Information processing device, radiation imaging system, information processing method, and program

Abstract

To provide a bone density measuring device capable of measuring accurate bone density and reducing a risk of outputting inaccurate bone density.SOLUTION: A bone density measuring device includes: a data acquisition unit for acquiring a bone density calibration value acquired by using first subject data obtained by imaging a first subject having predetermined bone density by a first imaging condition and first correction data obtained by imaging without arrangement of the first subject by the first imaging condition, second subject data obtained by imaging a second subject that is a bone density measurement target by a second imaging condition, and second correction data obtained by imaging without arrangement of the second subject by the second imaging condition; a generation unit for generating a bone image using the second subject data and the second correction data; and a bone density acquisition unit for acquiring bone density of the second subject using the bone image and the bone density calibration value.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to an information processing device, a radiation imaging system, an information processing method, and a program. [Background technology]

[0002] It is known that bone mineral density in bones is measured for the diagnosis of osteoporosis, etc. Dual-energy X-ray Absorptiometry (DXA) is used as a method for measuring bone mineral density. The DXA method is a technique for measuring bone density from the difference in X-ray absorption coefficient between soft tissue and bone tissue using two X-rays with different energy distributions. Conventionally, bone mineral density measuring devices using the DXA method used a line sensor to obtain data on the irradiated X-rays line by line. Therefore, conventional bone mineral density measuring devices using the DXA method required a long time for one imaging session, which was a burden on the patient.

[0003] In response to this, digital image diagnosis using X-ray images taken with a flat panel sensor (FPDe: Flat Panel Detector) has become widespread in recent years and is also being applied to bone density measurement. When taking bone density images using an FPDe, X-rays are irradiated onto the entire surface of the FPDe (cone beam imaging) to obtain an image, so each image is taken in a short time, making it possible to take images without placing a burden on the patient.

[0004] To calculate bone density from images acquired by FPDe, calibration is required between the value calculated from the output value of FPDe and the actual bone density value. By capturing an image of a bone density calibration phantom (Quality Control (QC) phantom) with known bone density, the bone density calibration value used for bone density calibration can be obtained. By calibrating the value calculated from the output value of FPDe with the bone density calibration value, a more appropriate bone density can be obtained.

[0005] Various methods have been proposed to improve the accuracy of bone density measurement. For example, Patent Document 1 proposes a method using machine learning to correctly extract areas such as the lumbar vertebrae and femur that are targets of bone density measurement. Patent Document 2 also proposes a method of analyzing correction data to prevent a decrease in measurement accuracy due to aging of the FPDe.

[0006] One of the causes of the decrease in measurement accuracy is the mismatch between the imaging conditions when imaging the patient and the imaging conditions when imaging the QC phantom. The imaging conditions include the X-ray generation conditions as well as the geometric arrangement of the X-ray tube and the FPDe.

[0007] In the DXA method, in addition to the images of the subject taken using two different X-rays, it is necessary to use a gain correction image (correction data) obtained when X-rays are irradiated without the subject placed on the imaging device in order to correct the conversion coefficient when converting the dose to pixel values. Specifically, the transmittance of each pixel of the high-energy X-ray image is calculated from the pixel values ​​of the correction data obtained by high-energy imaging and the pixel values ​​of the X-ray image of the subject. Similarly, the transmittance of each pixel of the low-energy X-ray image is calculated from the pixel values ​​of the correction data obtained by low-energy imaging and the pixel values ​​of the X-ray image of the subject. After that, the X-ray absorption coefficients of the soft tissue and bone tissue are derived from the two transmittances to generate a DXA image. Conventionally, the correction data is obtained when the QC phantom is imaged.

[0008] Here, in order to calculate the accurate transmittance at each pixel, the dose distribution irradiated to the FPDe must be consistent between the time when the correction data is acquired and the time when the X-ray image of the subject is acquired. When using a device dedicated to bone density measurement using the conventional DXA method, the relative position of the tube and the X-ray detector does not change significantly for each shot. Therefore, it was possible to calculate the transmittance at each pixel of the patient's X-ray image using the correction data acquired at the time of shooting the QC phantom.

[0009] In contrast, when imaging using a general FPDe, the tube and FPDe may be moved, and the relative positions of the tube and FPDe may change. If the imaging conditions change when imaging a patient, the dose distribution irradiated to the FPDe may change from the dose distribution when the QC phantom is imaged (when the correction data is acquired), making it impossible to calculate accurate transmittance and therefore correct bone density. Therefore, it is important to perform imaging of the patient under imaging conditions close to those used when imaging the QC phantom, and the reproducibility of the imaging conditions affects the accuracy of bone density. [Prior art documents] [Patent documents]

[0010] [Patent Document 1] JP 2021-037164 A [Patent Document 2] JP 2018-192054 A [Patent Document 3] JP 2020-25780 A Summary of the Invention [Problem to be solved by the invention]

[0011] By using FPDe, bone density measurement can be performed even with general radiography equipment. When bone density measurement is performed with general radiography equipment, general radiography may be performed between QC phantom radiography and patient radiography. When general radiography is performed, the radiography is performed with the geometric arrangement for general radiography. Therefore, when performing patient radiography for bone density measurement after general radiography, it is necessary to rearrange the X-ray tube and FPDe so that the geometric arrangement is the same as when the QC phantom radiography was performed before the patient radiography. However, since many of these settings are performed manually, the geometric arrangement may differ between the patient radiography and the QC phantom radiography, which may result in inaccurate bone density calculations.

[0012] Therefore, in one embodiment of the present disclosure, an information processing device is provided that can measure accurate bone density and reduce the risk of outputting inaccurate bone density. [Means for solving the problem]

[0013] An information processing device according to one embodiment of the present disclosure includes a data acquisition unit that acquires first subject data obtained by photographing a first subject having a predetermined bone density under first shooting conditions and first correction data obtained by photographing the first subject under the first shooting conditions without placing the first subject, second subject data obtained by photographing a second subject that is a bone density measurement target under second shooting conditions, and second correction data obtained by photographing the second subject under the second shooting conditions without placing the second subject, a generation unit that generates a bone image using the second subject data and the second correction data, and a bone density acquisition unit that acquires the bone density of the second subject using the bone image and the bone density calibration value. Effect of the Invention

[0014] According to an embodiment of the present disclosure, it is possible to measure accurate bone density while reducing the risk of outputting inaccurate bone density. [Brief description of the drawings]

[0015] [Figure 1] FIG. 1 is a configuration diagram of an example of a bone density measuring device according to a first embodiment. [Diagram 2] 1 illustrates an example of a hardware configuration of a radiation imaging system and a bone density measuring apparatus according to a first embodiment. [Diagram 3] 2 is a flowchart of a series of processes according to the first embodiment. [Figure 4] FIG. 1 is a diagram for explaining a method for measuring bone density by a DXA method. [Diagram 5] 1 is a flowchart of a bone mineral density calibration value calculation process according to the first embodiment. [Figure 6] FIG. 4 is a diagram for explaining a bone mineral density calibration value calculation process according to the first embodiment. [Figure 7]4 is a diagram for explaining a difference in X-ray dose distribution according to the first embodiment. FIG. [Figure 8] FIG. 2 is a diagram for explaining an overlapping region of irradiation fields according to the first embodiment. [Figure 9] 11 is a flowchart of a bone density calculation process according to the second embodiment. [Figure 10] 13 is a flowchart of a series of processes according to the third embodiment. [Figure 11] FIG. 11 is a configuration diagram of an example of a bone density measuring device according to a fourth embodiment. [Figure 12] 13 is a flowchart of a bone density calculation process according to the fourth embodiment. [Figure 13] FIG. 13 is a diagram for explaining a learning process according to the fourth embodiment. [Figure 14] FIG. 11 is a configuration diagram of an example of a bone density measuring device according to a fifth embodiment. [Figure 15] 13 shows an example of an output message according to the fifth embodiment. [Figure 16] 13 is a flowchart of a bone mineral density calculation process according to the sixth embodiment. [Figure 17] FIG. 13 is a configuration diagram of an example of a bone density measuring device according to Example 7. [Figure 18] 13 is a flowchart of an output control process according to a seventh embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] Hereinafter, exemplary embodiments for carrying out the present disclosure will be described in detail with reference to the drawings. However, the dimensions, materials, shapes, and relative positions of components described in the following embodiments are arbitrary and can be changed according to the configuration of the device to which the present disclosure is applied or various conditions. In addition, the same reference numerals are used in the drawings to indicate elements that are the same or functionally similar.

[0017] In the following embodiments, a radiography system using X-rays as an example of radiation will be described, but the radiography system according to the present disclosure may use other types of radiation. Here, the term radiation includes electromagnetic radiation such as X-rays and gamma rays, as well as particle radiation such as alpha rays, beta rays, particle beams, proton beams, heavy ion beams, and meson beams.

[0018] In the following, the machine learning model refers to a learning model based on a machine learning algorithm. Specific examples of machine learning algorithms include nearest neighbor methods, naive Bayes methods, decision trees, and support vector machines. Neural networks and deep learning may also be used. Appropriately, any of the above algorithms that can be used can be used and applied to the following embodiments. Furthermore, learning data (teacher data) refers to a data set used for learning the machine learning model, and is composed of a pair of input data input to the machine learning model and correct answer data that is the correct answer for the output result of the machine learning model.

[0019] A trained model refers to a model that has been trained in advance using appropriate training data for a machine learning model that follows any machine learning algorithm, such as deep learning. However, although a trained model has been obtained by training in advance using appropriate training data, this does not mean that no further training is performed, and additional training can be performed. Additional training can also be performed after the device is installed at the site of use.

[0020] Example 1 Hereinafter, with reference to Figs. 1 to 8, a bone density measuring device (information processing device) for measuring bone density from an image obtained using X-rays according to the first embodiment of the present disclosure, a radiography system, and a control method (information processing method) for the bone density measuring device will be described. As described above, when correction data (gain correction image) for an X-ray image of a patient is acquired during QC phantom imaging, if the imaging conditions during patient imaging are different from those during QC phantom imaging, the dose distribution differs between the correction data and the X-ray image of the patient. Therefore, an accurate transmittance cannot be calculated, and the correct bone density may not be calculated. In contrast, in this embodiment, correction data for the X-ray image of the patient is acquired at the timing of imaging the patient. As a result, the imaging conditions during patient imaging are the same as those during correction data acquisition, so that an accurate transmittance can be calculated, and a more correct bone density can be calculated.

[0021] In addition, the bone density calibration value obtained by imaging the QC phantom is a calibration value when an X-ray image of a patient is acquired under the same imaging conditions as when the QC phantom is imaged, and if the imaging conditions change when the patient is imaged, the correct bone density may not be calculated even if the bone density calibration value is used. In contrast, in this embodiment, the dose distribution of the image obtained when the patient is imaged is corrected based on the dose distribution of the image obtained when the QC phantom is imaged. As a result, the dose distribution of the image obtained when the patient is imaged becomes the same as the dose distribution of the image obtained when the QC phantom is imaged, so that accurate bone density can be calculated.

[0022] 1 is a configuration diagram of an example of a bone density measuring device according to the present embodiment. The bone density measuring device includes a first object data acquiring unit 101, a first correction data acquiring unit 102, a second object data acquiring unit 103, and a second correction data acquiring unit 104. The bone density measuring device also includes a bone image generating unit 105, a bone density calibration value calculating unit 106, a feature amount analyzing unit 107, and a bone density calculating unit 108.

[0023] The first object data acquisition unit 101 acquires data obtained by imaging a QC phantom under first imaging conditions of high energy and low energy as first object data. The first correction data acquisition unit 102 acquires data obtained by imaging under the first imaging conditions in the absence of an object as first correction data.

[0024] The second subject data acquisition unit 103 acquires data obtained by imaging a subject (patient) of bone density measurement under second imaging conditions of high energy and low energy as second subject data. The second correction data acquisition unit 104 acquires data obtained by imaging under the second imaging conditions in the absence of a subject as second correction data.

[0025] The bone image generating unit 105 generates a first bone image using the first object data and the first correction data. The bone image generating unit 105 generates a second bone image using the second object data and the second correction data. When generating the second bone image using the second object data and the second correction data, the bone image generating unit 105 corrects the dose distribution using a dose distribution correction coefficient obtained by the feature amount analyzing unit 107. The method of generating the bone image and the correction of the dose distribution will be described in detail later.

[0026] The bone mineral density calibration value calculation unit 106 calculates a bone mineral density calibration value using the first bone image obtained using the first object data and the first correction data and the known bone mineral density (true bone mineral density) of the QC phantom. The calculation method of the bone mineral density calibration value will be described later in detail.

[0027] The feature amount analysis unit 107 analyzes the first correction data and the second correction data to acquire image feature amounts. In this embodiment, the feature amount analysis unit 107 analyzes the first correction data to calculate a first dose distribution, and analyzes the second correction data to calculate a second dose distribution. The feature amount analysis unit 107 also acquires a dose distribution correction coefficient that converts the second dose distribution into the first dose distribution. Details of a method for estimating a dose distribution and a method for calculating a dose distribution correction coefficient will be described later.

