Diagnosis assistance device and method based on artificial intelligence processing of radiographic image

Abstract

The present invention relates to a diagnosis assistance device and method based on artificial intelligence processing of a radiographic image, wherein a disease diagnosis assistance device according to an embodiment of the present invention may comprise an information providing unit that constructs a diagnosis assistance model by learning a training radiographic image and additional information through a first feature processing unit, a second feature processing unit, and a feature fusion processing unit, determines a bone mineral density value and disease classification information with respect to a radiographic image to be diagnosed on the basis of the diagnosis assistance model, and provides disease diagnosis assistance information including the bone mineral density value, the disease classification information, and diagnosis basis information.

Description

AI-based diagnostic assistance device and method for radiographic image processing

[0001] The present invention relates to a diagnostic assistance device and method based on artificial intelligence processing of radiographic images, and more specifically, to a technology that provides diagnostic assistance information to prevent osteoporotic fractures and complications after bone-related surgery by constructing a diagnostic assistance model that predicts bone density through machine learning of radiographic images.

[0002] As people age, bone density gradually decreases, and as bone mineral density (BMD) declines in this way, they develop osteoporosis, in which holes form in the bones.

[0003] Osteoporosis can lead to very serious consequences, such as bones breaking even from minor impacts and failing to heal easily; therefore, periodic checkups are required to prevent osteoporosis or the worsening of symptoms.

[0004] Osteoporosis refers to a condition in which bone mass is excessively reduced compared to normal individuals, and it is a clinical condition accompanied by fractures and deformities of bone shape.

[0005] In other words, osteoporosis is a pathological condition characterized by an abnormal decrease in bone mass, accompanied by fractures of the spine and femur, as well as bone deformities.

[0006] Conventionally, bone density information can be obtained by a medical professional interpreting images of the patient, such as X-rays or ultrasounds.

[0007] However, if the medical professional interpreting the patient's X-ray images has low proficiency, there is a risk of misdiagnosis.

[0008] Conventional bone density measurement methods primarily target the lumbar spine and hip joints, making it difficult to accurately predict shoulder bone density.

[0009] In addition, low shoulder bone density can lead to osteoporotic humeral fractures and postoperative complications, but there was a lack of technical means to predict this in advance.

[0010] With the development of artificial intelligence, there is a need to consider technology that obtains bone density information by having AI interpret X-ray images of patients.

[0011] The present invention aims to provide diagnostic assistance information to prevent osteoporotic fractures and complications after bone-related surgery by constructing a diagnostic assistance model that predicts bone density using machine learning on radiographic images.

[0012] The present invention aims to improve the effectiveness of patient treatment, reduce medical costs, and provide information to assist doctors in establishing more accurate diagnoses and treatment plans by accurately predicting bone density in bone parts such as the shoulder, thereby preventing osteoporotic humeral fractures and postoperative complications in advance.

[0013] The present invention aims to enable medical personnel to obtain diagnostic assistance information to prevent osteoporotic fractures and complications after bone-related surgery solely through radiographic images.

[0014] A disease diagnosis assistance device according to one embodiment of the present invention may include an input processing unit that receives a learning radiographic image, additional information regarding the learning radiographic image, and a radiation image to be diagnosed; an artificial intelligence processing unit that learns the learning radiographic image and the additional information through a first feature processing unit, a second feature processing unit, and a feature fusion processing unit to construct a diagnosis assistance model and determines a bone mineral density (BMD) value and disease classification information for the radiation image to be diagnosed based on the diagnosis assistance model; and an information providing unit that provides disease diagnosis assistance information including at least one of the bone mineral density value, the disease classification information, and diagnosis basis information.

[0015] The first feature processing unit above can extract a first feature by learning either the learning radiation image or the diagnostic target radiation image.

[0016] The second feature processing unit above can extract a second feature by learning patient information including at least one of bone density evaluation, gender, age, case history, and whether or not surgery has been performed among the additional information.

