Diagnostic device

Abstract

To provide a diagnostic device that estimates a subject's age based on X-ray images.SOLUTION: In a diagnostic system in which a diagnostic device 101 is connected to a terminal via a communication network, the diagnostic device includes an image acquisition unit that acquires an X-ray image of a subject to be estimated, and an age estimation unit that estimates the age of the subject to be estimated, by setting the acquired X-ray image of the subject to be estimated as an input value, to a mathematical model that estimates age from an X-ray image.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to an analysis technique using X-ray images. [Background technology]

[0002] As the world's population ages, research on aging is becoming increasingly important. Aging is a complex change that is strongly associated with disease, and the impact of aging on individuals varies greatly. For example, even among elderly people of the same age, some are frail and require assistance in daily life, while others are active, independent, and youthful. Physiological indicators (biomarkers) are used in research to elucidate the relationship between aging and health. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Weinert, BT & Timiras, PS Invited review: Theories of aging. J Appl Physiol (1985) 95, 1706-1716, doi:10.1152 / japplphysiol.00288.2003 (2003). [Non-Patent Document 2] Higgins-Chen, AT, Thrush, KL & Levine, ME Aging biomarkers and the brain. Semin Cell Dev Biol 116, 180-193, doi:10.1016 / j.semcdb.2021.01.003 (2021). [Non-Patent Document 3] Jylhava, J., Pedersen, NL & Hagg, S. Biological Age Predictors. EBioMedicine 21, 29-36, doi:10.1016 / j.ebiom.2017.03.046 (2017).

Non-licensed Document 4

Non-licensed Document 5

Non-licensed Document 6

Non-licensed Document 7

Non-licensed Document 8

Non-licensed literature 9

[0004] By the way, one method of examining the lungs is by taking X-rays including the chest (hereafter referred to as "X-ray images"). X-ray images are minimally invasive and are frequently performed during regular health checkups. In addition, X-ray images clearly show the structures of various organs such as the stomach, lungs, and bones, making it possible to evaluate complex internal changes associated with aging. Due to these characteristics, X-ray images may be useful as a simple and quantitative biomarker of aging that can clarify the relationship between aging and disease.

[0005] The present invention was completed based on the above-mentioned viewpoint of the inventor, and its main purpose is to estimate the age of a subject by inputting an X-ray image into a mathematical model. [Means for solving the problem]

[0006] A diagnostic device in one embodiment of the present invention includes an image acquisition unit that acquires an X-ray image of a subject to be estimated, and an age estimation unit that estimates the age of the subject to be estimated by setting the X-ray image acquired of the subject to be estimated as an input value to a mathematical model that estimates age from an X-ray image. Effect of the Invention

[0007] According to the present invention, the age of a subject can be easily estimated. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is an overall configuration diagram of a diagnostic system. [Diagram 2] FIG. 2 is a functional block diagram of the diagnostic device. [Diagram 3] 1 is a flowchart showing a process of an embodiment of the present invention. [Figure 4] FIG. 1 is a conceptual diagram of an age estimation model. [Diagram 5] FIG. 1 is a schematic diagram showing an overview of electronic medical record selection. [Figure 6] FIG. 2 is a schematic diagram showing an overview of the process at the implementation stage. [Figure 7] FIG. 1 is a scatter plot showing the correlation between actual age and estimated age. [Figure 8] FIG. 4 is a schematic diagram of a first display screen. [Figure 9] FIG. 4 is a schematic diagram of a second display screen. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] First, an overview of the diagnostic system of the present invention will be described. Next, an overview of the age estimation model for estimating the age of a subject of the present invention will be described. Next, details of the processing performed in the present invention will be described in each embodiment of the selection stage, learning stage, and implementation stage.

