Information processing device, radiography system, method of operating the information processing device, and program

Abstract

To reduce the burden on the operator during bone density analysis. [Solution] An information processing device for calculating bone density using radiographic images generated based on radiation comprises: a patient information acquisition unit that acquires patient information input via an operation unit; an image information acquisition unit that acquires radiographic images taken in correspondence with the patient information; a region extraction unit that extracts a bone density analysis region from the acquired radiographic image; a storage unit that stores information about bones for which bone density has been previously calculated and patient information about the patient having said bones; and an analysis target determination unit that, when the acquired patient information corresponds to the stored patient information, determines a target region for calculating bone density in the extracted bone density analysis region based on the bone density analysis region linked to the stored patient information and the information about bones for which bone density has been calculated.

Description

Technical Field

[0001] The disclosure of this specification relates to an information processing apparatus, a radiation imaging system, an operation method of the information processing apparatus, and a program.

Background Art

[0002] For the diagnosis of osteoporosis and the determination of the efficacy of therapeutic drugs, bone density, which is the bone mass per 1 cm

[0005] , is measured. As a technique for measuring bone density, a method using an image obtained by the DXA (Dual energy X-ray Absorptiometry) method is known. In the DXA method, X-ray imaging is performed at two types of tube voltages, and based on the difference in the absorption rate of the X-ray transmission site, the bone region (bone region) and the soft tissue region (soft part region) in the image are discriminated. An image obtained using the DXA method is called an enesub (energy subtraction) image. Also, each image taken at two types of tube voltages is a single X-ray image.

[0003] For bone density measurement, it is necessary to further extract a target region for bone density measurement (hereinafter referred to as a bone density analysis region) from the bone region of the enesub image. For example, when measuring the bone density of the lumbar spine, generally, the vertebral bodies from the first lumbar vertebra (L1) to the fourth lumbar vertebra (L4), or the vertebral bodies from the second lumbar vertebra (L2) to the fourth lumbar vertebra (L4), become the bone density analysis regions. Also, when measuring the bone density of the femur, up to the proximal part of the femur such as the neck region becomes the bone density analysis region.

[0004] In an apparatus for measuring bone density, generally, an operator designates a bone density analysis region while viewing an image displayed on a screen. In the case of a lumbar spine examination, the bone region (overall contour) of the lumbar spine in the displayed image and the boundary lines of the vertebral bodies within the bone region are designated. In order to accurately measure bone density, when a vertebral body in which a metal member appears nearby or a fractured vertebral body is included, it is necessary to exclude that vertebral body from the bone density analysis region. Also, in order to perform follow-up observation of the bone density measurement results, the bone density analysis region is required to be equivalent to the region that was the target of the analysis region in the past.

[0005] The technology disclosed in Patent Document 1 displays past images including the bone density analysis region, allowing for easy visual comparison with the currently performed examination. This enables the operator to exclude unnecessary regions as needed, making it possible to perform measurements in the desired bone density analysis region. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Japanese Patent Publication No. 2023-121552 [Patent Document 2] Japanese Patent Publication No. 2024-82942 [Overview of the project] [Problems that the invention aims to solve]

[0007] The technology disclosed in Patent Document 1 mentioned above allows for easy visual comparison of which vertebrae have been excluded in the past. However, the task of excluding vertebrae is left to the operator each time an examination is performed. Therefore, one of the purposes of disclosure in this specification is to reduce the burden on the operator during bone density analysis. [Means for solving the problem]

[0008] In view of the above purpose, an information processing device relating to one aspect of this disclosure is: An information processing device that calculates bone density using radiographic images generated based on radiation, A patient information acquisition unit that acquires patient information entered via the control unit, An image information acquisition unit that acquires the radiographic image taken in accordance with the patient information, A region extraction unit for extracting bone density analysis regions from the acquired radiographic image, A storage unit that stores information about bones for which bone density has been previously calculated, and patient information about patients having such bones, The system includes an analysis target determination unit that, when the acquired patient information corresponds to the stored patient information, determines the target area for calculating bone density in the extracted bone density analysis area based on the bone density analysis area linked to the stored patient information and information regarding the bone area for which the bone density was calculated. [Effects of the Invention]

[0009] According to one aspect of this disclosure, the burden on the operator during bone density analysis can be reduced. [Brief explanation of the drawing]

[0010] [Figure 1] An example of the schematic configuration of the radiography system related to this disclosure is shown. [Figure 2] An example of the general configuration of the bone density measurement unit related to this disclosure is shown. [Figure 3] An example of the bone density calculation flow for Example 1 is shown. [Figure 4] A diagram illustrating the details of the bone density calculation method according to Example 1 is shown. [Figure 5] An example of the results of region extraction according to Example 1 is shown. [Figure 6] An example of the target of correction processing for the intervertebral region according to Example 1 is shown. [Figure 7] Another example of the target of the correction processing for the intervertebral region according to Example 1 is shown. [Figure 8] An example of a rotation correction method according to Example 1 is shown. [Figure 9] An example of a method for generating a straight line according to Example 1 is shown. [Figure 10] An example of how the bone density calculation results for Example 1 are displayed is shown. [Figure 11] An example of the stored information related to Example 1 is shown. [Figure 12] An example of the bone density calculation flow for Example 2 is shown. [Figure 13] A diagram illustrating the details of the bone density calculation method according to Example 2 is shown. [Figure 14]A diagram for explaining the details of the bone density calculation methods of Examples 3 and 4. [Figure 15] An example of the bone density calculation flow according to Examples 3 and 4 is shown. [Figure 16] An example of the update flow of the bone density analysis region according to Example 3 is shown. [Figure 17] A diagram for explaining the details of the update method of the bone density analysis region according to Example 3. [Figure 18] An example of the update flow of the bone density analysis region according to Example 4 is shown. [Figure 19] A diagram for explaining the details of the update method of the bone density analysis region according to Example 4.

Modes for Carrying Out the Invention

[0011] Hereinafter, exemplary embodiments for implementing the present disclosure will be described in detail with reference to the drawings. However, the dimensions, materials, shapes, relative positions of the components, etc. described in the following examples are arbitrary and can be changed according to the configuration of the apparatus to which the present disclosure is applied or various conditions. Also, in the drawings, the same reference numerals are used between the drawings to indicate the same or functionally similar elements. In addition, in each drawing, some components, members, and parts of the processing that are not important for explanation may be omitted from the display.

[0012] Note that hereinafter, as an example of radiation, a radiation imaging system using X-rays will be described. However, the radiation may be X-rays or other radiation. In the following examples, the term "radiation" can include, for example, electromagnetic radiation such as X-rays and γ-rays, and particle radiation such as α-rays, β-rays, particle beams, proton beams, heavy ion beams, and neutron beams.

[0013] Furthermore, in the following, "machine learning model" refers to a learning model based on a machine learning algorithm. Specific machine learning algorithms include nearest neighbors, naive Bayes, decision trees, and support vector machines. Deep learning, which uses neural networks to generate features and connection weights for learning, is also an example. As an algorithm using decision trees, methods employing gradient boosting, such as LightGBM and XGBoost, are also examples. Appropriately, any of the above algorithms that are available can be applied to the following examples and modifications. Also, "training data" refers to learning data, which consists of pairs of input and output data.

[0014] A pre-trained model refers to a machine learning model that follows any machine learning algorithm, such as deep learning, and has been trained (learned) in advance using appropriate training data. However, while a pre-trained model is obtained using appropriate training data in advance, it is not the case that it cannot be further trained; additional training can be performed. Additional training can be performed even after the device has been installed at the user's site.

[0015] (Example 1) As mentioned above, in order to accurately measure bone density, if metal components are visible nearby or if a fractured vertebra is included, it is necessary to exclude those vertebrae from the bone density analysis area. Furthermore, when monitoring bone density over time, if vertebral exclusion has been performed in the past, the same procedure must be repeated each time an examination is conducted. In Example 1, the identification and exclusion of vertebrae from the examination are made automatic, thereby reducing this workload for the operator and alleviating their burden.

[0016] The following describes the radiography system, information processing device, and operating method of the information processing device according to Embodiment 1 of this disclosure with reference to Figures 1 to 11. Figure 1 is a schematic diagram showing an example configuration of the radiography system 1 according to Embodiment 1 of this disclosure. In the following embodiments, the case in which X-rays are used as radiation will be described as an example, as described above. The radiography system 1 according to this embodiment comprises a radiography device 100 and a bone density measurement unit 200 which functions as an information processing device in this embodiment.

[0017] The radiography apparatus 100 includes a control unit 101, a radiation generator control unit 102, an X-ray tube 103, an FPD (Flat Panel Detector) 106, a display unit 107, and an operation unit 108. The radiation generator control unit 102 generates X-rays by applying a high-voltage pulse to the X-ray tube 103 when an exposure switch (not shown) is pressed, and the X-ray tube 103 irradiates with X-rays. When X-rays are irradiated from the X-ray tube 103, the X-rays pass through the subject (not shown) and grid 105 on the stand 104 and reach the FPD 106. The arriving X-rays are converted into visible light by the FPD 106 and output as an X-ray image to the control unit 101. The output X-ray image can be transferred via the control unit 101 to the bone density measurement unit 200, which will be described later.

[0018] The FPD 106 has an X-ray detection unit (not shown) equipped with a pixel array for generating signals corresponding to X-rays. The X-ray detection unit detects X-rays that have passed through the grid 105 as an image signal. The X-ray detection unit has pixels arranged in an array (two-dimensional region) that output signals corresponding to the incident light. The photoelectric conversion element of each pixel converts the X-rays, which have been converted into visible light by a phosphor, into an electrical signal and outputs it as an image signal. In this way, the X-ray detection unit is configured to detect X-rays that have passed through the grid 105 and acquire an image signal (X-ray image). The FPD 106 outputs the read-out image signal (X-ray image) to the control unit 101 according to instructions from the control unit 101. The X-ray detection unit may be an indirect conversion type detection unit that first converts X-rays into visible light using a scintillator or the like and then converts the visible light into an electrical signal using a photoelectric conversion element, or it may be a direct conversion type detection unit that directly converts the incident X-rays into an electrical signal.

