JP2022148729A - X-ray age estimation learning device, x-ray age estimation device, image age estimation learning device, image age estimation device, x-ray imaging system, evaluation device, x-ray age estimation learning method, x-ray age estimation method, and program - Google Patents

Abstract

To provide information useful for judging a state concerning health of a subject and determining a treatment policy.SOLUTION: An X-ray age estimation learning device includes a learning unit for learning an X-ray age estimation model showing correspondence between a feature quantity obtained from data of an X-ray image and an age of a subject by using data of an X-ray image of the subject and information on an age of the subject when the X-ray image is captured. An X-ray age estimation device includes an estimation unit for inputting the feature quantity obtained from the data of an estimation target X-ray image that is an X-ray image of the subject to be an age estimation target into a learned X-ray age estimation model to obtain estimation results of the age of the subject of an estimation target.SELECTED DRAWING: Figure 1

Description

The present invention relates to an X-ray age estimation learning device, an X-ray age estimation learning device, an image age estimation learning device, an image age estimation device, an X-ray imaging system, an evaluation device, an X-ray age estimation learning method, an X-ray age estimation method, and a program.

Conventionally, there has been a technique for measuring the vascular age of a living body from a subject's pulse wave (see Patent Document 1, for example).

WO2010/134233

Vascular age is currently widely used because it is intuitively easy to understand. However, vascular age is often not significantly changed by treatment. Therefore, vascular age was rarely used as information for physicians to change their clinical treatment policy.

In view of the above circumstances, the present invention provides an X-ray age estimation learning device, an X-ray age estimation device, an image age estimation learning device, and an image age estimation device that can provide useful information for judging health-related conditions and determining treatment policies. , an X-ray imaging system, an evaluation device, an X-ray age estimation learning method, an X-ray age estimation method and a program.

According to one aspect of the present invention, a characteristic amount obtained from the X-ray image data and the age of the subject are obtained by using X-ray image data of a subject and information on the age of the subject when the X-ray image was taken. and a learning unit for learning an X-ray age estimation model showing the correspondence with the X-ray age estimation learning device.

One aspect of the present invention is the X-ray age estimation learning device described above, wherein the X-ray image is a chest X-ray image.

One aspect of the present invention is an X-ray age estimation model that is learned using X-ray image data of a subject and information about the age of the subject when the X-ray image was taken, wherein from the X-ray image data Inputting the feature amount obtained from the data of the estimated target X-ray image, which is the X-ray image of the subject whose age is to be estimated, into the X-ray age estimation model showing the correspondence between the obtained feature amount and the age of the subject, and estimating the age. an estimating unit that obtains a result of estimating the age of the subject of interest.

One aspect of the present invention relates to the X-ray age estimating apparatus described above, which relates to a portion of the estimation target X-ray image from which the feature amount satisfying a predetermined condition for a degree of contribution to age estimation of the subject for age estimation is obtained. A contribution information acquisition unit that acquires contribution information, which is information, is further provided.

One aspect of the present invention is the above-described X-ray age estimation apparatus, wherein the information on the age in the estimation result by the estimation unit, the actual age of the subject whose age is to be estimated, and the age in the estimation result by the estimation unit. It further comprises an output unit that outputs one or more of difference information and the contribution information.

One aspect of the present invention is the X-ray age estimation device described above, wherein the X-ray image and the estimation target X-ray image are chest X-ray images.

According to one aspect of the present invention, data of a transmission image obtained by transmission imaging of the human body of a subject and information on the age of the subject when the transmission image was captured are used to perform the transmission imaging. The image age estimation learning device includes a learning unit that learns an age estimation model indicating correspondence between a feature value obtained from image data and the age of the subject.

