JP7595488B2 - Information processing device, information processing method, and computer program - Google Patents

Description

The present invention relates to an information processing device, an information processing method, and a computer program.

Daily inspection work of medical radiation imaging devices includes imaging a test object called a phantom for the purpose of calibrating the image quality of the radiation image (see, for example, paragraph 0026 of Patent Document 1). Here, the manager of the radiation imaging device acquires a reference phantom image (reference phantom image) in advance and acquires data such as the brightness of the region of interest of the reference phantom image. Then, when inspecting the radiation imaging device, the manager places the phantom in the radiation imaging device and acquires a phantom image (inspection phantom image), and compares data such as the brightness of the region of interest of the inspection phantom image with data such as the brightness of the region of interest of the reference phantom image to determine whether calibration of the radiation imaging device is required.

Japanese Patent Publication No. 2020-173267

When inspecting a radiation imaging device, phantom imaging does not necessarily use a single phantom, and imaging may be performed with multiple phantoms lined up. When multiple phantoms are used, the administrator of the radiation imaging device must line up the multiple phantoms on the radiation imaging device. Here, since the administrator manually arranges the phantoms on the radiation imaging device, there may be a shift in the arrangement of the phantoms each time they are arranged. In other words, there is a possibility that the arrangement of the phantoms when the reference phantom image is imaged may differ from the arrangement of the phantoms when the inspection phantom image is imaged. For this reason, if the administrator uses the position of the region of interest of the reference phantom image as it is when setting the position of the region of interest of the inspection phantom image, data with guaranteed accuracy cannot be obtained. Therefore, the administrator must set the region of interest of the inspection phantom image while visually inspecting the inspection phantom image.

As such, in conventional inspection work for radiographic imaging devices, the manager must set the position of the region of interest in the inspection phantom image himself, which takes time for the inspection work. In addition, if the setting of the region of interest in the inspection phantom image relies solely on human visual inspection, it is difficult to ensure the accuracy of data such as brightness required for calibration.

The present invention has been made in consideration of these circumstances, and aims to provide an information processing device, an information processing method, and a computer program that can reduce the work time required to inspect a radiation imaging device and improve the inspection accuracy.

According to the present invention, there is provided an information processing device including a shift amount acquisition unit and a position acquisition unit, the shift amount acquisition unit configured to acquire the shift amount between a reference phantom image and an inspection phantom image, the reference phantom image is acquired by imaging a plurality of separate phantoms, the inspection phantom image is acquired by imaging the same phantom as the reference phantom image, a region of interest is set in advance in the reference phantom image, and the position acquisition unit is configured to acquire the position of the region of interest in the inspection phantom image based on the shift amount and the position of the region of interest in the reference phantom image.

The present invention includes a shift amount acquisition unit and a position acquisition unit, and the position acquisition unit is configured to acquire the position of the region of interest of the inspection phantom image based on the shift amount acquired by the shift amount acquisition unit and the position of the region of interest of the reference phantom image. In this way, in the present invention, the information processing device automatically acquires the region of interest of the inspection phantom image, so that the work time during inspection of the radiation imaging device can be reduced and the accuracy of the inspection can be improved.

Various embodiments of the present invention will be described below. The embodiments described below can be combined with each other.
Preferably, the multiple phantoms include a first and a second phantom, the first and second phantoms are arranged side by side on a plane, the reference phantom image and the inspection phantom image each have a multiple image areas, the multiple image areas include a first and a second image area, the first and second image areas are acquired by imaging the first and second phantoms, respectively, and the area of interest is preset in each of the first and second image areas of the reference phantom image.
Preferably, the multiple phantoms further include a third phantom, the third phantom being arranged so as to overlap the first phantom, the multiple image regions including a third image region, the third image region being acquired by imaging the third phantom, and the region of interest being preset in the third image region of the reference phantom image.An information processing device is provided.
Preferably, an information processing device is provided which further comprises a feature pixel extraction unit, wherein each of the first and second image regions of the reference phantom image and each of the first and second image regions of the inspection phantom image have a feature pixel, the feature pixel extraction unit is configured to extract the feature pixels of the first and second image regions of the reference phantom image and the feature pixels of the first and second image regions of the inspection phantom image, the feature pixel extraction unit extracts the feature pixels based on the brightness values of pixels in the first and second image regions, and the deviation amount acquired by the deviation amount acquisition unit is the deviation amount between the positions of the feature pixels of the first and second image regions in the reference phantom image and the positions of the feature pixels of the first and second image regions in the inspection phantom image.
Preferably, an information processing device is provided which further includes a position setting unit, the position setting unit being configured to be able to set the region of interest in the reference phantom image, and the position acquisition unit acquiring the position of the region of interest in the inspection phantom image based on the amount of deviation and the position of the region of interest in the reference phantom image set by the position setting unit.
Preferably, the plurality of phantoms include ACR (American College of Radiology) phantoms.
Preferably, an information processing device is provided which further includes a region determination unit which binarizes each pixel in the reference phantom image and the inspection phantom image into a first and second value, and which determines first and second image regions based on the number of pixels in a second value pixel region, the second value pixel region being a pixel having the second value or a collection of pixels having the second value.
Preferably, an information processing device is provided, further comprising a region determination unit, wherein the region determination unit determines the multiple image regions based on a learning model that inputs the reference phantom image or the inspection phantom image and determines the multiple image regions.
Preferably, there is provided an information processing device, wherein the region determining section is configured to determine a third image region by executing a process of detecting a shape of a third phantom.

