Image processing device, method and program

Abstract

To enable correct evaluation of a state of a bone in an image processing device, method and program.SOLUTION: A processor specifies a target bone, which is a target of state evaluation of a bone, from a radiation image of a subject, determines whether the state evaluation of the bone is possible based on a specification result of the target bone, derives an evaluation result by performing the state evaluation of the bone using the target bone in a case where it is determined that the state evaluation of the bone is possible, and performs the state evaluation of the bone by changing an evaluation method in a case where it is determined that the state evaluation of the bone is not possible.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present disclosure relates to an image processing device, a method, and a program. [Background technology]

[0002] Bone diseases such as osteoporosis cannot be detected from the outside, and are only discovered when a bone fracture occurs. In particular, fractures of the femur and vertebrae, where osteoporosis is common, are likely to leave patients bedridden. It is known that the 5-year survival rate for patients who become bedridden is lower than the 5-year survival rate for cancer. For this reason, it is desirable to find osteoporosis patients who need treatment before a fracture occurs.

[0003] On the other hand, DXA (Dual X-ray Absorptiometry) is known as one of the representative bone mineral quantification methods used for the diagnosis of bone mineral density in bone disease. A method has been proposed in which a bone image showing the bones of a subject and a soft tissue image showing the soft tissue are obtained by energy subtraction processing, bone mineral content, i.e., bone mineral density, is derived from the bone image, muscle mass is derived from the soft tissue image, and the subject's fracture risk is evaluated based on the bone mineral content and muscle mass (see Patent Document 1). A method has also been proposed in which an evaluation value indicating the subject's fracture risk is calculated based on structural parameters that quantify the characteristics of the subject's bone mineral content and cancellous structure composed of bone trabeculae (see Patent Document 2). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] International Publication No. 2020 / 166561 [Patent Document 2] International Publication No. 2016 / 129682 Summary of the Invention [Problem to be solved by the invention]

[0005] When evaluating bone density, which is necessary for determining the risk of fracture, using a bone image, if an area of ​​bone that has already been fractured or an artificial object such as a screw used for treating a fracture is included, the patient's correct bone density cannot be calculated. For example, an area that has already been fractured will have a higher apparent bone density due to crushing. In addition, since artificial objects are often made of materials with a higher density than bone, for example, when calculating the average value of the bone density of a certain area, if the area contains an artificial object, the bone density will be overestimated. In this way, if the condition of the bone cannot be evaluated correctly, the risk of fracture and even the efficacy of a drug used to treat osteoporosis cannot be correctly determined. For this reason, it is possible to evaluate the condition of the bone by excluding fractures and artificial objects. However, if all or most of the bones to be evaluated, such as the lumbar vertebrae or proximal femur, are fractured or contain artificial objects, the condition of the bone cannot be correctly evaluated.

[0006] The present disclosure has been made in consideration of the above circumstances, and aims to enable correct evaluation of bone conditions. [Means for solving the problem]

[0007] The image processing device according to the present disclosure includes at least one processor, the processor identifying a target bone to be subjected to bone condition evaluation from a radiological image of a subject, determining whether or not the condition of the bone can be evaluated based on the result of identifying the target bone; If it is determined that bone condition evaluation is possible, an evaluation result is derived by performing a bone condition evaluation using the target bone; If it is determined that the bone condition cannot be evaluated, the evaluation method is changed and the bone condition is evaluated.

[0008] In the image processing device according to the present disclosure, the processor may specify the target bone by excluding fractures and artificial objects in the radiographic image.

[0009] In addition, in the image processing device according to the present disclosure, the processor may determine whether or not the condition of the bone can be evaluated based on whether or not a target bone not including a fracture or an artificial object is included in the radiological image to a degree equal to or greater than a predetermined standard.

[0010] In addition, in the image processing device according to the present disclosure, if it is determined that bone condition evaluation is not possible, the processor may change the evaluation method so as to evaluate the bone condition using an additional target bone of a different type from the target bone.

[0011] In addition, in the image processing device according to the present disclosure, when multiple types of additional target bones are prepared, the processor may change the evaluation method so as to perform bone condition evaluation using the additional target bones by adding different types of target bones according to a predetermined order.

[0012] In addition, in the image processing device according to the present disclosure, the processor may be configured to notify the user that bone condition evaluation is not possible when it is determined that bone condition evaluation is not possible for all of the multiple types of additional target bones.

[0013] In the image processing device according to the present disclosure, the radiation image may be a bone image in which the bones of the subject are emphasized.

[0014] In addition, in the image processing device according to the present disclosure, the processor may derive a bone image from a first radiographic image and a second radiographic image acquired by photographing the subject with radiation having different energy distributions.

[0015] In addition, in the image processing device according to the present disclosure, the processor may derive an evaluation value indicating the likelihood of osteoporosis or a judgment result as to whether or not osteoporosis occurs as an evaluation result based on image information of the region of the target bone in the radiation image.

[0016] In addition, in the image processing device according to the present disclosure, the processor may derive at least one of the bone density of the target bone, information on the microstructure of the target bone, and the relationship with adjacent bones of the target bone as an index representing the condition of the bone, and derive an evaluation result based on the index.

[0017] In addition, in the image processing device according to the present disclosure, the processor may derive a representative value of the bone density of the target bone as the index.

[0018] In addition, in the image processing device according to the present disclosure, the processor may derive a representative value of bone density for each of a plurality of lines in a predetermined direction in the target bone, and derive the representative value of the representative values ​​for each line as an index.

[0019] In addition, in the image processing device according to the present disclosure, the processor may display the radiographic image in which the target bone is enhanced and the evaluation result.

[0020] The image processing method according to the present disclosure includes a step of: identifying a target bone to be subjected to bone condition evaluation from a radiological image of a subject; determining whether or not the condition of the bone can be evaluated based on the result of identifying the target bone; If it is determined that bone condition evaluation is possible, a bone condition evaluation is performed using the target bone to derive an evaluation result; If it is determined that the bone condition cannot be evaluated, the evaluation method is changed and the bone condition is evaluated.

