Electronic Device Predicting the Occurrence of Vertebral Compression Fractures

Abstract

An electronic device for predicting a fracture according to an embodiment of the present invention may be configured to include a memory that stores a human X-ray image and an artificial intelligence model that has learned the correlation between the presence or absence of a compression fracture confirmed through the human MRI, and a processor that inputs the subject's X-ray image into the artificial intelligence model to predict the probability and timing of occurrence of a compression fracture in the subject.

Description

Technology Field

[0001] The present invention relates to an electronic device for predicting the occurrence of spinal compression fractures. Specifically, the present invention relates to an electronic device for predicting the occurrence of spinal compression fractures by utilizing abdominal X-ray images. Background Technology

[0002] A vertebral compression fracture refers to a condition in which the columns of the vertebrae fracture due to a strong external impact. Such fractures primarily occur in the elderly and patients with severe osteoporosis who are vulnerable to minor external impacts, while in young and middle-aged adults, they are mainly caused by traffic accidents or falls from heights.

[0003] If these spinal compression fractures are not diagnosed in the early stages, a lack of proper treatment often leads to the progression of scoliosis (curvature of the spine) or damage to surrounding ligaments and muscles, resulting in the development of other conditions. Therefore, a prompt diagnosis by a specialist is crucial for such spinal compression fractures.

[0004] However, in the case of such vertebral compression fractures, there is often no significant pain in the early stages of symptoms, and diagnosis is frequently missed because it is difficult to distinguish them using X-rays. Accordingly, various studies have been conducted to utilize artificial intelligence models to enable timely diagnosis of vertebral compression fractures.

[0005] Meanwhile, current compression fracture prediction models are limited to identifying the presence of a fracture using X-ray or MRI images. However, given the high clinical demand to predict future compression fractures beyond simply identifying their presence, research into artificial intelligence models capable of predicting such future events is necessary. Prior art literature

[0006] Korean Registered Patent No. 10-2214436 (Evaluation of bone density distribution using DXA imaging and method for predicting fractures using the same) The problem to be solved

[0007] The present invention is designed to provide a machine learning model that predicts the occurrence of future compression fractures by utilizing imaging findings from abdominal X-rays.

[0008] The purposes of the present disclosure are not limited to those mentioned above, and other purposes and advantages of the present disclosure not mentioned may be understood from the following description and will be more clearly understood from the embodiments of the present disclosure. Furthermore, it will be readily apparent that the purposes and advantages of the present disclosure can be realized by the means and combinations thereof set forth in the claims. means of solving the problem

[0009] An electronic device for predicting a fracture according to an embodiment of the present invention may be configured to include a memory that stores a human X-ray image and an artificial intelligence model that has learned the correlation between the presence or absence of a compression fracture confirmed through the human MRI, and a processor that inputs the subject's X-ray image into the artificial intelligence model to predict the probability and timing of occurrence of a compression fracture in the subject.

[0010] At this time, the artificial intelligence model may be a model that performs machine learning by setting the human X-ray image as input data, setting the difference between the time of taking the human X-ray image and the time of taking the human MRI as a variable input value, and setting information regarding whether a compression fracture is confirmed through the human MRI as output data.

[0011] And the above artificial intelligence model may be trained by setting X-ray images taken in the past, prior to the human MRI, as the input data.

[0012] And the above artificial intelligence model may be trained using at least one algorithm among Random survival forest, DeepSurv, Cox proportional hazard and Convolutional neural network.

[0013] And if the processor determines that the probability of a compression fracture in the subject is above a threshold, it may perform an operation to add visual elements to the problematic location in the subject's X-ray image based on Grad-CAM (Gradient-weighted Class Activation Mapping) (e.g., marking the problematic location with an arbitrary color). Subsequently, the processor may display the X-ray image with the marking of the problematic location on the display unit.

[0014] And the processor can obtain information regarding the time of shooting from the header information of the subject's X-ray image and provide guidance by outputting the time of occurrence of the compression fracture for the subject as the period elapsed since the time of shooting the X-ray.