[0010] First, an overview of a diagnostic system 100 according to the present invention will be described. FIG. 1 is a diagram showing the overall configuration of a diagnostic system 100. In the diagnostic system 100, the diagnostic device 101 is connected to a terminal 103 via a communication network 102. The diagnostic device 101 includes a selection device 104, an estimation device 105, and an evaluation device 106 (all of which will be described later). The terminal 103 is mainly assumed to be a general-purpose computer such as a workstation in a server room in a hospital or a laptop PC in a consultation room in a hospital, but may also be a portable communication terminal such as a tablet or a smartphone. The terminal 103 and the communication network 102 may be connected wirelessly or wired. It is assumed that the devices constituting the diagnostic system 100 are basically installed on the same premises of the same medical institution. The diagnostic system 100 may be provided as a so-called cloud service by installing the diagnostic device 101 at an external company or the like.

[0011] FIG. 2 is a functional block diagram of the diagnostic device 101. Each component of the diagnostic device 101 is realized by hardware including computing units such as a CPU (Central Processing Unit) and various coprocessors such as a GPU (Graphics Processing Unit) that provide auxiliary functions to the CPU, storage devices such as memory and storage, and wired or wireless communication lines connecting them, and software stored in the storage devices and supplying processing instructions to the computing units. The computer programs may be configured by device drivers, operating systems, various application programs located at higher layers than those, and libraries that provide common functions to these programs. Each block described below shows a functional block rather than a hardware-based configuration.

[0012] Diagnostic device 101 may be configured to include a web server, and terminal 103 may be configured to include a portable communication terminal and a web browser installed thereon.

[0013] The diagnostic device 101 includes a user interface processing unit 110 , a data processing unit 120 , a communication unit 130 , and a data storage unit 140 . The user interface processing unit 110 accepts operations from the user via an input interface such as a mouse, keyboard, or touch panel, and is also responsible for processing related to the user interface, such as image display and audio output. Note that it is assumed that the user of the diagnostic device 101 is mainly a doctor. The communication unit 130 is responsible for communication processing with the terminal 103 via the communication network 102. The data storage unit 140 stores various data such as the electronic medical record 200. Note that the X-ray image 300 is assumed to be listed as one of the items in the electronic medical record 200. The data processing unit 120 executes various processes based on the data acquired by the user interface processing unit 110 and the communication unit 130 and the data stored in the data storage unit 140.

[0014] The user interface processing unit 110 includes an input unit 111 that receives input from a user, and an output unit 112 that outputs various types of information such as images and sounds to the user.

[0015] The communication unit 130 includes a receiving unit 131 and a transmitting unit 132 . The receiving unit 131 receives the electronic medical record 200 and the actual age of the person whose age is to be estimated (hereinafter referred to as the "subject") from the terminal 103 via the communication network 102. The transmitting unit 132 transmits various output data to the terminal 103 via the communication network 102.

[0016] The data processing unit 120 includes an image acquisition unit 121 , an age estimation unit 122 , a learning unit 123 , a candidate selection unit 124 , a disease estimation unit 125 , and an output generation unit 126 . The image acquisition unit 121 acquires an X-ray image 300 from the electronic medical record 200 stored in the data storage unit 140. The age estimation unit 122 estimates the age of the subject by the age estimation model 400 based on the X-ray image 300 acquired by the image acquisition unit 121 (hereinafter, the age estimated by the age estimation model 400 is referred to as the "estimated age"). The learning unit 123 executes a learning process for the age estimation model 400 (described later) so that the estimated age of the subject can be calculated. The candidate selection unit 124 selects the electronic medical record 200 used for the learning process of the age estimation model 400. The method of selecting the electronic medical record will be described later. The disease estimation unit 125 calculates the difference (hereinafter, referred to as the "estimated age difference") obtained by subtracting the actual age of the subject from the estimated age of the subject calculated by the age estimation unit 122, and estimates the possibility of the subject suffering from a specific disease (hereinafter, referred to as the "possibility of suffering") based on this estimated age difference. In the present invention, the specific disease refers to all diseases listed in the ICD (International Classification of Diseases), but in particular, diseases caused by aging and lifestyle habits, such as hypertension, hyperuricemia, and COPD (Chronic Obstructive Pulmonary Disease), show a strong correlation. The output generation unit 126 generates images, characters, and voices for outputting information such as an estimated age to the user.