According to another aspect of an embodiment of the present invention, there is provided an information processing method including a shift amount acquisition step and a position acquisition step, in which the shift amount acquisition step acquires the shift amount between a reference phantom image and an inspection phantom image, the reference phantom image is acquired by imaging a plurality of separate phantoms, the inspection phantom image is acquired by imaging the same phantom as the reference phantom image, a region of interest is set in advance in the reference phantom image, and the position acquisition step acquires the position of the region of interest in the inspection phantom image based on the shift amount and the position of the region of interest in the reference phantom image.

According to another aspect of an embodiment of the present invention, there is provided a computer program for causing a computer to execute an information processing method, the information processing method including a shift amount acquisition step and a position acquisition step, in which the shift amount acquisition step acquires a shift amount between a reference phantom image and an inspection phantom image, the reference phantom image is acquired by imaging a plurality of separate phantoms, the inspection phantom image is acquired by imaging the same phantom as the reference phantom image, a region of interest is set in advance in the reference phantom image, and in the position acquisition step, the position of the region of interest in the inspection phantom image is acquired based on the shift amount and the position of the region of interest in the reference phantom image.

FIG. 1 is a diagram showing an example of the general configuration of a medical radiation imaging apparatus 100 including an information processing apparatus 1 according to an embodiment. FIG. 2 is an explanatory diagram of the data flow in the reference phantom image setting mode. FIG. 3 is an explanatory diagram of the data flow in the inspection phantom image setting mode. FIG. 4 is a schematic diagram of the first to third phantoms p1 to p3. Fig. 5A shows an example of the arrangement of the first to third phantoms p1, p21, p22, and p3 when capturing a reference phantom image. Fig. 5B is a schematic diagram of a reference phantom image acquired by capturing images of the first to third phantoms p1 to p3 shown in Fig. 5A with an imaging device. FIG. 6 is a schematic diagram of an image obtained by performing binarization processing on the reference phantom image shown in FIG. 5B. FIG. 7 is a schematic diagram of an image obtained by performing pixel number processing on the binarized image shown in FIG. Fig. 8A shows an example of a feature pixel px of the second image region R21 in the reference phantom image shown in Fig. 5B. Fig. 8B shows an example of a feature pixel px of the first image region R1 in the reference phantom image shown in Fig. 5B. Fig. 8C shows an example of a feature pixel px of the second image region R22 in the reference phantom image shown in Fig. 5B. Fig. 9A shows an example of the arrangement of the first to third phantoms p1 to p3 when capturing an inspection phantom image. Fig. 9B is a schematic diagram of an inspection phantom image acquired by capturing images of the first to third phantoms p1 to p3 shown in Fig. 9A with an imaging device. FIG. 10 is a schematic diagram of an image obtained by performing binarization processing on the inspection phantom image shown in FIG. 9B. FIG. 11 is a schematic diagram of an image obtained by performing pixel number processing on the binarized image shown in FIG. Fig. 12A shows an example of a feature pixel px of the second image region R21 in the inspection phantom image shown in Fig. 9B. Fig. 12B shows an example of a feature pixel px of the first image region R1 in the inspection phantom image shown in Fig. 9B. Fig. 12C shows an example of a feature pixel px of the second image region R22 in the inspection phantom image shown in Fig. 9B. Fig. 13A is a schematic diagram showing regions of interest Ra1, Ra21, Ra22, and Ra3 set on a reference phantom image. Fig. 13B is a schematic diagram showing regions of interest Ra1, Ra21, Ra22, and Ra3 set on an inspection phantom image by an information processing device according to an embodiment. FIG. 14 is a control flowchart of the information processing apparatus according to the embodiment. FIG. 15 is a diagram for explaining the histogram created for the binarized image shown in FIG.

The following describes the embodiments of the present invention with reference to the drawings. The various features shown in the embodiments below can be combined with each other. In addition, each feature can be an invention independently.

1. Description of Overall Configuration The overall configuration of a medical radiation imaging device 100 including an information processing device 1 according to an embodiment will be described. As shown in Fig. 1, the radiation imaging device 100 of this embodiment includes an imaging device 20, an image display device 30, and an information processing device 1. The information processing device 1 according to this embodiment is configured to be capable of automatically detecting an ROI position without changing a manual imaging technique or using a predetermined marker.

The imaging device 20 includes, for example, a radiation generating unit 21, a placement unit 22, a detection unit 23, and an image processing unit 24. The radiation generating unit 21 is configured to be able to emit radiation toward the placement unit 22 and the detection unit 23. The placement unit 22 is a unit for placing an object to be radiated with radiation, such as a phantom or a human body. The detection unit 23 is a unit for receiving radiation that has passed through an object to be radiated placed in the placement unit 22, and has a function of converting the radiation into a digital signal. The image processing unit 24 has a function of generating a phantom image based on the digital signal acquired from the detection unit 23. The image processing unit 24 also has a function of performing pre-processing on the phantom image, and specifically, the image processing unit 24 includes a size adjustment unit and a window level adjustment unit.

The resolution of the phantom image varies depending on the configuration of the imaging device 20. For this reason, the actual size (mm) per pixel of the phantom image may vary. Therefore, the size adjustment section of the image processing section 24 adjusts the size of the phantom image. The window level adjustment section of the image processing section 24 adjusts the window level of the phantom image. Window level adjustment is a process that improves the contrast of a specific gradation range of an image that has a wide range of gradation values. By performing window level adjustment, the visibility of the phantom image can be improved.

The image display device 30 is a display that displays various medical images, such as phantom images, captured by the imaging device 20 and subjected to image processing.

The information processing device 1 includes an area determination unit 2, a feature pixel extraction unit 3, a deviation amount acquisition unit 4, a position acquisition unit 5, an item value calculation unit 6, a mode setting unit 7, a position setting unit 8, and a memory unit 9. The area determination unit 2 also includes an area division unit 2A and a second determination unit 2B. The area division unit 2A further includes a binarization processing unit 2A1, a labeling unit 2A2, and a first determination unit 2A3.