[0017] The above feature fusion processing unit can construct a diagnostic assistance model that determines the bone density value and disease classification information determined based on the bone density value by fusion learning the first feature and the second feature in the case of the training radiographic image, and can determine the bone density value and disease classification information for the diagnostic target radiographic image based on the diagnostic assistance model.

[0018] The above feature fusion processing unit can fuse the first feature and the second feature using at least one layer among a concatenation layer, an average pooling layer, and a fully-connected (FC) layer with respect to the first feature and the second feature.

[0019] The above first feature processing unit can extract the first feature using a CNN (Convolutional Neural Network) and a vision transformer.

[0020] The above second feature processing unit may use a fully-connected (FC) layer corresponding to a number corresponding to the number of additional information.

[0021] The above input processing unit can adjust the size of the training radiation image and the diagnostic target radiation image and align them equally to either the left or the right side to perform normalization of the training radiation image and the diagnostic target radiation image.

[0022] The above diagnostic basis information may include a radiographic image indicating the portion from which the first feature is extracted in relation to the bone density value and the disease classification information in the above diagnostic target radiographic image.

[0023] A disease diagnosis assistance method according to an embodiment of the present invention may include: receiving a training radiographic image, additional information regarding the training radiographic image, and a radiation image to be diagnosed in an input processing unit; constructing a diagnosis assistance model in an artificial intelligence processing unit by learning the training radiographic image and the additional information through a first feature processing unit, a second feature processing unit, and a feature fusion processing unit, and determining a bone mineral density (BMD) value and disease classification information for the radiation image to be diagnosed based on the diagnosis assistance model; and providing disease diagnosis assistance information in an information providing unit, the information including at least one of the bone mineral density value, the disease classification information, and diagnosis basis information.

[0024] The step of constructing a diagnostic aid model by learning the training radiographic image and the additional information through the first feature processing unit, the second feature processing unit, and the feature fusion processing unit, and determining bone mineral density (BMD) values ​​and disease classification information for the diagnostic target radiographic image based on the diagnostic aid model may include: a step of extracting a first feature by learning either the training radiographic image or the diagnostic target radiographic image in the first feature processing unit; a step of extracting a second feature by learning patient information including at least one of bone mineral density evaluation, gender, age, case history, and whether surgery was performed in the second feature processing unit; and a step of constructing the diagnostic aid model by fusion learning the first feature and the second feature in the case of the training radiographic image in the feature fusion processing unit to determine the bone mineral density values ​​and disease classification information determined based on the bone mineral density values, and determining the bone mineral density values ​​and disease classification information for the diagnostic target radiographic image based on the diagnostic aid model.

[0025] The step of receiving the above-mentioned training radiation image, additional information regarding the above-mentioned training radiation image, and the radiation image to be diagnosed may perform normalization of the above-mentioned training radiation image and the radiation image to be diagnosed by adjusting the size of the above-mentioned training radiation image and the radiation image to be diagnosed and aligning them equally to either the left or the right side.

[0026] The above diagnostic basis information may include a radiographic image indicating the portion from which the first feature is extracted in relation to the bone density value and the disease classification information in the above diagnostic target radiographic image.

[0027] The present invention can provide diagnostic assistance information to prevent osteoporotic fractures and complications after bone-related surgery by constructing a diagnostic assistance model that predicts bone density through machine learning of radiographic images.

[0028] The present invention can improve the effectiveness of patient treatment, reduce medical costs, and provide helpful information to doctors to establish more accurate diagnoses and treatment plans by accurately predicting bone density in bone parts such as the shoulder, thereby preventing osteoporotic humeral fractures and postoperative complications in advance.

[0029] The present invention enables medical personnel to obtain diagnostic assistance information to prevent osteoporotic fractures and complications after bone-related surgery solely through radiographic images.

[0030] FIG. 1 is a diagram illustrating a diagnostic assistance device based on artificial intelligence processing of radiation images according to an embodiment of the present invention.

[0031] FIG. 2 is a diagram illustrating the image preprocessing configuration of a diagnostic assistance device based on artificial intelligence processing of radiation images according to an embodiment of the present invention.