[0017] FIG. 3 is a flow chart showing the flow of an embodiment of the present invention. The embodiment of the present invention is divided into three stages: a "selection stage," a "learning stage," and an "implementation stage." Prior to the learning stage, a selection stage is carried out to select data to be used for learning. In the selection stage, a large number of electronic medical records 200 are input into the diagnostic device 101, and the electronic medical records 200 are sorted into a group of healthy subjects and a group of unhealthy subjects (S10). In this embodiment, the selection was carried out on 101,000 X-ray images 300 collected from 70,000 people at multiple medical institutions in Osaka Prefecture.

[0018] In the learning stage, the X-ray images 300 of the electronic medical records 200 selected as the healthy subjects in the selection stage are input to the age estimation model 400 to perform learning processing (S12). Hereinafter, the subjects of the X-ray images 300 input to the age estimation model 400 for the learning processing are referred to as "learning subjects." As described above, the learning subjects in this embodiment are healthy subjects who have passed the selection stage. By training the age estimation model 400 with the X-ray images 300 collected from a large number of learning subjects, the age estimation model 400 becomes able to calculate an estimated age from the X-ray images 300.

[0019] In the implementation stage, the user (doctor) calculates the subject's estimated age using the diagnostic device 101. That is, the estimated age is calculated by inputting the X-ray image 300 of the subject to the age estimation model 400 that has undergone the learning stage (S14). Also, in the implementation stage, the diagnostic device 101 estimates the possibility of the subject suffering from each of a plurality of types of diseases based on the subject's estimated age calculated by the age estimation model 400 and the subject's actual age.

[0020] FIG. 4 is a conceptual diagram of an age estimation model 400 . In this embodiment, the age estimation model 400 is formed by a neural network. A neural network is a network formed by connecting together nodes that are mathematical models of the functions of human brain nerve cells.

[0021] The age estimation model 400 includes an input layer 410, an intermediate layer 420, and an output layer 430. The input layer 410 is provided with each pixel of the X-ray image 300 as input information. Specifically, each pixel of the X-ray image 300 is associated one-to-one with each node (hereinafter referred to as an "input node") included in the input layer 410. In FIG. 4, the input node is represented by "I", and the input node corresponding to the n-th pixel of the X-ray image 300 is represented by "In". In the embodiment of the present invention, the chest image electronic medical record 200 having a pixel count of 320×320=102400 pixels is used, but in FIG. 4, the pixel count is represented by "N" for simplicity.

[0022] The intermediate layer 420 receives input from the input layer 410 and performs calculation processing. Specifically, the nodes in the intermediate layer 420 (hereinafter referred to as "intermediate nodes") receive output values ​​from each input node as input, and determine their own output values ​​based on the weights set for the input values ​​and the thresholds set for themselves. In reality, in FIG. 4, the intermediate layer 420 is made up of multiple layers in which intermediate nodes are connected in multiple layers. In other words, the age estimation model 400 is a so-called deep learning model. The number of intermediate nodes can be set arbitrarily depending on the input / output targets of the age estimation model 400. The intermediate nodes determine their own output values ​​based on the output values ​​from the previous nodes such as the input layer.

[0023] Nodes included in the output layer 430 (hereinafter referred to as "output nodes") store the calculation results in the intermediate layer 420. In FIG. 4, the output node is represented by "O", and the output node corresponding to the nth type of output is represented by "On". In the present invention, an estimated age is output to output node O1. In FIG. 4, estimated age=X is output as the calculation result in the age estimation model 400. In the following, the selection phase, the learning phase, and the implementation phase will be described in detail in turn.