[0021] The image processing program according to the present disclosure includes a step of identifying a target bone that is a target for bone condition evaluation from a radiation image of a subject; A step of determining whether or not a bone condition can be evaluated based on the result of identifying the target bone; a step of deriving an evaluation result by performing a bone condition evaluation using a target bone when it is determined that bone condition evaluation is possible; If it is determined that the bone condition cannot be evaluated, the computer is caused to execute a procedure of changing the evaluation method and evaluating the bone condition. Effect of the Invention

[0022] According to the present disclosure, the condition of bones can be correctly evaluated. [Brief description of the drawings]

[0023] [Figure 1] FIG. 1 is a schematic block diagram showing a configuration of a radiation image capturing system to which an image processing device according to an embodiment of the present disclosure is applied; [Diagram 2] FIG. 1 is a diagram showing a schematic configuration of an image processing apparatus according to an embodiment of the present invention; [Diagram 3] FIG. 1 is a diagram showing a functional configuration of an image processing apparatus according to an embodiment of the present invention; [Figure 4] Figure showing bone images [Diagram 5] FIG. 1 is a diagram showing an example of a neural network used in the present embodiment; [Figure 6] Image showing bones including the compression fracture of the third lumbar vertebra [Figure 7] FIG. 1 shows an image of a bone including an artificial object. [Figure 8] A diagram showing the training data used to train trained model 23A. [Figure 9] Image showing bones including the compression fracture of the third lumbar vertebra [Figure 10] FIG. 1 shows an image of a bone including an artificial object. [Figure 11] A diagram showing the training data used to train trained model 26A. [Figure 12] A diagram showing the relationship between bone and soft tissue contrast and body thickness. [Figure 13] FIG. 1 shows a lookup table for obtaining correction coefficients. [Figure 14] FIG. 1 is a diagram for explaining derivation of a representative value of bone density for each line. [Figure 15] FIG. 1 is a diagram for explaining the derivation of the evaluation result of the positional relationship of bones. [Figure 16] A diagram showing the evaluation result display screen. [Figure 17] A flowchart showing the processing performed in this embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0024] Hereinafter, an embodiment of the present disclosure will be described with reference to the drawings. Fig. 1 is a schematic block diagram showing a configuration of a radiographic image capturing system to which an image processing device according to an embodiment of the present disclosure is applied. As shown in Fig. 1, the radiographic image capturing system according to this embodiment includes an image capturing device 1 and an image processing device 10 according to this embodiment.

[0025] The imaging device 1 is an imaging device for performing energy subtraction by a so-called one-shot method in which radiation such as X-rays emitted from a radiation source 3 and transmitted through a subject H is irradiated to a first radiation detector 5 and a second radiation detector 6 with different energies. During imaging, as shown in Fig. 1, the first radiation detector 5, a radiation energy converting filter 7 made of a copper plate or the like, and the second radiation detector 6 are arranged in this order from the side closest to the radiation source 3, and the radiation source 3 is driven. The first and second radiation detectors 5 and 6 are in close contact with the radiation energy converting filter 7.

[0026] As a result, the first radiation detector 5 acquires a first radiation image G1 of the subject H using low-energy radiation including so-called soft rays. Meanwhile, the second radiation detector 6 acquires a second radiation image G2 of the subject H using high-energy radiation from which the soft rays have been removed. The first and second radiation images are input to the image processing device 10.

[0027] The first radiographic image G1 and the second radiographic image G2 are, for example, front images including the chest and abdomen of the subject H. The image processing device according to this embodiment evaluates the bones of the subject H. The bones are evaluated using the lumbar vertebrae or the proximal femur region included in the radiographic images. Therefore, the first and second radiographic images G2 may be obtained by capturing an image of the lower abdomen of the subject H. Here, since the lumbar vertebrae are included in the front image including the chest and abdomen, the bones are evaluated using the lumbar vertebrae in this embodiment as well.

[0028] The first and second radiation detectors 5, 6 are capable of repeatedly recording and reading out radiation images, and may be so-called direct type radiation detectors that generate electric charges by directly receiving radiation, or so-called indirect type radiation detectors that convert radiation into visible light and then convert the visible light into an electric charge signal. As a method for reading out the radiation image signal, it is preferable to use a so-called TFT readout method in which the radiation image signal is read out by turning on and off a TFT (thin film transistor) switch, or a so-called optical readout method in which the radiation image signal is read out by irradiating the detector with readout light, but this is not limiting and other methods may also be used.

[0029] Next, an image processing device according to the present embodiment will be described. First, a hardware configuration of the image processing device according to the present embodiment will be described with reference to FIG. 2. As shown in FIG. 2, the image processing device 10 is a computer such as a workstation, a server computer, or a personal computer, and includes a CPU (Central Processing Unit) 11, a non-volatile storage 13, and a memory 16 as a temporary storage area. The image processing device 10 also includes a display 14 such as a liquid crystal display, an input device 15 such as a keyboard and a mouse, and a network I / F (InterFace) 17 connected to a network (not shown). The CPU 11, the storage 13, the display 14, the input device 15, the memory 16, and the network I / F 17 are connected to a bus 18. The CPU 11 is an example of a processor in the present disclosure.

[0030] The storage 13 is realized by a hard disk drive (HDD), a solid state drive (SSD), a flash memory, etc. The storage 13 as a storage medium stores the image processing program 12 installed in the image processing device 10. The CPU 11 reads out the image processing program 12 from the storage 13, loads it into the memory 16, and executes the loaded image processing program 12.

[0031] The image processing program 12 is stored in a storage device of a server computer connected to a network or in a network storage in an externally accessible state, and is downloaded and installed in a computer constituting the image processing device 10 upon request. Alternatively, the program is recorded on a recording medium such as a DVD (Digital Versatile Disc) or a CD-ROM (Compact Disc Read Only Memory) and distributed, and is installed in a computer constituting the image processing device 10 from the recording medium.

[0032] Next, the functional configuration of the image processing device according to this embodiment will be described. Fig. 3 is a diagram showing the functional configuration of the image processing device according to this embodiment. As shown in Fig. 3, the image processing device 10 includes an image acquisition unit 21, a bone image derivation unit 22, a target bone identification unit 23, a possible / probable determination unit 24, a change unit 25, an evaluation unit 26, and a display control unit 27. The CPU 11 executes the image processing program 12 to function as the image acquisition unit 21, the bone image derivation unit 22, the target bone identification unit 23, the possible / probable determination unit 24, the change unit 25, the evaluation unit 26, and the display control unit 27.

[0033] The image acquisition unit 21 causes the imaging device 1 to perform energy subtraction imaging of the subject H, thereby acquiring a first radiographic image G1 and a second radiographic image G2 of the subject H from the first and second radiation detectors 5, 6. When acquiring the first radiographic image G1 and the second radiographic image G2, imaging conditions such as imaging dose, radiation quality, tube voltage, SID (Source Image Receptor Distance) which is the distance between the radiation source 3 and the surfaces of the first and second radiation detectors 5, 6, SOD (Source Object Distance) which is the distance between the radiation source 3 and the surface of the subject H, and the presence or absence of an anti-scatter grid are set.

[0034] The SOD and SID are used to calculate the body thickness distribution as described below. The SOD is preferably obtained by, for example, a TOF (Time Of Flight) camera. The SID is preferably obtained by, for example, a potentiometer, an ultrasonic distance meter, a laser distance meter, or the like.

[0035] The imaging conditions may be set by the operator through the input device 15 .

[0036] The bone image derivation section 22 derives a bone image Gb in which the bones of the subject H are enhanced, from the first radiographic image G1 and the second radiographic image G2 acquired by the image acquisition section 21.