[0015] In addition, the above artificial intelligence model is characterized by being learned through a feature selection process, and the feature selection process may be performed by utilizing at least one of SelectKBest, RFE, and Grid search.

[0016] Furthermore, the method for predicting a fracture of an electronic device according to an embodiment of the present invention may include the steps of acquiring an X-ray image of a subject, inputting the X-ray image of the subject into an artificial intelligence model, and predicting the probability and timing of occurrence of a compression fracture of the subject according to the calculation of the artificial intelligence model, wherein the artificial intelligence model may be characterized as a model that has learned the correlation between a human X-ray image and the presence or absence of a compression fracture confirmed through the human MRI.

[0017] And a computer program stored on a computer-readable recording medium according to an embodiment of the present invention is executed by a processor of an electronic device, so that the electronic device can perform the method for predicting the fracture. Effects of the invention

[0018] The present invention can predict information regarding the occurrence of a spinal compression fracture solely from a subject's abdominal X-ray. Specifically, the present invention can predict the likelihood of a spinal compression fracture occurring and the timing of its occurrence solely from a subject's abdominal X-ray. Brief explanation of the drawing

[0019] FIG. 1 is a drawing for explaining an electronic device according to an embodiment of the present invention. FIG. 2 is a diagram illustrating the configuration of an electronic device according to an embodiment of the present invention. FIG. 3 is a diagram illustrating the configuration of a processor according to an embodiment of the present invention. FIG. 4 is a diagram illustrating the sequence of a method for predicting fractures according to an embodiment of the present invention. Specific details for implementing the invention

[0020] The advantages and features of the present invention and the methods for achieving them will become clear by referring to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below but can be implemented in various different forms. These embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the present invention, and the present invention is defined only by the scope of the claims.

[0021] The terms used in this specification are for describing embodiments and are not intended to limit the invention. In this specification, the singular form includes the plural form unless specifically stated otherwise in the text. The terms "comprises" and / or "comprising" used in this specification do not exclude the presence or addition of one or more other components in addition to the components mentioned. Throughout the specification, the same reference numerals refer to the same components, and "and / or" includes each of the mentioned components and all combinations of one or more. Although terms such as "first," "second," etc., are used to describe various components, these components are not limited by these terms. These terms are used merely to distinguish one component from another. Therefore, the first component mentioned below may be the second component within the technical scope of the invention.

[0022] Unless otherwise defined, all terms used herein (including technical and scientific terms) may be used in a meaning commonly understood by those skilled in the art to which the present invention pertains. Furthermore, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless explicitly and specifically defined otherwise.

[0023] The terms “part” or “module” as used in the specification refer to software or hardware components, such as FPGAs or ASICs, and “part” or “module” perform certain roles. However, “part” or “module” is not limited to software or hardware. “Part” or “module” may be configured to reside in an addressable storage medium or configured to run on one or more processors. Thus, by example, “part” or “module” includes components such as software components, object-oriented software components, class components, and task components, as well as processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided within the components and “parts” or “modules” may be combined into a smaller number of components and “parts” or “modules,” or further separated into additional components and “parts” or “modules.”

[0024] Spatially relative terms such as "below," "beneath," "lower," "above," and "upper" may be used to facilitate the description of the relationship between one component and other components as illustrated in the drawings. Spatially relative terms should be understood as encompassing different orientations of components during use or operation, in addition to the orientations depicted in the drawings. For example, if a component depicted in a drawing is inverted, a component described as "below" or "beneath" of another component may be placed "above" of that component. Therefore, the exemplary term "below" may encompass both the lower and upper directions. Components may also be oriented in other directions, and accordingly, spatially relative terms may be interpreted according to the orientation.

[0025] With reference to the drawings below, an electronic device according to an embodiment of the present invention will be described in detail.

[0026] FIG. 1 is a drawing for explaining an electronic device according to an embodiment of the present invention.