[0032] FIGS. 3a to 3c are drawings illustrating an artificial intelligence processing unit of a diagnostic assistance device based on artificial intelligence processing of radiation images according to an embodiment of the present invention.

[0033] FIGS. 4 and 5 are drawings illustrating diagnostic basis information provided by an artificial intelligence processing-based diagnostic aid for radiation images according to an embodiment of the present invention.

[0034] FIGS. 6 and 7 are drawings illustrating a diagnostic assistance method based on artificial intelligence processing of radiation images according to an embodiment of the present invention.

[0035] Hereinafter, various embodiments of this document are described with reference to the attached drawings.

[0036] The embodiments and the terms used therein are not intended to limit the technology described in this document to specific embodiments and should be understood to include various modifications, equivalents, and / or substitutions of said embodiments.

[0037] In describing various embodiments below, if it is determined that a detailed description of related known functions or configurations could unnecessarily obscure the essence of the invention, such detailed description will be omitted.

[0038] Furthermore, the terms described below are defined considering their functions in various embodiments, and these may vary depending on the intentions or practices of the user or operator. Therefore, their definitions should be based on the content throughout this specification.

[0039] In relation to the description of the drawings, similar reference numerals may be used for similar components.

[0040] A singular expression may include a plural expression unless the context clearly indicates otherwise.

[0041] In this document, expressions such as "A or B" or "at least one of A and / or B" may include all possible combinations of the items listed together.

[0042] Expressions such as "first," "second," "first," or "second" may modify the corresponding components regardless of order or importance, and are used merely to distinguish one component from another without limiting the components.

[0043] Where it is stated that a certain (e.g., first) component is "(functionally or telecommunicationally) connected" or "connected" to another (e.g., second) component, said certain component may be directly connected to said other component or connected through another component (e.g., third component).

[0044] In this specification, "configured to" may be used interchangeably with, depending on the context, for example, in hardware or software, "suitable for," "capable of," "modified to," "made to," "capable of," or "designed to."

[0045] In some situations, the expression "device configured to..." may mean that the device is "able to..." together with other devices or parts.

[0046] For example, the phrase “a processor configured (or set) to perform A, B, and C” may mean a dedicated processor for performing said operations (e.g., an embedded processor), or a general-purpose processor capable of performing said operations by executing one or more software programs stored in a memory device (e.g., a CPU or an application processor).

[0047] Also, the term 'or' means an inclusive or rather an exclusive or.

[0048] That is, unless otherwise noted or is not clear from the context, the expression 'x uses a or b' means any one of the natural inclusive permutations.

[0049] Terms such as '..bu', '..gi' used below refer to a unit that processes at least one function or operation, and this may be implemented in hardware or software, or a combination of hardware and software.

[0050] FIG. 1 is a diagram illustrating a diagnostic assistance device based on artificial intelligence processing of radiation images according to an embodiment of the present invention.

[0051] FIG. 1 illustrates the components of a diagnostic assistance device based on artificial intelligence processing of radiation images according to an embodiment of the present invention.

[0052] Referring to FIG. 1, a diagnostic assistance device (100) based on artificial intelligence processing of a radiation image according to one embodiment of the present invention includes an input processing unit (110), an artificial intelligence processing unit (120), and an information providing unit (130).

[0053] For example, the artificial intelligence processing unit (120) includes a first feature processing unit (121), a second feature processing unit (122), and a feature fusion processing unit (123).

[0054] According to one embodiment of the present invention, the input processing unit (110) receives a learning radiation image, additional information regarding the learning radiation image, and a radiation image to be diagnosed.

[0055] The input processing unit (110) adjusts the size of the learning radiation image and the diagnostic target radiation image.

[0056] In addition, the input processing unit (110) can perform normalization of the learning radiation image and the diagnostic target radiation image by aligning them equally to either the left or the right side.

[0057] The artificial intelligence processing unit (120) builds a diagnostic assistance model by learning the training radiation image and additional information through the first feature processing unit (121), the second feature processing unit (122), and the feature fusion processing unit (123).