[0037] Specifically, the bone image derivation unit 22 performs weighted subtraction between corresponding pixels of the first radiographic image G1 and the second radiographic image G2 as shown in the following formula (1) to derive a bone image Gb in which only the bones of the subject H included in the first radiographic image G1 and the second radiographic image G2 are extracted. In the following formula (1), μb is a weighting coefficient determined based on the attenuation coefficients of the soft tissue and bone, and x and y are the coordinates of each pixel of the bone image Gb. The bone image Gb derived by formula (1) includes not only bones but also objects whose attenuation coefficients are similar to those of bones. An example of an object whose attenuation coefficient is similar to that of bones is metal embedded in bones for treating a fracture or the like. Gb(x,y)=G1(x,y)-μb×G2(x,y) (1)

[0038] Fig. 4 is a diagram showing an example of a bone image Gb derived by the bone image derivation unit 22. Generally, the femoral neck and lumbar vertebrae are used for measuring bone density. In this embodiment, the first and second radiographic images G1 and G2 are front images of the subject's chest and abdomen. Therefore, the bone image Gb shown in Fig. 4 includes a part of the lumbar vertebrae.

[0039] Here, each of the first radiographic image G1 and the second radiographic image G2 contains scattered ray components based on radiation scattered within the subject H in addition to primary ray components of the radiation that has passed through the subject H. For this reason, it is preferable to remove the scattered ray components from the first radiographic image G1 and the second radiographic image G2. The method for removing the scattered ray components is not particularly limited, and for example, the method described in JP 2015-043959 A may be applied to remove the scattered ray components from the first radiographic image G1 and the second radiographic image G2. When using the method described in JP 2015-043959 A or the like, the body thickness distribution of the subject H and the scattered ray components for removing the scattered ray components are derived simultaneously.

[0040] Hereinafter, the removal of the scattered radiation components from the first radiographic image G1 will be described, but the removal of the scattered radiation components from the second radiographic image G2 can be performed in the same manner. First, the bone image derivation unit 22 acquires a virtual model of the subject H having an initial body thickness distribution T0(x,y). The virtual model is data virtually representing the subject H in which the body thickness according to the initial body thickness distribution T0(x,y) is associated with the coordinate position of each pixel of the first radiographic image G1. The virtual model of the subject H having the initial body thickness distribution T0(x,y) may be stored in advance in the storage 13 of the image processing device 10. In addition, the bone image derivation unit 22 may calculate the body thickness distribution T(x,y) of the subject H based on the SID and SOD included in the imaging conditions. In this case, the initial body thickness distribution T0(x,y) can be obtained by subtracting the SOD from the SID.

[0041] Next, the bone image derivation unit 22 generates an estimated image based on the virtual model by combining an estimated primary ray image obtained by estimating the primary ray image obtained by photographing the virtual model and an estimated scattered ray image obtained by estimating the scattered ray image obtained by photographing the virtual model, as an estimated image based on the first radiation image G1 obtained by photographing the subject H.

[0042] Next, the bone image derivation unit 22 corrects the initial body thickness distribution T0(x,y) of the virtual model so that the difference between the estimated image and the first radiographic image G1 becomes small. The bone image derivation unit 22 repeatedly generates the estimated image and corrects the body thickness distribution until the difference between the estimated image and the first radiographic image G1 satisfies a predetermined end condition. The bone image derivation unit 22 derives the body thickness distribution when the end condition is satisfied as the body thickness distribution T(x,y) of the subject H. In addition, the bone image derivation unit 22 removes the scattered ray components contained in the first radiographic image G1 by subtracting the scattered ray components when the end condition is satisfied from the first radiographic image G1. The bone image derivation unit 22 then derives the bone image Gb using the first and second radiographic images G1 and G2 from which the scattered ray components have been removed.

[0043] The target bone identifying unit 23 identifies a target bone to be subjected to bone condition evaluation from the bone image Gb. In this embodiment, the target bone identifying unit 23 identifies the target bone in the bone image Gb by excluding fractures and artificial objects. In this embodiment, the target bone identifying unit 23 identifies the lumbar vertebrae as the target bone. For this purpose, in this embodiment, a trained model 23A constructed by machine learning a neural network so as to extract normal lumbar vertebrae that do not include fractures and artificial objects is used. In this embodiment, the target bone identifying unit 23 has trained models 23B and 23C in addition to the trained model 23A. The trained models 23B and 23C will be described later.

[0044] When the target bone is the femur, a trained model constructed by machine learning a neural network to extract a normal femur that does not contain a fracture or an artificial object may be used. A fracture means a fractured part, i.e., a fractured portion, and "removing a fracture" includes removing the fractured part from the bone image Gb, removing an area including the fractured part, or removing the bone itself including the fractured part.

[0045] The neural network for constructing the trained model 23A used in this embodiment may be a simple perceptron, a multilayer perceptron, a deep neural network, a convolutional neural network, a deep belief network, a recurrent neural network, a probabilistic neural network, etc. For example, in this embodiment, a convolutional neural network is used as the neural network.

[0046] Fig. 5 is a diagram showing an example of a neural network used in this embodiment. As shown in Fig. 5, the neural network 30 includes an input layer 31, an intermediate layer 32, and an output layer 33. The intermediate layer 32 includes, for example, a plurality of convolution layers 35, a plurality of pooling layers 36, and a fully connected layer 37. In the neural network 30, the fully connected layer 37 exists in the stage before the output layer 33. In the neural network 30, the convolution layers 35 and the pooling layers 36 are alternately arranged between the input layer 31 and the fully connected layer 37.

[0047] The configuration of the neural network 30 is not limited to the example shown in Fig. 5. For example, the neural network 30 may include one convolution layer 35 and one pooling layer 36 between the input layer 31 and the fully connected layer 37.

[0048] The trained model 23A is constructed by machine learning a neural network using images of the lumbar vertebrae containing only normal bones as training data for detecting fractures and artifacts.

[0049] For example, when the third lumbar vertebra L3 has a compression fracture as shown in Fig. 6, the third lumbar vertebra L3 is deformed to become smaller in the body axis direction, and the bone density increases, so that the brightness becomes higher than that of a normal bone. Also, when the second lumbar vertebra L2 includes a bolt as shown in Fig. 7, the bolt image 38 becomes brighter than that of a normal bone because the bolt is made of metal.

[0050] The trained model 23A extracts normal lumbar vertebrae from the bone image Gb by learning the characteristics of normal bones. FIG. 8 is a diagram showing training data used for training the trained model 23A. As shown in FIG. 8, the training data 40 consists of training data 41 and correct answer data 42. The training data 41 is a bone image including only normal lumbar vertebrae. The correct answer data 42 is an image in which the normal lumbar vertebrae area in the training data 41 is masked. Then, by machine learning the neural network using a large amount of such training data 40, the trained model 23A is constructed so as to extract normal lumbar vertebrae when a bone image is input.

[0051] When lumbar vertebrae are extracted from the bone image Gb using such a trained model 23A, the extracted bones are normal bones that do not include fractures or artificial objects. That is, by using such a trained model 23A, lumbar vertebrae other than the third lumbar vertebra L3 are extracted from the bone image Gb shown in FIG. 6. Also, lumbar vertebrae other than the second lumbar vertebra L2 that includes artificial objects are extracted from the bone image Gb shown in FIG. 7. Therefore, the target bone identification unit 23 identifies the lumbar vertebrae detected by the trained model 23A among the lumbar vertebrae extracted from the bone image Gb as target bones that do not include fractures or artificial objects.