[0019] The display unit 107 (monitor) displays X-ray images (digital images), patient information, or user notifications received by the control unit 101 from the FPD 106. The operation unit 108 can control the display unit 107 through the control unit 101. The operation unit 108 can also be used to input examination information associated with the X-ray image, such as the date and time of examination, reception number, name of the referring physician, examination description, or name of the radiologist, via the user interface. Alternatively, the operation unit 108 can be used to input patient information associated with the X-ray image, such as patient ID, patient name, gender, date of birth, height, weight, and pregnancy status, or to input instructions to the FPD 106. The operation unit 108 may be configured using input devices such as a mouse or keyboard. An operation bar on the operation screen is also an example of the operation unit 108, and can be used to adjust the size of the processing, etc. The display unit 107 may also be configured as a touch panel display, in which case the display unit 107 can also be used as the operation unit 108.

[0020] The control unit 101 is connected to the radiation generator control unit 102, the FPD 106, the display unit 107, and the operation unit 108. As described above, the control unit 101 controls the irradiation of X-rays to the subject by controlling the radiation generator control unit 102 and can acquire the image signal of the subject by controlling the FPD 106. The control unit 101 also functions as a display control unit, displaying the X-ray image of the subject obtained from the image signal on the display unit 107, and can display information input from the operation unit 108 and images obtained from the bone density measurement unit 200, which will be described later. The control unit 101 also transfers the X-ray image acquired from the FPD 106 and information related to the X-ray image, such as patient information input by the operation unit 108, to the bone density measurement unit 200.

[0021] In this embodiment, the radiography apparatus 100 can generate energy sub-images for measuring bone density by performing X-ray imaging with two different tube voltages using a kV switching method. Alternatively, the energy sub-images may be generated from images of each layer obtained in a single exposure using a stacked sensor formed of two materials (phosphor and scintillator) with different X-ray absorption rates as the FPD 106. For example, these two materials may mainly contain CsI(Tl) (thallium-activated cesium iodide) and GOS (gadolinium sulfide). This stacked sensor may be configured such that the material on the side where the X-rays are incident (the subject side) contains CsI(Tl), and the material on the opposite side contains GOS.

[0022] Next, the bone density measurement unit 200, which functions as an information processing device in this embodiment, will be described with reference to Figure 2. Figure 2 is a schematic diagram showing an example of the configuration of the bone density measurement unit 200 according to this embodiment. The bone density measurement unit 200 includes an analysis area extraction unit 201, an analysis target determination unit 202, a storage unit 203, a bone density calculation unit 204, a bone density measurement display unit 205, an image information acquisition unit 206, and a patient information acquisition unit 207. The analysis area extraction unit 201 uses image information such as X-ray images acquired by the image information acquisition unit 206 from the control unit 101 to extract analysis areas in an X-ray image that can be used as analysis targets. Although the images to be acquired are assumed to be obtained from a single radiography device 100, the bone density measurement unit 200 may be connected to multiple radiography devices via a server or the like, and images may be obtained from each of them. Alternatively, images stored in an external storage device may be acquired. In this case, the patient information acquisition unit 207 can acquire patient information corresponding to images acquired from an external storage device, for example, via an input unit (not shown) provided in the bone density measurement unit 200. The analysis target determination unit 202 determines the area in which bone density will actually be measured within the area extracted as the analysis target by the analysis area extraction unit 201.

[0023] The bone density calculation unit 204 calculates the bone density for the region determined as the bone density measurement area (measurement target) by the analysis target determination unit 202. The storage unit 203 stores the image information, analysis target area, bone density calculation area, bone density, patient information of the analysis target, etc., in association with each other. The patient information acquisition unit 207 acquires information about the patient from whom the X-ray image is acquired, for example via the operation unit 108. The bone density measurement display unit 205 can be configured as a monitor provided on the bone density measurement unit 200, for example, and can display the extracted analysis area, the region determined as the measurement target, the calculated bone density, etc. Alternatively, the bone density measurement display unit 205 can be provided separately from the monitor and function as a display control unit to display the aforementioned regions, etc., on the monitor. Note that the bone density measurement display unit 205 does not have to be provided on the bone density measurement unit 200; instead, it may transmit the display targets described here to the control unit 101 and have the display unit 107 perform the function.

[0024] The specific operation of the bone density measurement unit 200 will be described below. The following explanation will use the measurement of bone density in the lumbar spine as an example. First, the analysis region extraction unit 201 takes the frontal X-ray image of the lumbar spine taken by the radiography device 100 as input and extracts the region from the first lumbar vertebra (L1) to the fourth lumbar vertebra (L4) (see Figure 4(a)). The analysis target determination unit 202 determines which vertebrae from the first lumbar vertebra (L1) to the fourth lumbar vertebra (L4) will be subject to bone density analysis. The vertebrae determined to be subject to analysis may be all vertebrae excluding those previously excluded from the bone density analysis region stored in the memory unit 203, or they may be based on manual correction by the operator via the operation unit 108, etc.

[0025] The memory unit 203 stores patient information such as the patient's name and patient ID entered by the operation unit 108, the vertebral body numbers (any of L1 to L4) that were not designated as bone density analysis areas by the analysis target determination unit 202, and images processed by the analysis area extraction unit 201. The memory unit 203 may be composed of, for example, ROM (Read Only Memory), RAM (Random Access Memory), etc. The memory unit 203 may also be an external storage device located outside the bone density measurement unit 200. In that case, the memory unit 203 may be composed of an external hard disk drive (HDD) or network-attached storage (NAS), etc. The bone density calculation unit 204 calculates bone density for the vertebral bodies designated as bone density analysis areas by the analysis target determination unit 202. The calculated bone density is displayed by the bone density measurement display unit 205.

[0026] Figure 4(a) shows an example of an anterior X-ray image of the lumbar spine acquired by the radiography device 100. In the same figure, for illustrative purposes, the lumbar vertebrae L1 to L4, which are an example of the bone density measurement target, are shown. In this embodiment, the analysis region extraction unit 201 takes the image in Figure 4(a) as an input image and extracts the bone density analysis region.

[0027] The input image may be an energy sub-image acquired by the DXA method. This energy sub-image may be generated from images obtained by X-ray imaging with two different tube voltages using a kV switching method. Alternatively, this energy sub-image may be generated from images of each layer obtained in a single exposure using a stacked sensor formed of two different materials (phosphors) with different X-ray absorption rates as the FPD106. Alternatively, a simple X-ray image acquired in a single exposure may be used as the input image. Furthermore, if the FPD106 is a stacked sensor, the image of the upper layer (subject side) may be used as the input image.

[0028] Next, with reference to Figure 3, the process of acquiring an X-ray image and calculating bone density in the X-ray image will be described. Figure 3 shows the processing flow performed by the radiography apparatus 100 and the bone density measurement unit 200 to calculate bone density in this embodiment. For example, when an instruction to take an X-ray image of a patient on the pedestal 104 and to perform an associated bone density examination is input to the control unit 101 via the operation unit 108, the process illustrated in Figure 3 is started, and the flow moves to step S100.

[0029] In step S100, the operator uses the control unit 108 to input examination information, including, for example, imaging conditions and patient information. Patient information may include, for example, patient ID, name, date of birth, height, weight, or a combination of these. Once the examination information is entered, the control unit 101 moves the flow back to step S101.

[0030] Next, in step S101, the radiography device 100 generates and acquires an X-ray image and transmits it to the bone density measurement unit 200. This operation may be performed automatically by, for example, the operator selecting a program to perform bone density measurement via the control unit 108 and issuing an instruction, or it may be performed by direct instruction from the operator.

[0031] Furthermore, if bone density measurement is instructed, the bone density measurement unit 200 may automatically execute steps S100 and S101. In this case, patient information entered via the operation unit 108 is acquired by the patient information acquisition unit 207. The image information acquisition unit 206 acquires image information acquired by the radiography device 100 as input images. After acquiring the X-ray images, the flow in the bone density measurement unit 200 proceeds to step S102.

[0032] Next, in step S102, the analysis region extraction unit 201 uses a trained model to extract (segment) the spinal region and the intervertebral region from the input image. Here, the spinal region refers to the area including the outline of the spine and its interior. Alternatively, only the outline (outer frame) of the spine may be treated as the spinal region. Similarly, the intervertebral region refers to the area including the outline of the intervertebral space and its interior. Alternatively, only the outline (outer frame) of the intervertebral space may be treated as the intervertebral region. Furthermore, the intervertebral region may also be a line. In this way, the analysis region extraction unit 201 functions as an example of a discriminator that extracts desired regions from the input image using a trained model.

[0033] Here, an example of the extracted results is shown in Figure 5. The white area in Figure 5(a) is the extracted result of the vertebral region, and the white area in Figure 5(b) is the extracted result of the intervertebral region. Figure 5(a) shows an example in which the transverse processes are not included in the vertebral region, but the transverse processes may be included.

[0034] The trained model may be stored in the memory unit 203 and recalled for use. Separate trained models may be provided for extracting the spinal region and for extracting the intervertebral region. These trained models may have the same structure, differing only in their weights. Alternatively, the trained model for extracting the spinal region and the trained model for extracting the intervertebral region may be a common model, with each channel representing the extraction results for the spinal region and the intervertebral region, respectively. The trained model in this case can be generated, for example, by training it with a combination of images near the lumbar spine and images labeled with the spinal and intervertebral regions within those images as training data.