[0024] [1. Selection stage] FIG. 5 is a schematic diagram showing an overview of selection of the electronic medical record 200. As shown in FIG. In the selection stage, the selection device 104 selects the electronic medical records 200 into a group of healthy people and a group of unhealthy people according to the presence or absence of a medical history for the subsequent learning stage. A medical history refers to a history that records a specific disease that has been suffered from in the past, the time of the disease, and medicines currently being taken due to the disease. In addition to the X-ray image 300, the electronic medical record 200 records a name, age (real age), medical history, and the like. Hereinafter, a person whose name is recorded in the electronic medical record 200 to be selected is referred to as a "candidate." In the present invention, a healthy person refers to a candidate who has never suffered from a specific disease, i.e., has no medical history. Also, an unhealthy person refers to a candidate who has suffered from a specific disease, i.e., has a medical history. Note that a person who suffers from a specific disease related to a part not visible in the X-ray image 300, such as the brain, or a person who suffers from a specific disease whose disease form cannot be seen in the X-ray image 300, such as high blood pressure, is also treated as an unhealthy person.

[0025] In the learning process of the age estimation model 400 executed in the learning stage described below, only X-ray images 300 selected from the healthy subject group are used. In the case of unhealthy subjects, unclear parts may appear in the X-ray images 300, which may result in an estimated age that is higher than the actual age being output. In other words, since the difference in estimated age is likely to be large, the estimation ability of the age estimation model 400 would be reduced if the learning process were performed using the X-ray images 300 of unhealthy subjects. The selection stage is a process for selecting from candidates learning subjects (healthy subjects) who can provide appropriate data for learning the age estimation model 400.

[0026] 5 shows how the selection device 104 selects the electronic medical records 200. Candidate No. 529 has no medical history, so the electronic medical record 200 is selected into the healthy subject group. As described above, candidates selected into the healthy subject group become learning subjects. In addition, candidate No. 530 is currently taking antihypertensive drugs due to high blood pressure, so the electronic medical record 200 is selected into the unhealthy subject group.

[0027] Here, let us assume that electronic medical record 200 No. 531 is to be selected. The candidate selection unit 124 of the selection device 104 acquires electronic medical record 200 No. 531 stored in the data storage unit 140. The candidate selection unit 124 checks the medical history recorded in the electronic medical record 200. The electronic medical record 200 No. 531 records that the candidate has suffered a stroke. Therefore, the candidate selection unit 124 selects electronic medical record 200 No. 0531 into the unhealthy subject group. In this way, the candidate selection unit 124 refers to all electronic medical records 200 and selects a learning subject (healthy subject) from a large number of candidates.

[0028] [2. Learning stage] In the learning stage, the learning unit 123 sets the X-ray image 300 of the learning subject as an input to the age estimation model 400. The X-ray image 300 input at this time is extracted from the electronic medical record 200 of the learning subject who was selected into the healthy subject group in the selection stage. The age estimation model 400 calculates the estimated age based on the X-ray image 300. The learning unit 123 compares the actual age of the learning subject recorded in the electronic medical record 200 with the estimated age. The learning unit 123 calculates the error between the estimated age and the actual age using MSE (mean square error) or the like, and adjusts the threshold value and weight of the intermediate node of the age estimation model 400 by the error backpropagation method so that this error becomes small. In this way, the learning stage aims to improve the prediction accuracy of the age estimation model 400. In this embodiment, 53,000 X-ray images 300 selected into the healthy subject group were used as input for learning. The learning stage is a process of performing learning to bring the estimated age (output value) by the age estimation model 400 closer to the actual age (correct answer) by setting the X-ray image 300 and actual age of the subject (healthy individual) as input and output values ​​of the age estimation model 400.

[0029] [3. Implementation Phase] FIG. 6 is a schematic diagram showing an overview of the process at the implementation stage. In the implementation stage, the X-ray image 300 of the subject is input to the age estimation model 400, which has achieved sufficient prediction accuracy through the learning stage, to output an estimated age. In addition, the subject's possibility of disease is estimated from the output estimated age. The processing in the implementation stage is executed by the estimation device 105 and the evaluation device 106.