[0052] The trained model 23A may be constructed so as to extract fractures or lumbar vertebrae including artificial objects. In this case, the target bone identification unit 23 uses the trained model 23A constructed in this way to identify lumbar vertebrae including fractures or artificial objects in the bone image Gb, and identifies other lumbar vertebrae as target bones. The trained model may be constructed so as to detect only artificial objects from the first radiographic image G1 or the second radiographic image G2. In this case, a region where an artificial object exists may be identified before generating the bone image Gb, and after generating the bone image Gb, the identified region may be excluded or bones including the identified region may be excluded to identify the target bone.

[0053] The target bone specifying unit 23 may specify the target bone based on the shape of the lumbar vertebrae. For example, when viewed from the front, the lumbar vertebrae are symmetrical with respect to the center line in the body axis direction, but as shown in FIG. 9, when the third lumbar vertebra L3 has a compression fracture, the symmetry is lost. For this reason, the target bone specifying unit 23 may derive an index representing the symmetry of the lumbar vertebrae extracted from the bone image Gb, specify the lumbar vertebrae with the loss of symmetry based on the index, and specify the lumbar vertebrae other than the specified lumbar vertebrae as the target bone. Here, the distance from the center line in the body axis direction of the extracted lumbar vertebra to the left and right outer contours of the lumbar vertebrae is derived, and the difference between the distance to the right and the distance to the left from the center line can be used as the index representing the symmetry. If the index representing the symmetry is greater than a predetermined threshold value, the lumbar vertebrae is highly likely to be fractured, and therefore it is possible not to specify the lumbar vertebrae as the target bone. Note that the difference between the areas of the bone regions on the left and right of the center line in the body axis direction may be used as the index representing the symmetry.

[0054] The target bone identification unit 23 may also derive the area difference between each lumbar vertebra extracted from the bone image Gb and at least one of the upper and lower lumbar vertebrae, and identify the target bone based on the difference. Here, the size of the lumbar vertebrae included in the bone image Gb is almost the same as the size of the lumbar vertebrae adjacent above and below. On the other hand, if a compression fracture occurs, the height of the lumbar vertebrae is smaller than the height of the lumbar vertebrae adjacent above and below. For this reason, the target bone identification unit 23 may derive the area difference between each lumbar vertebra extracted from the bone image Gb and, for example, the upper lumbar vertebra, and if the difference is equal to or less than a predetermined threshold value, the lumbar vertebrae is determined to have a compression fracture and is not identified as a target bone.

[0055] The target bone identification unit 23 may also derive the height-to-width ratio for each lumbar vertebrae extracted from the bone image Gb, and identify the target bone based on the result of comparison between the height-to-width ratios derived for the upper and lower lumbar vertebrae. Here, the size of the lumbar vertebrae included in the bone image Gb is almost the same as the shape of the lumbar vertebrae adjacent above and below. On the other hand, when a compression fracture occurs as shown in FIG. 6, the height-to-width ratio of the lumbar vertebrae is significantly different from the height-to-width ratio of normal lumbar vertebrae. For this reason, the target bone identification unit 23 may derive the difference between the ratios of each lumbar vertebrae extracted from the bone image Gb and the upper lumbar vertebrae, and when the difference is equal to or greater than a predetermined threshold value, the lumbar vertebrae is determined to have a compression fracture and is not identified as a target bone.

[0056] The feasibility determination unit 24 determines whether or not the bone condition can be evaluated based on the result of identifying the target bone. In this embodiment, a target bone that does not include a fracture or an artificial object is identified by the target bone identification unit 23. The feasibility determination unit 24 determines whether or not the bone condition can be evaluated based on whether or not the identified target bone is included in the radiographic image in an amount equal to or greater than a predetermined criterion.

[0057] 10, in a bone image Gb including the first lumbar vertebra L1 to the third lumbar vertebra L3 and a part of the fourth lumbar vertebra L4, if the first lumbar vertebra L1 to the third lumbar vertebra L3 are fixed by a metal fitting 45, only a part of the fourth lumbar vertebra L4 will be identified as the target bone by the target bone identifying unit 23. In this case, as will be described later, the bone condition evaluation is performed only on a part of the fourth lumbar vertebra L4, and therefore the bone condition cannot be evaluated with high accuracy.

[0058] For this reason, in the case where the lumbar vertebrae are the bone evaluation target, the feasibility determination unit 24 determines that the bone condition evaluation is possible when, for example, three or more of the five lumbar vertebrae are identified as target bones. On the other hand, the feasibility determination unit 24 determines that the bone condition evaluation is not possible when the bone image Gb contains two or less target bones. When there is one lumbar vertebra that is a target bone that does not contain a fracture or an artificial object as shown in FIG. 10, the feasibility determination unit 24 determines that the bone condition evaluation is not possible. On the other hand, when the bone image Gb contains four or more lumbar vertebrae that are target bones that do not contain a fracture or an artificial object as shown in FIG. 4, the feasibility determination unit 24 determines that the bone condition evaluation is possible.

[0059] The feasibility determination unit 24 may determine whether or not the bone state can be evaluated based on whether or not the area of ​​the region of the target bone in the bone image Gb is equal to or larger than a reference threshold value, instead of the number of lumbar vertebrae.

[0060] The change unit 25 changes the evaluation method when the feasibility determination unit 24 determines that the bone cannot be evaluated. Specifically, the change unit 25 instructs the target bone identification unit 23 to change the type of the target bone and identify the target bone. Also, the change unit 25 instructs the evaluation unit 26 to change the type of the target bone and evaluate the bone state, as described later.

[0061] In this embodiment, the bone image Gb includes the thoracic vertebrae, cervical vertebrae, and clavicle in addition to the lumbar vertebrae. In this embodiment, the target bone identification unit 23 has three trained models 23A to 23C. As described above, the trained model 23A is constructed by machine learning a neural network so as to extract normal lumbar vertebrae that do not include fractures and artifacts. The trained model 23B is constructed by machine learning a neural network so as to extract normal thoracic vertebrae that do not include fractures and artifacts. The trained model 23C is constructed by machine learning a neural network so as to extract normal clavicles that do not include fractures and artifacts. The learning of the neural network for constructing the trained models 23B and 23C is similar to that of the trained model 23A except that the teacher images are the thoracic vertebrae and clavicle, respectively, and therefore a detailed description is omitted here.

[0062] Here, when comparing the lumbar vertebrae, the thoracic vertebrae, and the clavicle, the accuracy of the bone condition evaluation decreases in the order of the lumbar vertebrae, the thoracic vertebrae, and the clavicle. Therefore, in this embodiment, when the feasibility determination unit 24 determines that the bone condition evaluation cannot be performed when the lumbar vertebrae are the target bones, the change unit 25 changes the evaluation method so that the thoracic vertebrae are the target bones instead of the clavicle. Furthermore, when the feasibility determination unit 24 determines that the bone condition evaluation cannot be performed when the thoracic vertebrae are the target bones, the change unit 25 changes the evaluation method so that the clavicle is the target bone.