[0035] Furthermore, preprocessing may be performed on the input image before the segmentation process. One example of preprocessing is a process to improve contrast. For example, a process to improve contrast could be linearly stretching the maximum brightness value in the image to 1 and the minimum brightness value to 0. Other methods include histogram equalization. Another example of preprocessing is a process to enhance edges. For example, a filtering process could be used to enhance edges. Another example of preprocessing is a standardization process that sets the average brightness value of the image to 0 and the variance to 1. Furthermore, a combination of the above preprocessing methods may be used.

[0036] The input image may be an image obtained by the radiography device 100 and then cropped by the operator using the operation unit 108 while viewing the image displayed on the display unit 107. The cropped image may also be resized to a specified image size. The specified image size may be, for example, the input image size of a trained model. The input image may also be an image that has been automatically cropped by the analysis region extraction unit 201 based on the imaging range, so as to leave the vicinity of the lumbar vertebrae. Furthermore, if there is a region extracted as an intervertebral region outside the extracted spinal region, this region may be considered a segmentation error and excluded. Once the analysis region extraction unit 201 has extracted the spinal region and intervertebral region, the flow proceeds to step S103.

[0037] Next, in step S103, the analysis region extraction unit 201 further extracts bone density analysis regions from the extraction results obtained in step S102 (Figure 5). Here, we will explain the case where the region of lumbar vertebrae L2 to L4, as exemplified in Figure 4(a), is extracted as the bone density analysis region. Note that the combination of bone density analysis regions is not limited to this, and for example, the region of the first lumbar vertebra L1 may also be extracted. In addition, in the case of a subject with more vertebral bodies than usual, additional regions such as L5 or L6 may be extracted.

[0038] As an example of a specific extraction method, first, the centroid position of each intervertebral region extracted in step S102 is calculated. Next, the vertical position (centroid Y-coordinate) of these centroid positions is calculated. Figure 6(a) shows an example of the intervertebral region extraction result, with the Y-coordinate of the centroid position indicated by a dotted line. Note that during the extraction of intervertebral regions, due to outliers and inference errors, intervertebral regions may be displayed as existing in locations where they do not actually exist, or conversely, intervertebral regions may be displayed as not existing in locations where they should exist. In such cases, based on the interval in the Y-coordinate of these centroid positions, the analysis region extraction unit 201 may perform the correction process described below as necessary to exclude inappropriate intervertebral regions or add appropriate intervertebral regions. Such correction processes will be described below.

[0039] As illustrated by H1 and H2 in Figure 6(a), if the distance in the Y coordinates between the centroid positions of the intervertebral regions is smaller than a set threshold, one of the corresponding intervertebral regions may be considered an outlier (or inference error) and excluded from the intervertebral region. For example, if the median of the distance in the Y coordinates between the centroid positions of each intervertebral region is ΔYm, the threshold set here may be, for example, ΔYm / 4. However, the method of setting the threshold is not limited to this and can be set by any method.

[0040] One example of exclusion methods here is to keep the larger intervertebral region (H1 side in Figure 6(a)) and exclude the smaller intervertebral region (H2 side in Figure 6(a)). Alternatively, you can keep the lower intervertebral region in the image and exclude the upper one. Conversely, you can keep the upper intervertebral region and exclude the lower one.

[0041] Furthermore, as illustrated in Figure 6(a) with H3 and H4, if the distance in the Y coordinate between the centroid positions of the intervertebral regions is greater than the set threshold, it may be considered an outlier (or inference error), and a new intervertebral region may be added at the midpoint of those centroid positions. For example, if the median of the distance in the Y coordinate between the centroid positions of each intervertebral region is ΔYm, the threshold set here may be, for example, ΔYm × 2. However, the method of setting the threshold is not limited to this and can be set by any method.

[0042] The intervertebral region added at this time may be a line segment with a width of 10 mm and a height of 1 mm, with the midpoint of the centroids of the H3 and H4 intervertebral regions as its centroid, as shown, for example, at the Y coordinate position of H3.5 in Figure 6(b). However, the size and shape of the intervertebral region here are not limited to this and can be set to any size or shape. Furthermore, the newly added intervertebral region may be a point at the above centroid position.

[0043] Another example of a correction process is explained using Figure 7. In this correction process, first, the vertical distance in the Y coordinate is calculated from the upper end of the extracted spinal region (H1 in Figure 7(a)) to the centroid position of the first intervertebral region counted medially from the upper end of the spine (H2 in Figure 7(b)). If this distance is greater than a threshold, an intervertebral region may be added at the height of the midpoint between the edge of the spinal region and the centroid position of the first intervertebral region counted medially (H1.5 in Figure 7(b)). The threshold set here may be, for example, ΔYm × 2, where ΔYm is the median value of the interval between the centroid positions of each intervertebral region in the Y coordinate. The intervertebral region added here may be a line segment with a width of 10 mm and a height of 1 mm, with the centroid position being the midpoint between the midpoint of the H1 row of the spinal region and the centroid position of the intervertebral region at H2, as shown in the Y coordinate position of H1.5 in Figure 7(b). However, the method for setting the threshold and the additional intervertebral region is not limited to this, and can be set in any way desired.

[0044] Similarly, at the Y-coordinate position, the vertical distance is calculated from the lower end of the extracted spinal region (H4 in Figure 7(a)) to the centroid of the first intervertebral region counting medially from the lower end of the spine (H3 in Figure 7(b)). If this distance is greater than the threshold, an intervertebral region may be added at the midpoint between the end of the spinal region and the centroid of the first intervertebral region counting medially, as shown at the Y-coordinate position of H3.5 in Figure 7(b).

[0045] For the intervertebral regions and their centroids that have been corrected in this way, the analysis region extraction unit 201 sets the centroid G1 of the fourth intervertebral region counting from the bottom (from the pelvic side) of the image, as shown in Figure 4(b). Similarly, it sets the centroid G2 of the third intervertebral region counting from the bottom, the centroid G3 of the second intervertebral region counting from the bottom, and the centroid G4 of the first intervertebral region counting from the bottom, respectively.

[0046] Next, for each of the centroids G1, G2, G3, and G4, horizontal lines LR1, LR2, LR3, and LR4 are set passing through them (the four dotted lines in Figure 4(b)). As illustrated, if the number of extracted intervertebral regions is less than four, four lines are drawn to divide the extracted spinal region vertically into five equal parts. These four lines may be extended to both the left and right edges of the image and set as lines LR1, LR2, LR3, and LR4, respectively, from the top of the image. If this process is performed, after extracting the bone density analysis region in the following process, a message indicating the possibility of a shift in the bone density analysis region or prompting the operator to manually correct it may be displayed on the bone density measurement display unit 205. Then, as shown in Figure 4(b), horizontal reference regions SR1, SR2, and SR3 with vertical widths are set between these horizontal lines LR1, LR2, LR3, and LR4.

[0047] Note that the extraction result of the spinal region in step S102 may be tilted as shown in Figure 8(a). In such cases, the rotation correction of the entire image may be performed using the methods illustrated in Figures 8(b) and 8(c). As a specific example of how to set the angle for rotation correction in the image, for example, the horizontal centroid position (black dot in Figure 8(b)) is determined for each row or thinned row of the extracted spinal region as shown in Figure 8(b). Next, the image is rotated so that the approximate lines (dotted lines in Figure 8(b)) become the vertical direction of the image as shown in Figure 8(c). By performing this processing, the extension direction of the spinal region can be made to match the extension direction of the Y coordinate.

[0048] Alternatively, instead of performing image rotation correction, lines LR1, LR2, LR3, LR4 and reference regions SR1, SR2, SR3 may be set diagonally based on the detected tilt. Furthermore, if the intervertebral region extracted in step S102 is tilted, lines LR1, LR2, LR3, LR4 may be drawn diagonally. An example of how to draw these lines is illustrated in Figure 9. Figure 9 shows one of the extracted intervertebral regions, assuming it is tilted as shown during extraction. In such a case, the tilt of this intervertebral region can be obtained by, for example, taking the midpoint vertically in each column or thinned-out column of the intervertebral region (black dots in Figure 9) and drawing approximation lines between them (dotted lines in Figure 9). Another example of how to draw these lines is to draw an approximation line of the spinal region near the intervertebral region, similar to Figure 8(b), and then draw a line perpendicular to this approximation line and passing through the centroid of the intervertebral region. Furthermore, horizontal reference regions SR1, SR2, and SR3, each with a vertical dimension, may be set between lines LR1, LR2, LR3, and LR4. Additionally, the image may be rotated and corrected based on the slope of the approximate line in these regions.

[0049] The analysis region extraction unit 201 extracts the spinal region (corresponding to the vertebral body) extracted within the reference region SR1 as the second lumbar vertebra L2. Similarly, it extracts the spinal region within the reference region SR2 as the third lumbar vertebra L3, and the spinal region within the reference region SR3 as the fourth lumbar vertebra L4. By combining the extraction results of the spinal region and the intervertebral region in this way, the bone density analysis region of the lumbar spine can be extracted without vertical displacement. As described above, the analysis region extraction unit 201 can function as an example of a detection means for detecting the bone density analysis region, which is the target region for bone density measurement of the lumbar spine, based on the extraction results of the discriminator.