Each component of the information processing device 1 and each component of the image processing unit 24 of the imaging device 20 may be realized by software or hardware. When realized by software, various functions can be realized by a CPU executing a computer program. The program may be stored in an internal storage unit or a non-transitory computer-readable recording medium. Also, a program stored in an external storage unit may be read and realized by so-called cloud computing. When realized by hardware, it can be realized by various circuits such as ASIC, FPGA, or DRP. In this embodiment, various information and concepts that include it are handled, but these are represented by high and low signal values as a binary bit collection consisting of 0 or 1, and communication and calculation can be performed by the above software or hardware aspects.

The information processing device 1 has a region determination function and an automatic region of interest setting function. The region determination function is a function that detects in which region of the phantom image the phantom depicted in the phantom image is located. In addition, the automatic region of interest setting function is a function that automatically sets the position and size of the region of interest of the inspection phantom image described below based on the region of interest of the reference phantom image described below.

The information processing device 1 is also configured to be able to selectively execute two setting modes. One setting mode is a reference phantom image setting mode, which is a mode for setting a region of interest in advance for a reference phantom image. The other setting mode is an inspection phantom image setting mode, which is a mode for automatically acquiring a region of interest in an inspection phantom image.
The above-mentioned region determination function is used in both the reference phantom image setting mode and the inspection phantom image setting mode, and the above-mentioned automatic region of interest setting function is used in the inspection phantom image setting mode.

2 Phantom 2-1 Example of Phantom Configuration In the embodiment, the phantom image will be described as a phantom image for mammography. Therefore, the phantom p used is also a phantom for mammography. As shown in Fig. 4, the phantom p in the embodiment has a first phantom p1, second phantoms p21 and p22, and a third phantom p3. That is, in the embodiment, a plurality of separate phantoms are simultaneously placed in the imaging device 20, and the phantoms are imaged.

The first phantom p1 is a rectangular plate material and has a pattern (not shown) formed thereon. The first phantom p1 is disposed between the second phantom p21 and the second phantom p22. The second phantoms p21 and p22 are also rectangular plate material and have a pattern (not shown) formed thereon. The second phantoms p21 and p22 are disposed so as to be aligned with each other on the same plane as the first phantom p1. The second phantoms p21 and p22 have gradual shading formed thereon. The third phantom p3 is a circular plate material and is disposed so as to overlap the first phantom p1.
The patterns formed on the first phantom p1 and the second phantoms p21 and p22 are, for example, a pattern simulating a tumor mass shadow, a pattern simulating a microcalcification shadow, a pattern simulating a linear shadow indicating a mammary gland or the like.

2-2 Phantom Image As described above, the information processing device 1 has an automatic setting function for a region of interest, but when the administrator uses this function, the information processing device 1 must acquire a reference phantom image (reference phantom image d1) and the region of interest must be set in advance for the reference phantom image d1. Here, the reference phantom image d1 and the inspection phantom image D1, and the layout of the phantom p corresponding to these images will be described.

For example, the administrator places the phantom p on the imaging device 20 in a layout as shown in FIG. 5A, and images the phantom p. As a result, the information processing device 1 acquires a reference phantom image d1 as shown in FIG. 5B. Then, as shown in FIG. 13A, the administrator manually sets a region of interest in this reference phantom image d1.

During inspection, the phantom p is not necessarily arranged in the same manner as the arrangement of the phantom p corresponding to the reference phantom image d1 (arrangement in FIG. 5A). In other words, the arrangement of the phantom p during inspection may deviate from the arrangement of the phantom p when the reference phantom image is captured. In the embodiment, as an example, the phantom p is arranged in the imaging device 20 in a layout as shown in FIG. 9A. When compared with the layout in FIG. 5A, the first phantom p1 shown in FIG. 9A is rotated about 2 degrees clockwise. Also, when compared with the layout in FIG. 5A, the second phantom p21 shown in FIG. 9A is rotated about 3 degrees clockwise, and the second phantom p22 shown in FIG. 9A is moved horizontally to the left. When the administrator captures the phantom p in the layout as shown in FIG. 9A, the information processing device 1 acquires an inspection phantom image D1 as shown in FIG. 9B.

3 Detailed Configuration of Information Processing Device 1 The information processing device 1 is configured to be capable of automatically setting a region of interest for the above-mentioned inspection phantom image D1. Note that in the embodiment, it is also possible to fine-tune the position of the region of interest that has been automatically set for the inspection phantom image D1, but it is optional whether or not the information processing device 1 has this fine-tuning function.

3-2 Region determination unit 2
The region determination unit 2 has a function of acquiring phantom images (reference phantom image d1 and inspection phantom image D1) from the image processing unit 24 and determining the region of each phantom in the phantom image. That is, the region determination unit 2 has a function of determining a region corresponding to the first phantom p1, a region corresponding to the second phantoms p21 and p22, and a region corresponding to the third phantom p3. The region determination unit 2 includes a region division unit 2A and a second determination unit 2B, and the region division unit 2A includes a binarization processing unit 2A1, a labeling unit 2A2, and a first determination unit 2A3. The configuration of these units will be described.

3-2-1 Area dividing section 2A
The region dividing unit 2A has a function of determining a region corresponding to the first phantom p1 and regions corresponding to the second phantoms p21 and p22 in the phantom image. This is determined not by region dividing section 2A but by second determining section 2B.
In this embodiment, the first phantom p1 is disposed so as to be sandwiched between the second phantoms p21 and p22. Therefore, once the region of the first phantom p1 is determined, the remaining region is determined as the second phantom p21. It is known in advance that the first phantom p1 has a larger area of high brightness compared to the second phantoms p21 and p22. Using this information, the region dividing section 2A can determine the region of the first phantom p1.

