of the middle and low frequency bands MLf in the same manner as in the first embodiment. On the other hand, in the fourth embodiment, no band tomographic image is selected in the high frequency band Hf for the tumor. Therefore, Figure 17 In the pixels P2 and P7 shown in FIG. 1 , the band tomographic images DH2 and DH3-1 are not selected. Therefore, when the processing of the first embodiment is performed in the fourth embodiment, with respect to the pixels P2 and P7, as shown in FIG. Figure 29 As shown, the synthesizing unit 34 derives the arithmetic mean of the pixel values of all the band tomographic images DH1 to DH6 similarly to the pixels P1, P13, and P15, and uses the arithmetic mean as the pixel value of pixels P2 and P7 of the synthesized band two-dimensional image CGH0 of the high frequency band Hf.

[0156] On the other hand, when the band tomographic image is selected and the processing of the second embodiment is performed as in the fourth embodiment, the synthesizing unit 34 generates a synthesized band two-dimensional image CGML21 of the middle and low frequency bands MLf of the tumor in the same manner as in the second embodiment. On the other hand, in the fourth embodiment, the band tomographic image is not selected in the high frequency band Hf of the tumor. Therefore, Figure 21 In the pixels P2 and P7 shown in FIG. 1 , the band tomographic images DH2 and DH3-1 are not selected. Therefore, when the processing of the second embodiment is performed in the fourth embodiment, with respect to the pixels P2 and P7, as shown in FIG. Figure 30 As shown, the synthesizing unit 34 derives the arithmetic mean value of the pixel values of all the band tomographic images DH1 to DH6 in the same manner as the pixels P1, P4 to P6, and P11 to P15, and uses the arithmetic mean value as the pixel value of the pixels P2 and P7 of the second synthesized band two-dimensional image CGH21 for the high-frequency band Hf of the tumor.

[0157] Furthermore, regarding the tumor, one band tomographic image that best represents the tumor can be selected for each pixel corresponding to a pixel of the synthesized two-dimensional image CG0 from among the multiple band tomographic images, both in the high-frequency band Hf and the low-medium frequency band MLf. This will be described below as a fifth embodiment.

[0158] In the fifth embodiment, regarding the tumor, the selection unit 33 selects one band tomographic image that best represents the tumor for each pixel corresponding to the pixel of the synthesized two-dimensional image CG0 in the medium and low frequency band MLf. Specifically, the selection unit 33 selects Figure 10 The band tomographic image DML4 is selected for the pixels P2 and P3, and the band tomographic image DML3 is selected for the pixels P7 to P10. In the fifth embodiment, the pixels P8 and P9 are not selected. Figure 10 Band tomographic images DML2 and DML4 are shown. In the high frequency band Hf, a band tomographic image is selected in the same manner as in the above-described embodiments.

[0159] When the band tomographic images are selected as in the fifth embodiment and the processing of the first embodiment is performed, the synthesizing unit 34 generates the synthesized band two-dimensional image CGH0 of the high frequency band Hf in the same manner as in the first embodiment. On the other hand, in the fifth embodiment, regarding the tumor, for each pixel corresponding to the pixel of the synthesized two-dimensional image CG0 among the plurality of band tomographic images in the middle and low frequency bands MLf, one band tomographic image that best represents the tumor is selected. Therefore, Figure 16 In the pixels P8 and P9 shown in FIG. 1 , only one band tomographic image DML3 is selected. Therefore, when the processing of the first embodiment is performed in the fifth embodiment, regarding the pixels P8 and P9, as shown in FIG. Figure 31 As shown, the synthesis unit 34 uses the pixel values of the pixels P8 and P9 of the band tomographic image DML3 as the pixel values of the pixels P8 and P9 of the synthesized band two-dimensional image CGML0 of the middle-low frequency band MLf.