[0028] The bone density calculation unit 108 calculates the bone density of the subject whose bone density is to be measured, using the second bone image and the bone density calibration value. The method of calculating the bone density will be described in detail later.

[0029] 2 shows an example of the hardware configuration of the radiography system and bone density measuring device according to the present embodiment. The radiography system according to the present embodiment includes an FPDe 202, an X-ray generator 203, a signal line 204, a display unit 205, a storage unit 206, a network I / F (interface) 207, an ion chamber 210, and an X-ray control unit 211. The radiography system also includes a control PC (Personal Computer) 201, an input unit 208, a display unit 209, an analysis PC 212, a display unit 219, and an input unit 218. Here, the bone density measuring device according to the present embodiment may include the control PC 201 and the analysis PC 212. The control PC 201 and the analysis PC 212 may be one PC, and in this case, the bone density measuring device may be configured by the one PC.

[0030] The control PC 201 and the FPDe 202 are connected via a signal line 204 such as a gigabit Ethernet network cable. The signal line 204 is not limited to a gigabit Ethernet cable, and may be, for example, a CAN (Controller Area Network) cable or an optical fiber cable. An X-ray generator 203, a display unit 205, a storage unit 206, a network I / F 207, an ion chamber 210, an X-ray control unit 211, and an analysis PC 212 are connected to the signal line 204.

[0031] The FPDe 202 functions as an example of a radiation detector that detects X-rays and generates an X-ray image. The FPDe 202 may be an indirect conversion type detector that converts radiation 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 may be a direct conversion type detector that directly converts incident radiation into an electrical signal.

[0032] The X-ray generating device 203 generates X-rays under the control of the X-ray control unit 211. The X-ray generating device 203 can be configured using a tube that generates radiation. The X-ray control unit 211 controls the X-ray generating device 203 based on imaging conditions set according to instructions from an operator, etc. The X-ray control unit 211 may be configured using circuits such as a dedicated PC or a control drive.

[0033] The ion chamber 210 outputs a signal corresponding to the intensity of the transmitted X-rays. By using the ion chamber 210, the dose of the irradiated X-rays can be monitored.

[0034] The network I / F 207 functions as an interface when the X-ray imaging system is connected to an external network, and can function as a communication unit of the X-ray imaging system. The network I / F 207 may be configured as an interface for a wired network, may be configured as an interface for a wireless network, or may be configured as both of these interfaces.

[0035] The storage unit 206 may be configured with any storage medium, for example, an optical disk or a solid state drive (SSD). The storage unit 206 can store various information related to imaging, patient information, various acquired images, acquired analysis results, and the like.

[0036] The display units 205, 209, and 219 may be configured using any monitor, and can display patient information, imaging information, various acquired images, analysis results such as bone density, and the like, under the control of the control PC 201 and the analysis PC 212. The control PC 201 and the analysis PC 212 can function as an example of a display control unit that controls the display by the display units 205, 209, and 219.

[0037] The input units 208 and 218 include input devices such as a keyboard, a mouse, a trackball, and a touch panel, and can input instructions to the control PC 201 and the analytical PC 212 by being operated by an operator.

[0038] The control PC 201 is provided with, for example, a CPU (Central Processing Unit) 2012, a RAM (Random Access Memory) 2013, a ROM (Read Only Memory) 2014, and a storage unit 2015. These components can be connected to one another via a bus 2011. In addition, an input unit 208 is connected to the control PC 201 using, for example, USB or PS / 2, and a display unit 209 is connected to the control PC 201 using, for example, DisplayPort or DVI.

[0039] The control PC 201 can function as a control device that controls the imaging system, and can send commands to the FPDe 202, the display unit 205, and the like. In the control PC 201, the processing contents for each imaging mode are stored in the storage unit 2015 as a software module, and are read into the RAM 2013 by an instruction means (not shown) and executed. Images acquired by the FPDe 202 and processed images can be sent to the storage unit 2015 in the control PC 201 or the storage unit 206 outside the control PC 201 and stored therein. The storage unit 2015 may be configured with any storage medium, and may be configured using, for example, an optical disk or an SSD.

[0040] The analysis PC 212 is provided with, for example, a CPU (Central Processing Unit) 2122, a RAM (Random Access Memory) 2123, a ROM (Read Only Memory) 2124, and a storage unit 2125. These components can be connected to one another via a bus 2121. In addition, an input unit 218 is connected to the analysis PC 212 using, for example, a USB or PS / 2, and a display unit 219 is connected to the analysis PC 212 using, for example, a DisplayPort or DVI.

[0041] The analytical PC 212 can function as an analysis device that analyzes data such as X-ray images. In the analytical PC 212, the processing contents for calculating bone density and the contents for creating a bone density analysis report are stored in a storage unit 2125 as software modules, and are read into a RAM 2124 and executed by an instruction means (not shown). The processed images and reports can be sent to a storage unit 206 outside the analytical PC 212 and stored therein. The storage unit 2125 may be configured with any storage medium, and may be configured using, for example, an optical disk or an SSD.

[0042] 1 are stored in a storage unit 2015 as software modules of a control PC 201. In addition, a bone image generation unit 105, a bone density calibration value calculation unit 106, a feature amount analysis unit 107, and a bone density calculation unit 108 are stored in a storage unit 2125 as software modules of an analysis PC 212.

[0043] When the control PC 201 and the analysis PC 212 are configured as one PC, these components may be implemented as software modules in the one PC. Also, each software module may be implemented as an integrated circuit performing a specific function, such as an ASIC, such as a dedicated image processing board. The configurations of the control PC 201 and the analysis PC 212 are not limited to the above configurations, and an optimal configuration may be adopted depending on the purpose.

[0044] In addition, although a CPU is used as the processor in this embodiment, the processor is not limited to this. The processor may be, for example, a micro processing unit (MPU), a graphical processing unit (GPU), a field-programmable gate array (FPGA), etc. Furthermore, the internal configuration of the control PC 201 and the analysis PC 212 may include a graphic control unit such as a GPU, a communication unit such as a network card, an input / output control unit such as a keyboard, a display, or a touch panel, etc.

[0045] In this embodiment, the control PC 201, the display unit 209, and the input unit 208 are described as separate devices, but they may be configured as an integrated unit. Similarly, the analysis PC 212, the display unit 219, and the input unit 218 may be configured as an integrated unit. For example, the display unit 209 may be configured as a touch panel display, and in this case, the display unit 209 can also be used as the input unit 208.

[0046] A series of processes according to this embodiment in the radiation imaging system and bone density measuring device having the above-mentioned configuration will be described with reference to Fig. 3(a) to Fig. 8. Hereinafter, the flow of the series of processes according to this embodiment will be described with reference to Fig. 3(a) and Fig. 3(b). Fig. 3(a) and Fig. 3(b) are flowcharts of the series of processes according to this embodiment.

[0047] First, a series of processes for calculating a bone density calibration value will be described with reference to FIG. 3(a). In step S301, the first object data acquisition unit 101 acquires a QC phantom image obtained by photographing a QC phantom as the first object data. Here, the QC phantom image is an example of calibration data used to calibrate bone density when calculating the bone density of a subject (patient) who is the subject of bone density measurement. In addition, in a bone density device using the DXA method, imaging is performed under two conditions, namely, imaging conditions in which X-rays are high energy and imaging conditions in which X-rays are low energy. Therefore, the first object data acquisition unit 101 acquires two images, a high energy QC phantom image and a low energy QC phantom image. Here, imaging under the high energy imaging condition and imaging under the low energy imaging condition are performed such that the conditions related to the energy of radiation are switched in a short period of time, so that imaging conditions related to the geometric arrangement of the tube, FPDe 202, etc. can be kept the same. For this reason, in the following, the imaging conditions for imaging the QC phantom are referred to as the first imaging conditions for high energy and low energy. The first imaging conditions include the arrangement conditions regarding the geometric arrangement of the tube, the FPDe202, etc.

[0048] Next, in step S302, the first correction data acquisition unit 102 acquires a first object-free image captured under the first imaging condition with the QC phantom removed as first correction data, and stores the first correction data in the memory unit 2015. The first correction data acquisition unit 102 may store the first correction data in the memory unit 2125 of the analysis PC 212. In a bone density measurement device using the DXA method, an object-free image is captured under the same imaging condition as when the QC phantom was captured. Therefore, the first correction data acquisition unit 102 acquires two images, a high-energy first object-free image and a low-energy first object-free image, captured under the first imaging condition of high energy and low energy.

[0049] In step S303, the bone image generating unit 105 generates a bone image (first bone image) using the high-energy QC phantom image, the low-energy QC phantom image, the high-energy first object-free image, and the low-energy first object-free image. The bone image is an image that represents the thickness of the bone.

[0050] A method for calculating a bone image will be described with reference to FIG. 4. FIG. 4 is a schematic diagram of X-rays passing through an object, which is composed of soft tissue 401 and bone 402. Here, the dose of incident X-rays 403 when irradiated with low-energy X-rays is defined as IL, and the dose of incident X-rays 403 when irradiated with high-energy X-rays is defined as IH. In addition, the dose of transmitted X-rays 404 when irradiated with low-energy X-rays is defined as IOL, and the dose of transmitted X-rays 404 when irradiated with high-energy X-rays is defined as IOH. In this case, the bone image dImg is expressed as in Equation 1.

[0051]

number

[0052] Next, in step S304, the bone mineral density calibration value calculation unit 106 calculates a bone mineral density calibration value and stores it in the storage unit 2125. The bone mineral density calibration value is a conversion formula that converts the bone mineral density calculated from the bone image into a more accurate bone mineral density.

[0053] Here, the calculation of the bone mineral density calibration value will be described with reference to Fig. 5 to Fig. 6(b). Fig. 5 is a flowchart of the bone mineral density calibration value calculation process according to this embodiment. Fig. 6(a) shows an example of a bone image, and a bone region and a base area extracted from the bone image. First, in step S501, the bone mineral density calibration value calculation unit 106 classifies vertebral bodies from the bone image 601. In this embodiment, the bone mineral density calibration value calculation unit 106 classifies the vertebral bodies into three regions, a bone region 602, a bone region 603, and a bone region 604, and extracts each region.

[0054] The bone region can be extracted using a rule-based or machine learning method. An example of a rule-based method is a method using Otsu's binarization. Since Otsu's binarization is a method that maximizes inter-class variance, it is possible to divide the bone region from other regions. The bone density calibration value calculation unit 106 can extract bone region 602, bone region 603, and bone region 604 from geometric information of the bone image divided by Otsu's binarization and other regions.

[0055] As a method using machine learning, for example, semantic segmentation may be performed. In this case, a trained model that recognizes bone regions by vertebrae can be generated by training a U-net type machine learning model or the like using training data in which bone images are input data and data classified by vertebrae is used as correct answer data (output data). The bone density calibration value calculation unit 106 can obtain bone regions 602, 603, and 604 output from the trained model by using the bone image 601 as an input of the trained model. In addition, the rule base and machine learning may be used in combination. The bone density calibration value calculation unit 106 can output a bone region mask in which the bone region is set to 1 and the other regions are set to 0 as an extraction result of the bone region.

[0056] After recognizing the bone region, in step S502, the bone density calibration value calculation unit 106 calculates the average value boVal of the pixel values ​​of the bone region. The bone density calibration value calculation unit 106 calculates the average value boVal of the pixel values ​​of the bone region using the bone image dImg according to Equation 2.

number

[0057] Next, in step S503, the bone mineral density calibration value calculation unit 106 extracts base areas 605, 606, and 607 corresponding to the classified bone regions 602, 603, and 604, respectively, from the bone image 601. The base areas are regions of soft tissue. The base areas can be extracted using a rule-based or machine learning method.

[0058] As a method using the rule base, for example, there is a method of extracting a fixed-size area at a certain distance from the extracted bone area as the base area. As a method using machine learning, for example, the base area may be extracted in the same manner as the extraction of the bone area. More specifically, a trained model that recognizes the base area for each vertebra can be generated by training a U-net type machine learning model using training data in which a bone image is used as input data and data in which the base area is classified for each vertebra is used as correct answer data. The bone density calibration value calculation unit 106 can obtain the base areas 605, 606, and 607 output from the trained model by using the bone image 601 as an input of the trained model. In addition, the rule base and machine learning may be used in combination. The bone density calibration value calculation unit 106 can output a base area mask in which the base area is set to 1 and the other areas are set to 0 as the base area extraction result.