[0030] Specifically, the image acquisition unit 121 acquires one X-ray image 300 from the electronic medical record 200 of the subject stored in the data storage unit 140. The age estimation unit 122 of the estimation device 105 inputs the X-ray image 300 of the subject acquired by the image acquisition unit 121 to the age estimation model 400. That is, each pixel of the X-ray image 300 is provided as input information to each input node of the input layer 410. Upon receiving this input, the intermediate layer 420 performs calculation processing. The output layer 430 outputs the estimated age of the subject as a calculation result, and the age estimation unit 122 acquires this. Thus, in the implementation stage, first, the X-ray image 300 of the subject (healthy or unhealthy) to be diagnosed is set as an input value of the age estimation model 400, and the estimated age of the subject is calculated by the age estimation model 400.

[0031] Next, the disease estimation unit 125 of the evaluation device 106 calculates an estimated age difference based on the estimated age acquired by the age estimation unit 122 and the actual age of the subject inputted at the terminal 103. The disease estimation unit 125 estimates the possibility of the subject having a disease according to the magnitude of the estimated age difference.

[0032] FIG. 7 is a scatter plot 500 showing the correlation between actual age and estimated age. A scatter diagram 500 shows the results of this embodiment, with the horizontal axis representing actual age and the vertical axis representing estimated age. The points on the scatter diagram 500 are distributed in an approximately linear fashion. In this embodiment, after the learning process, the estimated age was calculated using the age estimation model 400 to verify the results of the learning. In the learning process and its verification, a total of 67,000 X-ray images 300 were used, which were 36,000 people selected as the healthy subject group from the data of 70,000 people described above. More specifically, of the 67,000 X-ray images 300 obtained from the 36,000 people, 53,000 X-ray images 300 and actual ages collected from 29,000 healthy subjects (learning subjects) were input to the age estimation model 400, and the learning process was executed. Next, 14,000 X-ray images 300 collected from 7,000 healthy individuals who were not learning subjects (hereinafter referred to as "confirmation subjects") were input into the age estimation model 400, and the estimated age was compared with the actual age. The correlation between the actual age and estimated age for the confirmation subjects (healthy individuals) was confirmed and plotted, as shown in Figure 7. The calculated correlation coefficient was 0.96. From the above, since there is a very strong correlation between the actual age and the estimated age, it was verified that the age estimation model 400 can estimate the subject's age almost accurately if the subject is healthy.

[0033] FIG. 8 is a schematic diagram of the first display screen 600. As shown in FIG. After the age estimation unit 122 acquires the estimated age of the subject calculated by the age estimation model 400, the output generation unit 126 generates a first display screen 600 to display the estimated age on the output unit 112. A message indicating the estimated age of the subject is displayed on the first display screen 600. In the case of Fig. 8, it is displayed that the estimated age of the subject is X.

[0034] FIG. 9 is a schematic diagram of the second display screen 610. As shown in FIG. When the disease estimation unit 125 completes estimation of the subject's disease possibility, the output generation unit 126 generates a second display screen 610 to display the estimation result on the output unit 112. The estimation of disease possibility will be described later. The second display screen 610 displays the subject's estimated age, actual age, estimated age difference, disease possibility, etc. The disease estimation unit 125 estimates that a subject whose estimated age is higher than his actual age, i.e., whose estimated age difference is large in the positive direction, is more likely to develop a specific disease than the general level.

[0035] In FIG. 9, it is shown that the subject's estimated age is X, his actual age is Y, and the estimated age difference is Z. Here, Z is assumed to be a positive number. In other words, the subject's estimated age was assumed to be older than his actual age. Based on the estimated age difference being Z, the second display screen 610 also displays the possibility of disease. The possibility of disease is assumed to be ranked in descending order of "A", "B", "C", and "D". Since the subject's estimated age was older than his actual age, the disease estimation unit 125 evaluates the possibility of disease of hypertension and hyperuricemia as "B".