[0160] On the other hand, when the band tomographic images are selected and the processing of the second embodiment is performed as in the fifth embodiment, the synthesis unit 34 generates a second synthesized band two-dimensional image CGH21 of the high-frequency band Hf of the tumor in the same manner as in the second embodiment. On the other hand, in the fifth embodiment, in the middle and low frequency bands MLf, one band tomographic image that best represents the tumor is selected for each pixel corresponding to the pixel of the synthesized two-dimensional image CG0 among the plurality of band tomographic images. Therefore, in Figure 20 In the pixels P8 and P9 shown in FIG. 1 , only one band tomographic image DML3 is selected. Therefore, when the processing of the second embodiment is performed in the fifth embodiment, regarding the pixels P8 and P9, as shown in FIG. Figure 32As shown, the synthesizing unit 34 uses the pixel values of the pixels P8 and P9 of the band tomographic image DML3 as the pixel values of the pixels P8 and P9 of the synthesized band two-dimensional image CGML21 regarding the middle and low frequency band MLf of the tumor.

[0161] Furthermore, regarding the tumor, only in the low- to medium-frequency band MLf, one band tomographic image that best represents the tumor can be selected for each pixel corresponding to a pixel of the synthesized two-dimensional image CG0 among the multiple band tomographic images. This will be described below as a sixth embodiment.

[0162] In the sixth embodiment, regarding the tumor, the selection unit 33 selects one band tomographic image that best represents the tumor for each pixel corresponding to the pixel of the synthesized two-dimensional image CG0 only in the medium and low frequency bands MLf. Specifically, the selection unit 33 selects Figure 10 In the illustrated pixels P2 and P3, the band tomographic image DML4 is selected, and in the pixels P7 to P10, the band tomographic image DML3 is selected. On the other hand, in the sixth embodiment, no band tomographic image is selected in the high-frequency band Hf for a tumor.

[0163] When a band tomographic image is selected and the processing of the first embodiment is performed as in the sixth embodiment, the synthesizing unit 34 generates a synthesized band two-dimensional image CGH0 for the high-frequency band Hf, similarly to the fourth embodiment. Meanwhile, for the mid- and low-frequency bands MLf, the synthesizing unit 34 generates a synthesized band two-dimensional image CGML0 for the mid- and low-frequency bands MLf, similarly to the fifth embodiment.

[0164] On the other hand, when the band tomographic images are selected and the processing of the second embodiment is performed as in the sixth embodiment, the synthesizing unit 34 generates a second synthesized band two-dimensional image CGH21 for the high-frequency band Hf of the tumor, similarly to the fourth embodiment. On the other hand, for the medium- and low-frequency bands MLf, the synthesizing unit 34 generates a synthesized band two-dimensional image CGML21 for the medium- and low-frequency bands MLf of the tumor, similarly to the fifth embodiment.

[0165] Furthermore, in each of the above-described embodiments, when generating a composite band two-dimensional image from band tomographic images, the pixel value of a pixel where no structure of interest was detected was the arithmetic mean of the corresponding pixels in the band tomographic images, but this is not limiting. Furthermore, in the second and third embodiments, when generating the first composite two-dimensional image CG1, the pixel value of the corresponding pixels in the tomographic images Dj was also used, but this is not limiting. Furthermore, in the second and third embodiments, when generating the second composite band two-dimensional image CGML22 for the low- and medium-frequency bands MLf of burrs and calcifications, the pixel value of the corresponding pixels in the band tomographic images DMLj was also used, but this is not limiting. Other known techniques that use weighted averages, median values, or the like as pixel values can also be applied. Furthermore, a minimum projection method using the minimum value of corresponding pixels in each band tomographic image or each tomographic image, or a maximum projection method using the maximum value, can also be used. In this case, the band tomographic image or tomographic image having the pixel with the minimum value or the maximum value becomes the preset tomographic image of the present invention.

[0166] Furthermore, for pixels where no structure of interest is detected, an average value of the corresponding pixels in each band tomographic image or tomographic image can be derived. Pixels whose difference from the average value is less than a predetermined set value are considered noise pixels with a significant noise impact. The noise pixels are then excluded from the pixel values of the composite band 2D image or composite 2D image. Furthermore, for pixels where no structure of interest is detected, a distribution value of the pixel values within a predetermined region containing the pixel can be derived. Pixels with a distribution value less than the predetermined set value are considered noise pixels. The noise pixels are then excluded from the pixel values of the composite band 2D image or composite 2D image. In this case, the band tomographic image or tomographic image containing pixels that are not noise pixels becomes the predetermined tomographic image of the present invention. Furthermore, processing can be performed to detect the edges of structures contained in each band tomographic image or tomographic image. For pixels where no structure of interest is detected, the pixel values of the pixels containing the edge are used as the pixel values of the composite band 2D image or composite 2D image. In this case, the band tomographic image or tomographic image containing pixels containing the edge becomes the predetermined tomographic image of the present invention.