[0027] The electronic device according to an embodiment of the present invention may refer to the fracture prediction device (100) of FIG. 1. The fracture prediction device (100) (hereinafter, prediction device) may store an artificial intelligence model that calculates a predicted value regarding the probability of occurrence and the timing of occurrence of a spinal compression fracture of a subject from an abdominal X-ray image of the subject.

[0028] And the prediction device (100) can communicate with the X-ray imaging device (200) and can obtain an X-ray image of the subject from the X-ray imaging device (200).

[0029] And the prediction device (100) can calculate whether there is a possibility of a spinal compression fracture occurring in the subject and the numerical value of the possibility of occurrence by inputting the subject's X-ray image into the artificial intelligence model.

[0030] In addition, the prediction device (100) may input the subject's X-ray image into the artificial intelligence model and calculate a predicted value regarding the timing of the occurrence of the subject's spinal compression fracture as a result of the calculation of the artificial intelligence model. At this time, the timing of the occurrence of the subject's spinal compression fracture may be calculated as the period elapsed from the time the subject's X-ray image was taken. At this time, the X-ray image input into the artificial intelligence model may include information regarding the date of shooting extracted from the header information of the X-ray image.

[0031] According to various embodiments, the prediction device (100) may perform a preprocessing operation on an X-ray image of a subject input into an artificial intelligence model. The X-ray image input into the artificial intelligence model may have undergone a de-identification process. Additionally, according to various embodiments, when performing the preprocessing operation, the prediction device (100) may identify the time of capture of the X-ray image by extracting it from header information.

[0032] As such, the prediction device (100) inputs the subject's X-ray image into an artificial intelligence model that has been trained to calculate a prediction value for the subject's spinal compression fracture, and can calculate a prediction value for the subject's spinal compression fracture (probability of fracture occurrence, timing of fracture occurrence) according to the calculation of the artificial intelligence model.

[0033] And at this time, the artificial intelligence model may be a model trained based on information regarding whether a spinal compression fracture of the human was determined based on the human's past X-ray image and future MRI image, and the difference between the time of the X-ray image and the time of the MRI image.

[0034] For a more detailed explanation of the above prediction device (100), reference will be made to FIG. 2 and below.

[0035] FIG. 2 is a diagram illustrating the configuration of an electronic device (prediction device) according to an embodiment of the present invention.

[0036] The above prediction device (100) may be configured to include a processor (110), a memory (120), a communication unit (130), an input unit (140), and a display unit (150).

[0037] A more detailed description of the configuration of the above processor (110) will be given later with reference to FIG. 3.

[0038] The memory (120) can store an artificial intelligence model according to an embodiment of the present invention. Additionally, the memory (120) can store training data required for training the artificial intelligence model according to an embodiment of the present invention (an X-ray image of an arbitrary human and an MRI (information on whether there is a spinal compression fracture based on MRI reading)) and test data to be input into the artificial intelligence model (an X-ray image of a subject).

[0039] The above communication unit (130) can communicate with an external device, an X-ray imaging device (200), and can receive an X-ray image of the subject.

[0040] The above input unit (140) can receive an input signal for performing a predicted action for a spinal compression fracture of a subject.

[0041] The above display unit (150) can display prediction information regarding spinal compression fractures for the subject (whether (possibility) and timing of occurrence of spinal compression fractures).

[0042] And a more detailed description of the configuration of the above processor (110) will be described with reference to FIG. 3.

[0043] FIG. 3 is a diagram illustrating the configuration of a processor according to an embodiment of the present invention.

[0044] As illustrated in FIG. 3, the processor (110) may be configured to include a data management unit (111), a model learning unit (112), and a computation execution unit (113).

[0045] The above data management unit (111) can acquire X-ray images of subjects corresponding to data to be used for training an artificial intelligence model and input data to be input into an artificial intelligence model after training is complete, and can perform management thereon.

[0046] First, the data management unit (111) can acquire and manage data for training the artificial intelligence model.