[0059] After recognizing the base area, in step S504, the bone mineral density calibration value calculation unit 106 calculates the average value baVal of the pixel values ​​of the base area. The bone mineral density calibration value calculation unit 106 calculates the average value baVal of the pixel values ​​of the base area according to Equation 3 using the bone image dImg.

number

[0060] After the average pixel value of the bone region and the average pixel value of the base area are calculated, in step S505, the bone density calibration value calculation unit 106 removes the influence of the soft tissue by taking the difference between them as shown in Equation 4, and calculates the bone density densF obtained from the output value of FPDe. In this embodiment, since there are three bone regions, three bone densities are calculated. There are as many i as the number of extracted bone regions.

number

[0061] In step S506, the bone mineral density calibration value calculation unit 106 compares the bone mineral density denseF obtained from the output value of the FPDe202 with the true bone mineral density defined in the QC phantom, and calculates a calibration value for converting the bone mineral density obtained from the output value of the FPDe202 to the true bone mineral density. Since the bone mineral density of each vertebral body of the QC phantom is known, the bone mineral density of each vertebral body of the QC phantom is set as the true bone mineral density (predetermined bone mineral density). The bone mineral density calibration value calculation unit 106 performs least squares approximation using the bone mineral density denseF of each vertebral body obtained from the output value of the FPDe202 and the true bone mineral density of each vertebral body, and calculates an approximation formula 608 as shown in FIG. 6(b). The bone mineral density calibration value calculation unit 106 sets the coefficient k0, which is the intercept of the approximation formula 608, and the coefficient k1, which is the slope, as the bone mineral density calibration value. In this embodiment, the bone mineral density is calculated by removing the influence of soft tissue, but the bone mineral density may be calculated only from the pixel value of the bone region.

[0062] Next, the process of calculating the bone density of a patient will be described with reference to FIG. 3(b). In step S305, the second object data acquisition unit 103 acquires an object image obtained by imaging the object (patient) to be measured for bone density as second object data. In a bone density device using the DXA method, imaging is performed under two conditions, that is, imaging conditions in which X-rays are high energy and imaging conditions in which X-rays are low energy. Therefore, the second object data acquisition unit 103 acquires two images, an object image of high energy and an object image of low energy. Here, imaging under the high energy imaging condition and imaging under the low energy imaging condition are performed such that the conditions related to the energy of the radiation are switched in a short period of time, so that imaging conditions related to the geometric arrangement of the tube, FPDe 202, etc. can be kept the same. Therefore, hereinafter, the imaging conditions when imaging the patient are referred to as second imaging conditions of high energy and low energy. The second imaging conditions include arrangement conditions related to the geometric arrangement of the tube, FPDe 202, etc.

[0063] Next, in step S306, the second correction data acquisition unit 104 acquires a second subject-free image captured under the second imaging conditions without the subject as the second correction data. In a bone density measurement device using the DXA method, the subject-free image is captured under the same imaging conditions as when the subject (patient) was imaged. Therefore, the second correction data acquisition unit 104 acquires two images, a high-energy second subject-free image and a low-energy second subject-free image, captured under the second imaging conditions of high energy and low energy.

[0064] In step S307, the feature amount analysis unit 107 calculates a dose distribution correction coefficient for converting a dose distribution (second dose distribution) under the second imaging condition into a dose distribution (first dose distribution) under the first imaging condition. More specifically, the feature amount analysis unit 107 reads the first object-free image of high energy and the first object-free image of low energy from the storage unit 2015 or the storage unit 2125, and estimates (calculates) a first dose distribution (first pixel value distribution) under the first imaging condition using these. In addition, the feature amount analysis unit 107 estimates (calculates) a second dose distribution (second pixel value distribution) under the second imaging condition using the second object-free image of high energy and the second object-free image of low energy. Thereafter, the feature amount analysis unit 107 calculates a dose distribution correction coefficient for converting the second dose distribution into the first dose distribution.

[0065] In order to improve the accuracy of bone density measurement, it is desirable that the first and second imaging conditions are the same, but if general imaging is performed between the first and second imaging, the geometric arrangement of the tube and FPDe202 is set for general imaging. Therefore, even if the geometric arrangement of the tube and FPDe202 is returned to the geometric arrangement in the first imaging condition before patient imaging under the second imaging condition, it is difficult to completely match these geometric arrangements when performing the arrangement manually. For this reason, a difference in imaging conditions may occur between patient imaging under the second imaging condition and QC phantom imaging under the first imaging condition.

[0066] The feature amount analysis unit 107 estimates the difference in the imaging conditions from the distribution of pixel values ​​of the images obtained by each imaging, and calculates a correction coefficient for correcting the difference in the imaging conditions. The X-rays incident on the subject vary depending on the positions and distances between the X-ray tube and the FPDe 202. When a difference occurs in the geometric arrangement conditions, the dose distribution incident on the subject changes. Figures 7(a) and 7(b) are diagrams for explaining the difference in the X-ray dose distribution based on the geometric arrangement of the X-ray tube and the FPDe 202. Figure 7(a) shows an example of the X-ray dose distribution in imaging under a first imaging condition, and Figure 7(b) shows an example of the X-ray dose distribution in imaging under a second imaging condition.

[0067] 7(a), X-rays 702 irradiated from an X-ray tube 701 of the X-ray generator 203 under the first imaging condition reach the FPDe 202. Since the X-rays spread radially and attenuate in proportion to the square of the distance from the X-ray tube 701 to the FPDe 202, the dose distribution 703 of the X-rays on the FPDe 202 is distributed radially.

[0068] In contrast, in the example shown in FIG. 7(b), in the second imaging condition, the SID (Source Image Receptor Distance) is changed from the first imaging condition, and the X-ray tube 701 is shifted up, down, left, and right with respect to the FPDe 202. Here, the SID indicates the distance between the X-ray tube 701 and the FPDe 202. The attenuation of the X-rays 704 irradiated from the X-ray tube 701 and reaching the FPDe 202 is different from that of the X-rays 702 during imaging under the first imaging condition, based on the difference in the geometric arrangement conditions between the X-ray tube 701 and the FPDe 202. Therefore, the dose distribution of the X-rays under the second imaging condition becomes a dose distribution 705 different from the dose distribution 703 under the first imaging condition.

[0069] The feature amount analysis unit 107 calculates a dose distribution correction coefficient for correcting the influence on bone density based on the difference in the geometric arrangement condition by correcting the difference in the dose distribution of the X-ray. Since the dose distribution of the X-ray spreads radially, the feature amount analysis unit 107 fits the dose distribution of the image obtained under each imaging condition as a quadratic surface to generate a fitting image. After that, the feature amount analysis unit 107 converts the dose distribution of the X-ray under the second imaging condition to the dose distribution of the X-ray under the first imaging condition using the generated fitting image. Note that the fitting of the dose distribution of the X-ray may be performed by any known method, for example, the least squares method may be used. Here, if the fitting image is h′(x, y), the fitting image h′(x, y) can be expressed as in Equation 5.

number

[0070] Here, if the subject-free image is h(x, y), the fitting image h'(x, y) can be obtained by minimizing Equation 6.

number

[0071] Here, the image obtained by fitting the first object-free image of high energy is defined as a fitting image h1H, and the image obtained by fitting the first object-free image of low energy is defined as a fitting image h1L. The image obtained by fitting the second object-free image of high energy is defined as a fitting image h2H, and the image obtained by fitting the first object-free image of low energy is defined as a fitting image h2L. The feature amount analysis unit 107 divides the fitting image under the first imaging condition by the fitting image under the second imaging condition to calculate a dose distribution correction coefficient. The low-energy dose distribution correction coefficient CorL and the high-energy dose distribution correction coefficient CorH can be expressed as in Equation 7.

number

[0072] In step S308, the bone image generating unit 105 generates a bone image (second bone image) from the subject image obtained by imaging the patient. The bone image generating unit 105 generates the bone image using the high-energy subject image, the low-energy subject image, the second high-energy subject-free image, the second low-energy subject-free image, the low-energy dose distribution correction coefficient, and the high-energy dose distribution correction coefficient. The method of calculating the bone image may be substantially similar to the method of calculating the bone image in step S303, but dose distribution correction is added to Equation 1. Regarding the generation of the bone image from the subject image, the equation with dose distribution correction added can be expressed as Equation 8.

number

[0073] Finally, in step S309, the bone mineral density calculation unit 108 reads the bone mineral density calibration value from the storage unit 2125, and calculates the bone mineral density using the bone image generated by correcting the dose distribution in step S308 and the bone mineral density calibration value. The method of calculating the bone mineral density before applying the bone mineral density calibration value is the same as the procedure of calculating the bone mineral density using the bone image of the QC phantom. More specifically, the bone mineral density calculation unit 108 performs the same process as steps S501 to S505 on the bone image generated by correcting the dose distribution in step S308. After that, the bone mineral density calculation unit 108 calculates the final bone mineral density densR according to Equation 9 using the bone mineral density calibration values ​​k0 and k1 for the bone mineral density obtained from the output value of FPDe202 obtained by the same process as step S505.

number

[0074] As described above, the radiation imaging system according to this embodiment includes the X-ray generating device 203, the FPDe 202, the control PC 201, and the analysis PC 212. The X-ray generating device 203 functions as an example of a radiation generating device that generates radiation. The FPDe 202 functions as an example of a radiation detector that detects radiation and generates a radiation image. The control PC 201 and the analysis PC 212 function as an example of a bone density measuring device (information processing device) that measures the bone density of a subject.

[0075] The bone density measuring device includes a first subject data acquisition unit 101, a first correction data acquisition unit 102, a second subject data acquisition unit 103, a second correction data acquisition unit 104, a bone image generation unit 105, a bone density calibration value calculation unit 106, and a bone density calculation unit 108.

[0076] The first subject data acquisition unit 101 functions as an example of a data acquisition unit that acquires first subject data obtained by imaging a QC phantom having true bone density, which is an example of a first subject having a predetermined bone density, under a first imaging condition. The first correction data acquisition unit 102 functions as an example of a data acquisition unit that acquires first correction data obtained by imaging a first subject under a first imaging condition without placing the first subject. The second subject data acquisition unit 103 functions as an example of a data acquisition unit that acquires second subject data obtained by imaging a second subject to be measured for bone density under a second imaging condition. The second correction data acquisition unit 104 functions as an example of a data acquisition unit that acquires second correction data obtained by imaging a second subject under a second imaging condition without placing the second subject. The first subject data acquisition unit 101, the first correction data acquisition unit 102, the second subject data acquisition unit 103, and the second correction data acquisition unit 104 can be configured as one data acquisition unit. Moreover, the first imaging condition can include a first imaging condition of high energy and low energy, and the second imaging condition can include a second imaging condition of high energy and low energy.

[0077] The bone image generating unit 105 functions as an example of a generating unit that generates a first bone image using the first subject data and the first correction data, and generates a second bone image using the second subject data and the second correction data. The bone density calibration value calculating unit 106 functions as an example of a calibration value acquiring unit that acquires a bone density calibration value using the first bone image and a predetermined bone density of the first subject. The bone density calculating unit 108 functions as an example of a bone density acquiring unit that acquires the bone density of the second subject using the second bone image and the bone density calibration value.

[0078] With the above-mentioned configuration, the information processing device according to the present embodiment acquires second correction data corresponding to a radiation image of a subject to be bone density measured at the timing of imaging the subject. As a result, the imaging conditions when the subject is imaged are the same as the imaging conditions when the second correction data is imaged, so that an accurate transmittance can be acquired, and more accurate bone density can be obtained. In addition, the risk of outputting an inaccurate bone density can be reduced.

[0079] The information processing device according to this embodiment may further include a feature amount analysis unit 107 that functions as an example of an analysis unit that analyzes the first and second corrected data and acquires feature amounts of the first and second corrected data. In this case, the bone image generation unit 105 may generate a second bone image using the second subject data, the second corrected data, and the feature amounts.

[0080] More specifically, the feature amount analysis unit 107 can acquire a pixel value distribution (dose distribution) as the feature amount. Here, the pixel value distribution corresponds to the dose distribution. In this case, the bone image generation unit 105 can generate a second bone image using a correction coefficient for converting the pixel value distribution of the second correction data acquired using the feature amount into the pixel value distribution of the first correction data, the second subject data, and the second correction data. Here, the low-energy dose distribution correction coefficient CorL and the high-energy dose distribution correction coefficient CorH are examples of the correction coefficients.

[0081] According to this configuration, a bone image of a subject, which is a bone density measurement target, is generated with a corrected dose distribution based on the dose distribution when a bone density calibration image (QC phantom image) is captured. This makes the dose distribution of the image obtained when the subject is captured similar to the dose distribution of the image obtained when the calibration image is captured, so that the bone density can be more appropriately calibrated and more accurate bone density can be obtained.

[0082] In this embodiment, the bone image generating unit 105 uses the dose distribution correction coefficients CorL and CorH, which are correction coefficients, in Equation 8 for generating the second bone image. Alternatively, the bone image generating unit 105 may correct the second correction data using the correction coefficients, and generate the second bone image using the corrected second correction data and the second subject data.

[0083] In this embodiment, the first subject-free image is acquired after the QC phantom image is acquired, but the first subject-free image may be acquired prior to the QC phantom image. Similarly, the second subject-free image is acquired after the subject image of the bone density measurement target is acquired, but the second subject-free image may be acquired prior to the subject image.

[0084] Furthermore, in this embodiment, the bone density calibration value calculation unit 106 acquires a bone density calibration value using the first bone image and a predetermined bone density of the first subject. In contrast, the bone density calibration value may be calculated in advance and stored in the storage units 206, 2015, 2125 or an external storage device. In this case, the bone density calibration value calculation unit 106 may function as an example of a calibration value acquisition unit that acquires the bone density calibration value from these storage units. In addition, in this case, the calibration value acquisition unit may be included in a data acquisition unit that acquires the second subject data, the second correction data, and the like.