[0058] Additionally, the artificial intelligence processing unit (120) can determine bone mineral density (BMD) values ​​and disease classification information for a radiographic image to be diagnosed based on a diagnostic assistance model.

[0059] The information providing unit (130) may provide disease diagnosis assistance information including at least one of the bone density value, the disease classification information, and the diagnosis basis information.

[0060] The first feature processing unit (121) can extract a first feature by learning either a learning radiation image or a diagnostic target radiation image.

[0061] The second feature processing unit (122) can extract the second feature by learning patient information including at least one of bone density evaluation, gender, age, case history, and whether or not surgery was performed among additional information.

[0062] The feature fusion processing unit (123) fuses and learns the first feature and the second feature in the case of a learning radiographic image to build a diagnostic assistance model capable of determining bone density values ​​and disease classification information determined based on bone density values.

[0063] Additionally, the feature fusion processing unit (123) can determine bone density values ​​and disease classification information for a diagnostic target radiographic image based on a diagnostic assistance model.

[0064] The feature fusion processing unit (123) can fuse the first feature and the second feature using at least one of a concatenation layer, an average pooling layer, and a fully-connected (FC) layer for the first feature and the second feature.

[0065] The first feature processing unit (121) can extract the first feature using a CNN (Convolutional Neural Network) and a vision transformer.

[0066] The second feature processing unit (122) can use a fully-connected (FC) layer corresponding to a number of additional information.

[0067] Accordingly, the present invention can provide diagnostic assistance information to prevent osteoporotic fractures and complications after bone-related surgery by constructing a diagnostic assistance model that predicts bone density through machine learning of radiographic images.

[0068] Therefore, the present invention enables medical personnel to obtain diagnostic assistance information to prevent osteoporotic fractures and complications after bone-related surgery solely through radiographic images.

[0069] FIG. 2 is a diagram illustrating the image preprocessing configuration of a diagnostic assistance device based on artificial intelligence processing of radiation images according to an embodiment of the present invention.

[0070] FIG. 2 illustrates the image preprocessing configuration of a diagnostic assistance device based on artificial intelligence processing of radiation images according to an embodiment of the present invention.

[0071] Referring to FIG. 2, there is a difference in size and aspect between the radiation image (200) and the radiation image (210), and there is a difference in size between the radiation image (200) and the radiation image (220).

[0072] A diagnostic assistance device based on artificial intelligence processing of radiation images according to one embodiment of the present invention can increase the efficiency of artificial intelligence processing by performing a data preprocessing process to unify additional elements, such as the size and shooting direction of radiation images, according to the same standard.

[0073] In other words, the artificial intelligence processing-based diagnostic aid for radiation images according to one embodiment of the present invention can perform standardization of the training radiation image and the diagnostic radiation image by adjusting the sizes of the training radiation image and the diagnostic target radiation image and aligning them equally to either the left or the right side.

[0074] FIGS. 3a to 3c are drawings illustrating an artificial intelligence processing unit of a diagnostic assistance device based on artificial intelligence processing of radiation images according to an embodiment of the present invention.

[0075] Referring to FIG. 3a, an artificial intelligence processing unit (300) according to one embodiment of the present invention is composed of a first feature extraction unit (301), a second feature extraction unit (302), and a feature fusion processing unit (303).

[0076] A first feature extraction unit (301) according to an embodiment of the present invention extracts a first feature from a radiation image.

[0077] For example, the second feature extraction unit (302) extracts the second feature based on the FC layer corresponding to the number of additional information.

[0078] The feature fusion processing unit (303) fuses the first feature and the second feature using a connecting layer that connects the first feature and the second feature.

[0079] Referring to FIG. 3b, an artificial intelligence processing unit (310) according to one embodiment of the present invention is composed of a first feature extraction unit (311), a second feature extraction unit (312), and a feature fusion processing unit (313).

[0080] A first feature extraction unit (311) according to one embodiment of the present invention extracts a first feature from a radiation image.

[0081] For example, the second feature extraction unit (312) extracts the second feature based on the FC layer corresponding to the number of additional information.