[0167] Furthermore, while all structures of interest, including tumors, spurs, and calcifications, are detected in the above-described embodiments, this is not a limitation. The technology of the present invention can also be applied when detecting at least one structure of interest among tumors, spurs, and calcifications. Furthermore, when only one structure of interest is detected, it is sufficient to select a band tomographic image based solely on the frequency band.

[0168] Furthermore, the radiation in each of the above-mentioned embodiments is not particularly limited, and α-rays, γ-rays, and the like can also be applied in addition to X-rays.

[0169] Furthermore, in each of the above-described embodiments, for example, various processors described below can be used as the hardware configuration of the processing units (processing units) that perform various processes, such as the image acquisition unit 30, the structure-of-interest detection unit 31, the frequency decomposition unit 32, the selection unit 33, the synthesis unit 34, and the display control unit 35. As described above, the various processors described above include general-purpose processors (CPUs) that execute software (programs) to function as various processing units, as well as processors such as FPGAs (Field Programmable Gate Arrays) whose circuit configuration can be modified after manufacturing, programmable logic devices (PLDs), and ASICs (Application Specific Integrated Circuits) that have circuit configurations specifically designed to perform specific processes, such as dedicated circuits.

[0170] A single processing unit may be composed of one of these various processors, or a combination of two or more processors of the same or different types (e.g., a combination of multiple FPGAs or a combination of a CPU and an FPGA). Furthermore, a single processor may constitute multiple processing units.

[0171] As an example of a single processor constituting multiple processing units, there is a method in which a single processor is composed of a combination of one or more CPUs and software, as exemplified by computers such as clients and servers, and the processor functions as multiple processing units. A second method is a method in which a processor is used that implements the functions of an entire system including multiple processing units using a single IC (Integrated Circuit) chip, as exemplified by a system-on-chip (SoC). In this manner, various processing units are constructed as a hardware structure using one or more of the aforementioned processors.

[0172] Furthermore, more specifically, as the hardware configuration of these various processors, circuits (Circuitry) combining circuit elements such as semiconductor elements can be utilized.

[0173] Explanation of symbols

[0174] 1- Mammography apparatus, 2- Console, 3- Image storage system, 4- Image processing apparatus, 11- Rotation axis, 12- Arm, 13- Radiographic table, 14- Radiation irradiation unit, 15- Radiation detector, 15A- Detection surface, 16- Radiation source, 17- Compression plate, 21- CPU, 22- Image processing program, 23- Storage device, 24- Display, 25- Input device, 26- Memory, 27- Network I / F, 28- General Line, 30-image acquisition unit, 31-structure of interest detection unit, 32-frequency decomposition unit, 33-selection unit, 34-synthesis unit, 35-display control unit, 36-frequency decomposition unit, 40-indicator, 42A, 42B-calcification area, 50-display screen, 100-radiation imaging system, CG0-synthesized two-dimensional image, CG1-first synthesized two-dimensional image, CG11, CG12-intermediate synthesized two-dimensional image, CGML0-synthesized band two-dimensional image of the middle and low frequency band, CGML21-second synthesized band two-dimensional image of the middle and low frequency band of the tumor, CGH0-synthesized band two-dimensional image of the high frequency band, CGH21-second synthesized band two-dimensional image of the high frequency band of the tumor, CGH22-second synthesized band two-dimensional image of the high frequency band of the burr, CGH23-second synthesized band two-dimensional image of the high frequency band of the calcification, Dj (j=1~m)-tomographic image, DMLj-band tomographic image of the middle and low frequency band Image, DHj-band tomographic image of high-frequency band, Gi (i=1~n)-projection image, Hf-high-frequency band, K13, K21, K31, K32, K41a, K41b, K42, K52, K63-structure of interest, M-breast, MLf-medium and low-frequency bands, P1~P15, P100~P102-pixels, R1~R3-detection results, Si (i=1~n)-ray source position, Sc-reference ray source position, X0-optical axis.