[0085] In this embodiment, the feature amount analysis unit 107 performs linear fitting of the dose distribution with the dose distribution as a quadratic surface, but the degree of the surface used in fitting does not have to be quadratic. Therefore, the feature amount analysis unit 107 may obtain the pixel value distribution using multidimensional surface fitting. The feature amount analysis unit 107 may also perform nonlinear fitting. When performing nonlinear fitting, fitting parameters may be optimized by iterative processing or the like.

[0086] Furthermore, in this embodiment, the dose distribution correction coefficients are acquired for each energy type, high energy and low energy, but they may be acquired for only one of them. Also, the dose distribution correction coefficients may be acquired for each energy type, and one of the acquired dose distribution correction coefficients may be used. In this case, the dose distribution correction coefficients with higher fitting accuracy may be selected, for example, by selecting the one with the smaller error between the pixel values ​​of the original image on which fitting is performed and the pixel values ​​of the fitting image from among the acquired dose distribution correction coefficients. In this way, when one of the dose distribution correction coefficients acquired for each energy type is used, both energies are corrected with the same coefficient.

[0087] In this embodiment, the image feature is acquired to correct the dose distribution, but the dose distribution correction may be switched ON / OFF depending on the accuracy of acquiring the image feature. For example, if the error between the pixels of the original image on which fitting is performed and the pixels of the fitting image is equal to or greater than a threshold, the bone image may be output without correcting the dose distribution. Therefore, if the fitting error of the curved surface fitting is equal to or greater than a threshold, the bone image generating unit 105 may not use the feature to generate the second bone image. In this case, the bone image is generated using a correction coefficient with unfavorable fitting accuracy, thereby preventing the accuracy of the bone density measurement from decreasing.

[0088] In this embodiment, when the X-ray irradiation field is narrowed, the dose distribution may be corrected only in the irradiation field area. Figure 8 shows an example of the deviation of the irradiation field when the X-ray tube moves up and down and left and right with respect to the FPDe 202 between the first imaging condition and the second imaging condition. In Figure 8, the irradiation field under the first imaging condition is the irradiation field area 801, and the irradiation field under the second imaging condition is the irradiation field area 802.

[0089] If an area where radiation is blocked by a collimator is included in either the exposure field area 801 or the exposure field area 802, the dose distribution cannot be obtained accurately. Therefore, the influence of the collimator can be suppressed by performing dose distribution correction and bone density acquisition in an overlap area 803 where there is no area where radiation is blocked by a collimator in both the exposure field area 801 and the exposure field area 802. Therefore, the feature amount analysis unit 107 may analyze the first corrected data and the second corrected data in the exposure field area of ​​the second corrected data and in the exposure field area of ​​the first corrected data.

[0090] The method of recognizing the irradiation field can use a rule-based or machine learning method. As an example of a rule-based method, a method using a Hough transform can be mentioned. By performing a Hough transform, it is possible to extract straight line components in an image. The bone image generating unit 105 can recognize the irradiation field by narrowing down the straight line components in the image while taking into account pixel values ​​and geometric arrangements related to the QC phantom image, the subject image, and the subject-free image as conditions.

[0091] As a method using machine learning, for example, semantic segmentation may be performed. More specifically, a QC phantom image, a subject image, and a subject-free image are used as input data, and a U-net type machine learning model or the like is trained using training data in which the irradiation field area is used as correct answer data, thereby generating a trained model that recognizes the irradiation field. In addition, a rule base and machine learning may be used in combination. For example, Patent Document 3 discloses a method for improving the accuracy of the recognition of the irradiation field by recognizing the approximate irradiation field area using machine learning and then processing it using a rule base.

[0092] Example 2 Next, a bone density measuring device according to a second embodiment of the present disclosure will be described with reference to Fig. 9. Note that the radiation imaging system and the bone density measuring device according to this embodiment have the same configuration as the radiation imaging system and the bone density measuring device according to the first embodiment, and therefore the same reference numerals are used and the description thereof will be omitted.

[0093] In the first embodiment, in order to avoid the X-ray image of the subject (patient) and the second correction data corresponding to the X-ray image being affected by the difference in dose distribution based on the difference between the first and second imaging conditions, the second correction data captured under the second imaging condition is acquired. Here, for the subject-free image that is the correction data, usually, a plurality of subject-free images under the same conditions are acquired and averaged to prevent the S / N ratio from deteriorating in the division of Equation 1 when generating the bone image. In this regard, for the first subject-free image acquired during the QC phantom imaging under the first imaging condition, since the patient is not placed, many images can be acquired. On the other hand, when the patient is imaged under the second imaging condition, the number of second subject-free images acquired is less than the number of first subject-free images acquired during the QC phantom imaging, since it affects the workflow.

[0094] Therefore, the feature amount analysis unit 107 according to this embodiment estimates the difference between the number of the first object-free images and the number of the second object-free images, and generates a filter for adjusting the difference in the amount of noise caused by the difference in the number of images. Furthermore, when generating a bone image using the second object data, which is the object image, and the second correction data, the bone image generation unit 105 adjusts the amount of noise using the filter obtained by the feature amount analysis unit 107.

[0095] Hereinafter, a series of processes according to this embodiment will be described with reference to Fig. 9, showing the flow of bone mineral density calculation processing. Fig. 9 is a flowchart of the bone mineral density calculation processing according to this embodiment. Note that the series of processes for calculating bone mineral density calibration values ​​according to this embodiment are similar to the series of processes for calculating bone mineral density calibration values ​​shown in Fig. 3(a) according to the first embodiment, and therefore the description thereof will be omitted. Also, steps S905 and S906 in Fig. 9 are similar to steps S305 and S306 in Fig. 3(b) according to the first embodiment, and therefore the description thereof will be omitted.

[0096] In step S907, the feature amount analysis unit 107 reads the high energy first object-free image and the low energy first object-free image from the memory unit 2015 or the memory unit 2125, and estimates a first amount of noise under the first shooting conditions. The feature amount analysis unit 107 also estimates a second amount of noise under the second shooting conditions from the high energy second object-free image and the low energy second object-free image. Thereafter, the feature amount analysis unit 107 calculates a filter that adjusts the second amount of noise to the first amount of noise.

[0097] More specifically, the feature amount analysis unit 107 calculates the standard deviation in the irradiation field region from the first object-free image of high energy and the first object-free image of low energy, and sets the standard deviation as the noise amount N1. The region to be calculated may be further limited to the image center region in the irradiation field region or the region where the QC phantom is arranged. Similarly, the feature amount analysis unit 107 calculates the standard deviation in the irradiation field region from the second object-free image of high energy and the second object-free image of low energy, and sets the standard deviation as the noise amount N2.

[0098] When the number of the acquired high-energy first object-free images and the acquired low-energy first object-free images is m1, the noise amount in the first average image obtained by averaging the m1 images is 1 / √m1 before averaging. On the other hand, when the number of the acquired high-energy second object-free images and the acquired low-energy second object-free images is m2, the noise amount in the second average image obtained by averaging the m2 images is 1 / √m2 before averaging. At this time, the feature amount analysis unit 107 performs spatial filtering on the second average image to match the noise amount of the second average image to the noise amount of the first average image. When a one-dimensional moving average filter is applied as the spatial filtering, the feature amount analysis unit 107 calculates a filter coefficient f as shown in Equation 10. Here, n is the number of filter taps of the moving average and is an integer.

number

[0099] In step S908, the bone image generating unit 105 generates a bone image using the high energy object image, the low energy object image, the second high energy object-free image, the second low energy object-free image, and a moving average filter. The method of calculating the bone image may be substantially the same as the method of calculating the bone image in step S303, but the bone image generating unit 105 uses Equation 11 in which noise amount adjustment is added to Equation 1. The noise amount adjustment is performed by filtering using the filter coefficient calculated in step S907, and corresponds to the function G in Equation 11.

number

[0100] The bone mineral density calculation process in step S909 is similar to the bone mineral density calculation process in step S309 in the first embodiment, and therefore a description thereof will be omitted.

[0101] The feature amount analysis unit 107 according to this embodiment can calculate the amount of noise as the feature amount. In this case, the bone image generation unit 105 generates a second bone image using a noise adjustment filter for converting the amount of noise in the second corrected data calculated using the feature amount into the amount of noise in the first corrected data, the second subject data, and the second corrected data. Note that the noise adjustment filter may be, for example, a one-dimensional moving average filter, a two-dimensional moving average filter, or a low-pass filter such as a Gaussian filter.

[0102] According to this configuration, when generating a bone image of a subject for which bone density is to be calculated, bone image generating unit 105 adjusts the amount of noise in the second subject-free image based on the amount of noise in the first subject-free image captured under the first imaging conditions. This makes it possible to prevent the S / N ratio from deteriorating when generating the bone image, and to calculate a more accurate bone density.

[0103] In this embodiment, the bone image generating unit 105 adjusts the noise amount of the second object-free image, but the adjustment of the noise amount may be performed together with the correction of the dose distribution according to the first embodiment. More specifically, the feature amount analyzing unit 107 can further acquire a pixel value distribution as a feature amount. In addition, the bone image generating unit 105 can generate a second bone image using a correction coefficient for converting the pixel value distribution of the second correction data into the pixel value distribution of the first correction data, which is acquired using the pixel value distribution of the first correction data and the pixel value distribution of the second correction data, the second object data, the second correction data, and a noise adjustment filter. With this configuration, it is possible to suppress deterioration of the S / N ratio when generating the bone image, and it is possible to more appropriately calibrate the bone density, and it is possible to acquire a more accurate bone density.

[0104] When performing dose distribution correction after adjusting the amount of noise, the bone image generating unit 105 can calculate the bone image dImg according to Equation 12.

number

[0105] On the other hand, when the dose distribution correction is performed prior to the adjustment of the amount of noise, the bone image generating unit 105 can process the no-subject image and the object image separately. In this case, the bone image generating unit 105 corrects the dose distribution for the second no-subject image of low energy and the second no-subject image of high energy. After that, the bone image generating unit 105 can average the no-subject images using the first no-subject image of low energy and the first no-subject image of high energy by the difference between the number of acquisitions of m2 and m1, thereby adjusting the amount of noise.

[0106] Example 3 Next, a bone density measuring device according to a third embodiment of the present disclosure will be described with reference to Fig. 10(a) and Fig. 10(b). Since the radiation imaging system and the bone density measuring device according to this embodiment have the same configuration as the radiation imaging system and the bone density measuring device according to the first embodiment, the same reference numerals are used and the description will be omitted.

[0107] In the first embodiment, when generating a bone image of a subject (patient), the dose distribution under the second imaging condition is corrected to match the dose distribution under the first imaging condition. In contrast, in the present embodiment, the bone density measuring device acquires a QC phantom image and a first subject-free image in advance for each of a plurality of dose modes, and calculates a bone density calibration value for each dose mode. Furthermore, the bone density measuring device calibrates the bone density of the subject to be measured using a bone density calibration value corresponding to a dose mode in which the output value of the FPDe202 related to the first imaging condition is close to the output value of the FPDe202 related to the second imaging condition.

[0108] A series of processes according to this embodiment will be described below with reference to Fig. 10(a) and Fig. 10(b), which are flowcharts of a series of processes according to this embodiment.

[0109] First, a series of processes for calculating the bone mineral density calibration value will be described with reference to FIG. 10(a). In step S1001, the first object data acquisition unit 101 acquires a QC phantom image obtained by capturing an image of a QC phantom as first object data. In a bone mineral density device using the DXA method, imaging is performed under two conditions, namely, imaging conditions in which X-rays are high energy and imaging conditions in which X-rays are low energy. Therefore, the first object data acquisition unit 101 acquires two images, a high energy QC phantom image and a low energy QC phantom image. Here, imaging under the high energy imaging condition and imaging under the low energy imaging condition are performed such that the conditions related to the energy of the radiation are switched in a short period of time, so that imaging conditions related to the geometric arrangement of the tube, FPDe 202, etc. can be kept the same. Therefore, hereinafter, the imaging conditions when capturing an image of a QC phantom are referred to as first imaging conditions of high energy and low energy. The first imaging conditions include arrangement conditions related to the geometric arrangement of the tube, FPDe 202, etc.

[0110] In this embodiment, a plurality of dose modes are prepared as bone density measurement modes. The dose mode is a mode in which the mAs value at the time of imaging is used as a parameter. The dose mode may be a mode in which both the mAs value corresponding to high energy and the mAs value corresponding to low energy are used as parameters. For example, a mAs value corresponding to high energy and a mAs value corresponding to low energy may be set for one dose mode. The first subject data acquisition unit 101 acquires a high-energy QC phantom image and a low-energy QC phantom image for each dose mode.

[0111] In step S1002, the first correction data acquisition unit 102 acquires a first object-free image captured under the first imaging condition with the QC phantom removed as first correction data, and stores the first correction data in the memory unit 2015 or the memory unit 2125. In a bone density device using the DXA method, an object-free image is captured under the same imaging condition as when the QC phantom was captured. Therefore, the first correction data acquisition unit 102 acquires two images, a first object-free image of high energy and a first object-free image of low energy. In this embodiment, the first object-free image of high energy and the first object-free image of low energy are acquired for each dose mode in the same manner as when the QC phantom was captured, and stored in the memory unit 2015 or the memory unit 2125.