[0082] The feature fusion processing unit (313) fuses the first feature and the second feature by calculating the average of the first feature and the second feature using an average pooling layer for the first feature and the second feature.

[0083] Referring to FIG. 3c, an artificial intelligence processing unit (320) according to one embodiment of the present invention is composed of a first feature extraction unit (321), a second feature extraction unit (322), and a feature fusion processing unit (323).

[0084] A first feature extraction unit (321) according to an embodiment of the present invention extracts a first feature from a radiation image.

[0085] For example, the second feature extraction unit (322) extracts the second feature based on the FC layer corresponding to the number of additional information.

[0086] The feature fusion processing unit (323) fuses the first feature and the second feature by additionally applying an FC layer to the first feature and the second feature.

[0087] Accuracy for Model Bone Density Classification Image 0.85 Image+Additional Info (Age) 0.86 Image+Additional Info (Gender) 0.83 Image+Additional Info (Age)+Additional Info (Gender) 0.86

[0088] According to one embodiment of the present invention, fusing the first feature and the second feature increases the AUC (area under the cuve) corresponding to the degree of bone density prediction for a radiation image, which can be summarized as shown in Table 1 above.

[0089] According to one embodiment of the present invention, an AI-based diagnostic assistance device for radiation images can increase the classification accuracy of bone density and osteoporosis by extracting image features for a radiation image as a first feature and extracting additional information, such as bone density numerical information evaluated by an expert for the image features, as a second feature and performing fusion processing.

[0090] FIGS. 4 and 5 are drawings illustrating diagnostic basis information provided by an artificial intelligence processing-based diagnostic aid for radiation images according to an embodiment of the present invention.

[0091] Referring to FIG. 4, a diagnostic aid device based on artificial intelligence processing of a radiation image according to one embodiment of the present invention provides diagnostic basis information as a result of a radiation image determined to be osteoporosis.

[0092] The portion of the image (400) (401) and the portion of the image (410) (411) where the first feature is extracted are indicated and shown.

[0093] The parts used according to the degree of color intensity for the partial area (401) and the partial area (411) are indicated.

[0094] Referring to FIG. 5, a diagnostic aid based on artificial intelligence processing of a radiation image according to one embodiment of the present invention provides diagnostic basis information as a result of a radiation image that is determined not to be osteoporosis.

[0095] The portion of the image (500) (501) and the portion of the image (510) (511) where the first feature is extracted are indicated and shown.

[0096] The parts used according to the degree of color intensity for the partial area (501) and the partial area (511) are indicated.

[0097] The diagnostic basis information may include a radiographic image indicating the portion from which a first feature is extracted in relation to bone density values ​​and disease classification information in the radiographic image to be diagnosed.

[0098] FIGS. 6 and 7 are drawings illustrating a diagnostic assistance method based on artificial intelligence processing of radiation images according to an embodiment of the present invention.

[0099] FIG. 6 illustrates a procedure for a diagnostic assistance method based on artificial intelligence processing of a radiation image according to an embodiment of the present invention, which builds a diagnostic assistance model that predicts bone density by machine learning on a radiation image and provides diagnostic assistance information for determining osteoporosis.

[0100] Referring to FIG. 6, in step (S601), the diagnostic assistance method based on artificial intelligence processing of a radiation image according to an embodiment of the present invention receives either training data or diagnostic data.

[0101] That is, the artificial intelligence processing-based diagnostic assistance method for radiation images according to one embodiment of the present invention can receive a training radiation image, additional information regarding the training radiation image, and a radiation image to be diagnosed.

[0102] In step (S602), the diagnostic assistance method based on artificial intelligence processing of a radiation image according to an embodiment of the present invention constructs a diagnostic assistance model based on training data and learns diagnostic data based on the diagnostic assistance model to determine diagnostic assistance information.

[0103] That is, the artificial intelligence processing-based diagnostic assistance method for a radiation image according to one embodiment of the present invention constructs a diagnostic assistance model by learning a training radiation image and additional information through a first feature processing unit, a second feature processing unit, and a feature fusion processing unit, and can determine bone mineral density (BMD) values ​​and disease classification information for a radiation image to be diagnosed based on the diagnostic assistance model.