[0047] According to one embodiment, the data management unit (111) can acquire data corresponding to input data, output data, and time variable input values ​​required for the learning of the artificial intelligence model. The data management unit (111) can determine whether the acquired data corresponding to input data, output data, and time variable input values ​​satisfies conditions suitable for use in the learning of the artificial intelligence model.

[0048] Specifically, the data management unit (111) can acquire an abdominal X-ray image of a human and can acquire MRI image data taken after the time of X-ray image capture of the same human.

[0049] According to various embodiments, the data management unit (111) can check whether the time of taking the X-ray image of the human and the time of taking the MRI of the same human exceed a reference value, and if the reference is satisfied, it can determine that it is usable as training data. Alternatively, the data management unit (111) can acquire multiple pairs of X-ray images and MRI images of humans obtained as training data for the model, and classify the image pairs into groups according to the level of difference between the time of taking each image so that the difference between the time of taking each image can be configured in various ways.

[0050] And, in order to perform a prediction operation through an artificial intelligence model, the data management unit (111) can obtain an X-ray image of a subject to be input into the artificial intelligence model from an external device, an X-ray imaging device (200).

[0051] The above data management unit (111) can extract information related to the time when the X-ray image was taken (shot date information) from the X-ray image of the subject, similar to training data management.

[0052] At this time, medical images, including X-ray images acquired by the data management unit (111), can be implemented in the DICOM (Digital Imaging and Communications in Medicine) format. For reference, the DICOM format is a standard developed to store, transmit, and share medical image data in a standardized format. In addition, the DICOM can include not only pixel data but also various metadata such as patient information, examination date, device information, and shooting conditions. Accordingly, the data management unit (111) can extract the shooting date corresponding to the metadata from the acquired medical images.

[0053] In addition, the data management unit (111) can extract patient information (e.g., age, etc.) from acquired medical images (e.g., X-ray of an arbitrary human, X-ray of a subject, etc.).

[0054] The above model learning unit (112) can support the learning of an artificial intelligence model according to an embodiment of the present invention.

[0055] According to one embodiment, the model learning unit (112) can train a survival analysis-based artificial intelligence model to predict the occurrence of future spinal compression fractures in a subject.

[0056] The above model learning unit (112) can train the artificial intelligence model by setting X-ray image data of multiple humans as input data. At this time, the X-ray images used for training may be data taken at a relatively past point in time (compared to MRI). At this time, the X-ray images may refer to X-ray images of the human abdomen.

[0057] And the above model learning unit (112) can train the artificial intelligence model by setting information regarding whether a spinal compression fracture occurred (Yes / No) confirmed through an MRI of the same human taken relatively in the future compared to the time when the human X-ray image was taken as output data.

[0058] And the above model learning unit (112) can train the artificial intelligence model by setting information on the period of time elapsed from the time of taking the human X-ray to the time of taking the same human MRI (difference in time of taking) as a time variable input value.

[0059] At this time, the model learning unit (112) can identify the time of X-ray imaging and the time of MRI imaging by extracting from metadata included in each medical image (X-ray, MRI) data. Information regarding the time of medical image imaging can be extracted from the header information of the medical image data formed in DICOM format.

[0060] And the above model learning unit (112) can perform learning based on a plurality of algorithms related thereto to perform analysis of X-ray images and prediction operations based on survival analysis models during the computation process of the artificial intelligence model. The above model learning unit (112) can perform learning of the artificial intelligence model by utilizing, for example, at least one of Random survival forest, DeepSurv, Cox proportional hazard and Convolutional neural network (CNN).

[0061] In this case, the above Random survival forest corresponds to a survival analysis model that predicts a survival curve in each tree and combines those predictions to create a final survival curve, and can be helpful in evaluating the importance of features and verifying the contribution of each feature. For reference, in artificial intelligence models, a survival analysis model corresponds to a model that analyzes the time taken for an event of interest, such as death or disease occurrence, to occur. Such a survival analysis model predicts whether an event will occur (e.g., indicating 1 for event occurrence and 0 for no event occurrence), and if an event occurs, the time taken until the event occurs (survival time) can be set as a key variable.