[0112] In step S1003, the bone image generating unit 105 generates a bone image for each dose mode using the high-energy QC phantom image, the low-energy QC phantom image, the high-energy first object-free image, and the low-energy first object-free image. The method of generating the bone image may be the same as the method of generating the bone image in step S303 according to the first embodiment.

[0113] In step S1004, the bone mineral density calibration value calculation unit 106 calculates a bone mineral density calibration value for each dose mode, and stores the calculated value in the storage unit 2125. The method of calculating the bone mineral density calibration value may be the same as the method of calculating the bone mineral density calibration value in step S304 according to the first embodiment.

[0114] Next, a process of calculating a patient's bone mineral density according to this embodiment will be described with reference to Fig. 10(b). Steps S1005 and S1006 in Fig. 10(b) are similar to steps S305 and S306 according to the first embodiment, and therefore will not be described.

[0115] In step S1007, the feature amount analysis unit 107 reads the high-energy first object-free image and the low-energy first object-free image from the storage unit 2015 or the storage unit 2125. Here, the feature amount analysis unit 107 can read the high-energy first object-free image and the low-energy first object-free image in a dose mode associated with a mAs value similar to the mAs value corresponding to the high energy and the mAs value corresponding to the low energy associated with the second imaging condition. The feature amount analysis unit 107 compares the output value of the FPDe202 between the first imaging condition and the second imaging condition using the high-energy first object-free image and the low-energy first object-free image and the high-energy second object-free image and the low-energy second object-free image.

[0116] As described above, when imaging a subject under the second imaging condition after general imaging using the FPDe202, the geometric arrangement condition between the tube and the FPDe202 may change from the geometric arrangement condition under the first imaging condition. In relation to this, for example, if the distance (SID) between the tube and the FPDe202 changes, the dose of X-rays detected by the FPDe202 changes, and the output value of the FPDe202 changes. In order to improve the accuracy of bone density measurement, it is desirable that the first imaging condition and the second imaging condition are the same.

[0117] Therefore, in this embodiment, more appropriate bone density calibration is performed using a bone density calibration value related to the first imaging condition in a dose mode related to an output value of FPDe202 that is close to the output value of FPDe202 when imaging the subject. In relation to this, in step S1007, the feature amount analysis unit 107 compares the output value of FPDe202 between the first imaging condition and the second imaging condition, and identifies a dose mode that is close to the output value of FPDe202 under the second imaging condition.

[0118] First, the feature amount analysis unit 107 calculates a first average value, which is the average value of pixel values ​​of a specific region from the first object-free image of low energy or the first object-free image of high energy. The specific region may be, for example, a rectangular region including the center of the image, but is not limited thereto and may be arbitrarily set according to a desired configuration. However, it is preferable that the rectangular region does not include the region outside the irradiation field.

[0119] Next, feature amount analysis unit 107 calculates a second average value, which is the average value of pixel values ​​in the specific region used to calculate the first average value, from a second object-free image of energy (high energy or low energy) corresponding to the energy of the first object-free image used to calculate the first average value. Note that the energy value itself of the second object-free image used to calculate the second average value and the energy of the first object-free image used to calculate the first average value do not need to be the same, as long as they are the same in whether they are high energy or low energy.

[0120] The feature amount analysis unit 107 compares the first average value with the second average value, and if the second average value is greater than the first average value, reads out a first object-free image in a dose mode higher than the dose mode of the first imaging condition, and uses the first object-free image to calculate a third average value in the same manner. On the other hand, if the first average value is compared with the second average value and the second average value is smaller than the first average value, reads out a first object-free image in a dose mode lower than the dose mode of the first imaging condition, and uses the first object-free image to calculate a fourth average value in the same manner. At this time, the energy of the first object-free image to be newly read out may be energy (high energy or low energy) corresponding to the energy of the first object-free image used in the calculation of the first average value. Here, the energy value itself of the first object-free image to be newly read out and the energy of the first object-free image used in the calculation of the first average value do not need to be the same, as long as they are the same in whether they are high energy or low energy.

[0121] In step S1008, the bone density calculation section 108 generates a bone image according to Equation 1 using the high-energy object image, the low-energy object image, the second high-energy object-free image, and the second low-energy object-free image.

[0122] In step S1009, the bone density calculation unit 108 reads the bone density calibration value from the storage unit 2125, and calculates the bone density using the bone image and the bone density calibration value. When the second average value is greater than the first average value in step S1007, the bone density calculation unit 108 reads the bone density calibration value of the dose mode higher than the dose mode of the second imaging condition in addition to the bone density calibration value corresponding to the dose mode of the second imaging condition. Here, the bone density calibration value corresponding to the dose mode of the second imaging condition corresponds to the bone density calibration value corresponding to the irradiation dose used when capturing the second correction data. On the other hand, when the second average value is smaller than the first average value, the bone density calculation unit 108 reads the bone density calibration value of the dose mode lower than the dose mode of the second imaging condition in addition to the bone density calibration value corresponding to the dose mode of the second imaging condition.

[0123] The bone density calculation unit 108 calculates bone density in each dose mode from Equation 9 using the bone density calibration value in the dose mode of the first imaging condition and the bone density calibration value in the other dose mode that has been read. More specifically, the bone density calculation unit 108 calculates bone density in the first imaging condition using the bone image generated in step S1008 and the bone density calibration value corresponding to the dose mode of the second imaging condition. In addition, the bone density calculation unit 108 calculates bone density in the other dose mode that has been read using the bone image generated in step S1008 and the bone density calibration value in the other dose mode that has been read.

[0124] Thereafter, the bone density calculation unit 108 outputs the result of weighting and adding the bone density under the first imaging condition and the bone density under the other dose mode that was read based on the average value calculated in step S1007 as the final bone density. As an example, when the second average value is greater than the first average value, the bone density under the first imaging condition is densRN, and the bone density in the dose mode higher than the first imaging condition is densRH, the calculated bone density densR can be expressed as in Equation 13.

[0125]

number

[0126] As described above, in the bone density measuring device according to the present embodiment, the first object data acquiring unit 101 and the first correction data acquiring unit 102 can acquire a plurality of first object data and a plurality of first correction data captured under a first imaging condition using different doses. The bone image generating unit 105 can generate a plurality of first bone images using the plurality of first object data and the plurality of first correction data. The bone density calibration value calculating unit 106 can acquire a plurality of bone density calibration values ​​according to the doses using the plurality of first bone images and a predetermined bone density of the first object. In addition, the feature amount analyzing unit 107 can analyze the first correction data and the second correction data, and acquire an average value of pixel values ​​of the first correction data and an average value of pixel values ​​of the second correction data. In addition, the feature amount analyzing unit 107 can compare an average value of pixel values ​​of the first correction data with an average value of pixel values ​​of the second correction data.

[0127] The bone density calculation unit 108 can obtain a first bone density by using a bone density calibration value corresponding to an exposure dose used when capturing the second correction data, among the plurality of bone density calibration values, and the second bone image. The bone density calculation unit 108 can obtain a second bone density by using a bone density calibration value corresponding to a dose selected according to a comparison result between an average pixel value of the first correction data and an average pixel value of the second correction data, among the plurality of bone density calibration values, and the second bone image. Furthermore, the bone density calculation unit 108 can perform a weighted addition of the first bone density and the second bone density to calculate the bone density of the bone density measurement subject.

[0128] According to the bone density measuring device of the present embodiment, more appropriate bone density calibration can be performed using the bone density calibration value associated with the first imaging condition in the dose mode associated with the output value of the FPDe202 close to the output value of the FPDe202 when the subject is imaged, and therefore more accurate bone density can be calculated.

[0129] In this embodiment, it is assumed that there are dose modes higher and lower than the dose mode of the second imaging condition. On the other hand, if the dose mode of the second imaging condition is the highest dose mode or the lowest dose mode, extrapolation may be performed using other dose modes for the dose modes higher and lower than the dose mode of the second imaging condition. In this case, the third average value or the fourth average value and the bone density corresponding to the dose modes higher and lower than the second imaging condition for which extrapolation is performed may be calculated.

[0130] In the present embodiment, the bone density is corrected using the average pixel value of the no-subject image under each imaging condition of the bone density measuring device, but a bone density correction coefficient may be obtained in advance for each average value. The correction coefficient may be obtained, for example, using the actually measured bone density and the average pixel value of the no-subject image. In this case, the feature amount analysis unit 107 obtains a correction coefficient from the average pixel value of the second no-subject image under the second imaging condition, and the bone density calculation unit 108 corrects the obtained bone density using the correction coefficient.

[0131] In this embodiment, step S1008 is performed after step S1007. However, step S1008 may be performed before step S1007 or in parallel with step S1007.

[0132] Example 4 A bone density measuring device according to a fourth embodiment of the present disclosure will be described with reference to Fig. 11 and Fig. 12. The bone density measuring device according to this embodiment uses a trained model that has been subjected to machine learning in advance to acquire a bone image of a subject that is a bone density measurement target acquired under a second imaging condition, as if the bone image was acquired under a first imaging condition.

[0133] 11 is a configuration diagram of an example of a bone density measuring device according to the present embodiment. The bone density measuring device according to the present embodiment is provided with a first object data acquiring unit 1101, a first correction data acquiring unit 1102, a second object data acquiring unit 1103, and a second correction data acquiring unit 1104. Here, the first object data acquiring unit 1101 and the first correction data acquiring unit 1102 are respectively similar to the first object data acquiring unit 101 and the first correction data acquiring unit 102 according to the first embodiment. Moreover, the second object data acquiring unit 1103 and the second correction data acquiring unit 1104 are respectively similar to the second object data acquiring unit 103 and the second correction data acquiring unit 104 according to the first embodiment.

[0134] The bone density measuring device according to the present embodiment further includes a bone image generating unit 1105, a bone density calibration value calculating unit 1106, a bone image correcting unit 1107, and a bone density calculating unit 1108. Here, the bone density calibration value calculating unit 1106 and the bone density calculating unit 1108 are the same as the bone density calibration value calculating unit 106 and the bone density calculating unit 108 according to the first embodiment, respectively. The bone image generating unit 1105 according to the present embodiment generates a bone image using the first subject data and the first correction data. In addition, the bone image generating unit 1105 generates a bone image using the second subject data and the second correction data. The bone image correcting unit 1107 acquires a corrected bone image using the first correction data, the second correction data, and the bone image of the subject to be subjected to bone density measurement as inputs of a trained model. The bone image generating unit 1105 and the bone image correcting unit 1107 are stored in the storage unit 2125 as software modules of the analysis PC 212.

[0135] Hereinafter, a series of processes according to this embodiment will be described with reference to Fig. 12, showing the flow of bone mineral density calculation processing. Fig. 12 is a flowchart of the bone mineral density calculation processing according to this embodiment. Note that the series of processes for calculating bone mineral density calibration values ​​according to this embodiment are similar to the series of processes for calculating bone mineral density calibration values ​​shown in Fig. 3(a) according to the first embodiment, and therefore the description thereof will be omitted. Also, steps S1205 and S1206 in Fig. 12 are similar to steps S305 and S306 in Fig. 3(b) according to the first embodiment, and therefore the description thereof will be omitted.

[0136] In step S1207, the bone image generating unit 1105 generates a bone image according to Equation 1 using the high-energy object image, the low-energy object image, the second high-energy object-free image, and the second low-energy object-free image.

[0137] In step S1208, the bone image correction unit 1107 uses the bone image generated in step S1207, the first correction data, and the second correction data as inputs to the trained model to obtain a corrected bone image output from the trained model. Here, the obtained corrected bone image corresponds to a corrected bone image in which differences in dose distribution based on differences between the first and second imaging conditions, etc., have been corrected.

[0138] In this embodiment, a trained machine learning model (trained model) is used as a method for generating a corrected bone image. Hereinafter, learning of the machine learning model according to this embodiment will be described with reference to Figs. 13(a) and 13(b). Figs. 13(a) and 13(b) show a learning method when a CNN (Convolutional Neural Network) is used as the machine learning model. Fig. 13(a) shows a learning flow according to this embodiment. Fig. 13(b) is a diagram for explaining learning using learning data (teacher data).

[0139] The learning data includes a set of data in which bone image 1311, object-free image 1312, and object-free image 1313 are used as input data, and correct bone image 1317 is used as correct data (output data). There are high-energy and low-energy object-free images, and both or either one may be used as learning data. In this embodiment, high-energy object-free images are used.