3-2-1-1 Binarization processing unit 2A1
The binarization processing unit 2A1 is configured to perform binarization processing on the phantom image. In this binarization processing, when a pixel in the phantom image is less than the threshold, the pixel is set to a first value (0 in the embodiment) and is shown in black. When a pixel in the phantom image has a brightness equal to or greater than a predetermined threshold, the pixel is set to a second value (1 in the embodiment) and is shown in white. Each region shown by the second value (white) is referred to as a second-value pixel region. When the binarization processing unit 2A1 performs binarization processing on, for example, the reference phantom image d1, the binarization processing unit 2A1 generates a binarized processed image shown in FIG. 6. Similarly, when the binarization processing unit 2A1 performs binarization processing on, for example, the inspection phantom image D1, the binarization processing unit 2A1 generates a binarized processed image shown in FIG. 10.

3-2-1-2 Labeling section 2A2
The labeling unit 2A2 has a function of labeling the second-value pixel region of the binarized phantom image. For example, as shown in Fig. 5A and Fig. 6, the labeling unit 2A2 labels the regions g1, g21, g22, and g3 of the image obtained by binarizing the reference phantom image d1. The region g1 corresponds to the first phantom p1, the region g21 corresponds to the second phantom p21, the region g22 corresponds to the second phantom p22, and the region g3 corresponds to the third phantom p3.

3-2-1-3 First decision section 2A3
The first determination unit 2A3 determines the first and second image regions based on the number of pixels in the labeled second-value pixel regions (regions g1, g21, g22, and g3). The first determination unit 2A3 has a first function of determining the largest area among the second-value pixel areas and determining an area corresponding to the first phantom p1 based on the determination result. 2, the region corresponding to the first phantom p1 is defined as the first image region R1. In addition, the first determination unit 2A3 determines the first image region R1, thereby determining the second phantoms p21 and p22. The regions corresponding to the second phantoms p21 and p22 are defined as second image regions R21 and R22, respectively.

First, the first function will be described.
The largest region among the regions g1, g21, g22, and g3 is the region g1. Therefore, the first determination unit 2A3 determines that the region g1 is the largest region. Here, the first determination unit 2A3 sets a rectangle circumscribing the region g1. Specifically, the smallest rectangle among the rectangles that include the region g1 is set. This smallest rectangle corresponds to the region sandwiched between the upper and lower virtual lines L1 in the reference phantom image as shown in FIG. 7. Also, in the inspection phantom image, it corresponds to the region sandwiched between the upper and lower virtual lines L1 as shown in FIG. 11. In the embodiment, the region sandwiched between the upper and lower virtual lines L1 is not the first image region R1, but the region sandwiched between the virtual line L2 that is a predetermined number of pixels away from the virtual line L1 is the first image region R1.
In addition, when the threshold value in the binarization process is not set to an appropriate value, the end portion of the region g1 is expected to be determined as the second value, but there is a possibility that it is determined as the first value. In that case, when the first determination unit 2A3 executes the process related to the first function, the end portion of the region g1 belongs to an area outside the two virtual lines L1, and the region g1 is divided by the virtual lines L1. Therefore, in the embodiment, the area sandwiched between the upper and lower virtual lines L1 is not the first image region R1, but the area sandwiched between the virtual lines L2 that are a predetermined number of pixels away from the virtual line L1 is the first image region R1. In the embodiment, the value of the predetermined pixel is described as 100 pixels as an example, but is not particularly limited. The value of the predetermined pixel can be set to, for example, the number of pixels equivalent to 5 (mm) of the actual size.

Note that, as shown in FIG. 9B, the layout of the phantom p during inspection is such that the first phantom p1 is rotated clockwise by about 2 degrees. Therefore, the smallest rectangle in the inspection phantom image is larger than the smallest rectangle in the reference phantom image. In other words, the area sandwiched between the virtual lines L1 in the inspection phantom image is larger than the area sandwiched between the virtual lines L1 in the reference phantom image. Therefore, the area sandwiched between the virtual lines L2 in the inspection phantom image (first image region R1) is larger than the area sandwiched between the virtual lines L2 in the reference phantom image (first image region R1). In this way, if the position and size of the first image region R1 can be roughly grasped, there is no problem in terms of accuracy, so in the embodiment, differences in the size of the first image region R1 are allowed.

Next, the second function will be described.
The first determination unit 2A3 determines the regions of the phantom image divided vertically by the first image region R1 as the second image regions R21 and R22. Here, the overall size of the reference phantom image and the overall size of the inspection phantom image are the same. Meanwhile, as described above, the first image region R1 in the inspection phantom image is larger than the first image region R1 in the reference phantom image. Therefore, the second image regions R21 and R22 in the inspection phantom image are smaller than the second image regions R21 and R22 in the reference phantom image. For the same reason as above, in the embodiment, the difference in size between the second image regions R21 and R22 is allowed.

3-2-2 Second deciding section 2B
The second determination unit 2B executes a process for detecting the shape of the third phantom p3 for the phantom images (the reference phantom image and the inspection phantom image) to determine a region of the phantom image corresponding to the third phantom p3. Here, the region corresponding to the third phantom p3 is defined as a third image region R3. The third image region R3 is disposed within the first image region R1. In the embodiment, The shape of the third phantom p3 is circular. Therefore, in order to detect the circular shape, the second determination unit 2B performs a Hough transform on the phantom image to determine the third image region R3. By performing this Hough transformation, the second determination unit 2B obtains the center coordinates of the third image region R3 for each of the reference phantom image and the inspection phantom image.