[0063] Even when the target bone is specified as a thoracic vertebra, the possibility determination unit 24 determines whether or not the bone condition can be evaluated based on whether or not the specified target bone is included in the radiographic image by a predetermined criterion or more, as in the case of specifying a lumbar vertebra. The criterion may be changed depending on whether the target bone is a lumbar vertebra or a thoracic vertebra. For example, there are five lumbar vertebrae, whereas there are twelve thoracic vertebrae. For this reason, if the number is used as the criterion, it may be determined that the bone condition can be evaluated when, for example, eight or more lumbar vertebrae not including fractures or artificial objects are included in the bone image Gb.

[0064] Here, if the feasibility determination unit 24 determines that the bone condition cannot be evaluated even when the thoracic vertebrae are the target bones, the change unit 25 changes the evaluation method to change the target bone to the clavicle. There are two clavicles, one on the left and one on the right of the human body. Therefore, when the clavicle is the target bone, the feasibility determination unit 24 may determine that the bone condition can be evaluated when one or more clavicles that do not contain fractures or artificial objects are identified as the target bone. If the feasibility determination unit 24 determines that the bone condition cannot be evaluated even when the clavicle is the target bone, the change unit 25 outputs information indicating that the bone condition cannot be evaluated to the display control unit 27.

[0065] The evaluation unit 26 derives an evaluation result representing the state of the bones of the subject H based on the target bone. In this embodiment, the evaluation unit 26 has trained models 26A to 26C that output an evaluation value representing the likelihood of osteoporosis when image information of the region of the target bone in the bone image Gb, i.e., the distribution of pixel values ​​of the target bone, is input. The trained model 26A outputs an evaluation value when the target bone is a lumbar vertebra. The trained model 26B outputs an evaluation value when the target bone is a thoracic vertebra. The trained model 26C outputs an evaluation value when the target bone is a clavicle.

[0066] The trained model 26A is constructed by machine learning a neural network using images of lumbar vertebrae (hereinafter referred to as images of normal lumbar vertebrae) extracted from bone images of patients without osteoporosis and images of lumbar vertebrae of patients with osteoporosis as training data. FIG. 11 is a diagram showing an example of training data for training the trained model 26A. As shown in FIG. 11, first training data 51 representing normal lumbar vertebrae and second training data 52 representing osteoporotic lumbar vertebrae are used as training data for training the trained model 26A. Normal lumbar vertebrae have dense microstructure of bone trabeculae, while osteoporotic lumbar vertebrae have coarser microstructure of bone trabeculae compared to normal lumbar vertebrae.

[0067] The neural network is machine-trained so that the output when first teacher data 51 is input is 0, and the output when second teacher data 52 is input is 1. As a result, trained model 26A is constructed so that when an input image of a lumbar spine indicates osteoporosis, an evaluation value close to 1 is output. In this case, the neural network learns the distribution of pixel values ​​of the lumbar spine image and the microstructure of the bone trabeculae, and trained model 26A is constructed so as to output an evaluation value representing the likelihood of osteoporosis from the distribution of pixel values ​​and the microstructure of the lumbar spine image.

[0068] On the other hand, the trained model 26B is constructed by machine learning a neural network using an image of a thoracic vertebra extracted from a bone image of a patient without osteoporosis and an image of a thoracic vertebra of a patient with osteoporosis as training data. The trained model 26C is constructed by machine learning a neural network using an image of a clavicle extracted from a bone image of a patient without osteoporosis and an image of a clavicle of a patient with osteoporosis as training data.

[0069] The evaluation unit 26 may derive the evaluation value itself output by the trained models 26A to 26C as the evaluation result, or may derive the presence or absence of osteoporosis as the evaluation result by comparing the evaluation value with a threshold value. Even in normal bones that do not have osteoporosis, the bone density differs in the lumbar vertebrae, thoracic vertebrae, and clavicle due to differences in load. For this reason, the threshold value to be compared with the evaluation value may be set to be different for each of the lumbar vertebrae, thoracic vertebrae, and clavicle.

[0070] The evaluation unit 26 may also derive the bone density of the target bone and derive the evaluation result based on the bone density. In this case, the evaluation unit 26 derives the bone density B(x,y) corresponding to each pixel by converting each pixel value Gb(x,y) of the region of the target bone in the bone image Gb into a pixel value of the bone image when acquired under the reference imaging condition. Specifically, the evaluation unit 26 derives the bone density B(x,y) for each pixel by correcting each pixel value Gb(x,y) of the region of the target bone in the bone image Gb using a correction coefficient acquired from a lookup table (not shown) described later.

[0071] Here, the higher the tube voltage in the radiation source 3 and the higher the energy of the radiation emitted from the radiation source 3, the smaller the contrast between soft tissue and bones in the radiological image. In addition, in the process of the radiation passing through the subject H, low-energy components of the radiation are absorbed by the subject H, causing beam hardening, in which the radiation becomes more energetic. The increase in the energy of the radiation due to beam hardening becomes greater as the body thickness of the subject H increases.

[0072] FIG. 12 is a diagram showing the relationship between the contrast between bones and soft tissues and the body thickness of the subject H. Note that FIG. 12 shows the relationship between the contrast between bones and soft tissues and the body thickness of the subject H at three tube voltages of 80 kV, 90 kV, and 100 kV. As shown in FIG. 12, the higher the tube voltage, the lower the contrast. Also, when the body thickness of the subject H exceeds a certain value, the larger the body thickness, the lower the contrast. Note that the larger the pixel value Gb(x, y) of the bone region in the bone image Gb, the higher the contrast between the bones and soft tissues. Therefore, the relationship shown in FIG. 12 shifts to the high contrast side as the pixel value Gb(x, y) of the bone region in the bone image Gb increases.

[0073] In this embodiment, a lookup table (not shown) for acquiring correction coefficients for correcting the difference in contrast in the bone image Gb according to the tube voltage at the time of imaging and the decrease in contrast due to the influence of beam hardening is stored in the storage 13. The correction coefficients are coefficients for correcting each pixel value Gb(x, y) of the bone image Gb.

[0074] FIG. 13 is a diagram showing an example of a lookup table stored in the storage 13. In FIG. 13, a lookup table LUT1 in which a reference imaging condition is set to a tube voltage of 90 kV is illustrated. As shown in FIG. 13, in the lookup table LUT1, a larger correction coefficient is set as the tube voltage increases and as the subject's body thickness increases. In the example shown in FIG. 13, since the reference imaging condition is a tube voltage of 90 kV, when the tube voltage is 90 kV and the body thickness is 0, the correction coefficient is 1. Note that, although the lookup table LUT1 is shown in two dimensions in FIG. 13, the correction coefficient differs depending on the pixel value of the bone region. Therefore, the lookup table LUT1 is actually a three-dimensional table to which an axis representing the pixel value of the bone region is added.