Claims

1. An image processing device comprising at least one processor, The processor is configured to: detecting a structure of interest from a plurality of tomographic images representing a plurality of tomographic planes of the object, selecting a tomographic image from the plurality of tomographic images according to a frequency band in a region where the structure of interest is detected, generating a synthetic two-dimensional image using the selected tomographic image in a region where the structure of interest is detected, and generating a synthetic two-dimensional image using a preset tomographic image in a region where the structure of interest is not detected, By performing frequency decomposition on the plurality of tomographic images, a plurality of frequency band tomographic images are derived for each of a plurality of frequency bands. selecting, for each pixel corresponding to a pixel of the synthesized two-dimensional image from among the plurality of band tomographic images, a band tomographic image corresponding to a tomographic image in which the structure of interest is detected, based on the frequency band; The synthesized two-dimensional image is generated using the selected band tomographic images in the region where the structure of interest is detected.

2. The image processing apparatus according to claim 1, wherein: The processor is configured to select, according to the frequency band, different numbers of frequency band tomographic images corresponding to the tomographic image in which the structure of interest is detected from the plurality of frequency band tomographic images.

3. The image processing device according to claim 1 or 2, wherein: The plurality of frequency bands include a first frequency band and a second frequency band lower than the first frequency band, The processor is configured to select a smaller number of the frequency band tomographic images in the first frequency band than in the second frequency band.

4. The image processing apparatus according to claim 3, wherein: The processor is configured to select, in the second band, all of the band tomographic images including the structure of interest for each pixel corresponding to a pixel of the synthesized two-dimensional image among the plurality of band tomographic images.

5. The image processing apparatus according to claim 3, wherein: The processor is configured to select, in the second band, one of the plurality of band tomographic images that best represents the structure of interest for each pixel corresponding to a pixel of the synthesized two-dimensional image. The image processing apparatus according to claim 3 , wherein: The processor is configured to select, in the first band, one of the plurality of band tomographic images that best represents the structure of interest for each pixel position corresponding to a pixel position of the synthesized two-dimensional image.

7. The image processing apparatus according to claim 5, wherein: The one band tomographic image that best represents the structure of interest is the band tomographic image in which the structure of interest is largest or the band tomographic image in which the likelihood of detecting the structure of interest is the highest.

8. The image processing apparatus according to claim 1 or 2, wherein: The processor further selects the band tomographic image according to the type of the structure of interest.

9. The image processing apparatus according to claim 8, wherein: The structures of interest are tumors, spurs, and calcifications.

10. The image processing apparatus according to claim 1 or 2, wherein: The processor is configured to generate a composite band two-dimensional image for each of the bands using the selected band tomographic images in pixels of the band tomographic images corresponding to the structure of interest, and to perform frequency synthesis on the composite band two-dimensional images to generate the composite two-dimensional image. The image processing apparatus according to claim 10 , wherein: The processor is configured to, when a plurality of band tomographic images are selected, generate the composite band two-dimensional image having pixel values of the band tomographic images determined based on a preset priority of a structure of interest, among pixels corresponding to pixels of the composite band two-dimensional image.

12. The image processing apparatus according to claim 8, wherein: The processor is configured to generate a first synthesized two-dimensional image by synthesizing the plurality of tomographic images. In the pixels of the band tomographic image corresponding to the structure of interest, for each type of the structure of interest, a synthetic band two-dimensional image is generated according to each of the frequency bands using the selected band tomographic image, the synthetic band two-dimensional image is frequency synthesized to generate a second synthetic two-dimensional image for each type of the structure of interest, and the second synthetic two-dimensional image generated for each type of the structure of interest is synthesized with the first synthetic two-dimensional image, thereby generating the synthetic two-dimensional image.