[0036] For example, suppose that the diagnostic device 101 measures an estimated age based on the X-ray image 300 of a certain subject S1. As a result of the measurement, the second display screen 610 outputs a message indicating that the estimated age is 40 years old. It should be noted that the actual age of the subject S1 is 40 years old. In other words, since the physical condition of the subject S1 is determined to be appropriate for his or her actual age, the subject S1 is estimated to be healthy. Therefore, the disease estimation unit 125 estimates that the possibility of the subject S1 being afflicted is low.

[0037] On the other hand, suppose that when the diagnostic device 101 measures the estimated age based on the X-ray image 300 of another subject S2 whose actual age is 30, the second display screen 610 outputs that the estimated age is 70 years old. In other words, since the body of the subject S2 is determined to be in a weakened state compared to his actual age, the subject S2 is estimated to be unhealthy. Therefore, the disease estimation unit 125 estimates that the possibility of the subject S2 being afflicted is high. In this way, the estimated age difference, which is the difference between the subject's actual age and the estimated age calculated from the subject's physical characteristics, is linked to the possibility of afflicting the subject.

[0038] The validity of the estimation of the probability of developing various diseases based on the estimated age difference was confirmed by the inventor through a verification work using logistic regression analysis of the estimated age difference and the probability of developing diseases. In this verification, it was found that the odds ratio (probability ratio) of developing a specific disease increases exponentially for every 1 year increase in the estimated age difference. For example, in the case of hypertension, it was found that the odds ratio increases by 1.04 times for every 1 year increase in the estimated age difference.

[0039] For example, suppose that the actual age of a certain subject S3 is 30 years old, but the estimated age of the subject S3 based on the X-ray image 300 is 50 years old. In other words, the estimated age difference is +20 years. In this case, the probability that the subject S3 has high blood pressure is 2.19 (1.04 to the 20th power) times higher than a subject whose estimated and actual ages are both 30 years old, i.e., whose estimated age difference is 0 years. Based on the odds ratios thus obtained for various specific diseases, the disease estimation unit 125 evaluates the possibility of the subject having the disease. For example, the judgment criteria are set in advance, such as "D" if the probability of high blood pressure is within 1.1 times compared to a healthy person (estimated age difference of zero), and "C" if it is greater than 1.1 times and within 1.3 times.

[0040] The output generation unit 126 may generate the information to be displayed on the first display screen 600 and the second display screen 610 as sound, and output it from the output unit 112. The output generation unit 126 may also transfer the sound and the first display screen 600 and the second display screen 610 to the transmission unit 132, and output it from the terminal 103 via the communication network 102.

[0041] [Summary] The diagnostic device 101 has been described above based on the embodiments of the selection stage, learning stage, and implementation stage. When the diagnostic device 101 is given an X-ray image 300 as an input, an estimated age can be obtained by calculation using the age estimation model 400. The diagnostic device 101 of the present invention extracts only the X-ray image 300 of a healthy person to train the age estimation model 400, so that it is possible to calculate an estimated age with high accuracy. Note that a biomarker is a concept that is originally premised on being created based on data of a healthy person. The age estimation model 400 of this embodiment trains using the X-ray image 300 of a healthy person with no medical history as an input. In other words, the age estimation model 400 realizes correct biomarker estimation by excluding data of unhealthy people from the data to be learned. In addition, the diagnostic device 101 can estimate the possibility of suffering from a specific disease based on the estimated age.

[0042] According to the present invention, the risk of a disease for which the subject himself / herself has no subjective symptoms, i.e., a so-called pre-disease, can be estimated based on the estimated age difference. That is, since the possibility of contracting a specific disease increases with an increase in the estimated age difference, the magnitude of the risk of contracting a specific disease can be quantitatively estimated.