[0104] In step (S603), the diagnostic assistance method based on artificial intelligence processing of a radiation image according to an embodiment of the present invention provides diagnostic assistance information.

[0105] That is, the artificial intelligence processing-based diagnostic assistance method for radiation images according to one embodiment of the present invention can provide disease diagnostic assistance information including at least one of bone density values, disease classification information, and diagnostic basis information.

[0106] FIG. 7 illustrates a procedure for constructing a diagnostic assistance model based on artificial intelligence processing of a radiation image according to an embodiment of the present invention.

[0107] Referring to FIG. 7, in step (S701), the diagnostic assistance method based on artificial intelligence processing of a radiation image according to an embodiment of the present invention processes inputs for a radiation image and additional information.

[0108] That is, the artificial intelligence processing-based diagnostic assistance method for radiation images according to one embodiment of the present invention can process input for a training radiation image and additional information regarding the training radiation image.

[0109] In step (S702), the AI ​​processing-based diagnostic assistance method for a radiation image according to an embodiment of the present invention extracts a first feature from the radiation image.

[0110] That is, the artificial intelligence processing-based diagnostic assistance method for a radiation image according to one embodiment of the present invention can extract a first feature by learning either a training radiation image or a radiation image to be diagnosed.

[0111] In step (S703), the AI ​​processing-based diagnostic assistance method for a radiation image according to an embodiment of the present invention extracts a second feature from additional information.

[0112] That is, the diagnostic assistance method based on artificial intelligence processing of a radiation image according to one embodiment of the present invention can extract a second feature by learning patient information including at least one of bone density evaluation, gender, age, case history, and whether or not surgery has been performed among additional information.

[0113] In step (S704), the artificial intelligence processing-based diagnostic assistance method for a radiation image according to an embodiment of the present invention determines the fusion learning result of the first feature and the second feature.

[0114] That is, the artificial intelligence processing-based diagnostic assistance method for radiation images according to one embodiment of the present invention can determine bone density values ​​and disease classification information determined based on bone density values ​​by fusion learning of the first feature and the second feature in the case of a training radiation image.

[0115] In step (S705), the diagnostic assistance method based on artificial intelligence processing of a radiation image according to an embodiment of the present invention provides diagnostic assistance information.

[0116] That is, the diagnostic assistance method based on artificial intelligence processing of radiation images according to one embodiment of the present invention can provide diagnostic assistance information by constructing a diagnostic reporting model that provides bone density values, disease classification information, and diagnostic basis information.

[0117] Therefore, by accurately predicting bone density for bone parts such as the shoulder, the present invention can prevent osteoporotic humeral fractures and postoperative complications in advance, thereby enhancing the effectiveness of patient treatment, reducing medical costs, and providing helpful information to doctors to establish more accurate diagnoses and treatment plans.

[0118] The device described above may be implemented as a hardware component, a software component, and / or a combination of a hardware component and a software component. For example, the device and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable array (FPA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. In addition, other processing configurations, such as parallel processors, are also possible.

[0119] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or command the processing unit independently or collectively. Software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium or device, or transmitted signal wave so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and may be stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0120] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can make various modifications and variations from the description above. For example, suitable results can be achieved even if the described techniques are performed in a different order than described, and / or the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0121] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below.

Claims

1. An input processing unit that receives a learning radiation image, additional information regarding the learning radiation image, and a radiation image to be diagnosed; An artificial intelligence processing unit that constructs a diagnostic assistance model by learning the training radiographic image and the additional information through a first feature processing unit, a second feature processing unit, and a feature fusion processing unit, and determines bone mineral density (BMD) values ​​and disease classification information for the diagnostic target radiographic image based on the diagnostic assistance model; and Characterized by including an information providing unit that provides disease diagnosis assistance information including at least one of the above bone density value, the above disease classification information, and the above diagnostic basis information. Disease diagnosis assistance device.