3-3 Feature pixel extraction unit 3
The feature pixel extraction unit 3 extracts feature pixels in the phantom images (reference phantom image and inspection phantom image) by performing a feature pixel extraction process. The feature pixel extraction process is a process for extracting feature pixels from pixels in the phantom image based on the luminance values of pixels in the first and second image regions R1, R21, and R22. Here, the feature pixels are used to obtain the amount of deviation between the reference phantom image and the inspection phantom image. In the embodiment, the feature pixel extraction unit 3 can extract, for example, pixels in the phantom image in which the difference in luminance between adjacent pixels exceeds a predetermined threshold value as feature pixels.

As shown in Figs. 8A to 8C, the feature pixel extraction unit 3 performs feature pixel extraction processing on each of the first image region R1, the second image region R21, R22, and the third image region R3 of the reference phantom image d1. Similarly, as shown in Figs. 12A to 12C, the feature pixel extraction unit 3 performs feature pixel extraction processing on each of the first image region R1, the second image region R21, R22, and the third image region R3 of the inspection phantom image. As a result, feature pixels px are extracted from the first image region R1, the second image region R21, R22, and the third image region R3. In the embodiment, the third phantom p3 is a phantom with uniform shading, so the corresponding third image region R3 has uniform brightness. Therefore, in the embodiment, the feature pixel extraction unit 3 sets the edge portion of the third image region R3 as a feature pixel. The method of extracting feature pixels is not limited to the content described here. In other words, it is sufficient if the position of the reference pixel can be specified.

3-4 Deviation amount acquisition unit 4
The deviation amount acquiring unit 4 is configured to acquire the deviation amount between the reference phantom image and the inspection phantom image. In the embodiment, the deviation amount acquiring unit 4 associates the feature pixels of the reference phantom image with the feature pixels of the inspection phantom image, and acquires an affine matrix corresponding to the amount of movement and rotation of these feature pixels. Information corresponding to the deviation amount is added to the affine matrix.

In an embodiment, for each image region, an affine matrix is obtained as follows.
The deviation amount acquiring unit 4 acquires an affine matrix based on the feature pixel data d3 of the first image region R1 of the reference phantom image and the feature pixel data D3 of the first image region R1 of the inspection phantom image. The feature pixel data d3 includes information on the position of the feature pixel px of the first image region R1 of the reference phantom image as shown in FIG. 8B and feature amount information (e.g., intensity, etc.) of the feature pixel px. The same is true for the feature pixel data D3. As shown in FIG. 8B, in the first image region R1, not only the feature pixel px corresponding to the first phantom but also the feature pixel px corresponding to the third phantom (a plurality of feature pixels forming a circle in the third image region R3 of FIG. 8B) exist. The feature pixel px corresponding to the third phantom becomes a disturbance when acquiring the affine matrix related to the first image region R1, so the deviation amount acquiring unit 4 calculates the affine matrix excluding the feature pixel px corresponding to the third phantom.

The deviation amount acquisition unit 4 also acquires an affine matrix based on the feature pixel data d3 of the second image region R21 of the reference phantom image and the feature pixel data D3 of the second image region R21 of the inspection phantom image. Note that the feature pixel data d3 includes information on the position of the feature pixel px of the second image region R21 of the reference phantom image as shown in FIG. 8A, and feature amount information (e.g., intensity, etc.) of the feature pixel px. The same is true for the feature pixel data D3.

The deviation amount acquisition unit 4 also acquires an affine matrix based on the feature pixel data d3 of the second image region R22 of the reference phantom image and the feature pixel data D3 of the second image region R22 of the inspection phantom image. Note that the feature pixel data d3 includes information on the position of the feature pixel px of the second image region R22 of the reference phantom image as shown in FIG. 8C, and feature amount information of the feature pixel px (e.g., intensity, etc.). The same applies to the feature pixel data D3.

The deviation amount acquisition unit 4 does not acquire an affine matrix for the third image region R3. The deviation amount acquisition unit 4 acquires the movement amount (deviation amount) of the third image region R3 using the center coordinates of the third image region R3 acquired by the second determination unit 2B. Specifically, the deviation amount acquisition unit 4 acquires the movement amount (deviation amount) of the third image region R3 based on the center coordinates of the third image region R3 of the reference phantom image and the center coordinates of the third image region R3 of the inspection phantom image. That is, in the embodiment, since the third phantom p3 is circular, the deviation amount acquisition unit 4 does not need to acquire information on the amount of rotation when acquiring the deviation amount of the third phantom p3. In other words, the deviation amount acquisition unit 4 acquires the deviation amount of the third phantom p3 by taking into account only the parallel movement of the third image region R3.

3-5 Position Acquisition Unit 5
The position acquisition unit 5 acquires the position of the region of interest in the inspection phantom image based on the above-mentioned amount of deviation and the position of the region of interest in the reference phantom image set by the position setting unit 8, which will be described later.

In this embodiment, the position acquisition unit 5 acquires the position of the region of interest in the inspection phantom image by performing affine transformation on each image region as described below.
As shown in Figures 13A and 13B, the position acquisition unit 5 acquires the position of the region of interest Ra1 in the first image region R1 of the inspection phantom image by performing an affine transformation using the affine matrix acquired by the deviation acquisition unit 4 on the position of the region of interest Ra1 in the first image region R1 of the reference phantom image.
In addition, the position acquisition unit 5 acquires the position of the region of interest Ra21 in the second image region R21 of the inspection phantom image by performing an affine transformation using the affine matrix acquired by the deviation acquisition unit 4 on the position of the region of interest Ra21 in the second image region R21 of the reference phantom image.
In addition, the position acquisition unit 5 acquires the position of the region of interest Ra22 in the second image region R22 of the inspection phantom image by performing an affine transformation using the affine matrix acquired by the deviation acquisition unit 4 on the position of the region of interest Ra22 in the second image region R22 of the reference phantom image.
Furthermore, the position acquisition unit 5 acquires the position of the region of interest Ra3 in the third image region R3 of the inspection phantom image based on the amount of parallel movement of the third image region R3 acquired by the displacement amount acquisition unit 4.