[0043] According to the present invention, it is also possible to estimate the importance of postoperative care for patients who have undergone surgery in an ICU (intensive care unit). For example, assume that a patient P1 whose actual age is 20 years old and a patient P2 whose actual age is 50 years old undergo surgery due to the same specific disease related to the large intestine. First, X-ray images 300 of the patients P1 and P2 are acquired before the surgery. Each X-ray image 300 is input to the diagnostic device 101, and the estimated ages of the patients P1 and P2 are output. At this time, it is assumed that the estimated age of the patient P1 is 30 years old. That is, since the estimated age difference is +10 years, it can be estimated that not only the large intestine but also the organs of the chest are weakened compared to a healthy 20-year-old. On the other hand, it is assumed that the estimated age of the patient P2 is 50 years old. That is, since the estimated age difference of the patient P2 is 0 years, it is estimated that the lungs of the patient P2 are as healthy as a healthy 50-year-old. The patients P1 and P2 suffer from the same specific disease, and the patient P1 is younger in terms of actual age. However, because the calculation of the estimated age confirms that patient P1 is physically weak, the doctor determines that it is necessary to closely monitor the postoperative progress of patient P1. In other words, the doctor determines that patient P1 should receive thorough postoperative care.

[0044] For patients suffering from particularly serious diseases, X-ray images 300 are acquired before surgery, and the diagnostic device 101 outputs the estimated age and the likelihood of the disease. For patients with a large estimated age difference, i.e., patients with a high likelihood of the disease, based on the output results, the doctor provides intensive postoperative care. In this way, by estimating the importance of postoperative care based on the output results from the diagnostic device 101, efficient use of human and physical medical resources, i.e., efficient triage of postoperative care, can be achieved.

[0045] According to the present invention, it is also possible to diagnose subjects in remote locations. For example, if a clinic in a remote location such as a rural area or an isolated island has a device capable of capturing an X-ray image 300, the X-ray image 300 captured by this device can be sent to a hospital where a diagnostic device 101 is installed or to a cloud or the like via a communication network 102, thereby making it possible to diagnose the possibility of the subject being afflicted with a disease. In addition, by diagnosing the possibility of afflicting a subject using the diagnostic device 101, new knowledge regarding the subject's physical condition can be obtained, such as the possibility that the subject may actually be afflicted with a chronic disease. This allows doctors in remote locations who lack the equipment and facilities of their clinics, or doctors who lack medical knowledge and diagnostic experience, to accurately grasp the physical condition of the subject. Doctors can also easily recognize abnormalities in patients that would not normally be noticed by referring to the diagnosis of the diagnostic device 101 in a remote location as a second opinion, so to speak. In other words, collaboration between doctors and AI (artificial intelligence) such as that of the present invention can raise the level of medical care nationwide. Furthermore, the doctor and the subject can easily check the diagnostic results by sending them to a terminal 103 at a remote clinic via the communication network 102. Also, based on the diagnostic results, if the doctor determines that the subject is highly likely to be infected, he or she can advise the subject to undergo detailed testing at a larger medical institution.

[0046] The diagnostic device 101 of the present invention can estimate the age of a subject only if it has an X-ray image 300. For example, suppose that the police rescue an unidentified lost person, but the lost person is unable to tell his / her identity due to a decline in cognitive function. Therefore, the lost person is sent to a medical institution where an apparatus capable of capturing an X-ray image 300 and the diagnostic device 101 are installed, and an X-ray image 300 of the lost person is captured. The captured X-ray image 300 is input to the diagnostic device 101, which calculates an estimated age. The estimated age thus obtained can be used as a trigger to progress the identification of the lost person by the police.

[0047] [Variations] In the present invention, the X-ray image 300 has been described as including the chest, but the main subject may be something other than the chest. For example, the subject of the X-ray image 300 may be the whole body or the head.

[0048] In the present invention, it has been described that each pixel of the X-ray image 300 of the subject is given as an input to the age estimation model 400. As a modified example, in addition to each pixel of the X-ray image 300, patient information such as the subject's medical history, medical treatment history, smoking history, age, surgical history, sex, race, blood test results, physiological test results, and pathological test results may also be given as an input. These patient information may be acquired by the input unit 111 of the diagnostic device 101. A doctor collects and inputs patient information through various tests. Alternatively, the input unit 111 may read patient information such as medical history from an electronic medical record or the like. According to the knowledge of the present inventor, the patient information also has a clear correlation with the estimated age. By inputting the patient information to the age estimation model 400 together with the X-ray image 300, the estimation accuracy of the estimated age is further improved.