[0075] The evaluation unit 26 extracts a correction coefficient C0(x,y) for each pixel according to the imaging conditions including the body thickness distribution T(x,y) of the subject H and the set value of the tube voltage stored in the storage 13 from the lookup table LUT1. Then, as shown in the following formula (2), the evaluation unit 26 multiplies each pixel value Gb(x,y) of the region of the target bone in the bone image Gb by the correction coefficient C0(x,y) to derive a bone density B(x,y) for each pixel of the region of the target bone. As a result, a bone density image B having the bone density B(x,y) as a pixel value is derived. The bone density B(x,y) derived in this way represents the pixel value of the bone part of the bone region included in the radiographic image obtained by imaging the subject H with a tube voltage of 90 kV, which is the standard imaging condition, and from which the influence of beam hardening has been removed. B(x,y)=C0(x,y)×Gb(x,y) (2)

[0076] The evaluation unit 26 may derive the low-frequency components of each pixel of the bone image Gb, and derive the bone density B(x, y) based on the low-frequency components.

[0077] The evaluation unit 26 then derives a representative value of the bone density B(x, y) of each pixel in the target bone region as the evaluation result. The representative value may be a sum, an average value, a median value, a maximum value, or a minimum value. When there are multiple target bones, an evaluation result may be derived for each target bone, or a representative value of the representative values ​​for each target bone may be derived as the evaluation result. The representative value is an example of an index representing the bone condition.

[0078] Also, the result of determining the presence or absence of osteoporosis using the representative value of bone density may be used as the evaluation result. In this case, the evaluation unit 26 may determine osteoporosis when the representative value of bone density is equal to or less than a predetermined threshold value. Here, when the bone density is equal to or less than 70% of the YAM value (Young Adult Mean), osteoporosis is determined. Therefore, a value equivalent to 70% of the YAM value may be used as the threshold value. Even in normal bones that do not have osteoporosis, the bone density differs in each of the lumbar vertebrae, thoracic vertebrae, and clavicle due to differences in load. Therefore, the threshold value used when determining whether or not osteoporosis occurs based on the representative value of bone density is set to be different for each of the lumbar vertebrae, thoracic vertebrae, and clavicle.

[0079] Here, when the sum of the bone density of each pixel in the target bone region is derived as the evaluation result of the target bone, the evaluation result corresponds to the bone mass of the target bone. However, if the local bone contour of the target bone deviates from the original contour, the sum of the bone density fluctuates. On the other hand, when the average of the bone density of each pixel in the target bone region is derived, even if the local contour of the target bone deviates from the original contour, the effect on the evaluation result is small. However, if the positioning during shooting is shifted and the lumbar vertebrae rotate around the central axis in the body axis direction with respect to the shooting direction, the area of ​​the lumbar vertebrae in the bone image Gb changes and the pixel value changes, so the average value changes.

[0080] Therefore, the evaluation unit 26 may derive a representative value of bone density for each line in the x-direction or y-direction in the target bone region, and derive the representative value of the representative value for each line as the evaluation result. For example, as shown in FIG. 14, in the x-direction indicated by the line 55 in the second lumbar vertebra L2, the sum of bone density may be derived for each line in order to reduce the influence of area variation due to the deviation of positioning in the body axis direction, i.e., the y-direction, and in the y-direction, i.e., the direction perpendicular to the line 55, the average of the sum of bone density for each line 55 in the x-direction may be derived to reduce the influence of deviation of the contour, and this may be used as the evaluation result. When deriving the average of the sum of bone density for each line, outliers that are significantly different from the sum of bone density for each line may be excluded. In addition, if the sum of bone density for each line is too large or too small, there is a possibility that the target bone has not been detected correctly, so for example, 20% above and below the sum of bone density for each line may be excluded from the calculation of the average.

[0081] In addition, when constructing the trained models 26A to 26C, a bone density image may be used as training data. In this case, the trained models 26A to 26C may be constructed so as to output an evaluation value representing the likelihood of osteoporosis by inputting a bone density image of a target bone region.

[0082] The evaluation unit 26 may also extract a cancellous bone region of the target bone from the bone density image and derive the variance of bone density in the cancellous bone region as the evaluation result. Here, the variance of bone density in the cancellous bone region is an evaluation value that represents the degree of density of the microstructure of the bone trabeculae. The variance of bone density is an example of an index that represents the state of the bone.

[0083] The evaluation unit 26 may derive the positional relationship between the target bone and the bone adjacent to the target bone as an evaluation result. FIG. 15 is a diagram showing a bone image of the lower abdomen of a patient with scoliosis. As shown in FIG. 15, the target bone is the third lumbar vertebra L3 for the bone image Gb of the patient with scoliosis. The evaluation unit 26 derives the center of gravity g3 of the third lumbar vertebra L3 and the center of gravity g2 of the second lumbar vertebra L2 above the third lumbar vertebra L3, and derives a difference value of the positions of the centers of gravity g2 and g3 in the x direction. If the difference value of the positions is equal to or greater than a predetermined threshold value, the evaluation result may be output that the risk of fracture is high because the force is not evenly applied from the vertebrae at the top of the target bone. The difference value itself may be used as an evaluation result of the positional relationship. The difference value is an example of an index representing the state of the bone.

[0084] The evaluation unit 26 may derive an evaluation value representing the likelihood of osteoporosis as an evaluation result based on the representative value of bone density, the variance value of bone density, and the evaluation result of the positional relationship (i.e., the difference value between the positions of the centers of gravity g2 and g3 in the x direction). In this case, a method can be used in which the representative value of bone density, the variance value of bone density, and the evaluation result of the positional relationship are used as a multidimensional vector, and a boundary on a multidimensional plane that distinguishes whether or not there is osteoporosis is set by machine learning. In this case, a trained model for determining whether or not there is osteoporosis from the multidimensional vector can be constructed by machine learning a neural network using the representative value of bone density, the variance value of bone density, and the evaluation result of the positional relationship in the case of osteoporosis, and the representative value of bone density, the variance value of bone density, and the evaluation result of the positional relationship in the case of normality as teacher data.

[0085] The display control unit 27 displays the evaluation results derived by the evaluation unit 26 on the display 14. FIG. 16 is a diagram showing a display screen of the evaluation results. As shown in FIG. 16, a bone density image 61 and evaluation results 62 are displayed on a display screen 60. A bone image Gb may be displayed instead of the bone density image 61. The evaluation results 62 include the presence or absence of osteoporosis and a representative value of the bone density (unit: g / cm) of the target bone. 2 ) is displayed.

[0086] As shown in Fig. 16, the target bone from which the evaluation result was derived is emphasized in the bone density image 61. That is, the lumbar vertebrae other than the third lumbar vertebra L3, which has a compression fracture, are emphasized. Note that in Fig. 16, the target bone is emphasized by applying hatching to it, but this is not limiting. The target bone may also be emphasized by surrounding it with a line, applying color, or the like.

[0087] As described above, when information indicating that bone condition evaluation is not possible is output from the change unit 25, the display control unit 27 notifies the user that bone condition evaluation is not possible by displaying text such as "Bone condition evaluation is not possible" on the display 14. Note that instead of or in addition to a display, the user may be notified that bone condition evaluation is not possible by audio.