According to one aspect of the present invention, learning is performed using transmission image data, which is an image obtained by transmission imaging of the human body of a subject, and age information of the subject when the transmission imaging image is captured. An age estimating model, wherein the age estimating model indicates the correspondence between the feature value obtained from the data of the transmission image and the age of the subject, and the estimation target image, which is the transmission image of the subject whose age is to be estimated. The image age estimation apparatus includes an estimation unit that inputs a feature value obtained from data and obtains an estimation result of the age of the subject whose age is to be estimated.

According to one aspect of the present invention, the X-ray age estimation apparatus described above performs X-ray imaging of a subject, and the X-ray image data of the subject obtained by the X-ray imaging is sent to the X-ray age estimation apparatus as data of an estimation target X-ray image. An X-ray imaging system comprising: an X-ray imaging device for inputting;

One aspect of the present invention is based on the result of performing a predetermined calculation using the estimation result of the subject's age by the above-described X-ray age estimation apparatus, and the information on the subject's biological body or lifestyle habits or a numerical value representing the information. and an evaluation unit that evaluates the subject's disease or condition.

According to one aspect of the present invention, a characteristic amount obtained from the X-ray image data and the age of the subject are obtained by using X-ray image data of a subject and information on the age of the subject when the X-ray image was taken. and a learning step of learning a Roentgen age estimation model showing a correspondence between the roentgen age estimation learning method.

One aspect of the present invention is an X-ray age estimation model that is learned using X-ray image data of a subject and information about the age of the subject when the X-ray image was taken, wherein from the X-ray image data Inputting the feature amount obtained from the data of the estimated target X-ray image, which is the X-ray image of the subject whose age is to be estimated, into the X-ray age estimation model showing the correspondence between the obtained feature amount and the age of the subject, and estimating the age. obtaining an estimate of the subject's age of interest.

One aspect of the present invention is a program for causing a computer to operate as any of the X-ray age estimation learning devices described above.

One aspect of the present invention is a program for causing a computer to operate as any of the above-described X-ray age estimation devices.

INDUSTRIAL APPLICABILITY According to the present invention, it is possible to provide useful information for judging health-related conditions and determining treatment strategies.

1 is a block diagram showing the configuration of an X-ray age estimation system according to one embodiment of the present invention; FIG. 1 is a block diagram showing the configuration of an X-ray age estimation learning device; FIG. It is a block diagram which shows the structure of an X-ray age estimation apparatus. It is a figure which shows the hardware constitutions of the computer used as a Roentgen age estimation learning apparatus or a Roentgen age estimation apparatus. FIG. 4 is a flowchart showing processing of the X-ray age estimation learning device; It is a flowchart which shows the process of a radiography age estimation apparatus. FIG. 10 is a diagram showing an overview of an embodiment using the Roentgen age estimation system; FIG. 2 is a diagram showing the age structure indicated by the data used in the X-ray age estimation system; FIG. 10 is a diagram showing the estimation accuracy of the Roentgen age estimation model; FIG. 10 is a diagram showing the correspondence between chest X-ray age and actual age obtained using evaluation data as input. FIG. 10 is a diagram showing the correspondence between chest X-ray age and actual age obtained using test data as input. FIG. 10 is a diagram showing a heat map; FIG. 10 is a diagram showing the relationship between the age difference and the odds ratio for those with findings. FIG. 10 is a diagram showing the relationship between age difference and heart failure outcome. 1 is a diagram showing the configuration of an X-ray imaging system; FIG. It is a figure which shows the structure of an evaluation system.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a diagram showing an X-ray age estimation system 1. As shown in FIG. The Roentgen age estimation system 1 has an Roentgen age estimation learning device 10 and an Roentgen age estimation device 50 . The X-ray age estimation learning device 10 and the X-ray age estimation device 50 respectively perform image age estimation when X-ray image data obtained by performing X-ray photography is used as image data obtained by transmissively photographing the interior of the human body. It is an example of a learning device and an image age estimation device. Although only one X-ray age estimation device 50 is shown in the figure, the number of X-ray age estimation devices 50 is arbitrary. Alternatively, the Roentgen age estimation learning device 10 and the Roentgen age estimation device 50 may be integrated into the same device.