3-2 Item value calculation unit 6
The radiation imaging device 100 is inspected by calculating values of predetermined items. The items can be arbitrarily determined according to the type of the radiation imaging device 100, etc. For example, items such as (1) the average value of the luminance of pixels in the regions of interest Ra1, Ra21, Ra22, and Ra3, (2) the standard deviation of the luminance of pixels in the region of interest Ra1, and (3) the difference between the average value of the luminance of pixels in the region of interest Ra1 and the average value of the luminance of pixels in the region of interest Ra3 can be adopted. The item value calculation unit 6 is configured to calculate these values (1)-(3).

3-2 Mode setting section 7
The mode setting unit 7 is configured to be able to set one of a reference phantom image setting mode and an inspection phantom image setting mode as a mode to be implemented by the information processing device 1. The mode setting unit 7 accepts the setting of the mode via, for example, an operation unit (for example, a mouse or a keyboard) attached to the information processing device 1.
The reference phantom image setting mode is a mode for acquiring the reference phantom image, the first image region R1, the second image region R21, R22, and the third image region R3 of the reference phantom image, the characteristic pixels px of these image regions, and the regions of interest Ra1, Ra21, Ra22, and Ra3 of these image regions.
The inspection phantom image setting mode is a mode for acquiring the inspection phantom image, the first image region R1, the second image region R21, R22, and the third image region R3 of the inspection phantom image, the feature pixels px of these image regions, and the regions of interest Ra1, Ra21, Ra22, and Ra3 of these image regions.

3-8 Position setting unit 8
In the reference phantom image setting mode, the administrator needs to manually set the regions of interest Ra1, Ra21, Ra22, and Ra3 on the reference phantom image via an operation unit attached to the information processing device 1. The position setting unit 8 accepts the setting of the positions of the regions of interest Ra1, Ra21, Ra22, and Ra3 via this operation unit.
In the inspection phantom image setting mode, the information processing device 1 automatically acquires the regions of interest Ra1, Ra21, Ra22, and Ra3 of the inspection phantom image. Therefore, the administrator does not need to set the positions of the regions of interest Ra1, Ra21, Ra22, and Ra3 of the inspection phantom image via the operation unit. However, the position setting unit 8 is configured to be able to fine-tune the positions of the regions of interest Ra1, Ra21, Ra22, and Ra3 of the acquired inspection phantom image. This allows the inspection to be carried out with the utmost care.

3-9 Memory Unit 9
The storage unit 9 stores various data and programs, such as data related to the reference phantom image and the inspection phantom image (data of each image region, data of feature pixels, and data of regions of interest).

4. Operation Description The operation flow of the radiation imaging apparatus 100 shown in FIG. 14 will be described together with the various data flows shown in FIG. 2 and FIG.

3-1 Step S1: Phantom imaging The administrator places the phantom p in the placement section 22 of the imaging device 20. Then, the administrator operates the radiation generating unit 21 under predetermined conditions to image the phantom p with the imaging device 20. Note that the operating conditions of the imaging device 20 when acquiring a reference phantom image are the same as the operating conditions of the imaging device 20 when acquiring an inspection phantom image.

3-2 Step S2: Pre-processing The image processing unit 24 performs pre-processing. That is, the image processing unit 24 performs size matching and window level adjustment on the acquired phantom image. The contents of the pre-processing of the reference phantom image and the contents of the pre-processing of the inspection phantom image are the same.

3-3 Step S3: Determination of Each Image Region The region determination unit 2 acquires a phantom image from the image processing unit 24.
When the setting mode is set to the reference phantom image setting mode, the phantom image is the reference phantom image d1, and when the setting mode is set to the inspection phantom image setting mode, the phantom image is the inspection phantom image D1.

Then, the region division section 2A of the region determination section 2 determines the first image region R1 and the second image regions R21, R22. In determining these regions, the above-mentioned binarization process, labeling process, and processes corresponding to the first and second functions are performed. In addition, the second determination section 2B of the region determination section 2 determines the third image region R3 by executing a process to detect the shape of the above-mentioned third phantom p3.

3-4 Step S4: Extraction of Feature Pixels The feature pixel extractor 3 obtains image region data from the region determiner 2.
Specifically, in the reference phantom image setting mode shown in Fig. 2, the feature pixel extraction unit 3 acquires data (first and second image region data d21) including the positions of the first image region R1 and the second image regions R21, R22 of the reference phantom image d1 from the region division unit 2A. In addition, the feature pixel extraction unit 3 acquires data (third image region data d22) including the position of the third image region R3 of the reference phantom image d1 from the second determination unit 2B. Then, the feature pixel extraction unit 3 extracts feature pixels px (corresponding to the feature pixel data d3 shown in Fig. 2) of each image region (the first image region R1, the second image regions R21, R22, and the third image region R3) of the reference phantom image d1.