[0050] In the above, each vertebral body was identified by counting the vertebral region medially from the lower end of the spine. Next, an example of identification when the lower end of the spine is not included in the input image will be described. In this case, the operator specifies the vertebral body number to be analyzed in advance to the analysis target determination unit 202. The operator operates the analysis target determination unit 202 to trim the input image so that the vicinity of the specified vertebral body is included. The analysis region extraction unit 201 extracts the vertebral region and intervertebral region from the trimmed image using a trained model. The analysis region extraction unit 201 obtains the vertebral body region based on the extracted results. Each extracted vertebral body region can be identified by assigning a vertebral body number based on the vertebral body number specified in advance by the operator. Furthermore, the trimming method has been described here assuming that the operator always performs it. However, for example, the information trimmed by the operator may be stored in the storage unit 203, and in subsequent examinations, the analysis region extraction unit 201 may trim to the same position as previously trimmed positions. Once the bone density analysis region is extracted in step S103, the flow proceeds to step S104.

[0051] Next, in step S104, background regions are set for calculating bone transmittance. An example of how to set background regions is described below. The background regions are shown as BG11, BG12, BG21, BG22, BG31, and BG32 in Figure 4(c), for example. These regions are set on the left and right sides of the lumbar vertebrae L2, L3, and L4 within the reference regions SR1, SR2, and SR3 set in step S103. An example of how to search for these background regions is described below.

[0052] The vertical dimensions of background areas BG11, BG12, BG21, BG22, BG31, and BG32 can be set to the same value as the vertical dimensions of their respective reference areas SR1, SR2, and SR3. The horizontal dimensions of background areas BG11, BG12, BG21, BG22, BG31, and BG32 can be set to a pre-configured value. For example, this width can be set to 20 mm. The specified value for the width may be stored in the memory unit 203 beforehand. Alternatively, it may be specified by the operator using the operation unit 108.

[0053] The position of background region BG11 may be automatically selected in the area to the left of the second lumbar vertebra L2 within the reference region SR1, such that the dispersion of luminance values ​​within the background region is minimized. Similarly, the position of background region BG12 may be automatically selected in the area to the right of the second lumbar vertebra L2 within the reference region SR1, such that the dispersion of luminance values ​​within the background region is minimized. Furthermore, background regions BG21 and BG22 may be selected in the left and right areas of the third lumbar vertebra L3, and background regions BG31 and BG32 may be selected in the left and right areas of the fourth lumbar vertebra L4.

[0054] In this case, the transverse process region may be excluded so as not to be selected as a background region. For example, when searching for background region BG11, the area designated as the transverse process region to the left of the region extracted as the second lumbar vertebra L2 may be excluded from the search range. For example, a range of 15 mm may be specified as the transverse process region. Similarly, when searching for background regions BG12, BG21, BG22, BG31, and BG32, the left and right transverse process regions of lumbar vertebrae L2, L3, and L4 may be excluded. Note that the method of setting the background region is not limited to this and can be set in any way. For example, the operator may specify the background region by instructing the analysis target determination unit 202 via the operation unit 108.

[0055] The bone density analysis region and background region extracted as described above may be displayed on the bone density measurement display unit 205. Furthermore, based on the bone density analysis region and background region displayed on the bone density measurement display unit 205, the operator may instruct the analysis target determination unit 202 to modify and determine the bone density analysis region and background region. Similarly, the operator may instruct the analysis target determination unit 202 to modify and determine the centroid and straight lines that appear during the above calculation process. The background region does not necessarily have to be displayed on the bone density measurement display unit 205. Once the background region is extracted, the flow proceeds to step S105.

[0056] Next, in step S105, the analysis target determination unit 202 retrieves past examination information from the memory unit 203. At this time, the analysis target determination unit 202 uses the patient information entered by the operation unit 108 as a search key to identify vertebrae that were previously excluded from the bone density analysis area.

[0057] Figure 11 shows an example of data stored in the memory unit 203. Note that Figure 11 is just an example; for example, the patient ID column may be other patient information, and the excluded vertebral body number may be the vertebral body number that was the subject of the bone density analysis area. Also, the stored data may be the most recent data from past examinations, or data from earlier periods.

[0058] According to the example data, if the patient ID currently being examined is 1, the memory unit 203 retrieves the vertebral body number "L3" that corresponds to patient ID "1" and is to be excluded. If the patient ID currently being examined is "2", the memory unit 203 attempts to retrieve the excluded vertebral body number corresponding to patient ID "2", but there are no vertebral bodies that have been excluded from the bone density analysis area in the past. Therefore, the excluded vertebral body number is not retrieved. If the patient ID currently being examined is "4", no past examination information for that patient is stored, so the excluded vertebral body number is not retrieved. Once the analysis target determination unit 202 retrieves past examination information, the flow proceeds to step S106.

[0059] Next, in step S106, the analysis target determination unit 202 determines whether or not past examination information obtained in step S105 exists. If there are vertebrae that were previously excluded from the bone density analysis area, the determination in step S106 is "Yes", and the flow proceeds to step S107. If there is no past examination information obtained in step S105, or if there are no vertebrae that were previously excluded from the bone density analysis area, the determination in step S106 is "No", and the flow proceeds to step S108.

[0060] Next, in step S107, the analysis target determination unit 202 excludes vertebral regions that were previously excluded from the bone density analysis region from the bone density analysis region extracted in step S103. For example, if the excluded vertebral number stored in the memory unit 203 is "L3", the analysis target determination unit 202 excludes the third lumbar vertebra L3 in Figure 4(c), which was extracted by the analysis region extraction unit 201, from the bone density analysis region. After excluding the previously excluded vertebral regions from the bone density analysis region extracted this time, the flow proceeds to step S108.

[0061] Next, in step S108, the analysis target determination unit 202 determines the vertebrae that were not excluded through step S106 or step S107 as targets for analysis. At this time, the operator can check the bone density analysis area displayed on the bone density measurement display unit 205 and, if necessary, instruct the analysis target determination unit 202, for example via the operation unit 108, to modify the bone density analysis area. Once the vertebrae to be analyzed have been determined, the flow proceeds to step S109.

[0062] Next, in step S109, the bone density calculation unit 204 calculates the bone density in each bone density analysis region by referring to the bone density analysis region and the background region. Bone density can be calculated by known methods, for example, by obtaining the thickness of bone and soft tissue in the image using the DXA method. Alternatively, bone density may be calculated by obtaining the atomic number and surface density using the DXA method. Once the bone density is calculated, the flow proceeds to step S110.

[0063] The calculated bone density is displayed on the bone density measurement display unit 205, and the operator confirms the bone density in step S110. Figure 10 shows an example of the bone density calculation results displayed on the bone density measurement display unit 205. Figure 10(a) shows an example of the display when all extracted vertebrae are included in the bone density calculation. The example display screen includes an ID display unit 300 that displays patient information, an examination information display unit 301 that displays information such as the date and time of the examination, a score display unit 302 that displays the bone density score, an image display unit 303 that displays the bone density analysis area, and an exit button 304.

[0064] The content displayed in the ID display unit 300, the examination information display unit 301, and the score display unit 302 is merely an example and is not limited to this; other information may also be included. For example, the ID display unit 300 may display patient ID, patient name, gender, date of birth, height, weight, examination site, pregnancy status, etc. The examination information display unit 301 may display examination date and time, reception number, referring physician's name, examination description, and interpreting physician's name, etc. In addition, the bone density score displayed in the score display unit 302 may include evaluation values ​​such as %AGE, Z score, %YAM, and T score, in addition to the site, bone density, bone mineral density, and bone area. Furthermore, based on the calculated bone density, any other values ​​useful for diagnosis may be displayed in the bone density measurement display unit 205. For example, a value compared to the average bone density of the same age group or other age groups may be displayed in the bone density measurement display unit 205. Also, for example, time-series data compared to past measurement results may be displayed in the bone density measurement display unit 205.

[0065] Figure 10(b) shows an example of bone density calculation results when the third lumbar vertebra L3 is excluded from the analysis area. When the extracted area includes vertebrae excluded from bone density calculation, the excluded vertebrae can be displayed in gray, as shown in the example of the score display unit 305, and they can be excluded from the calculation of the average value. Note that graying is just one example; the display method is not limited to this example, as long as the operator can recognize that the vertebrae were excluded from bone density calculation by using strikethrough or hiding them. Also, when the analysis area includes vertebrae excluded from bone density calculation, as shown in the image display unit 306, the diagonal lines indicating the corresponding vertebrae are not displayed so that it is clear that they were excluded from bone density calculation for the extracted area. Note that hiding the diagonal lines is just one example; as long as the operator can recognize that the vertebrae were excluded from bone density calculation, the outer frame of the vertebrae can be highlighted, for example, or text indicating that the vertebrae have been excluded can be displayed. The operator checks the bone density calculation results, and if there are no problems, presses the exit button 304. Upon pressing the exit button 304, the flow moves to step S111.

[0066] Next, in step S111, the bone density calculation unit 204 stores the vertebral body number of the bone density analysis area that was excluded from the bone density calculation and the patient information in the storage unit 203. For example, in the examination of a patient with patient ID 3, if the vertebral body number that was not included in the bone density analysis area was "L4", the storage unit 203 stores "3" as the patient ID and "L4", the vertebral body number that was excluded from the bone density calculation, as shown in the third row of Figure 11. As a result, in subsequent examinations of the patient with patient ID "3", it becomes possible to exclude the fourth lumbar vertebra, which corresponds to vertebral body number "L4", from the bone density analysis area.

[0067] In this embodiment, the bone density calculation unit 204 calculates the bone density in each bone density analysis region by referring to the vertebral body region adjusted as described above. Bone density can be calculated by known methods, for example, by obtaining the thickness of bone and soft tissue in the image using the DXA method. Alternatively, bone density may be calculated by obtaining the atomic number and surface density using the DXA method. The calculated bone density may be displayed on the bone density measurement display unit 205. In this embodiment, for the sake of explanation, the display unit 107 and the bone density measurement display unit 205 are configured differently, but they may be combined into the same configuration.