The X-ray age estimation learning device 10 learns an X-ray age estimation model using learning data including X-ray image data and age information of a subject. The X-ray image data is X-ray image data obtained by X-raying a subject. Age information indicates the actual age of the subject when the radiograph was taken. The X-ray age estimation model is a model that indicates the correspondence between the feature value obtained from the X-ray image data and the X-ray age of the subject. The roentgen age is the age estimated based on the roentgen image. The roentgen age estimation device 50 uses the roentgen age estimation model learned by the roentgen age estimation learning device 10 to obtain the roentgen age of the age estimation target subject from the roentgen image data of the age estimation target subject. In this embodiment, chest X-ray image data is used as the X-ray image data. Chest X-ray image data is chest X-ray image data obtained by X-raying the chest of a subject. The radiographic age estimated based on chest radiographic images is referred to as chest radiographic age.

Chest radiography is the most commonly performed examination. A chest X-ray image obtained by radiography can be obtained quickly and easily, and contains a lot of useful information. Therefore, chest X-ray images are used not only for routine medical care but also for medical examinations. Chest X-ray images are very important for diagnosing and monitoring cardiovascular and pulmonary diseases. However, interpretation of chest X-ray images requires specialized experience and knowledge, and results in a wide range of findings. Therefore, chest radiographs have not been used as a constant quantitative marker. This has made it difficult to incorporate information from chest radiographs into recent big data-based healthcare reforms.

On the other hand, conventionally, it has been known that chest X-ray images differ depending on the subject's age. However, it has not been clarified what kind of characteristics the difference appears in concrete terms, and it has been difficult even for an experienced doctor to accurately estimate the subject's age from the chest X-ray image. The X-ray age estimating apparatus 50 of the present embodiment obtains the chest X-ray age, which is the result of estimating the age of the subject whose age is to be estimated, using only the chest X-ray image data of the subject whose age is to be estimated.

The radiographic age estimator 50 also presents the portion of the chest radiographic image that contributed to the estimation of the chest radiographic age. Since the chest X-ray age estimated by the X-ray age estimating device 50 is highly accurate as described later, if the estimated chest X-ray age is higher than the actual age and there is a large difference between them, some kind of abnormal finding is suggested. be. In such a case, the portion of the chest X-ray image that greatly contributed to the estimation of the chest X-ray age is estimated to be the location of the abnormal findings. In addition, the difference between chronological age and chest radiograph age (hereinafter also referred to as age difference) is also associated with adverse outcomes such as heart failure. These findings suggest that chest radiographic age is not only a novel cardiovascular biomarker, but also useful for clinicians to predict, prevent, and manage heart disease in patients.

FIG. 2 is a block diagram showing the configuration of the X-ray age estimation learning device 10. As shown in FIG. The X-ray age estimation learning device 10 includes a model storage unit 11 , a data storage unit 12 , a conversion unit 13 , a learning data generation unit 14 , a learning unit 15 , a selection unit 18 and an output unit 19 . Note that the data storage unit 12 may be located outside the X-ray age estimation learning device 10 .

The model storage unit 11 stores an X-ray age estimation model. A DNN (Deep Neural Network) such as a CNN (Convolutional neural network) can be used for the Roentgen age estimation model, for example. The model storage unit 11 may store a plurality of Roentgen age estimation models having different structures.

The data storage unit 12 stores learning data and verification data. The learning data is used for learning the Roentgen age estimation model. Validation data is used to validate the accuracy of the trained Roentgen age estimation model. The learning data and validation data include the subject's X-ray image data and the subject's age information. The X-ray image data used in the present embodiment is, for example, chest X-ray image data obtained by X-raying the chest of a subject from the front. The age information included in the training data and validation data indicates the actual age of the subject when the radiograph was taken. Information indicating the date of birth of the subject and the date on which the X-ray was taken may be used as the age information.