In the inspection phantom image setting mode shown in FIG. 3, the feature pixel extraction unit 3 acquires data (first and second image region data D21) including the positions of the first image region R1 and the second image regions R21, R22 of the inspection phantom image D1 from the region division unit 2A. In addition, the feature pixel extraction unit 3 acquires data (third image region data D22) including the position of the third image region R3 of the inspection phantom image D1 from the second determination unit 2B. Then, the feature pixel extraction unit 3 extracts feature pixels px (corresponding to the feature pixel data D3 shown in FIG. 3) from each image region (first image region R1, second image regions R21, R22, and third image region R3) of the inspection phantom image D1.

3-5 Step S5: Determination of Setting Mode The mode setting unit 7 determines whether the set mode is the reference phantom image setting mode or the inspection phantom image setting mode. If it is the reference phantom image setting mode, the process proceeds to step S6. If it is the inspection phantom image setting mode, the process proceeds to step S7.

3-6 Step S6: Setting the region of interest in the reference phantom image As shown in FIG. 13A, the administrator sets the positions of the regions of interest Ra1, Ra21, Ra22, and Ra3 (corresponding to the region of interest data dR shown in FIG. 3) for each image region via the operation unit of the information processing device 1.

3-7 Step S7: Obtaining the amount of deviation (obtaining the affine matrix and the amount of translation)
The deviation amount acquisition unit 4 acquires each affine matrix (corresponding to the deviation amount data D4 shown in Figure 3) that takes into account information corresponding to the deviation amount for the first image region R1 and the second image regions R21 and R22 based on the positions of the characteristic pixels of each image region in the reference phantom image and the positions of the characteristic pixels of each image region in the inspection phantom image.
In addition, the shift amount acquisition unit 4 acquires the amount of parallel movement of the third image region R3 (corresponding to the shift amount data D4 shown in Figure 3) based on the center coordinates of the third image region R3 in the reference phantom image and the center coordinates of the third image region R3 in the inspection phantom image.

3-8 Step S8: Obtaining the positions of the regions of interest in the inspection phantom image The position acquisition unit 5 acquires the positions of the regions of interest Ra1, Ra21, Ra22, Ra3 in the inspection phantom image D1 (region of interest data D5 shown in Figure 3) based on the positions (region of interest data dR) of the regions of interest Ra1, Ra21, Ra22, Ra3 in the reference phantom image d1, the affine matrix (shift amount data D4) and the parallel translation amount (shift amount data D4).

3-9 Step S9: Fine adjustment by administrator The administrator fine-adjusts the positions of the regions of interest Ra1, Ra21, Ra22, and Ra3 of the acquired inspection phantom image as necessary. In other words, when the position setting unit 8 receives an instruction for fine adjustment via the operation unit, it changes the positions of the regions of interest Ra1, Ra21, Ra22, and Ra3 of the inspection phantom image.

3-10 Step S10: Calculate, display, and record item values The item value calculation unit 6 calculates predetermined item values for the regions of interest Ra1, Ra21, Ra22, and Ra3. The calculation results are displayed on the image display device 30 and stored in the storage unit 9.

4. Modification 4-1 Determination of Image Area 1
In the embodiment, the first determination unit 2A3 of the area division unit 2A of the area determination unit 2 is described as determining the first image area R1 and the second image areas R21, R22 by performing processing such as determining the largest area among the areas labeled by the labeling unit 2A2 (area g1, area g21, area g22 and area g23), but is not limited to this.
The first determination unit 2A3 may create a histogram for the binarized image (see FIG. 6 and FIG. 10) and determine the first image region R1 and the second image regions R21 and R22 using the histogram. As shown in FIG. 15, the histogram is represented by two axes. Note that FIG. 15 shows an example in which a histogram is created for the image shown in FIG. 6. The vertical axis of FIG. 15 corresponds to the vertical direction of the binarized image shown in FIG. 6, and the horizontal axis of FIG. 15 corresponds to the total number (frequency) of pixels whose luminance is the second value among the pixels of the binarized image shown in FIG. 6. Then, scanning may be started from the center of the vertical axis (scanning start position), and the position where the frequency of the histogram falls below the threshold for the first time may be determined as the first image region R1. In FIG. 15, the positions where the frequency of the histogram falls below the threshold for the first time correspond to the upper limit position and the lower limit position. Then, as described in the embodiment, once the first image region R1 is determined, the remaining second image regions R21 and R22 can also be determined.

4-2 Image area determination 2
The region determining unit 2 may determine the image region by machine learning. In other words, the region determining unit 2 may be configured to determine the image region based on a learning model that inputs the reference phantom image d1 or the inspection phantom image D1 and determines the image region (the first image region R1, the second image regions R21 and R22, and the third image region R3). Note that the input image may be the phantom image itself, or may be a binarized image.

4-3 Regarding the amount of deviation In the embodiment, the amount of deviation is obtained by extracting the characteristic pixels of each image region, but the present invention is not limited to this. For example, the region determination unit 2 may obtain the amount of deviation by using template matching. That is, the region determination unit 2 may calculate the sum of absolute pixel differences (SAD) between each image region of the reference phantom image and each image region of the inspection phantom image while shifting each image region of the reference phantom image up, down, left, and right. Then, the region determination unit 2 may use the amount of movement of each image region of the reference phantom image that minimizes the SAD as the amount of deviation.

4-4 Phantom The radiation imaging apparatus 100 can also be applied to an ACR (American College of Radiology) phantom. That is, the phantom p may include an ACR phantom.

4-5 Feature Pixel Extraction Method In the embodiment, the feature pixel extraction unit 3 extracts, as feature pixels, pixels in the phantom image in which the difference in luminance between adjacent pixels exceeds a predetermined threshold, but this is not limited to this. The feature pixel extraction unit 3 may also use a known feature detector such as SIFT or KAZE to extract feature pixels.