[0088] Next, the processing performed in this embodiment will be described. Fig. 17 is a flow chart showing the processing performed in this embodiment. First, the image acquisition unit 21 causes the imaging device 1 to perform imaging and acquires first and second radiographic images G1, G2 having different energy distributions (step ST1). Next, the bone image derivation unit 22 derives a bone image Gb representing the bone region of the subject H from the first radiographic image G1 and second radiographic image G2 acquired by the image acquisition unit 21 (step ST2).

[0089] Next, the target bone identification unit 23 identifies a target bone to be evaluated in the bone image Gb, excluding fractures and artificial objects (step ST3). Next, the feasibility determination unit 24 determines whether or not the bone condition can be evaluated based on the target bone identification result (step ST4). If bone condition evaluation is possible (step ST4: YES), the evaluation unit 26 derives an evaluation result representing the bone condition of the subject H based on the target bone (step ST5). Furthermore, the display control unit 27 displays the evaluation result (step ST6) and ends the process.

[0090] On the other hand, if bone condition evaluation is not possible (step ST4: NO), the feasibility determination unit 24 determines whether it has been determined that bone condition evaluation is not possible for all prepared types of target bones (all target bones not possible?: step ST7). If step ST7 is negative, the change unit 25 changes the evaluation method (step ST8). Specifically, it instructs the target bone identification unit 23 to change the type of target bone and identify the target bone. It also instructs the evaluation unit 26 to change the type of target bone and evaluate the bone condition, as described below. Then, it returns to step ST3, and the processes from step ST3 onwards are repeated.

[0091] If step ST7 is positive, the change unit 25 outputs information representing that the bone condition cannot be evaluated to the display control unit 27. As a result, the display control unit 27 notifies the user that the bone condition cannot be evaluated by displaying the information on the display 14 (step ST9), and ends the process.

[0092] In this manner, in this embodiment, a target bone to be subjected to bone condition evaluation is specified, and whether or not the bone condition can be evaluated is determined based on the result of specifying the target bone. If it is determined that the bone condition can be evaluated, the evaluation result is derived by performing bone condition evaluation using the target bone, and if it is determined that the bone condition cannot be evaluated, the evaluation method is changed to perform bone condition evaluation. Therefore, even if all or most of the target bones to be evaluated are fractured or contain artificial objects, in such cases, the evaluation method can be changed, for example, by changing the target bone, to perform bone condition evaluation. Therefore, the bone condition of the subject H can be correctly evaluated without being affected by fractures and artificial objects.

[0093] In addition, if it is determined that bone condition evaluation is not possible for all of the multiple types of additional target bones, a notification to that effect is given so that the user can take measures such as re-photographing subject H in accordance with the notification.

[0094] The evaluation unit 26 may evaluate the evaluation results for the same subject over time. For example, for a patient taking medicine for osteoporosis, the efficacy of the medicine may be evaluated by comparing the past evaluation results with the latest evaluation results.

[0095] The evaluation unit 26 may further derive a fracture risk as an evaluation result. In this case, for example, a method of deriving a fracture risk using a trained model as described in JP-A-09-508813 can be used. In addition, a method of deriving a fracture probability using an arithmetic formula as described in WO 2020 / 054738 can be used.

[0096] In the above embodiment, the bone image Gb is derived by energy subtraction processing, but the present invention is not limited to this. For example, the bone image Gb may be derived from one radiographic image using a trained model that has been machine-trained to enhance the bone region in the radiographic image.

[0097] In the above embodiment, the first and second radiographic images G1, G2 are acquired by a one-shot method when performing the energy subtraction process, but the present invention is not limited to this. The first and second radiographic images G1, G2 may be acquired by a so-called two-shot method in which imaging is performed twice using only one radiation detector. In the case of the two-shot method, the position of the subject H contained in the first radiographic image G1 and the second radiographic image G2 may be shifted due to the body movement of the subject H. For this reason, it is preferable to perform the process of this embodiment after aligning the position of the subject in the first radiographic image G1 and the second radiographic image G2.

[0098] In the above embodiment, the visceral fat mass distribution is derived using the first and second radiographic images acquired in a system that captures the image of the subject H using the first and second radiation detectors 5 and 6. However, the visceral fat mass distribution may be derived using the first and second radiographic images G1, G2 acquired using a stimulable phosphor sheet instead of a radiation detector. In this case, two stimulable phosphor sheets are stacked and irradiated with radiation that has passed through the subject H, and the radiographic image information of the subject H is accumulatively recorded on each stimulable phosphor sheet, and the radiographic image information is photoelectrically read from each stimulable phosphor sheet to acquire the first and second radiographic images G1, G2. The two-shot method may also be used when acquiring the first and second radiographic images G1, G2 using the stimulable phosphor sheets.

[0099] Moreover, the radiation in the above embodiment is not particularly limited, and in addition to X-rays, α-rays, γ-rays, etc. can be used.

[0100] In the above embodiment, the following various processors can be used as the hardware structure of the processing unit that executes various processes, such as the image acquisition unit 21, bone image derivation unit 22, target bone identification unit 23, possibility determination unit 24, change unit 25, evaluation unit 26, and display control unit 27. As described above, the above various processors include a CPU, which is a general-purpose processor that executes software (programs) and functions as various processing units, as well as a programmable logic device (PLD), which is a processor whose circuit configuration can be changed after manufacture, such as an FPGA (Field Programmable Gate Array), and a dedicated electric circuit, such as an ASIC (Application Specific Integrated Circuit), which is a processor having a circuit configuration designed specifically for executing specific processes.

[0101] A single processing unit may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (e.g., a combination of multiple FPGAs or a combination of a CPU and an FPGA). Also, multiple processing units may be configured with a single processor.

[0102] As an example of configuring multiple processing units with one processor, first, there is a form in which one processor is configured with a combination of one or more CPUs and software, as typified by computers such as client and server, and this processor functions as multiple processing units. Second, there is a form in which a processor is used that realizes the functions of the entire system including multiple processing units with one IC (Integrated Circuit) chip, as typified by System On Chip (SoC), etc. In this way, the various processing units are configured using one or more of the above various processors as a hardware structure.

[0103] Furthermore, more specifically, the hardware structure of these various processors can be an electric circuit that combines circuit elements such as semiconductor elements.