[0068] As described above, the bone density measurement unit 200, which functions as an example of an information processing device according to this disclosure, calculates bone density using a radiographic image (X-ray image) generated based on radiation such as X-rays. The illustrated bone density measurement unit 200 comprises a patient information acquisition unit 207, an image information acquisition unit 206, a region extraction unit (analysis region extraction unit 201), a storage unit 203, and an analysis target determination unit 202. The patient information acquisition unit 207 acquires patient information input via the operation unit 108. The image information acquisition unit 206 acquires a radiographic image taken of a patient corresponding to the patient information input via the operation unit 108. The analysis region extraction unit 201 extracts a bone density analysis region from the radiographic image acquired by the image information acquisition unit 206. The storage unit 203 stores information about bone regions for which bone density was calculated in previously taken radiographic images, and patient information about patients who have those bone regions and for whom radiographic imaging was performed for bone density measurement, linked together. The analysis target determination unit 202 determines whether the acquired patient information corresponds to the stored patient information. In the embodiment described above, the patient information is stored in the storage unit 203. However, instead of the storage unit 203, the patient information may be stored in an external storage device, and the patient information may be acquired from the external storage device when the patient information acquisition unit 207 acquires patient information, and the analysis target determination unit 202 will then make the determination. If there is a correspondence, the analysis target determination unit 202 determines the target area for calculating bone density in the extracted (obtained in the current imaging) bone density analysis area based on the bone density analysis area linked to the stored patient information and the information on the bone area from which bone density was calculated. By providing such an analysis target determination unit 202, the area that was targeted for bone density calculation in previous bone density measurements for the same patient is automatically determined, reducing the complexity of the operator having to set the target area for bone density calculation each time. The image from which the bone density analysis area is extracted may be a DXA image or a radiographic image. Furthermore, the information regarding the bone may include, for example, the previously excluded vertebral body numbers illustrated in Figure 11, or the vertebral body numbers that were previously included in the calculation of bone density.

[0069] More specifically, the bone density measurement unit 200 automatically excludes areas from which bone density is not calculated from the bone density analysis area extracted by the analysis area extraction unit 201. In other words, if the information regarding the bone area from which bone density has been calculated includes the existence of areas that were excluded from bone density calculation in the stored bone density analysis area, the analysis target determination unit 202 decides to exclude the area corresponding to the area excluded from the extracted bone density analysis area.

[0070] The bone density measurement unit 200 described above may further include a bone density measurement display unit 205. The bone density measurement display unit 205 can also function as an example of a display control unit that further commands the display unit 107 to display the extracted bone density analysis area and the bone density calculation result on the display unit 107. Furthermore, if the bone density measurement display unit 205 has a monitor, it may function as a display control unit that displays the extracted bone density analysis area and the bone density calculation result on the monitor. When measuring bone density, the analysis target determination unit 202 may determine that the stored bone density analysis area includes areas that have been excluded. In that case, the bone density measurement display unit 205 displays the extracted bone density analysis area on the display unit 107 in a display manner that allows the determined excluded areas to be distinguished from other areas of the extracted bone density analysis area, for example, as illustrated in Figure 10(b).

[0071] The bone density measurement unit 200 is connected to the operation unit 108, and the operation unit 108 can function as an input unit to receive instructions from the operator. Furthermore, the bone density measurement unit 200 can also display an image along with a touch panel on the bone density measurement display unit 205, thereby allowing the bone density measurement display unit 205 to function as an input unit using the touch panel. In this case, the analysis region extraction unit 201 can modify the bone density analysis region in response to instructions from the operator received via the input unit. Alternatively, the analysis region extraction unit 201 may extract the bone density analysis region using a trained model. For example, in the above embodiment, the image to be extracted is an image of a vertebral body, and the analysis region extraction unit 201 extracts the vertebral body using a trained model, resulting in the lumbar spine being the bone region for which bone density is calculated.

[0072] As a result, it becomes possible to automatically exclude vertebrae previously excluded from the bone density analysis area from the bone density calculation, thereby reducing the burden on the operator during bone density analysis. Furthermore, when excluding extracted vertebrae from the bone density calculation, errors or omissions in exclusion selections may occur by the operator. In addition, the reproducibility of corrections to the bone density analysis area is also left to the operator's skill, which can lead to cumbersome work and, in some cases, a decrease in accuracy. In the bone density calculation process according to this embodiment, by automatically identifying the current bone density analysis area based on the previous bone density analysis area as described above, it is expected that errors or omissions in exclusion selections and a decrease in reproducibility when correcting the bone density analysis area will be prevented.

[0073] (Example 2) In Example 1, the analysis region extraction unit 201 extracted the spinal region and intervertebral region from the input image using a trained model, and identified the first lumbar vertebra L1 to the fourth lumbar vertebra L4. In this example, the operator specifies the bone region and vertebral body boundary line to identify these lumbar vertebrae L1 to L4. In addition, the excluded vertebral body regions are highlighted on the displayed image based on the vertebral body numbers stored in the memory unit 203. The operation method of the bone density measurement device according to this example will be described below. In the following description, the same configuration, function, and operation as in Example 1 will be shown in the figures using the same reference numbers, etc., and explanations related to these will be omitted. The following mainly describes the differences from Example 1.

[0074] The following describes the process for calculating bone density in this embodiment, with reference to Figure 12. Figure 12 shows the processing flow executed by the bone density measurement unit 200 in this embodiment. Steps that perform the same processing as those described in Figure 3 are denoted by the same reference numerals, and detailed explanations are omitted here. The process according to this embodiment begins, and the flow moves to step S200.

[0075] In step S200, the trained model extracts the vertebral bodies and displays them on the bone density measurement display unit 205. At this time, the operator can roughly specify the bone region to be used for bone density calculation in the analysis target determination unit 202 while looking at the image displayed on the bone density measurement display unit 205. Alternatively, the bone region may be specified by painting it with a predetermined color using, for example, a pen tool or eraser tool. The color used at this time may be semi-transparent. After that, the analysis region extraction unit 201 may initially display the bone region of the vertebral bodies that have been automatically extracted using the trained model, and the operator may make modifications to it.

[0076] Figure 13(a) shows an example of an X-ray image of the vicinity of lumbar vertebrae L2-L4 acquired by the radiography device 100. In this embodiment, the operator roughly specifies the bone region to be used for bone density calculation in advance. However, it is also possible to use a pre-trained model from the beginning, for example, and use the image in Figure 13(a) as the input image to identify each vertebra in steps S200 and S201. In this case, the input image may be one of two X-ray images acquired by the DXA method, and may be the X-ray image with the lower tube voltage. Furthermore, if the FPD 106 is a stacked sensor, an image from either layer (for example, the upper layer on the subject side) may be used as the input image.

[0077] Furthermore, the input image here may be an image obtained from the radiography device 100, with the area near the lumbar spine extracted from it. Extraction of the area near the lumbar spine may be performed by specifying the target region in the analysis target determination unit 202. Alternatively, the area near the lumbar spine may be automatically extracted from the irradiation field and the position of the lumbar spine using a trained model or similar method. Once the bone region in the image is specified, the flow proceeds to step S201.

[0078] Next, in step S201, the operator specifies the boundary lines of the vertebral bodies. Figure 13(b) shows the display of the vertebral body boundary lines, with the boundary lines superimposed on the displayed image. The boundary lines of the vertebral bodies can be specified by dragging the cursor with the mouse on the image displayed on the bone density measurement display unit 205. If the bone density measurement display unit 205 and the analysis target determination unit 202 are integrated as a touch panel display, the operator can specify and adjust the boundary lines of the vertebral bodies by tracing their finger on the display screen of the bone density measurement display unit 205. Alternatively, the vertebral body boundary lines may be initially displayed using a trained model or the like, and then the operator may fine-tune them.

[0079] By performing the above process, the region of each vertebra can be extracted based on the set analysis region. After region extraction, the flow moves to step S105. Note that the processes performed in steps S106 and S105 are the same as in Example 1, so their explanation is omitted here. If there are vertebrae that have been previously excluded in step S106, the flow moves to step S202.

[0080] Next, in step S202, the bone density measurement display unit 205 highlights the vertebral regions that were previously excluded from the bone density analysis area, as obtained in step S105. For example, if the excluded vertebral number stored in the memory unit 203 is "L3", the bone density measurement display unit 205 displays the region of vertebral number "L3" by filling it in, as shown in Figure 13(c), to distinguish it from the bone density analysis area. Note that the highlighting method shown in Figure 13(c) is just one example; instead of filling it in, the outer frame of the vertebral region may be highlighted, or text indicating that it was previously excluded may be displayed. In addition, previously highlighted images may also be displayed.

[0081] In this embodiment, the bone density calculation unit 204 calculates the bone density in each bone density analysis region by referring to the vertebral body region adjusted as described above. Bone density can be calculated by known methods, for example, by obtaining the thickness of bone and soft tissue in the image using the DXA method. Alternatively, bone density may be calculated by obtaining the atomic number and surface density using the DXA method. The calculated bone density may be displayed on the bone density measurement display unit 205. In this embodiment, for the sake of explanation, the display unit 107 and the bone density measurement display unit 205 are configured differently, but they may be combined into the same configuration. Similarly, the operation unit 108 and the analysis target determination unit 202 may also be combined into the same configuration.

[0082] In this embodiment, the bone density measurement unit 200 is connected to the operation unit 108, and the operation unit 108 can function as an input unit to receive instructions from the operator. Alternatively, the bone density measurement unit 200 can display an image along with a touch panel on the bone density measurement display unit 205, thereby allowing the bone density measurement display unit 205 to function as an input unit using the touch panel. In this case, the analysis target determination unit 202 can modify the determined target area for bone density calculation in response to instructions from the operator received via the input unit.