[0049] In the present invention, the disease estimation unit 125 of the diagnostic device 101 is described as calculating the estimated age difference based on the estimated age calculated by the age estimation unit 122 and the actual age of the subject inputted by the terminal 103, and estimating the possibility of the subject being afflicted. As a modified example, the diagnostic device 101 may include a prognosis estimation unit (not shown). The prognosis estimation unit may calculate the estimated age difference based on the estimated age calculated by the age estimation unit 122 and the actual age of the subject inputted by the terminal 103, and estimate the prognosis of the subject. The "prognosis" here means the possibility that the condition of the subject will worsen during a predetermined period. Specifically, the prognosis estimation unit may estimate the subject's remaining life expectancy, the rate of serious illness, the possibility of being discharged from the hospital, etc., based on the estimated age difference. In addition, based on this information as well as the estimated age difference and the possibility of afflicting, the output generation unit 126 may generate a report as a deliverable summarizing the estimation results. [Explanation of symbols]

[0050] 100 diagnostic system, 101 diagnostic device, 102 communication network, 103 terminal, 104 selection device, 105 estimation device, 106 evaluation device, 110 user interface processing unit, 111 input unit, 112 output unit, 120 data processing unit, 121 image acquisition unit, 122 age estimation unit, 123 learning unit, 124 candidate selection unit, 125 disease estimation unit, 126 output generation unit, 130 communication unit, 131 receiving unit, 132 transmitting unit, 140 data storage unit, 200 electronic medical record, 300 X-ray image, 400 age estimation model, 410 input layer, 420 intermediate layer, 430 output layer, 500 scatter plot, 600 first display screen, 610 second display screen

Claims

1. an image acquisition unit for acquiring an X-ray image of a subject to be estimated; A diagnostic device comprising: an age estimation unit that estimates the age of a subject to be estimated by setting an X-ray image obtained about the subject to be estimated as an input value to a mathematical model that estimates age from an X-ray image.

2. 2. The diagnostic device according to claim 1, further comprising a learning unit that adjusts the mathematical model by using an X-ray image of a subject as an input value and an actual age of the subject as an output value.

3. The diagnostic device according to claim 2 , further comprising a candidate selection unit that acquires electronic medical records of candidates to be experimental subjects, and selects from a plurality of candidates a candidate who does not have a predetermined medical history by referring to the electronic medical records as a subject to be subjected to the experiment.

4. The diagnostic device according to claim 1, further comprising a disease estimation unit that calculates the age difference between the estimated age of the subject to be estimated based on the mathematical model and the actual age of the subject to be estimated, and estimates the possibility of the subject to be estimated having a specific disease based on the age difference.

5. The diagnostic device according to claim 4 , wherein the disease estimation unit determines that the greater the age difference, the higher the possibility of disease.

6. connected to a communication terminal via a communication network, a receiving unit for receiving an X-ray image and an actual age of a subject to be predicted from the communication terminal; The diagnostic device according to claim 1 , further comprising a transmission unit that transmits to the communication terminal data indicating a possibility of contracting a specific disease estimated from the estimated age based on the received X-ray image and the received actual age.

7. The diagnostic device according to claim 1, further comprising an input unit that acquires one or more of the medical history, medical treatment history, smoking history, age, surgical history, gender, race, blood test results, physiological test results, and pathological test results of the subject to be estimated as input values ​​for the mathematical model.

8. The diagnostic device according to claim 1 , further comprising a prognosis estimation unit that calculates an age difference between an estimated age of the subject to be estimated based on the mathematical model and an actual age of the subject to be estimated, and estimates a prognosis of the subject to be estimated based on the age difference.

Images