[0140] The bone image 1311 and the no-subject image 1312 are two pieces of data acquired under the same shooting conditions, and the bone image 1311 is an image generated using the no-subject image 1312. On the other hand, the correct bone image 1317 and the no-subject image 1313 are two pieces of data acquired under the same shooting conditions, and the correct bone image 1317 is an image generated using the no-subject image 1313. However, the shooting conditions of the correct bone image 1317 and the no-subject image 1313 are different from the shooting conditions of the bone image 1311 and the no-subject image 1312. In addition, the shooting conditions of the correct bone image 1317 and the no-subject image 1313 can be the same as the first shooting conditions. In contrast, the shooting conditions of the bone image 1311 and the no-subject image 1312 can be shooting conditions including geometric arrangement conditions different from the first shooting conditions. Note that the subject of the bone image 1311 and the correct bone image 1317 may be the same, and may be a QC phantom or a patient. Also, the bone image 1311 and the correct bone image 1317 may be a phantom such as a commonly used lumbar phantom whose bone density is unknown. Note that when using bone images of such phantoms, it is not necessary to consider the exposure of the patient, and therefore learning data can be generated efficiently.

[0141] In this embodiment, in addition to a bone image and a no-subject image with the same imaging conditions, a no-subject image with different imaging conditions is input, and a network is constructed that outputs a bone image with the different imaging conditions. In other words, a network is constructed that inputs two no-subject images with different dose distributions and a bone image corresponding to one of the no-subject images, and outputs a bone image corresponding to the other no-subject image with a different dose distribution.

[0142] Here, the flow of the learning process according to this embodiment will be described along the flow of FIG. 13(a). In this embodiment, the bone image correction unit 1107 can function as a learning unit that performs the learning process, but the learning unit may be provided separately from the bone density measurement device, and the bone image correction unit 1107 may use a trained model. In step S1301, when the bone image 1311, the object-free image 1312, and the object-free image 1313 are input to the CNN 1314 in the middle of learning, an inference process is performed using the parameters of the CNN 1314, and a corrected bone image 1316 is output as the inference result. Here, the CNN 1314 has a structure in which a large number of processing units 1315 are arbitrarily connected. Examples of the processing units 1315 include units that perform convolution calculation processing, normalization processing, or processing using an activation function such as ReLU or Sigmoid, and each unit has a parameter group for describing the respective processing contents. The multiple processing units 1315 can be connected in layers of about three to several hundred sets that perform processing in sequence, such as convolution operation, normalization, and activation function, and can have a variety of structures.

[0143] Next, in step S1302, a loss function is calculated from the corrected bone image 1316, which is the inference result, and the correct bone image 1317. As the loss function, any function such as squared error or cross entropy error can be used.

[0144] In step S1303, backpropagation is performed using the loss function calculated in step S1302 as a starting point, and the parameter set of the CNN 1314 is updated.

[0145] Finally, in step S1304, the end of learning is determined. If learning is to be continued, the process proceeds to step S1301. The processes of steps S1301 to S1303 are repeated while changing the bone image 1311, the object-free image 1312, the object-free image 1313, and the correct bone image 1317. This repeats parameter update of the CNN 1314 so as to reduce the loss function, and the accuracy of the bone image output by the trained model can be improved. If the learning progresses sufficiently and it is determined in step S1304 that the learning is ended, the process is completed. The end of learning can be determined based on a determination criterion set according to a desired configuration, such as, for example, the accuracy of the inference result is equal to or greater than a certain value without overlearning, or the loss function is equal to or less than a certain value.

[0146] By using the trained model in which such learning has been performed, the bone image correction unit 1107 can acquire a corrected bone image as if it were acquired under the first imaging condition, based on the bone image acquired under the second imaging condition. More specifically, the bone image correction unit 1107 inputs the high-energy first object-free image, the high-energy second object-free image, and the bone image generated in step S1207 to the trained model, thereby acquiring a corrected bone image output from the trained model. Here, the output corrected bone image corresponds to a bone image obtained by analyzing the difference in dose distribution between the first imaging condition and the second imaging condition and absorbing the dose.

[0147] In step S1209, the bone density calculation unit 1108 reads the bone density calibration value from the storage unit 2125, and calculates the bone density using the corrected bone image acquired in step S1208 and the bone density calibration value. Note that the method of calculating the bone density may be the same as that in step S309 according to the first embodiment, except that the corrected bone image is used.

[0148] As described above, the bone density measuring device according to the present embodiment further includes a bone image correction unit 1107 that functions as an example of a correction unit that acquires a corrected second bone image using the first correction data, the second correction data, and the second bone image as input data of the trained model. The bone density calculation unit 108 can acquire bone density using the corrected second bone image and the bone density calibration value. Here, the trained model can be acquired using training data in which the third correction data obtained under the third shooting condition, the fourth correction data obtained under the fourth shooting condition, and the fourth bone image obtained by shooting the subject under the fourth shooting condition are used as input data, and the third bone image obtained by shooting the subject under the third shooting condition is used as correct data. Here, the subject-free image 1313 may be an example of the third correction data, the subject-free image 1312 may be an example of the fourth correction data, the bone image 1311 may be an example of the fourth bone image, and the correct bone image 1317 may be an example of the third bone image.

[0149] According to the bone density measuring device of this embodiment, by using the trained model, it is possible to obtain a corrected bone image in which the difference in dose distribution due to the difference between the first and second imaging conditions is corrected based on the bone image obtained under the second imaging condition, etc. Therefore, it is possible to more appropriately calibrate the bone density and obtain a more accurate bone density.

[0150] In this embodiment, bone image 1311, subject-free image 1312, and subject-free image 1313 are used as input data for the learning data. Alternatively, instead of subject-free image 1312 and subject-free image 1313, a difference image or a division image between subject-free image 1312 and subject-free image 1313 may be used as input data for the learning data. In this case, similarly in step S1208, bone image corrector 1107 may input a difference image or a division image between the first subject-free image and the second subject-free image.

[0151] The CNN used in this embodiment may be a U-net type machine learning model having an encoder function consisting of multiple layers including multiple downsampling layers and a decoder function consisting of multiple layers including multiple upsampling layers. In the U-net type machine learning model, the position information (spatial information) obscured in multiple layers configured as an encoder is configured (for example, by using skip connections) so that it can be used in layers of the same dimension (layers corresponding to each other) in multiple layers configured as a decoder.

[0152] In this embodiment, the error backpropagation method is used for learning. In contrast, a method (dropout) of randomly inactivating units (each neuron or each node) constituting the CNN may be used for learning. Furthermore, the trained model used by the bone image correction unit 1107 may be generated using transfer learning. In this case, for example, transfer learning may be performed on a machine learning model trained using bone images of subjects of different types, etc., to generate a trained model to be used for the correction process of the bone image. By performing such transfer learning, a trained model can be efficiently generated even for subjects for which it is difficult to obtain a large amount of training data. The subjects of different types, etc. referred to here may be, for example, animals, plants, products, etc.

[0153] Here, the GPU can perform efficient calculations by processing more data in parallel. Therefore, when performing learning multiple times using a machine learning model using CNN as described above, it is effective to perform processing using the GPU. Therefore, the learning unit according to this embodiment uses a GPU in addition to a CPU. Specifically, when executing a learning program including a machine learning model, the CPU and the GPU work together to perform calculations to perform learning. Note that in the learning process, calculations may be performed only by the CPU or the GPU.

[0154] Although the configuration of the machine learning model has been described above, the machine learning model is not limited to the model using CNN as described above. The learning of the machine learning model used in this embodiment may be any model similar to machine learning using a model that can extract (express) the feature amount of learning data such as an image by itself through learning.

[0155] Example 5 A bone density measuring device according to a fifth embodiment of the present disclosure will be described with reference to Figs. 14 and 15. Fig. 14 is a configuration diagram of an example of a bone density measuring device according to this embodiment. As shown in Fig. 14, the bone density measuring device according to this embodiment is provided with a message output unit 1409 in addition to the components shown in Fig. 1. The message output unit 1409 is stored in the storage unit 2015 as a software module of the control PC 201. The message output unit 1409 is also stored in the storage unit 2125 as a software module of the analysis PC 212. The output message is displayed on the display unit 209 of the control PC 201 or the display unit 219 of the analysis PC 212.

[0156] A series of processes according to this embodiment is substantially similar to the series of processes according to the first embodiment, and therefore, the series of processes according to this embodiment will be described with reference to Fig. 3(a) and Fig. 3(b). The processes from step S301 to step S305 are similar to the processes in the same steps according to the first embodiment, and therefore the description thereof will be omitted.

[0157] In step S306, the second correction data acquisition unit 104 acquires a second object-free image captured under the second imaging conditions, excluding the object, as the second correction data. In a bone density device using the DXA method, imaging is performed under the same imaging conditions as when the object image was captured. Therefore, the second correction data acquisition unit 104 acquires two images, a high-energy second object-free image and a low-energy second object-free image, captured under the second imaging conditions of high energy and low energy.

[0158] In addition, the second correction data acquisition unit 104 checks whether the high energy subject image, the low energy subject image, the high energy second subject-free image, and the low energy second subject-free image are all present. If it is determined that at least one of these images is insufficient, the message output unit 1409 can display a message indicating that data used for bone density measurement is insufficient on the display unit 209 or the display unit 219. In this case, the bone density measurement device can inform the examiner that necessary data is insufficient. For example, if correction data cannot be acquired, the message output unit 1409 can output a message 1501, as shown in FIG. 15, indicating that data used for bone density measurement is insufficient and urging the examiner to acquire (or reacquire) correction data.

[0159] The method of checking whether all data necessary for bone density measurement is available may be to check whether acquisition of imaging data according to a prepared protocol is complete. The second correction data acquisition unit 104 may also perform image analysis on the acquired image to check whether subject data and non-subject data are available.

[0160] In step S307, the feature amount analysis unit 107 reads the first object-free image of high energy and the first object-free image of low energy from the storage unit 2015 or the storage unit 2125, and estimates a first dose distribution under the first imaging condition. The feature amount analysis unit 107 also estimates a second dose distribution under the second imaging condition from the second object-free image of high energy and the second object-free image of low energy. The feature amount analysis unit 107 also calculates a linear distribution correction coefficient that converts the second dose distribution into the first dose distribution. The calculation method of the linear distribution correction coefficient may be the same as that of step S307 according to the first embodiment.

[0161] Furthermore, the feature amount analysis unit 107 obtains the calculation accuracy of the dose distribution coefficient. Here, the calculation accuracy of the dose distribution coefficient may be determined based on, for example, an error (fitting accuracy) between the pixel value of the fitting image used when calculating the dose distribution coefficient and the pixel value of the original image on which the fitting is performed. For example, when the error is equal to or greater than a predetermined threshold, the calculation accuracy of the dose distribution coefficient may be determined to be equal to or less than the threshold. When the calculation accuracy of the dose distribution coefficient is equal to or less than the threshold, the message output unit 1409 can display a message 1502 indicating that calibration as shown in FIG. 15 is required on the display unit 209 or the display unit 219. The calibration may be a calibration of the arrangement of the X-ray tube and the FPDe 202, may be a calibration of these arrangements regarding the first imaging condition, or may be a calibration of these arrangements regarding the second imaging condition. By performing the calibration of these arrangements regarding the first imaging condition, imaging after calibration can be performed without considering the exposure dose of the subject, etc.

[0162] Furthermore, if the processing is continued as is after displaying message 1502, message output unit 1409 can display message 1503 on display unit 209 or display unit 219, indicating that there is a possibility that the accuracy of bone density may decrease and indicating whether or not to continue the examination. Furthermore, if analytical PC 212 is independent of control PC 201, it is difficult to perform re-imaging at the timing after displaying message 1502. Therefore, message output unit 1409 can display message 1504 on display unit 209 or display unit 219, indicating that there is a possibility that the accuracy of bone density may decrease, and can only inform the user that the accuracy of bone density will decrease.

[0163] The processes in steps S308 and S309 are similar to those in the first embodiment, and therefore the description thereof will be omitted.

[0164] As described above, the bone density measuring device according to this embodiment may further include a message output unit 1409 that outputs a message indicating that data to be used for bone density measurement is insufficient when at least one of the second correction data and the second object data cannot be acquired. In addition, when the acquisition accuracy of the feature amount acquired by the feature amount analysis unit 107 is equal to or lower than a threshold, the message output unit 1409 may output a message indicating that the bone density accuracy is reduced or a message prompting reacquisition of the first object data and the first correction data.

[0165] According to the bone density measuring device of the present embodiment, it is possible to inform the examiner that the data required for bone density measurement is insufficient, and to prevent inappropriate processing. Also, it is possible to urge the examiner to reacquire the data to be used for bone density measurement. Furthermore, it is possible to urge the examiner to perform calibration based on the deviation of the geometric arrangement of the tube or the FPDe 202, or to inform the examiner of the deterioration of the measurement accuracy of the bone density, and thus to improve the convenience of the bone density measuring device.

[0166] Example 6 A bone density measuring device according to a sixth embodiment of the present disclosure will be described with reference to Fig. 16. Note that the radiation imaging system and the bone density measuring device according to this embodiment have the same configurations as the radiation imaging system and the bone density measuring device according to the first embodiment, and therefore the same reference numerals are used and the description thereof will be omitted.

[0167] In the first embodiment, the feature amount analysis unit 107 calculates a dose distribution correction coefficient for correcting the dose distribution under the second imaging condition to match the dose distribution under the first imaging condition, and the bone image generation unit 105 generates a bone image of the subject using the dose distribution correction coefficient. In contrast, in the present embodiment, the feature amount analysis unit 107 corrects the bone density calibration value based on the difference in the magnitude of the dose between imaging under the first imaging condition and imaging under the second imaging condition. The bone density calculation unit 108 calibrates the bone density using the corrected bone density calibration.