[0104] The following are appended claims to this disclosure. (Additional note 1) at least one processor; The processor, Identifying a target bone to be evaluated for bone condition from the radiological image of the subject; determining whether or not a state of the bone can be evaluated based on the result of identifying the target bone; if it is determined that the bone condition evaluation is possible, deriving an evaluation result by performing the bone condition evaluation using the target bone; When it is determined that the bone condition cannot be evaluated, the image processing device changes the evaluation method and evaluates the bone condition. (Additional note 2) 2. The image processing device according to claim 1, wherein the processor identifies the target bone by excluding fractures and artificial objects in the radiation image. (Additional note 3) The image processing device according to claim 2, wherein the processor determines whether or not to evaluate the condition of the bone based on whether or not the target bone, excluding the fracture and the artificial object, is included in the radiological image to a degree equal to or greater than a predetermined criterion. (Additional note 4) The image processing device according to any one of claims 1 to 3, wherein the processor, when it is determined that the condition of the bone cannot be evaluated, changes the evaluation method so as to evaluate the condition of the bone using an additional target bone of a different type from the target bone. (Additional note 5) The image processing device according to claim 4, wherein the processor changes the evaluation method when multiple types of additional target bones are prepared, so as to evaluate the condition of the bone using the additional target bones by adding the different types of target bones in a predetermined order. (Additional note 6) The image processing device according to claim 5, wherein the processor notifies the user that bone condition evaluation is not possible when it is determined that bone condition evaluation is not possible for all of the multiple types of additional target bones. (Additional note 7) 7. The image processing device according to claim 1, wherein the radiation image is a bone image in which bones of the subject are emphasized. (Additional note 8) 8. The image processing device according to claim 7, wherein the processor derives the bone image from a first radiographic image and a second radiographic image acquired by photographing the subject with radiation having different energy distributions. (Additional note 9) The image processing device according to any one of claims 1 to 8, wherein the processor derives an evaluation value representing the likelihood of osteoporosis or a judgment result as to whether or not the bone is osteoporotic as the evaluation result based on image information of the region of the target bone in the radiological image. (Additional note 10) The image processing device according to any one of claims 1 to 8, wherein the processor derives at least one of the bone density of the target bone, information on the microstructure of the target bone, and a relationship between the target bone and adjacent bones as an index representing the condition of the bone, and derives the evaluation result based on the index. (Additional note 11) The image processing device according to claim 10, wherein the processor derives a representative value of the bone density in the target bone as the index. (Additional note 12) The image processing device described in Appendix 11, wherein the processor derives the representative value of the bone density for each of a plurality of lines in a predetermined direction in the target bone, and derives the representative value of the representative values ​​for each of the lines as the index. (Additional note 13) 13. The image processing device according to claim 1, wherein the processor displays the radiographic image in which the target bone is enhanced and the evaluation result. (Additional note 14) The computer identifies a target bone to be evaluated for bone condition from the radiological image of the subject, determining whether or not a state of the bone can be evaluated based on the result of identifying the target bone; if it is determined that the bone condition evaluation is possible, deriving an evaluation result by performing the bone condition evaluation using the target bone; an image processing method for changing the evaluation method and evaluating the bone condition when it is determined that the bone condition cannot be evaluated; (Additional note 15) A step of identifying a target bone to be subjected to bone condition evaluation from a radiological image of a subject; determining whether or not a state of the bone can be evaluated based on a result of identifying the target bone; a step of deriving an evaluation result by performing the bone condition evaluation using the target bone when it is determined that the bone condition evaluation is possible; and if it is determined that the bone condition cannot be evaluated, changing the evaluation method and evaluating the bone condition. [Explanation of symbols]

[0105] 1. Imaging device 3 Radiation source 5, 6 Radiation detectors 7 Radiation Energy Conversion Filter 10 Image processing device 11 CPU 12 Image Processing Programs 13. Storage 14 Display 15 Input Devices 16 Memory 17 Network Interface 18 Bus 21 Image acquisition unit 22 Bone image extraction unit 23 Target bone identification section 23A~23C Trained models 24 Possibility judgment section 25 Changes 26 Evaluation Section 26A~26C Trained models 27 Display control section 30 Neural Networks 31 Input Layer 32 Middle Class 33 Output layer 35 Convolutional Layer 36 Pooling Layer 38 Bolt Statue 40 Teacher Data 41 Training Data 42 Correct data 45 Metal fittings 51 First Teacher Data 52 Second Teacher Data 55 Line 60 Evaluation result display screen 61 Bone Density Imaging 62 Evaluation Results G1 First radiological image G2 Second radiographic image Gb bone image g2,g3 center of gravity H Subject L2 2nd lumbar vertebra L3 3rd lumbar vertebra

Claims

1. At least one processor; The processor, Identifying a target bone to be evaluated for bone condition from the radiological image of the subject; determining whether or not a state of the bone can be evaluated based on the result of identifying the target bone; if it is determined that the bone condition evaluation is possible, deriving an evaluation result by performing the bone condition evaluation using the target bone; When it is determined that the bone condition cannot be evaluated, the image processing device changes the evaluation method and evaluates the bone condition.

2. The image processing apparatus according to claim 1 , wherein the processor identifies the target bone by excluding fractures and artificial objects in the radiographic image.

3. The image processing device according to claim 2 , wherein the processor determines whether or not the condition of the bone can be evaluated based on whether or not the target bone, excluding the fracture and the artificial object, is included in the radiographic image in an amount equal to or greater than a predetermined criterion.

4. The image processing device according to claim 1 or 2, wherein the processor, when it is determined that the condition of the bone cannot be evaluated, changes the evaluation method so as to evaluate the condition of the bone using an additional target bone of a different type from the target bone.

5. 5. The image processing device according to claim 4, wherein when multiple types of additional target bones are prepared, the processor changes the evaluation method so as to evaluate the condition of the bone using the additional target bones by adding the different types of target bones in a predetermined order.

6. The image processing device according to claim 5 , wherein the processor issues a notification to the effect that bone condition evaluation is not possible for all of the plurality of types of additional target bones when it is determined that the bone condition evaluation is not possible.

7. 3. The image processing apparatus according to claim 1, wherein the radiation image is a bone image in which bones of the subject are emphasized.

8. The image processing device according to claim 7 , wherein the processor derives the bone image from a first radiographic image and a second radiographic image acquired by photographing the subject with radiation having different energy distributions.

9. The image processing device according to claim 1 , wherein the processor derives, as the evaluation result, an evaluation value indicative of likelihood of osteoporosis or a determination result as to whether or not osteoporosis exists, based on image information of the region of the target bone in the radiation image.

10. 3. The image processing device according to claim 1, wherein the processor derives at least one of the bone density of the target bone, information on the microstructure of the target bone, and a relationship between the target bone and adjacent bones as an index representing the condition of the bone, and derives the evaluation result based on the index.

11. The image processing device according to claim 10 , wherein the processor derives a representative value of the bone density in the target bone as the index.

12. The image processing device according to claim 11 , wherein the processor derives the representative value of the bone density for each of a plurality of lines in a predetermined direction in the target bone, and derives the representative value of the representative values ​​for each of the lines as the index.

13. The image processing apparatus according to claim 1 , wherein the processor displays the radiographic image in which the target bone is enhanced and the evaluation result.

14. The computer identifies a target bone to be evaluated for bone condition from the radiological image of the subject, determining whether or not a state of the bone can be evaluated based on the result of identifying the target bone; if it is determined that the bone condition evaluation is possible, deriving an evaluation result by performing the bone condition evaluation using the target bone; an image processing method for changing the evaluation method and evaluating the bone condition when it is determined that the bone condition cannot be evaluated;

15. A step of identifying a target bone to be subjected to bone condition evaluation from a radiological image of a subject; determining whether or not a state of the bone can be evaluated based on a result of identifying the target bone; a step of deriving an evaluation result by performing the bone condition evaluation using the target bone when it is determined that the bone condition evaluation is possible; and if it is determined that the bone condition cannot be evaluated, changing the evaluation method and evaluating the bone condition.

Images