13. The image processing apparatus according to claim 12, wherein: The processor is configured to replace pixel values of the structure of interest in the first synthesized two-dimensional image with pixel values of the structure of interest in the second synthesized two-dimensional image, thereby synthesizing the second synthesized two-dimensional image with the first synthesized two-dimensional image. The image processing apparatus according to claim 13 , wherein: The processor is configured to generate the synthesized two-dimensional image having pixel values of the second synthesized two-dimensional image determined according to a preset priority of the structure of interest when a plurality of the structures of interest are included in corresponding pixels between the plurality of the second synthesized two-dimensional images.

15. The image processing apparatus according to claim 8, wherein: The processor is configured to generate a first synthesized two-dimensional image by synthesizing the plurality of tomographic images. extracting a region of a predetermined specific type of structure of interest from the first synthesized two-dimensional image, For other structures of interest other than the specific type of structure of interest, in the pixels of the band tomographic image corresponding to the other structures of interest, a synthetic band two-dimensional image is generated according to each frequency band for each type of the other structures of interest using the selected band tomographic image, the synthetic band two-dimensional image is frequency synthesized to generate a second synthetic two-dimensional image for each type of the other structures of interest, the second synthetic two-dimensional image for the other structures of interest is synthesized into the first synthetic two-dimensional image, and the area of the specific type of structure of interest is synthesized on the first synthetic two-dimensional image synthesized with the second synthetic two-dimensional image, thereby generating the synthetic two-dimensional image.

16. The image processing apparatus according to claim 15, wherein: The specific type of structure of interest is calcification, and the other structures of interest are tumors and burrs.

17. The image processing apparatus according to claim 15, wherein: The processor is configured to replace pixel values of the structure of interest in the first synthesized two-dimensional image with pixel values of the structure of interest in the second synthesized two-dimensional image, thereby synthesizing the second synthesized two-dimensional image with the first synthesized two-dimensional image.

18. The image processing apparatus according to claim 17, wherein: The processor is configured to generate the synthesized two-dimensional image having pixel values of the second synthesized two-dimensional image determined according to a priority of a preset structure of interest when a plurality of the other structures of interest are included in corresponding pixels between the plurality of the second synthesized two-dimensional images.

19. The image processing apparatus according to claim 15, wherein: The processor is configured to replace the pixel values of the structure of interest in the first synthesized two-dimensional image synthesized with the second synthesized two-dimensional image with the pixel values of the area of the specific type of structure of interest, thereby synthesizing the area of the specific type of structure of interest on the first synthesized two-dimensional image synthesized with the second synthesized two-dimensional image.

20. An image processing method, wherein: detecting a structure of interest from a plurality of tomographic images representing a plurality of tomographic planes of the object, selecting a tomographic image from the plurality of tomographic images according to a frequency band in a region where the structure of interest is detected, generating a synthetic two-dimensional image using the selected tomographic image in a region where the structure of interest is detected, and generating a synthetic two-dimensional image using a preset tomographic image in a region where the structure of interest is not detected, By performing frequency decomposition on the plurality of tomographic images, a plurality of frequency band tomographic images are derived for each of a plurality of frequency bands. selecting, for each pixel corresponding to a pixel of the synthesized two-dimensional image from among the plurality of band tomographic images, a band tomographic image corresponding to a tomographic image in which the structure of interest is detected, based on the frequency band; The synthesized two-dimensional image is generated using the selected band tomographic images in the region where the structure of interest is detected.

21. A recording medium capable of being read by a computer, the recording medium recording an image processing program, the image processing program causing the computer to execute the following steps: detecting a structure of interest from a plurality of tomographic images representing a plurality of tomographic planes of the object; selecting a tomographic image from the plurality of tomographic images according to a frequency band in a region where the structure of interest is detected; generating a synthetic two-dimensional image using the selected tomographic image in a region where the structure of interest is detected, and generating a synthetic two-dimensional image using a preset tomographic image in a region where the structure of interest is not detected; deriving a plurality of frequency band tomographic images for each of a plurality of frequency bands by frequency decomposing the plurality of tomographic images; selecting, for each pixel corresponding to a pixel of the synthesized two-dimensional image among the plurality of band tomographic images, a band tomographic image corresponding to a tomographic image in which the structure of interest is detected, according to the frequency band; as well as The synthesized two-dimensional image is generated using the selected band tomographic images in the region where the structure of interest is detected.

Images