[0168] Hereinafter, a series of processes according to this embodiment will be described with reference to FIG. 16, showing the flow of bone mineral density calculation processing. FIG. 16 is a flowchart of the bone mineral density calculation processing according to this embodiment. Note that the series of processes for calculating bone mineral density calibration values ​​according to this embodiment are similar to the series of processes for calculating bone mineral density calibration values ​​shown in FIG. 3(a) according to the first embodiment, and therefore the description thereof will be omitted. Also, steps S1605 and S1606 in FIG. 16 are similar to steps S305 and S306 in FIG. 3(b) according to the first embodiment, and therefore the description thereof will be omitted.

[0169] In step S1607, the feature amount analysis unit 107 reads the high-energy first object-free image from the memory unit 2015 or the memory unit 2125, and reads the bone mineral density calibration value from the memory unit 2125. Thereafter, the feature amount analysis unit 107 corrects the bone mineral density calibration value using the high-energy first object-free image and the high-energy second object-free image.

[0170] The bone density calibration value is a conversion formula for converting the bone density obtained from the output value of the FPDe202 into the true bone density. If there is a deviation in the geometric arrangement of the tube, FPDe202, etc., between the imaging of the subject to be bone density measured and the imaging of the QC phantom, the dose distribution will be different, and therefore a deviation will also occur in the bone density obtained from the output value of the FPDe202. Therefore, in this embodiment, the bone density calibration value is corrected to a bone density calibration value that absorbs the difference in the magnitude of the dose.

[0171] The average value of pixel values ​​in the central portion of the first high-energy object-free image is m1, and the average value of pixel values ​​in the central portion of the second high-energy object-free image is m2. The feature amount analysis unit 107 divides the average value m1 by the average value m2 to calculate the dose ratio mr. Since the bone density obtained from the output value of the FPDe 202 is calculated logarithmically, the dose ratio mr is added as an offset. Therefore, by correcting the value k0 used in Equation 9 with a correction amount corresponding to the dose ratio mr, it is possible to calculate a bone density calibration value k0' that absorbs the influence of the deviation in the geometric arrangement between the first imaging condition and the second imaging condition. Here, the correction amount corresponding to the dose ratio can be calculated in advance. The correction amount corresponding to the dose ratio can be determined, for example, by checking the bone density while adjusting the correction amount for the actually measured dose ratio. A table or a function may be prepared for the correction amount corresponding to the dose ratio, and the feature amount analysis unit 107 may specify the correction amount according to the dose ratio by referring to the table or the like.

[0172] Steps S1608 and S1609 are similar to steps S308 and S309 in Fig. 3(b) according to the first embodiment, and therefore the description thereof will be omitted. However, in step S1609, the bone mineral density calculation unit 108 uses the bone mineral density calibration value k1 and the bone mineral density calibration value k0' corrected in step S1607 as the bone mineral density calibration value.

[0173] As described above, the feature amount analysis unit 107 according to this embodiment can analyze the first and second correction data and obtain the dose ratio between the first and second correction data. In this case, the bone density calculation unit 108 can obtain the bone density using the bone density calibration value obtained by correcting the obtained bone density calibration value according to the dose ratio and the second bone image.

[0174] According to the bone density measuring device of the present embodiment, it is possible to obtain a bone density calibration value that absorbs the influence of the deviation in the geometric arrangement between the first imaging condition and the second imaging condition, and therefore it is possible to more appropriately calibrate the bone density and obtain a more accurate bone density.

[0175] In this embodiment, the bone mineral density calibration value was corrected using the high-energy first object-free image and the high-energy second object-free image. In contrast, the bone mineral density calibration value may be corrected using the low-energy first object-free image and the low-energy second object-free image. In addition, the area from which the average pixel value is obtained may be limited to the bone area of ​​the QC phantom.

[0176] Example 7 A bone density measuring device according to a seventh embodiment of the present disclosure will be described with reference to Fig. 17 and Fig. 18. The bone density measuring device according to this embodiment outputs a subject image obtained by imaging a subject (patient) who is the target of bone density measurement and information indicating the imaging conditions of the subject image, and manages the radiation exposure of the subject. On the other hand, the bone density measuring device according to this embodiment does not output a subject-free image (first correction data and second correction data) that is not necessary for managing the radiation exposure, or information indicating the imaging conditions of the subject-free image.

[0177] Fig. 17 is a configuration diagram of an example of a bone density measuring device according to this embodiment. As shown in Fig. 17, the bone density measuring device according to this embodiment is provided with an output control unit 1709 in addition to the components shown in Fig. 1.

[0178] The control PC 201 according to this embodiment sends the images acquired by the FPDe 202, the processed images, and information indicating the imaging conditions to the storage unit 206 outside the control PC 201, and stores them. The analysis PC 212 according to this embodiment sends a report indicating the processing contents of the bone density calculation and the bone density measurement result to the storage unit 206 outside the analysis PC 212, and stores them. The control PC 201 collects subject images and bone images obtained by imaging the subject, reports or report images related to bone density, and information indicating the imaging conditions corresponding to the images, and stores them in the storage unit 206, thereby managing the radiation dose of the patient.

[0179] The output control unit 1709 is stored in the storage unit 2015 as a software module of the control PC 201. The output control unit 1709 is also stored in the storage unit 2125 as a software module of the analysis PC 212. The output control unit 1709 controls the output so as not to output a subject-free image that is unnecessary for managing the radiation exposure of the subject when the control PC 201 and the analysis PC 212 output various images, reports on bone density, information indicating the imaging conditions corresponding to the images, and the like.

[0180] Fig. 18 is a flowchart of the output control process according to this embodiment. The output control process will be described below with reference to Fig. 18. Note that the series of processes for calculating the bone mineral density calibration value and the bone mineral density measurement process according to this embodiment may be similar to those according to the first embodiment, and therefore the description thereof will be omitted.

[0181] In step S1801, the output control unit 1709 determines whether or not an image is a no-subject image when outputting various images from the control PC 201 or the analysis PC 212. If the image is a no-subject image, it is not added to the output targets because it is not necessary for the patient's subsequent examination and dose management, and the process proceeds to step S1803. On the other hand, if the image is an image obtained by photographing a subject, the process proceeds to step S1802.

[0182] The determination of whether an image is a no-subject image can be made based on image information. For example, by specifying information indicating the presence or absence of a subject at the time of shooting and including the information indicating the presence or absence of a subject in the image information, the output control unit 1709 can determine whether an image is a no-subject image based on the image information of each image. In addition, the output control unit 1709 may determine whether an image is a subject image, a bone image, or a report image by performing image analysis.

[0183] In step S1802, the output control unit 1709 adds the image and information indicating the imaging conditions corresponding to the image to the output target. The imaging conditions corresponding to the image may include information regarding the irradiation dose. The information regarding the irradiation dose may include, for example, information such as the tube voltage set at the time of imaging. In addition, when the imaging conditions corresponding to the image include information indicating the geometric arrangement, for example, information indicating the geometric arrangement of the tube and the FPDe 202 specified based on the collimator lamp included in the captured image and the markers placed at the time of imaging may be included.

[0184] In step S1803, the output control unit 1709 judges whether or not all images in the examination to be output have been checked. If it is judged that all images have been checked, the process proceeds to step S1804. On the other hand, if it is judged that all images have not been checked, the process proceeds to step S1801.

[0185] In step S1804, the output control unit 1709 outputs the image to be output, information on the shooting conditions corresponding to the image, and a report to the storage unit 206. When the output process ends in step S1804, the output control process according to this embodiment ends.

[0186] As described above, the bone density measuring device according to this embodiment further includes an output control unit 1709 that controls output so as not to output the first correction data and the second correction data when outputting an image. According to the bone density measuring device according to this embodiment, when outputting various images and bone density measurement results, object-free images unnecessary for managing the amount of radiation exposure of a patient are not output, thereby making it possible to efficiently manage information used for managing the amount of radiation exposure of a patient.

[0187] In this embodiment, the output control unit 1709 controls the output so as not to output a subject-free image when outputting various images. In contrast, the output control unit 1709 may control the output so as not to output information indicating imaging conditions related to a subject-free image when outputting a report on bone density measurement. In this case as well, the bone density measurement device can efficiently manage information used for managing the patient's radiation exposure by not outputting information indicating imaging conditions unnecessary for managing the patient's radiation exposure.

[0188] In this embodiment, the output control unit 1709 is configured to control the output so as not to output a subject-free image, but may be configured to control the output so as not to output a QC phantom image or information indicating the imaging conditions of the QC phantom image. Even in this case, by not outputting information unnecessary for managing the radiation exposure of the subject (patient) who is the subject of bone density measurement, it is possible to efficiently manage information used for managing the radiation exposure of the patient. The output control unit 1709 may also use image information or image analysis processing to determine whether or not the image is a QC phantom image. In contrast, the determination of whether or not the image is a QC phantom image may be performed based on the examination type that is included in the data related to the QC phantom image and that defines whether the imaging is for examining a patient or for obtaining a bone density calibration value.

[0189] In the above-mentioned first to seventh embodiments, the first object data acquisition unit 101, the first correction data acquisition unit 102, the second object data acquisition unit 103, and the second correction data acquisition unit 104 acquired various X-ray images from the FPDe 202 connected to them in a communicable manner. In contrast, these data acquisition units may acquire various X-ray images from an external storage device connected to the control PC 201 or the analysis PC 212. In addition, these data acquisition units may acquire various X-ray images from a radiation detection device or an external storage device connected to the control PC 201 or the analysis PC 212 via an arbitrary network.

[0190] Moreover, the various calculation processes in the above-mentioned first to seventh embodiments are not limited to processes of calculating various values ​​by calculation. Such processes may be processes of acquiring various values ​​by referring to a lookup table or the like determined in advance by actual measurements or the like. For example, in the bone density calibration value calculation process, the bone density calibration value may be acquired by referring to a lookup table using the first bone image and the true bone density of the QC phantom. In addition, in the bone density calculation unit, the bone density may be acquired by referring to a lookup table using the second bone image and the bone density calibration value. Similarly, in the feature amount analysis process, the feature amount may be acquired by referring to a lookup table using the first correction data and the second correction data. Similarly, in other calculation processes, various values ​​may be acquired by referring to a lookup table using a value used in the calculation process.

[0191] In addition, in the various trained models mentioned above, it is believed that the magnitude of the brightness values ​​of the input data images, the order and slope of light and dark areas, position, distribution, continuity, etc. are extracted as part of the features and used in the inference process.

[0192] The various trained models described above can be provided in the bone density analysis device. The trained model may be configured, for example, by a software module executed by a processor such as a CPU, MPU, GPU, or FPGA, or may be configured by a circuit that performs a specific function such as an ASIC. The trained model may be provided in another server device connected to the bone density analysis device. In this case, the bone density analysis device can use the trained model by connecting to a server or the like equipped with the trained model via any network such as the Internet. Here, the server equipped with the trained model may be, for example, a cloud server, a fog server, an edge server, or the like. In addition, when configuring a network within a facility, a site including a facility, or an area including multiple facilities to be capable of wireless communication, the reliability of the network may be improved by configuring the network to use radio waves of a dedicated wavelength band allocated exclusively to the facility, site, area, or the like. In addition, the network may be configured by wireless communication capable of high-speed communication, large-capacity communication, low-latency communication, and multiple simultaneous connections.

[0193] (Other Examples) The present disclosure can also be realized by a process in which a program for implementing one or more functions of the above-described embodiments is provided to a system or device via a network or a storage medium, and one or more processors in the computer of the system or device read and execute the program. It can also be realized by a circuit (e.g., ASIC) for implementing one or more functions. A computer may have one or more processors or circuits, and may include separate computers or a network of separate processors or circuits for reading and executing computer-executable instructions.

[0194] The processor or circuitry 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), and the processor or circuitry may include a digital signal processor (DSP), a data flow processor (DFP), or a neural processing unit (NPU).