[0062] Furthermore, the aforementioned DeepSurv is a neural network-based model that can learn complex patterns from input data and estimate survival functions, and can integrate CNN layers to handle image data.

[0063] And Cox proportional hazard can model the proportional hazard of the survival function through interpretable weights and can be learned by considering the relationship between the time variable and the feature.

[0064] And Convolutional Neural Networks (CNNs) handle image data, extract features from images, and utilize them to perform predictions.

[0065] According to various embodiments, the model learning unit (112) may train the artificial intelligence model to perform a prediction operation by combining the Random survival forest, DeepSurv, Cox proportional hazard and Convolutional neural network (CNN) or through an ensemble of models.

[0066] In addition, the model learning unit (112) can perform feature selection and hyperparameter tuning when learning the artificial intelligence model.

[0067] According to one embodiment, the model learning unit (112) can perform a feature selection or hyperparameter tuning process by utilizing at least one of SelectKBest, Recursive Feature Elimination (RFE), and Grid Search.

[0068] The above-mentioned operation execution unit (113) can produce prediction information regarding the subject's spinal compression fracture through the operation of an artificial intelligence model learned through the above-mentioned model learning unit (112).

[0069] Specifically, when the above operation execution unit (113) obtains an X-ray image of the subject's abdomen, it can input the obtained X-ray image of the subject into the artificial intelligence model.

[0070] And the above operation execution unit (113) can perform the operation of the artificial intelligence model.

[0071] First, the operation execution unit (113) can perform a preprocessing operation on the X-ray image of the subject, which is data input to the artificial intelligence model.

[0072] Specifically, the operation execution unit (113) can perform preprocessing operations such as image resizing, normalization (normalizing pixel values ​​of the image), contrast enhancement, and noise removal on data input to the artificial intelligence model.

[0073] Subsequently, the operation execution unit (113) can perform an operation to extract features based on a pre-trained artificial intelligence model from the X-ray image of the subject for which the preprocessing operation has been completed. To this end, the operation execution unit (113) passes the input X-ray image of the subject through a layer such as a CNN (Convolutional Neural Network) used for image analysis, and in the process, extracts features corresponding to various patterns, shapes, and structures.

[0074] And the above operation execution unit (113) can combine information extracted from the X-ray image by the artificial intelligence model and detect features related to whether there is a fracture (vertebral compression fracture).

[0075] Afterwards, the above-mentioned operation execution unit (113) can perform a prediction operation to predict whether there is a spinal compression fracture based on the operation of the above-mentioned artificial intelligence model.

[0076] At this time, the above-mentioned calculation unit (113) can perform a prediction operation regarding the possibility of a spinal compression fracture in the subject and the time when the spinal compression fracture will occur. The above-mentioned calculation unit (113) can determine that there is a possibility of occurrence if the possibility of a spinal compression fracture occurring is greater than or equal to a threshold value, and determine that there is no possibility of occurrence if it is less than the threshold value. At this time, the above-mentioned calculation unit (113) can set a reference period and measure and provide guidance on the probability of a spinal compression fracture occurring within that period.

[0077] In addition, the above calculation unit (113) can calculate and provide a predicted value for the time when a spinal compression fracture occurs, and this can be calculated based on the time of the subject's X-ray imaging and the elapsed period therefrom.

[0078] According to various embodiments, the operation performing unit (113) may perform an operation of adding a mark for a problematic location (a location corresponding to a feature with high contribution to the final judgment) in the X-ray image of the subject based on Grad-CAM (Gradient-weighted Class Activation Mapping). At this time, the mark may be, for example, a border drawn in an arbitrary color for the problematic area.

[0079] And the above operation execution unit (113) can perform the operation of adding visual elements to the subject's X-ray image by the above Grad-CAM only when it is determined that the possibility of a compression fracture in the subject is greater than or equal to a threshold value.