2. In Paragraph 1, The first feature processing unit above learns either of the learning radiation image and the diagnostic target radiation image to extract a first feature, and The second feature processing unit is characterized by extracting a second feature by learning patient information including at least one of bone density evaluation, gender, age, case history, and whether or not surgery was performed among the additional information. Disease diagnosis assistance device.

3. In Paragraph 2, The feature fusion processing unit is characterized by constructing a diagnostic assistance model that determines the bone density value and disease classification information determined based on the bone density value by fusion learning the first feature and the second feature in the case of the training radiographic image, and determining the bone density value and disease classification information for the diagnostic target radiographic image based on the diagnostic assistance model. Disease diagnosis assistance device.

4. In Paragraph 3, The feature fusion processing unit is characterized by fusing the first feature and the second feature using at least one layer among a concatenation layer, an average pooling layer, and a fully-connected (FC) layer with respect to the first feature and the second feature. Disease diagnosis assistance device.

5. In Paragraph 2, The above-mentioned first feature processing unit is characterized by extracting the first feature using a CNN (Convolutional Neural Network) and a vision transformer. Disease diagnosis assistance device.

6. In Paragraph 2, The above second feature processing unit is characterized by using a fully-connected (FC) layer corresponding to a number corresponding to the number of additional information. Disease diagnosis assistance device.

7. In Paragraph 1, The above input processing unit is characterized by adjusting the size of the training radiation image and the diagnostic target radiation image, and aligning them equally to either the left or the right side to perform normalization on the training radiation image and the diagnostic target radiation image. Disease diagnosis assistance device.

8. In Paragraph 1, The above diagnostic basis information is characterized by including a radiographic image that indicates the portion where the first feature is extracted in relation to the bone density value and the disease classification information in the above diagnostic target radiographic image. Disease diagnosis assistance device.

9. A step of receiving a training radiation image, additional information regarding the training radiation image, and a radiation image to be diagnosed in an input processing unit; A step of constructing a diagnostic assistance model by learning the training radiographic image and the additional information through a first feature processing unit, a second feature processing unit, and a feature fusion processing unit in an artificial intelligence processing unit, and determining bone mineral density (BMD) values ​​and disease classification information for the diagnostic target radiographic image based on the diagnostic assistance model; and The information providing unit is characterized by including the step of providing disease diagnosis assistance information comprising at least one of the bone density value, the disease classification information, and the diagnosis basis information. Disease diagnosis assistance method.

10. In Paragraph 9, The step of constructing a diagnostic assistance model by learning the training radiographic image and the additional information through the first feature processing unit, the second feature processing unit, and the feature fusion processing unit, and determining bone mineral density (BMD) values ​​and disease classification information for the diagnostic target radiographic image based on the diagnostic assistance model, In the first feature processing unit above, a step of extracting a first feature by learning either the learning radiation image and the diagnosis target radiation image; In the second feature processing unit above, a step of extracting a second feature by learning patient information including at least one of bone density evaluation, gender, age, case history, and whether or not surgery has been performed among the additional information; and The method is characterized by including the step of constructing a diagnostic assistance model in the above feature fusion processing unit for the case of the training radiographic image by fusion learning the first feature and the second feature to determine the bone density value and disease classification information determined based on the bone density value, and determining the bone density value and disease classification information for the diagnostic target radiographic image based on the above diagnostic assistance model. Disease diagnosis assistance method.

11. In Paragraph 9, The step of receiving the above-mentioned learning radiation image, additional information regarding the above-mentioned learning radiation image, and the radiation image to be diagnosed is, The method is characterized by including the step of adjusting the size of the training radiation image and the diagnostic target radiation image, and aligning them equally to either the left or the right side to perform normalization on the training radiation image and the diagnostic target radiation image. Disease diagnosis assistance method.

12. In Paragraph 9, The above diagnostic basis information is characterized by including a radiographic image that indicates the portion where the first feature is extracted in relation to the bone density value and the disease classification information in the above diagnostic target radiographic image. Disease diagnosis assistance method.

Images