[0195] The above disclosure includes the following configurations, methods, and programs. (Configuration 1) a data acquisition unit that acquires a bone density calibration value obtained using first subject data obtained by photographing a first subject having a predetermined bone density under a first photographing condition and first correction data obtained by photographing the first subject under the first photographing condition without placing the first subject, second subject data obtained by photographing a second subject to be subjected to bone density measurement under a second photographing condition, and second correction data obtained by photographing the second subject under the second photographing condition without placing the second subject; a generation unit that generates a bone image using the second object data and the second corrected data; a bone density acquisition unit that acquires a bone density of the second subject using the bone image and the bone density calibration value; An information processing device comprising: (Configuration 2) The data acquisition unit further acquires the first correction data, The information processing device further includes an analysis unit that analyzes the first corrected data and the second corrected data and acquires feature amounts of the first corrected data and the second corrected data, 2. The information processing device according to configuration 1, wherein the generation unit generates the bone image by using the second object data, the second correction data, and the feature amount. (Configuration 3) 3. The information processing device according to configuration 2, wherein the analysis unit acquires a pixel value distribution as the feature amount. (Configuration 4) The information processing device according to configuration 3, wherein the generation unit generates the bone image using a correction coefficient that converts a pixel value distribution of the second corrected data into a pixel value distribution of the first corrected data obtained using the feature amount, the second subject data, and the second corrected data. (Configuration 5) 5. The information processing device according to configuration 4, wherein the generation unit corrects the second correction data using the correction coefficient, and generates the bone image using the corrected second correction data and the second subject data. (Configuration 6) 6. The information processing device according to any one of configurations 3 to 5, wherein the analysis unit acquires the pixel value distribution using multidimensional surface fitting. (Configuration 7) 7. The information processing device according to configuration 6, wherein the generation unit does not use the feature amount for generating the bone image when a fitting error of the curved surface fitting is equal to or greater than a threshold. (Configuration 8) 8. The information processing device according to any one of configurations 2 to 7, wherein the analysis unit performs the analysis within an irradiation field region of the second corrected data and within an irradiation field region of the first corrected data. (Configuration 9) 3. The information processing device according to configuration 2, wherein the analysis unit acquires an amount of noise as the feature amount. (Configuration 10) The information processing device according to configuration 9, wherein the generation unit generates the bone image by using a noise adjustment filter for converting an amount of noise in the second corrected data acquired using the feature amount into an amount of noise in the first corrected data, the second subject data, and the second corrected data. (Configuration 11) The analysis unit further acquires a pixel value distribution as the feature amount, The information processing device of configuration 10, wherein the generation unit generates the bone image using a correction coefficient for converting a pixel value distribution of the second correction data into a pixel value distribution of the first correction data, the second subject data, the second correction data, and the noise adjustment filter, the correction coefficient being obtained using the pixel value distributions of the first correction data and the second correction data. (Configuration 12) an analysis unit that analyzes the first correction data and the second correction data and obtains an average value of pixel values ​​of the first correction data and an average value of pixel values ​​of the second correction data, the data acquisition unit acquires a plurality of bone mineral density calibration values ​​corresponding to a dose, the bone mineral density calibration values ​​being obtained using a plurality of first object data and a plurality of first correction data ... correction data and The analysis unit compares an average value of pixel values ​​of the first corrected data with an average value of pixel values ​​of the second corrected data, The bone density acquisition unit includes: obtaining a first bone density using a bone density calibration value corresponding to an exposure dose used in capturing the second correction data and the bone image among the plurality of bone density calibration values; acquiring a bone density calibration value corresponding to a dose selected from the plurality of bone density calibration values ​​according to a comparison result between an average pixel value of the first correction data and an average pixel value of the second correction data, and a second bone density using the bone image; 2. The information processing device according to configuration 1, wherein the bone density is obtained by performing a weighted addition of the first bone density and the second bone density. (Configuration 13) The data acquisition unit further acquires the first correction data, A correction unit is further provided that uses the first correction data, the second correction data, and the bone image as input data for a trained model to obtain a corrected bone image, 2. The information processing device according to configuration 1, wherein the bone density acquisition unit acquires the bone density by using the corrected bone image and the bone density calibration value. (Configuration 14) The information processing device according to configuration 13, wherein the trained model is a trained model obtained using training data in which third correction data obtained under third shooting conditions, fourth correction data obtained under fourth shooting conditions, and a bone image obtained by photographing a subject under the fourth shooting conditions are used as input data, and the bone image obtained by photographing the subject under the third shooting conditions is used as correct data. (Configuration 15) 15. The information processing device according to any one of configurations 1 to 14, further comprising a message output unit that outputs a message indicating that data to be used for bone density measurement is insufficient when at least one of the second correction data and the second subject data cannot be acquired. (Configuration 16) The information processing device according to any one of configurations 2 to 9, further comprising a message output unit that outputs, when the acquisition accuracy of the feature amount is equal to or lower than a threshold, a message indicating that the accuracy of bone density is decreasing, or a message prompting the user to re-acquire the first subject data and the first correction data. (Configuration 17) The data acquisition unit further acquires the first correction data, An analysis unit that analyzes the first correction data and the second correction data to obtain a dose ratio of the first correction data and the second correction data, 2. The information processing device according to configuration 1, wherein the bone density acquisition unit acquires the bone density using a calibrated bone density value obtained by correcting the calibrated bone density value in accordance with the dose ratio and the bone image. (Configuration 18) 18. The information processing device according to any one of configurations 1 to 17, further comprising an output control unit that controls output so as not to output the first correction data and the second correction data when outputting an image. (Configuration 19) An information processing device according to any one of configurations 1 to 17, further comprising an output control unit that controls output so as not to output information indicating shooting conditions related to the first correction data and the second correction data when outputting a report including the bone density. (Configuration 20) 20. The information processing device according to any one of configurations 1 to 19, wherein the first imaging conditions include a first imaging condition of high energy and low energy, and the second imaging conditions include a second imaging condition of high energy and low energy. (Configuration 21) the data acquisition unit further acquires the first object data and the first correction data; the generating unit generates a bone image using the first object data and the first correction data; The information processing device of any one of configurations 1 to 20, further comprising a calibration value acquisition unit that acquires the bone density calibration value using a bone image generated using the first subject data and the first correction data and the specified bone density of the first subject. (Configuration 22) A radiation generating device that generates radiation; a radiation detector that detects the radiation and generates a radiation image; An information processing device according to any one of configurations 1 to 21; A radiation imaging system comprising: (Method 1) acquiring a bone density calibration value using first subject data obtained by photographing a first subject having a predetermined bone density under a first photographing condition and first correction data obtained by photographing the first subject under the first photographing condition without placing the first subject; acquiring second subject data obtained by imaging a second subject, which is a bone density measurement target, under second imaging conditions, and second correction data obtained by imaging the second subject under the second imaging conditions without placing the second subject; generating a bone image using the second object data and the second corrected data; acquiring a bone density of the second subject using the bone image and the bone density calibration value; An information processing method comprising: (Program 1) A program that, when executed by a computer, causes the computer to execute each step of the information processing method described in Method 1.

[0196] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above embodiments. Inventions modified within the scope of the present disclosure and inventions equivalent to the present disclosure are also included in the present disclosure. In addition, the above-mentioned embodiments can be appropriately combined within the scope of the present disclosure. [Explanation of symbols]

[0197] 103: second subject data acquisition unit (data acquisition unit), 104: second correction data acquisition unit (data acquisition unit), 105: bone image generation unit (generation unit), 106: bone density calibration value calculation unit (data acquisition unit), 108: bone density calculation unit (bone density acquisition unit)

Claims

1. a data acquisition unit that acquires a bone density calibration value obtained using first subject data obtained by photographing a first subject having a predetermined bone density under a first photographing condition and first correction data obtained by photographing the first subject under the first photographing condition without placing the first subject, second subject data obtained by photographing a second subject to be subjected to bone density measurement under a second photographing condition, and second correction data obtained by photographing the second subject under the second photographing condition without placing the second subject; a generation unit that generates a bone image using the second object data and the second corrected data; a bone density acquisition unit that acquires a bone density of the second subject using the bone image and the bone density calibration value; An information processing device comprising:

2. The data acquisition unit further acquires the first correction data, The information processing device further includes an analysis unit that analyzes the first corrected data and the second corrected data and acquires feature amounts of the first corrected data and the second corrected data, The information processing apparatus according to claim 1 , wherein the generating section generates the bone image by using the second object data, the second corrected data, and the feature amount.

3. The information processing apparatus according to claim 2 , wherein the analysis unit acquires a pixel value distribution as the feature amount.

4. 4. The information processing device according to claim 3, wherein the generation unit generates the bone image using a correction coefficient that converts a pixel value distribution of the second correction data into a pixel value distribution of the first correction data obtained using the feature amount, the second subject data, and the second correction data.

5. The information processing apparatus according to claim 4 , wherein the generation section corrects the second corrected data by using the correction coefficient, and generates the bone image by using the corrected second corrected data and the second object data.

6. The information processing device according to claim 3 , wherein the analysis unit acquires the pixel value distribution using multidimensional surface fitting.

7. The information processing apparatus according to claim 6 , wherein the generating unit does not use the feature amount for generating the bone image when a fitting error of the surface fitting is equal to or greater than a threshold value.

8. The information processing apparatus according to claim 2 , wherein the analysis unit performs the analysis within an exposure field region of the second corrected data and within an exposure field region of the first corrected data.

9. The information processing apparatus according to claim 2 , wherein the analysis unit acquires an amount of noise as the amount of features.

10. 10. The information processing device according to claim 9, wherein the generation unit generates the bone image using a noise adjustment filter for converting an amount of noise in the second corrected data acquired using the feature amount into an amount of noise in the first corrected data, the second subject data, and the second corrected data.

11. The analysis unit further acquires a pixel value distribution as the feature amount, 11. The information processing device according to claim 10, wherein the generation unit generates the bone image using a correction coefficient for converting a pixel value distribution of the second correction data into a pixel value distribution of the first correction data, the second subject data, the second correction data, and the noise adjustment filter, the correction coefficient being obtained using the pixel value distributions of the first correction data and the second correction data.

12. an analysis unit that analyzes the first correction data and the second correction data and obtains an average value of pixel values ​​of the first correction data and an average value of pixel values ​​of the second correction data, the data acquisition unit acquires a plurality of bone mineral density calibration values ​​corresponding to a dose, the bone mineral density calibration values ​​being obtained using a plurality of first object data and a plurality of first correction data ... correction data and The analysis unit compares an average value of pixel values ​​of the first corrected data with an average value of pixel values ​​of the second corrected data, The bone density acquisition unit includes: obtaining a first bone density using a bone density calibration value corresponding to an exposure dose used in capturing the second correction data and the bone image among the plurality of bone density calibration values; acquiring a bone density calibration value corresponding to a dose selected from the plurality of bone density calibration values ​​according to a comparison result between an average pixel value of the first correction data and an average pixel value of the second correction data, and a second bone density using the bone image; The information processing apparatus according to claim 1 , wherein the bone density is obtained by performing a weighted addition of the first bone density and the second bone density.

13. The data acquisition unit further acquires the first correction data, A correction unit is further provided that uses the first correction data, the second correction data, and the bone image as input data for a trained model to obtain a corrected bone image, The information processing apparatus according to claim 1 , wherein the bone density acquisition unit acquires the bone density by using the corrected bone image and the bone density calibration value.

14. 14. The information processing device according to claim 13, wherein the trained model is a trained model obtained using training data in which third corrected data obtained under third shooting conditions, fourth corrected data obtained under fourth shooting conditions, and a bone image obtained by photographing a subject under the fourth shooting conditions are used as input data, and the bone image obtained by photographing the subject under the third shooting conditions is used as correct data.

15. The information processing device according to claim 1 , further comprising a message output unit that outputs a message indicating that there is insufficient data to be used for bone density measurement when at least one of the second correction data and the second subject data cannot be acquired.

16. The information processing device according to claim 2, further comprising a message output unit that outputs a message indicating that bone density accuracy is decreasing or a message prompting the user to re-acquire the first subject data and the first correction data when the acquisition accuracy of the feature amount is below a threshold value.

17. The data acquisition unit further acquires the first correction data, An analysis unit that analyzes the first correction data and the second correction data to obtain a dose ratio of the first correction data and the second correction data, The information processing apparatus according to claim 1 , wherein the bone density acquisition unit acquires the bone density by using a calibrated bone density value obtained by correcting the calibrated bone density value in accordance with the dose ratio and the bone image.

18. The information processing apparatus according to claim 1 , further comprising an output control unit that controls output so as not to output the first correction data and the second correction data when outputting an image.

19. The information processing device according to claim 1 , further comprising an output control unit that controls output so as not to output information indicating shooting conditions related to the first correction data and the second correction data when outputting a report including the bone density.

20. The information processing apparatus according to claim 1 , wherein the first imaging condition includes a first imaging condition of high energy and low energy, and the second imaging condition includes a second imaging condition of high energy and low energy.

21. the data acquisition unit further acquires the first object data and the first correction data; the generating unit generates a bone image using the first object data and the first correction data; The information processing device according to claim 1 , further comprising a calibration value acquisition unit that acquires the bone density calibration value using a bone image generated using the first subject data and the first correction data and the specified bone density of the first subject.

22. A radiation generating device that generates radiation; a radiation detector that detects the radiation and generates a radiation image; An information processing device according to any one of claims 1 to 21; A radiation imaging system comprising:

23. acquiring a bone density calibration value using first subject data obtained by photographing a first subject having a predetermined bone density under a first photographing condition and first correction data obtained by photographing the first subject under the first photographing condition without placing the first subject; acquiring second subject data obtained by imaging a second subject, which is a bone density measurement target, under second imaging conditions, and second correction data obtained by imaging the second subject under the second imaging conditions without placing the second subject; generating a bone image using the second object data and the second corrected data; acquiring a bone density of the second subject using the bone image and the bone density calibration value; An information processing method comprising:

24. A program which, when executed by a computer, causes the computer to execute each step of the information processing method according to claim 23.

Images