1: Information processing device 2: Area determination section 2A: Area division section 2A1: Binarization processing section 2A2: Labeling section 2A3: First determination section 2B: Second determination section 3: Feature pixel extraction section 4: Displacement amount acquisition section 5: Position acquisition section 6: Item value calculation section 7: Mode setting section 8: Position setting section 9: Memory section 20: Imaging device 21: Radiation generation section 22: Arrangement section 23: Detection section 24: Image processing section 30: Image display device 100: Radiation imaging device D1: Inspection phantom image D21: Second image region data D22: Third image region data D3: Feature pixel D4: Displacement amount data D5: Region of interest data d1: Reference phantom image d21: Second image region data d22: Third image region data d3: Feature pixel dR: Region of interest data L1: Virtual line L2: Virtual line R1 : First image region R21 : Second image region R22 : Second image region R3 : Third image region Ra1 : Region of interest Ra21 : Region of interest Ra22 : Region of interest Ra3 : Region of interest g1 : Region g21 : Region g22 : Region g3 : Region p : Phantom p1 : First phantom p21 : Second phantom p22 : Second phantom p3 : Third phantom px : Characteristic pixel

Claims (11)

A displacement amount acquisition unit and a position acquisition unit are provided,
the deviation amount acquisition unit is configured to acquire a deviation amount between a reference phantom image and an inspection phantom image,
The reference phantom image is acquired by imaging a plurality of separate phantoms (excluding phantoms having markers for position detection formed by combining materials having different radiation transmittances with respect to a base material),
The inspection phantom image is acquired by imaging the same phantom as the reference phantom image,
A region of interest is preset in the reference phantom image,
The information processing device, wherein the position acquisition unit is configured to acquire a position of a region of interest in the inspection phantom image based on the amount of deviation and a position of the region of interest in the reference phantom image. 2. The information processing device according to claim 1,
the plurality of phantoms includes a first and a second phantom;
the first and second phantoms are arranged side by side on a plane,
Each of the reference phantom image and the inspection phantom image has a plurality of image regions;
the plurality of image regions includes first and second image regions;
the first and second image regions are acquired by imaging a first and a second phantom, respectively;
An information processing apparatus, wherein the region of interest is preset in each of the first and second image regions of the reference phantom image. 3. The information processing device according to claim 2,
The plurality of phantoms further includes a third phantom,
The third phantom is arranged so as to overlap the first phantom,
the plurality of image regions includes a third image region;
The third image region is acquired by imaging the third phantom;
The information processing device, wherein the region of interest is preset in a third image region of the reference phantom image. 4. The information processing device according to claim 2,
The image processing apparatus further includes a feature pixel extraction unit,
Each of the first and second image regions of the reference phantom image and each of the first and second image regions of the inspection phantom image have feature pixels;
The feature pixel extraction unit is configured to extract the feature pixels of the first and second image regions of the reference phantom image and the feature pixels of the first and second image regions of the inspection phantom image, and the feature pixel extraction unit extracts the feature pixels based on brightness values of pixels of the first and second image regions,
The information processing device, wherein the amount of deviation acquired by the deviation acquisition unit is the amount of deviation between the positions of the feature pixels of the first and second image regions in the reference phantom image and the positions of the feature pixels of the first and second image regions in the inspection phantom image. An information processing device according to any one of claims 1 to 4,
Further comprising a position setting unit,
the position setting unit is configured to be able to set the region of interest of the reference phantom image,
The position acquisition unit acquires the position of the region of interest in the inspection phantom image based on the amount of deviation and the position of the region of interest in the reference phantom image set by the position setting unit. An information processing device according to any one of claims 1 to 5,
The information processing apparatus, wherein the plurality of phantoms include an ACR (American College of Radiology) phantom. 4. The information processing device according to claim 2,
A region determining unit is further provided,
The region determination unit binarizes each pixel in the reference phantom image and the inspection phantom image into a first value and a second value, and the region determination unit determines the first and second image regions based on the number of pixels in a second value pixel region;
An information processing device, wherein the second value pixel region is a pixel having a second value or a collection of pixels having the second value. 4. The information processing device according to claim 2,
A region determining unit is further provided,
The region determination unit determines the plurality of image regions based on a learning model that determines the plurality of image regions by inputting the reference phantom image or the inspection phantom image. An information processing apparatus according to claim 7 dependent on claim 3,
The information processing device, wherein the region determining unit is configured to determine a third image region by executing a process of detecting a shape of a third phantom. The method includes a deviation amount acquisition step and a position acquisition step,
In the shift amount acquisition step, a shift amount between the reference phantom image and the inspection phantom image is acquired,
The reference phantom image is acquired by imaging a plurality of separate phantoms (excluding phantoms having markers for position detection formed by combining materials having different radiation transmittances with respect to a base material),
The inspection phantom image is acquired by imaging the same phantom as the reference phantom image,
A region of interest is preset in the reference phantom image,
An information processing method, in which the position acquisition step acquires a position of the region of interest in the inspection phantom image based on the amount of deviation and the position of the region of interest in the reference phantom image. A computer program for causing a computer to execute an information processing method,
The information processing method includes a displacement amount acquiring step and a position acquiring step,
In the shift amount acquisition step, a shift amount between the reference phantom image and the inspection phantom image is acquired,
The reference phantom image is acquired by imaging a plurality of separate phantoms (excluding phantoms having markers for position detection formed by combining materials having different radiation transmittances with respect to a base material),
The inspection phantom image is acquired by imaging the same phantom as the reference phantom image,
A region of interest is preset in the reference phantom image,
A computer program, in the position acquisition step, for acquiring a position of the region of interest in the inspection phantom image based on the amount of deviation and the position of the region of interest in the reference phantom image.