[0080] For reference, the above Grad-CAM is an example of Explainable AI. The above Grad-CAM is used to visually interpret the predictions of a deep learning model and can perform visualization operations on which parts of specific input data the model mainly focuses on and which features it utilized to make predictions.

[0081] And the above operation execution unit (113) can generate an X-ray image of the subject with visual elements added to the problem area and display it on the display unit.

[0082] FIG. 4 is a diagram illustrating the sequence of a method for predicting fractures according to an embodiment of the present invention.

[0083] As illustrated in FIG. 4, an electronic device (100) according to an embodiment of the present invention can perform step 410 of acquiring an abdominal X-ray image of a subject.

[0084] And the electronic device (100) can proceed with step 420, which inputs the acquired abdominal X-ray image of the subject into an artificial intelligence model.

[0085] And the electronic device (100) can proceed with 430 steps of predicting when a compression fracture will occur in the subject based on the operation of the artificial intelligence model.

[0087] In summary, an electronic device for predicting a fracture according to an embodiment of the present invention may be configured to include a memory that stores an artificial intelligence model that has learned the correlation between a human X-ray image and whether a compression fracture is confirmed through an MRI of the human, and a processor that inputs the subject's X-ray image into the artificial intelligence model to predict the probability and timing of occurrence of a compression fracture in the subject.

[0088] At this time, the artificial intelligence model may be a model that performs machine learning by setting the human X-ray image as input data, setting the difference between the time of taking the human X-ray image and the time of taking the human MRI as a variable input value, and setting information regarding whether a compression fracture is confirmed through the human MRI as output data.

[0089] And the above artificial intelligence model may be trained by setting X-ray images taken in the past, prior to the human MRI, as the input data.

[0090] And the above artificial intelligence model may be trained using at least one algorithm among Random survival forest, DeepSurv, Cox proportional hazard and Convolutional neural network.

[0091] And if the processor determines that the probability of a compression fracture in the subject is above a threshold, it may perform an operation to add visual elements to the problematic location in the subject's X-ray image based on Grad-CAM (Gradient-weighted Class Activation Mapping) (e.g., marking the problematic location with an arbitrary color). Subsequently, the processor may display the X-ray image with the marking of the problematic location on the display unit.

[0092] And the processor can obtain information regarding the time of shooting from the header information of the subject's X-ray image and provide guidance by outputting the time of occurrence of the compression fracture for the subject as the period elapsed since the time of shooting the X-ray.

[0093] In addition, the above artificial intelligence model is characterized by being learned through a feature selection process, and the feature selection process may be performed by utilizing at least one of SelectKBest, RFE, and Grid search.

[0094] Furthermore, the method for predicting a fracture of an electronic device according to an embodiment of the present invention may include the steps of acquiring an X-ray image of a subject, inputting the X-ray image of the subject into an artificial intelligence model, and predicting the probability and timing of occurrence of a compression fracture of the subject according to the calculation of the artificial intelligence model, wherein the artificial intelligence model may be characterized as a model that has learned the correlation between a human X-ray image and the presence or absence of a compression fracture confirmed through the human MRI.

[0095] And a computer program stored on a computer-readable recording medium according to an embodiment of the present invention is executed by a processor of an electronic device, so that the electronic device can perform the method for predicting the fracture.

[0097] An electronic device (100) according to an embodiment of the present invention may include a processor (110), a memory (120), a communication unit (130), etc.

[0098] Memory can store various programs and data necessary for the operation of electronic devices. Memory can be implemented as non-volatile memory, volatile memory, flash memory, hard disk drives (HDD), or solid-state drives (SSD).