[0083] In the above embodiment, the system highlights vertebrae that have been previously excluded from the bone density analysis area to the operator, allowing the operator to easily recognize the vertebrae excluded from the bone density calculation and reducing the operator's burden during bone density analysis. Furthermore, by making it easy to identify the current bone density analysis area in the bone density calculation process according to this embodiment, it is expected that errors or omissions in exclusion selections will be prevented, and the decrease in reproducibility when correcting the bone density analysis area will be improved.

[0084] (Example 3) For example, during lumbar spine examinations, implants made of metal components may be placed in specific lumbar vertebrae as part of fracture treatment. In such cases, the operator excludes the implant from the image during the examination, and bone density is measured in that state. Therefore, during follow-up examinations, it is necessary to reproduce the state with the implant excluded to a certain extent, and then calculate the bone density by specifying the bone density analysis region. Even this exclusion process places a significant burden on the operator, and its reduction is desired.

[0085] This embodiment aims to achieve the objective of the present invention by automatically reflecting the results of any previous modifications made by the operator to the vertebral region in relation to implants or the like, in the current region extraction results. In the following description, the same configuration, function, and operation as in Embodiment 1 will be shown in the figures using the same reference numerals, and explanations related to these will be omitted. The following mainly describes the differences from Embodiment 1.

[0086] Figure 14(a) shows an example of an X-ray image of the lumbar spine targeted by this embodiment. In this example, a metal implant 402 is attached to the vertebral body 401. In such cases, the discriminator by the analysis region extraction unit 201 may extract the vertebral body region including the implant. Therefore, the operator needs to perform a correction process to exclude the implant portion 403 using the pen tool or eraser tool, as shown in Figure 14(b). Performing such a correction process every time an X-ray is taken is very cumbersome for the operator. On the other hand, as shown in Figures 14(a) and 14(c), the shooting conditions and positioning are not exactly the same for each shot, so the results of the previous correction process cannot be directly applied. Figure 14(c) shows an example of an X-ray image of the same vertebral body 401 taken at different times. Specifically, the direction of extension of the lumbar spine is different in both figures, and the width is also not the same. Therefore, in this embodiment, the region extraction result for the current X-ray image is corrected based on the amount of difference between the previous and current X-ray images.

[0087] The following describes the process for calculating bone density in this embodiment, with reference to Figure 15. Figure 15 shows the processing flow executed by the bone density measurement unit 200 in this embodiment. Steps that perform the same processing as those described in Figure 3 are denoted by the same reference numerals, and detailed explanations are omitted here. The process according to this embodiment begins, and the flow moves to step S100.

[0088] First, the analysis region extraction unit 201 performs the processing steps S100 to S104 in the same manner as in Example 1 to extract the bone density analysis region and the background region. After these extractions, the flow proceeds to step S300.

[0089] Next, in step S300, the analysis target determination unit 202 retrieves past examination information from the memory unit 203. At this time, the analysis target determination unit 202 retrieves the previous X-ray image and bone density analysis area using the patient information entered by the operation unit 108 as a search key. Here, the bone density analysis area is data stored as a binary image in which the bone density analysis area on the X-ray image is set to 1 and everything else to 0, and the analysis target determination unit 202 retrieves this data. The memory unit 203 records the bone density analysis area (bone density calculation area before correction) that was automatically extracted by the analysis area extraction unit 201 in the previous imaging, and the bone density calculation area (bone density calculation area after correction) in which the bone density was actually calculated. The analysis target determination unit 202 retrieves both of these data from the memory unit 203.

[0090] In this embodiment, the bone density calculation region is stored as a binary image, but it is not limited to this; it may also be stored as a multi-level image with each lumbar vertebra (L2-L4) individually labeled. It is also possible to store it as coordinate data instead of an image. Once past examination information is obtained, the flow proceeds to step S301.

[0091] Next, in step S301, the analysis target determination unit 202 determines whether the bone density calculation area has been modified based on the previous examination information obtained in step S300. This determination can be made, for example, by checking whether there is a difference between the binary images before and after the modification. If it is determined that a modification has been made, the determination in step S302 will be "Yes," and the flow will proceed to step S302. In step S302, the bone density analysis area is updated, but the detailed operation of this will be described later. If the determination is made that no modification has been made, the determination in step S303 will be "No," and the flow will proceed to step S303.

[0092] Next, in step S303, the bone density analysis area is displayed on the bone density measurement display unit 205, and it is determined whether or not the determined bone density analysis area needs to be modified. This determination can be made, for example, by the operator pressing a button (not shown) displayed on the bone density measurement display unit 205, or by dragging a cursor with a mouse on the image. If the operator indicates to the bone density measurement display unit 205 that modification is necessary, the flow moves to step S304. In step S304, the operator modifies the bone density analysis area as needed, for example, using a pen tool or eraser tool. After modification, if the operator performs a specific operation, or if a predetermined amount of time has elapsed, the flow moves to step S109, assuming that the modification is complete. Alternatively, in step S303, if the operator performs a specific operation, or if a predetermined amount of time has elapsed, it is determined that no modification is needed, and the flow moves to step S109.

[0093] In the following steps S109 to S110, the same processing as in Example 1 is performed. The calculated bone density is then displayed on the bone density measurement display unit 205. After the results are displayed, the flow moves to step S305, where the bone density analysis area before and after the modification is linked to the patient information and stored in the storage unit 203.

[0094] Here, the bone density analysis area update process performed in step S302 will be described below with reference to Figure 16. Figure 16 is a processing flow showing the details of the bone density analysis area update process performed by the analysis target determination unit 202. When the processing flow shown in Figure 15 moves to step S302, the processing in step S400 begins.

[0095] First, in step S400, a histogram of the bone density calculation region before and after the previous correction is generated. The histogram is obtained by calculating the frequency of each pixel value only from the region corresponding to the bone density calculation region of the previously acquired X-ray image. Note that the X-ray image used to generate the histogram can be either of the two types of tube voltages obtained using a kV switching method or the like.

[0096] Figure 17(a) shows an example of histogram generation. Implants and other metal components do not transmit X-rays as well as bone tissue, so they appear in the region 501 with low pixel values ​​in the histogram, as shown in Figure 17(a). On the other hand, soft tissues transmit X-rays more well than bone tissue, so they appear in the region 502 with high pixel values ​​in the histogram, as shown in Figure 17(a). Therefore, in a histogram that excludes regions other than bone, the frequencies of regions 501 and 502 decrease, resulting in a unimodal histogram centered on the average value of the bone region. Thus, by excluding the pixel value range TH1~TH2 in Figure 17(a) from the bone density calculation region, only the bone region can be extracted. After histogram generation, the flow proceeds to step S401.

[0097] In step S401, the threshold pixel values ​​TH1 and TH2 are calculated to identify the range of the bone density analysis region. Specifically, the frequencies are checked in order from the lowest pixel values ​​for the histograms before and after correction, and the pixel value TH1 is calculated as the first pixel value where the number of pixels not excluded (frequency after correction) exceeds the number of pixels excluded by correction (frequency before correction - frequency after correction). Similarly, the frequencies are checked in order from the highest pixel values, and the pixel value TH2 is calculated as the first pixel value where the number of pixels not excluded exceeds the number of pixels excluded. Once pixel values ​​TH1 and TH2 are calculated, the flow proceeds to step S402.

[0098] Next, in step S402, the amount of histogram shift is calculated from the previous and current X-ray images. Figure 17(b) shows an example of the histogram shift. The histogram shape in the bone density analysis region is approximately the same when the tube voltage used for imaging is the same, with little influence from positioning. On the other hand, if the imaging dose is different, a lateral shift occurs in the histogram. Figure 17(b) shows the case where the current imaging dose is less than the previous imaging dose. In this case, the histogram of the current imaging shifts to the left, towards lower pixel values, compared to the previous image.

[0099] Therefore, in this embodiment, this displacement amount ΔV is calculated. Specifically, values ​​(mean, mode, median, etc.) that show the central trend of the bone density calculation area before correction are calculated from the histogram, and the difference between them is calculated as ΔV. Note that if the tube voltage is not the same in the previous and current imaging, the shape of the histogram will change. In such cases, instead of the shift amount, a polynomial that minimizes the frequency error of both histograms should be found, and that polynomial should be used as a transformation formula to correct the displacement amount. Once the displacement amount is calculated, the flow proceeds to step S403.

[0100] Next, in step S403, the threshold corresponding to the bone density analysis region is determined. Specifically, the threshold is calculated by shifting the threshold (pixel values ​​TH1, TH2) obtained in step S401 by ΔV (TH1-ΔV, TH2-ΔV). As shown in Figure 17(b), the threshold calculated in this way is shifted by the amount of histogram shift, making it a suitable threshold for excluding areas other than the bone region from the bone density calculation region. After updating the threshold, the flow proceeds to step S404.

[0101] Next, in step S404, the areas with pixel values ​​below the updated threshold TH1 and areas with pixel values ​​above TH2 are excluded from the bone density analysis area. Note that in threshold processing, noise that cannot be completely excluded, such as isolated points, may remain around the excluded boundaries. In such cases, isolated points may be excluded using morphological operations or other methods.