[0099] The communication unit can communicate with external devices. In particular, the communication unit may include various communication chips such as Wi-Fi chips, Bluetooth chips, wireless communication chips, NFC chips, and low-power Bluetooth chips (BLE chips). In this case, the Wi-Fi chip, Bluetooth chip, and NFC chip perform communication using LAN, Wi-Fi, Bluetooth, and NFC methods, respectively. When using a Wi-Fi chip or a Bluetooth chip, various connection information such as SSID and session key is transmitted and received first, and after establishing a communication connection using this information, various information can be transmitted and received. A wireless communication chip refers to a chip that performs communication according to various communication standards such as IEEE, Zigbee, 3G (3rd Generation), 3GPP (3rd Generation Partnership Project X-ray), and LTE (Long Term Evolution).

[0100] The processor can control the overall operation of a user device using various programs stored in memory. The processor may be composed of RAM, ROM, a graphics processing unit, a main CPU, first to n interfaces, and a bus. At this time, the RAM, ROM, graphics processing unit, main CPU, first to n interfaces, etc., may be connected to each other through a bus.

[0101] RAM stores the operating system and application programs. Specifically, when an electronic device boots up, the operating system is stored in RAM, and various application data selected by the user can be stored in RAM.

[0102] The ROM stores a set of instructions for booting the system, etc. When a turn-on command is input and power is supplied, the main CPU copies the O / S stored in memory (200) to RAM according to the instructions stored in the ROM, and runs the O / S to boot the system. When booting is complete, the main CPU copies various application programs stored in memory to RAM, and runs the application programs copied to RAM to perform various operations.

[0103] The main CPU accesses memory and performs operations, including booting and execution, using the OS stored in memory. Additionally, the main CPU performs various operations using various programs, content, and data stored in memory.

[0104] The first to n interfaces are connected to the various components described above. One of the first to n interfaces may be a network interface connected to an external device through a network.

[0105] Furthermore, the processor can control the artificial intelligence model. In this case, the control unit may, of course, include a dedicated graphics processor (e.g., GPU) for controlling the artificial intelligence model.

[0106] The processor may include one or more cores (not shown) and a graphics processing unit (not shown) and / or a connection channel (e.g., a bus, etc.) for transmitting and receiving signals with other components.

[0107] A processor according to one embodiment performs the method described in connection with the present invention by executing one or more instructions stored in memory.

[0108] Meanwhile, the processor may further include RAM (Random Access Memory, not shown) and ROM (Read-Only Memory, not shown) for temporarily and / or permanently storing signals (or data) processed within the processor. Additionally, the processor (130) may be implemented in the form of a system-on-chip (SoC) that includes at least one of a graphics processing unit, RAM, and ROM.

[0109] Memory can store programs (one or more instructions) for processing and controlling the processor. Programs stored in the storage unit can be divided into multiple modules according to their function.

[0110] The steps of the method or algorithm described in connection with embodiments of the present invention may be implemented directly in hardware, implemented as a software module executed by hardware, or implemented by a combination thereof. The software module may reside in RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (EleX-rayrically Erasable Programmable ROM), flash memory, a hard disk, a removable disk, a CD-ROM, or any form of computer-readable recording medium well known in the art to which the present invention belongs.

[0111] The components of the present invention may be implemented as a program (or application) and stored on a medium to be executed in combination with a computer, which is hardware. The components of the present invention may be implemented as software programming or software elements, and similarly, embodiments may be implemented in programming or scripting languages ​​such as C, C++, Java, assembler, etc., including various algorithms implemented as combinations of data structures, processes, routines, or other programming configurations. Functional aspects may be implemented as algorithms executed on one or more processors.

[0112] Although the present invention has been described in detail with reference to the examples above, those skilled in the art may make modifications, changes, and variations to these examples without departing from the scope of the invention. In short, it should be noted that in order to achieve the intended effect of the present invention, it is not necessary to separately include all functional blocks shown in the drawings or follow all sequences shown in the drawings exactly as shown, and that such matters may fall within the technical scope of the invention as described in the claims. Explanation of the symbols

[0113] 100 : Server 110 : Processor 111 : Data Management Department 112 : Model learning section 113 : Operation execution unit 200 : X-ray imaging device

Claims

Claim 1 An electronic device for performing a fracture prediction, comprising: a memory for storing an artificial intelligence model that has learned the correlation between a human X-ray image and whether a compression fracture is confirmed through the human MRI; The electronic device comprises: a processor that inputs an X-ray image of a subject into the artificial intelligence model to predict the probability and timing of occurrence of a compression fracture in the subject; wherein the artificial intelligence model receives the human X-ray image taken in the past prior to the human MRI from the electronic device as input data, and the input data is data in which the difference between the time of taking the human X-ray image and the time of taking the human MRI is set as a variable input value; wherein the artificial intelligence model is a model in which machine learning is performed to output information regarding whether a compression fracture is confirmed through the human MRI; and wherein, when the probability of a compression fracture in the subject is determined to be greater than a threshold value, the processor identifies a problematic location in the subject's X-ray image based on Grad-CAM (Gradient-weighted Class Activation Mapping), obtains information regarding the time of taking from the header information of the subject's X-ray image, and outputs the timing of occurrence of the compression fracture in the subject as the period elapsed since the time of taking the X-ray. Claim 2 delete Claim 3 delete Claim 4 An electronic device according to claim 1, characterized in that the artificial intelligence model is trained using at least one of Random survival forest, DeepSurv, Cox proportional hazard, and Convolutional neural network. Claim 5 delete Claim 6 delete Claim 7 An electronic device according to claim 1, wherein the artificial intelligence model is learned through a feature selection process, and the feature selection process is performed by utilizing at least one of SelectKBest, RFE, and Grid search. Claim 8 delete Claim 9 In a system for predicting fractures, an X-ray imaging device that generates an X-ray image of a subject's abdomen through imaging; A system comprising: an electronic device that acquires an X-ray image of a subject from the X-ray imaging device and inputs the X-ray image of the subject into an artificial intelligence model to predict the probability and timing of occurrence of a compression fracture in the subject; wherein the artificial intelligence model receives the human X-ray image, which was taken in the past prior to the human MRI, as input data from the electronic device, and the input data is data in which the difference between the time of taking the human X-ray image and the time of taking the human MRI is set as a variable input value; wherein the artificial intelligence model is a model in which machine learning is performed to output information regarding whether a compression fracture is confirmed through the human MRI; and wherein, when the electronic device determines that the probability of a compression fracture in the subject is above a threshold value, it identifies a problematic location in the subject's X-ray image based on Grad-CAM (Gradient-weighted Class Activation Mapping), acquires information regarding the time of taking from the header information of the subject's X-ray image, and outputs the timing of occurrence of the compression fracture in the subject as the period elapsed since the time of taking the X-ray. Claim 10 A method for predicting a fracture of an electronic device comprises: a step of acquiring an X-ray image of a subject; a step of inputting the X-ray image of the subject into an artificial intelligence model; a step of predicting the probability and timing of occurrence of a compression fracture of the subject according to the calculation of the artificial intelligence model; and a step of identifying a problematic location in the X-ray image of the subject based on Grad-CAM (Gradient-weighted Class Activation Mapping) if it is determined that the probability of a compression fracture of the subject is greater than or equal to a threshold value; wherein the step of predicting the timing of occurrence of the compression fracture includes a step of acquiring information regarding the time of shooting from the header information of the X-ray image of the subject; The method comprises the step of predicting the timing of the occurrence of a compression fracture in the subject based on the elapsed period from the time of taking the X-ray, wherein the artificial intelligence model receives the human X-ray image taken in the past compared to the human MRI from the electronic device as input data, and the input data is data in which the difference between the time of taking the human X-ray image and the time of taking the human MRI is set as a variable input value, and the artificial intelligence model is a model in which machine learning is performed to output information regarding whether a compression fracture is confirmed through the human MRI. Claim 11 A computer program stored on a computer-readable recording medium, wherein the computer program is executed by a processor of an electronic device to cause the electronic device to perform the method of claim 10.

Images