[0102] As described above, the bone density measurement unit 200 according to this embodiment may further include an input unit and a modification unit. The bone density measurement unit 200 may also display a touch panel along with an image on the bone density measurement display unit 205. This allows the bone density measurement display unit 205 to function as an example of an input unit using the touch panel, and the operator can input instructions to the bone density measurement unit 200 using this input unit. The bone density measurement unit 200, more specifically the analysis target determination unit 202, receives instructions from the operator via this input unit. In this embodiment, the analysis target determination unit 202 can also function as an example of a modification unit that performs modifications to exclude areas corresponding to implants, for example. In this case, the analysis target determination unit 202 performs modifications to exclude areas from which bone density calculation is not performed from the bone density analysis area extracted by the analysis area extraction unit 201, in response to instructions from the operator received via the input unit. Furthermore, the bone information linked to the patient information stored in the memory unit 203 includes information about the bone area for which bone density was calculated, and information about the region that has been modified to be excluded.

[0103] In this embodiment, the analysis target determination unit 202 determines to exclude the region corresponding to the region that has been modified to be excluded (403 in Figure 14(b)) from the target region for calculating bone density in the extracted bone density analysis region. In this embodiment, the analysis target determination unit 202 also determines the region that has been modified to be excluded based on the histogram of the bone density analysis region before and after modification, which is linked to the stored patient information. Then, based on the determined region that has been modified to be excluded, it determines the corresponding region to be excluded. More specifically, the analysis target determination unit 202 calculates a conversion parameter that matches the histogram of the bone density analysis region linked to the stored patient information with the histogram of the extracted bone density analysis region. Then, based on this conversion parameter, it determines the corresponding region to be excluded.

[0104] In summary, this embodiment reduces the operator's cumbersome work by automatically reflecting any previous modifications made to the vertebral region in the current region extraction results. Furthermore, the bone density calculation process in this embodiment is also expected to improve the reproducibility that occurs when modifying the bone density analysis region.

[0105] (Example 4) This embodiment automatically incorporates the results of past vertebral region modifications performed by the operator into the current region extraction results, using a method different from that of Embodiment 3. Specifically, it adapts the previous bone density analysis region to the current imaging by warping it. In this embodiment, step S302 in Figure 15 is performed according to the processing flow shown in Figure 18, but the rest of the process is the same as that performed in Embodiment 3, so these explanations are omitted here.

[0106] The bone density analysis region update process, a characteristic of this embodiment, which is performed by the analysis target determination unit 202, will be explained below using Figure 18. Figure 18 is a processing flow showing the details of the bone density analysis region update process. When the processing flow shown in Figure 15 moves to step S302, the processing in step S500 begins.

[0107] First, in step S500, keypoints are detected from the previous and current X-ray images. Keypoints are characteristic points in the image, and areas with a large feature quantity representing the degree of characteristic are detected as feature points. Commonly used feature quantities can be used for this detection, such as A-KAZE, KAZE, SIFT, SURF, BRISK, FAST, ORB, etc. Figures 19(a) and 19(b) show examples of keypoint detection from lumbar spine X-ray images. In these figures, the points marked with an "x" as example point 601 represent the respective keypoints. Figure 19(a) is an example of the X-ray image taken previously, and Figure 19(b) is an example of the X-ray image taken this time. If the subject is the same, it is possible to detect keypoints at approximately the same positions even if there are differences in positioning.

[0108] Here, the range in which keypoint detection is performed may be the entire captured X-ray image, or it may be limited to only the bone density analysis region extracted by the analysis region extraction unit 201. Alternatively, keypoint detection may be performed from a binary image showing the bone density analysis region, in which case keypoint detection should be performed using the shape of the boundary between the bone density analysis region and the other regions as a characteristic feature. Once keypoints are detected, the flow proceeds to step S501.

[0109] Next, in step S501, keypoints detected from the previous and current X-ray images are matched. Specifically, the pairwise distance between the feature vectors of the keypoints calculated from the previous and current X-ray images is calculated, and those with similar distances are selected as corresponding keypoints. Here, the sum of squared differences or the Hamming distance can be used as the distance measure between features. Figure 19(c) shows an example of the matching process, with solid lines (602, 603, etc.) indicating the matched keypoints. Once the keypoint matching is complete, the flow moves to step S502.

[0110] Next, in step S502, geometric transformation parameters (projection matrices) are calculated from the matched keypoints. Specifically, based on the corresponding points shown in Figure 19(c), the matrix for projecting the image shown in Figure 19(a) onto the image shown in Figure 19(b) is estimated using the Direct Linear Transform (DLT) algorithm or the RANSAC algorithm. Note that the solid line 603 in Figure 19(c) shows an example of an incorrect match, but even with such outliers, accurate estimation is possible by using the RANSAC algorithm. Furthermore, if nonlinear deformation is to be considered, the image may be divided into blocks and the projection transformation matrix may be estimated for each local region. After the calculation of the geometric transformation parameters, the flow proceeds to step S503.

[0111] Next, in step S503, the previously corrected bone density analysis region (binary image) is warped (transformed based on the projection matrix). After the warping is complete, the flow moves to step S504. The result of the warping is a binary image showing the corrected bone density calculation region that matches the current X-ray image. In step S504, the bone density analysis region is corrected based on this data. Specifically, the corrected bone density calculation region is defined as the area where both the warped bone density calculation region (binary image) and the bone density calculation region (binary image) extracted by the analysis region extraction unit 201 are 1 (bone region).

[0112] As described above, in this embodiment, the analysis target determination unit 202 calculates geometric transformation parameters to match the image from which the bone density analysis region linked to the stored patient information has been extracted with the acquired image. Based on these geometric transformation parameters, the bone density analysis region, which has been modified to exclude regions for which bone density calculation is not performed, is geometrically transformed to determine the target region for bone density calculation. At that time, key points can be detected from both images, and geometric transformation parameters can be calculated from the correspondence between the key points.

[0113] Furthermore, in this embodiment, the analysis target determination unit 202 may calculate geometric transformation parameters to match the image showing the bone density analysis region linked to the stored patient information with the extracted image showing the bone density analysis region. In this case, the analysis target determination unit 202 can determine the target region for bone density calculation by geometrically transforming the bone density analysis region, which has been modified to exclude regions where bone density is not to be calculated, based on the geometric transformation parameters calculated in this way. In this case, the image showing the bone density analysis region can be an image in which the pixel values ​​of regions other than the bone density analysis region are masked to 0. Alternatively, the image showing the bone density analysis region may be a binary image in which the pixel values ​​of the bone density analysis region are set to 1 and the pixel values ​​of regions other than the bone density analysis region are set to 0. Furthermore, the geometric transformation parameters may be calculated by detecting keypoints from both images and determining the correspondence between the keypoints.

[0114] Furthermore, in the present invention, the analysis region extraction unit 201 of the bone density measurement unit 200 can extract the bone density analysis region by inputting an image from which the bone density analysis region has been extracted into a trained model obtained by machine learning using training data containing information on the bone density analysis region. In this case, the bone density analysis region linked to the stored patient information and the extracted bone density analysis region are obtained by the same trained model.

[0115] As described above, the above embodiment reduces the operator's cumbersome work by automatically reflecting any previous modifications made to the vertebral region in the current region extraction results. Furthermore, the bone density calculation process in this embodiment is also expected to improve the reproducibility that occurs when modifying the bone density analysis region.

[0116] Although this embodiment uses the lumbar spine as an example, the same method can be applied to the femur, for example. For instance, a method for automatically extracting the bone density calculation region for the femur is known, as disclosed in Patent Document 2. Therefore, by replacing the operation of the analysis region extraction unit 201 with the method described in Patent Document 2, the same process can be implemented in the case of the femur.

[0117] (Other examples) The disclosure can also be implemented by supplying a program that implements one or more of the functions of the embodiments described above to a system or device via a network or storage medium, and by having one or more processors in the computer of that system or device read and execute the program. Furthermore, the disclosure can also be implemented by a circuit (e.g., an ASIC) that implements one or more functions. A computer may have one or more processors or circuits and may include a plurality of separate computers or a network of a plurality of separate processors or circuits for reading and executing computer executable instructions.

[0118] A processor or circuit may include a central processing unit (CPU), a microprocessing unit (MPU), a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), or a field-programmable gateway (FPGA). Furthermore, a processor or circuit may include a digital signal processor (DSP), a dataflow processor (DFP), or a neural processing unit (NPU).

[0119] This disclosure includes the following configurations, methods, and programs. (Composition 1) An information processing device that calculates bone density using images generated based on radiation, A patient information acquisition unit that acquires patient information entered via the control unit, An image information acquisition unit that acquires the image taken in accordance with the patient information, A region extraction unit for extracting bone density analysis regions from the acquired image, An information processing apparatus comprising: an analysis target determination unit that, when the acquired patient information corresponds to patient information relating to a patient having the bones that has been previously linked to information relating to the bones for which bone density has been calculated and stored in a storage unit, determines a target area for calculating bone density in the extracted bone density analysis area based on the bone density analysis area linked to the stored patient information and the information relating to the bones for which bone density has been calculated; (Configuration 2) The information processing device according to Configuration 1, wherein if the information regarding the bone portion for which bone density has been calculated includes the existence of a region excluded from the bone density calculation in the stored bone density analysis region, the analysis target determination unit determines to exclude the region corresponding to the excluded region from the extracted bone density analysis region. (Composition 3) The system further includes a display control unit that displays the extracted bone density analysis area and the bone density calculation result on a display unit. The information processing apparatus according to configuration 2, wherein, if the stored bone density analysis area includes the presence of the excluded area, the display control unit causes the extracted bone density analysis area to be displayed on the display unit in a display manner that distinguishes the determined excluded area from other areas of the extracted bone density analysis area. (Composition 4) It is further equipped with an input unit that receives instructions from the operator, The information processing device according to any one of configurations 1 to 3, wherein the analysis target determination unit modifies the determined target area in response to instructions from the operator received via the input unit. (Composition 5) The aforementioned image is an image of a vertebral body, The region extraction unit extracts vertebral bodies using a trained model, The information processing device according to any one of configurations 1 to 4, wherein the bone portion from which the bone density is calculated is the lumbar vertebrae. (Composition 6) An input unit that receives instructions from the operator, The system further includes a modification unit that, in response to instructions from the operator received via the input unit, performs modifications to remove areas from the bone density analysis area extracted by the area extraction unit where bone density calculation is not performed, The information processing device according to any one of configurations 1 to 4, wherein the information regarding the bone portion for which the bone density has been calculated includes information regarding the region that has been modified to be excluded. (Composition 7) The information processing apparatus according to configuration 6, wherein the analysis target determination unit determines, in the extracted bone density analysis region, to exclude the region corresponding to the region to be modified from the target region for calculating bone density. (Composition 8) The information processing device according to configuration 6 or 7, wherein the analysis target determination unit determines the region to be excluded based on the histogram of the bone density analysis region before and after modification, which is linked to the stored patient information, and determines the corresponding region to be excluded based on the determined region to be excluded. (Composition 9) The information processing device according to configuration 8, wherein the analysis target determination unit calculates conversion parameters to match the histogram of the bone density analysis region linked to the stored patient information with the histogram of the extracted bone density analysis region, and determines the corresponding region to be excluded based on the conversion parameters. (Composition 10) The information processing apparatus according to configuration 6 or 7, wherein the analysis target determination unit calculates geometric transformation parameters to match the image extracted from the bone density analysis region linked to the stored patient information with the acquired image, performs a geometric transformation on the bone density analysis region that has been modified to exclude regions for which bone density calculation is not performed based on the geometric transformation parameters, and determines the target region for which bone density is calculated. (Composition 11) The information processing device according to configuration 10 detects keypoints from both images and calculates the geometric transformation parameters from the correspondence between the keypoints. (Composition 12) The information processing apparatus according to configuration 6 or 7, wherein the analysis target determination unit calculates geometric transformation parameters to match the image showing the bone density analysis region linked to the stored patient information with the image showing the extracted bone density analysis region, performs a geometric transformation on the bone density analysis region that has been modified to exclude regions for which bone density is not to be calculated based on the geometric transformation parameters, and determines the target region for which bone density is to be calculated. (Composition 13) The information processing device according to configuration 12, wherein the image showing the bone density analysis region is an image in which the pixel values ​​outside the bone density analysis region are masked to 0. (Composition 14) The information processing device according to configuration 12, wherein the image showing the bone density analysis region is a binary image in which the pixel values ​​of the bone density analysis region are set to 1 and the pixel values ​​of the pixels outside the bone density analysis region are set to 0. (Composition 15) The information processing device according to configuration 12 detects keypoints from both images and calculates the geometric transformation parameters from the correspondence between the keypoints. (Composition 16) The region extraction unit extracts the bone density analysis region by inputting the image into a trained model obtained by machine learning using training data that includes information on the bone density analysis region, according to any one of configurations 6 to 15. (Composition 17) The information processing device according to configuration 16, wherein the bone density analysis region linked to the stored patient information and the extracted bone density analysis region are obtained by the same trained model. (Composition 18) An information processing device as described in any of configurations 1 to 17, A radiography system comprising: an information processing device connected to the aforementioned information processing device; a radiography device that irradiates a subject with radiation and generates an image based on the radiation; and a radiography system. (Method 1) A method for operating an information processing device that calculates bone density using an image generated based on radiation, To acquire patient information entered via the control unit, To acquire the image taken in accordance with the aforementioned patient information, Extracting the bone density analysis region from the acquired image, A method for operating an information processing device, comprising: when the acquired patient information corresponds to stored patient information relating to a patient having the bone portion linked to information about the bone portion for which bone density was previously calculated, determining a target area for calculating bone density in the extracted bone density analysis area based on the bone density analysis area linked to the stored patient information and the information about the bone portion for which bone density was calculated. (program) A program that, when executed by a computer, causes the computer to perform each step of the operation method of the information processing device described in Method 1.

[0120] Although preferred embodiments of the present invention have been described above, it goes without saying that the present invention is not limited to these embodiments, and various modifications and changes are possible within the scope of its essence. [Explanation of Symbols]

[0121] 100 Radiography equipment 101 Control Unit 102 Radiation Generator Control Unit 103 X-ray tube 104 Stand 105 grid 106 FPD 107 Display section 108 Operation section 200 Bone density measurement section 201 Analysis area extraction part 202 Analysis Target Determination Unit 203 Storage section 204 Bone density calculation section 205 Bone density measurement display section

Claims

1. An information processing device that calculates bone density using images generated based on radiation, A patient information acquisition unit that acquires patient information entered via the control unit, An image information acquisition unit that acquires the image taken in accordance with the patient information, A region extraction unit for extracting bone density analysis regions from the acquired image, An information processing apparatus comprising: an analysis target determination unit that, when the acquired patient information corresponds to patient information relating to a patient having the bones that has been previously linked to information relating to the bones for which bone density has been calculated and stored in a storage unit, determines a target area for calculating bone density in the extracted bone density analysis area based on the bone density analysis area linked to the stored patient information and the information relating to the bones for which bone density has been calculated;

2. If the information relating to the bone portion for which bone density has been calculated includes the existence of a region that has been excluded from the calculation of bone density in the stored bone density analysis region, the analysis target determination unit determines to exclude the region corresponding to the excluded region from the extracted bone density analysis region, as described in claim 1.

3. The system further includes a display control unit that displays the extracted bone density analysis area and the bone density calculation result on a display unit. The information processing apparatus according to claim 2, wherein, if the stored bone density analysis area includes the presence of the excluded area, the display control unit causes the extracted bone density analysis area to be displayed on the display unit in a display manner that distinguishes the determined excluded area from other areas of the extracted bone density analysis area.

4. It is further equipped with an input unit that receives instructions from the operator, The information processing apparatus according to claim 1, wherein the analysis target determination unit modifies the determined target area in response to instructions from the operator received via the input unit.

5. The aforementioned image is an image of a vertebral body, The region extraction unit extracts vertebral bodies using a trained model, The information processing apparatus according to claim 1, wherein the bone portion from which the bone density is calculated is the lumbar vertebrae.

6. An input unit that receives instructions from the operator, The system further includes a modification unit that, in response to instructions from the operator received via the input unit, performs modifications to remove areas from the bone density analysis area extracted by the area extraction unit where bone density calculation is not performed, The information processing apparatus according to claim 1, wherein the information relating to the bone portion for which the bone density has been calculated includes information relating to the region to be excluded by the modification.

7. The information processing apparatus according to claim 6, wherein the analysis target determination unit determines, in the extracted bone density analysis region, to exclude the region corresponding to the region on which the exclusion modification was performed from the target region for calculating the bone density.

8. The information processing apparatus according to claim 6, wherein the analysis target determination unit determines the region to be excluded based on the histogram of the bone density analysis region before and after modification, which is linked to the stored patient information, and determines the corresponding region to be excluded based on the determined region to be excluded.

9. The information processing apparatus according to claim 8, wherein the analysis target determination unit calculates conversion parameters to match the histogram of the bone density analysis region linked to the stored patient information with the histogram of the extracted bone density analysis region, and determines the corresponding region to be excluded based on the conversion parameters.

10. The information processing apparatus according to claim 6, wherein the analysis target determination unit calculates geometric transformation parameters to match the image extracted from the bone density analysis region linked to the stored patient information with the acquired image, performs a geometric transformation on the bone density analysis region that has been modified to exclude regions for which bone density calculation is not performed based on the geometric transformation parameters, and determines the target region for which bone density is calculated.

11. The information processing apparatus according to claim 10, wherein the geometric transformation parameters are calculated from the correspondence between key points detected from both images.

12. The information processing apparatus according to claim 6, wherein the analysis target determination unit calculates geometric transformation parameters to match the image showing the bone density analysis region linked to the stored patient information with the image showing the extracted bone density analysis region, performs a geometric transformation on the bone density analysis region that has been modified to exclude regions for which bone density is not to be calculated based on the geometric transformation parameters, and determines the target region for which bone density is to be calculated.

13. The information processing apparatus according to claim 12, wherein the image showing the bone density analysis region is an image in which the pixel values ​​outside the bone density analysis region are masked to 0.

14. The information processing apparatus according to claim 12, wherein the image showing the bone density analysis region is a binary image in which the pixel values ​​of the bone density analysis region are set to 1 and the pixel values ​​of the pixels outside the bone density analysis region are set to 0.

15. The information processing apparatus according to claim 12, wherein the geometric transformation parameters are calculated from the correspondence between key points detected from both images.

16. The information processing apparatus according to claim 6, wherein the region extraction unit extracts the bone density analysis region by inputting the image into a trained model obtained by machine learning using training data that includes information on the bone density analysis region.

17. The information processing apparatus according to claim 16, wherein the bone density analysis region linked to the stored patient information and the extracted bone density analysis region are obtained by the same trained model.

18. An information processing device according to any one of claims 1 to 17, A radiography system comprising: an information processing device connected to the aforementioned information processing device; a radiography device that irradiates a subject with radiation and generates an image based on the radiation; and a radiography system.

19. A method for operating an information processing device that calculates bone density using an image generated based on radiation, To acquire patient information entered via the control unit, To acquire the image taken in accordance with the aforementioned patient information, Extracting the bone density analysis region from the acquired image, A method for operating an information processing device, which includes: when the acquired patient information corresponds to stored patient information relating to a patient having the bone portion linked to information about the bone portion for which bone density was previously calculated, determining a target area for calculating bone density in the extracted bone density analysis area based on the bone density analysis area linked to the stored patient information and the information about the bone portion for which bone density was calculated.

20. A program that, when executed by a computer, causes the computer to perform each step of the operation method of the information processing device described in claim 19.

Images