The conversion unit 13 performs predetermined data conversion on the X-ray image data so as to obtain the feature amount to be input to the X-ray age estimation model. Note that the X-ray age estimation learning device 10 does not need to have the conversion unit 13 when data conversion of the X-ray image data is unnecessary. The learning data generation unit 14 generates new learning data using the learning data stored in the data storage unit 12 and writes the generated learning data to the data storage unit 12 .

The learning unit 15 learns the Roentgen age estimation model using the learning data. The learning unit 15 has an estimating unit 16 and an updating unit 17 . The estimating unit 16 acquires feature amounts from the X-ray image data included in the learning data. A feature amount is, for example, a pixel value. The estimating unit 16 inputs the acquired feature amount to the X-ray age estimation model stored in the model storage unit 11, and obtains an estimation result of the X-ray age. The updating unit 17 updates the X-ray age estimation model so that the error between the X-ray age obtained based on the X-ray image data included in the learning data and the actual age of the subject indicated by the learning data is reduced. When the Roentgen age estimation model is a DNN, updating the Roentgen age estimation model means updating the value of the weight between neurons, that is, the value of the parameter used in the function representing the connection relationship between neurons constituting the DNN. It is to be.

When there are a plurality of X-ray age estimation models learned by the learning unit 15, the selection unit 18 inputs the feature amount obtained from the X-ray image data included in the verification data to each X-ray age estimation model to determine the X-ray age. Get the estimated result. The selection unit 18 may cause the estimation unit 16 to calculate the X-ray age. The selection unit 18 calculates the error between the X-ray age obtained based on the X-ray image data included in the verification data and the actual age of the subject indicated by the verification data. The selection unit 18 selects the Roentgen age estimation model with the smallest statistical error from a plurality of Roentgen age estimation models. If there is only one Roentgen age estimation model, the Roentgen age estimation learning device 10 may not have the selection unit 18 . In addition, the selection unit 18 outputs the statistic value of the error calculated for each Roentgen age estimation model by the output unit 19, and the information of the Roentgen age estimation model selected by the user based on the output statistic value of the error is not shown. It may be input to the X-ray age estimation learning device 10 through the input unit.

The output unit 19 outputs various data. The output may be displayed on a display, printed by a printer, written on a recording medium, or transmitted to another device connected to the X-ray age estimation learning device 10 via a network. The output unit 19 outputs the Roentgen age estimation model selected by the selector 18 by transmitting it to the Roentgen age estimation device 50 or another device, or by writing it in a recording medium.

FIG. 3 is a block diagram showing the configuration of the X-ray age estimation device 50. As shown in FIG. The Roentgen age estimation device 50 includes an input unit 51 , a model storage unit 52 , a conversion unit 53 , an estimation unit 54 , a contribution information acquisition unit 55 and an output unit 56 .

The input unit 51 inputs data. The input may be received from another device connected to the Roentgen age estimation device 50 via a network, or may be read from a recording medium. The input unit 51 inputs the Roentgen age estimation model learned by the Roentgen age estimation learning device 10 and writes it to the model storage unit 52 . Further, the input unit 51 inputs X-ray image data to be used for estimating the X-ray age, and outputs the X-ray image data to the conversion unit 53 .

The model storage unit 52 stores an X-ray age estimation model. As the model storage unit 52, the model storage unit 11 included in the X-ray age estimation learning device 10 may be used. The conversion unit 53 performs data conversion similar to the data conversion performed by the conversion unit 13 of the X-ray age estimation learning device 10 on the X-ray image data input by the input unit 51 . Note that the X-ray age estimation device 50 may not have the conversion unit 53 if data conversion of the X-ray image data is not required. The estimating unit 54 acquires a feature amount from the X-ray image data converted by the converting unit 53 in the same manner as the estimating unit 16 of the X-ray age estimation learning device 10 . The estimating unit 54 inputs the acquired feature amount to the X-ray age estimation model stored in the model storage unit 52 to obtain the X-ray age.

The contribution information acquisition unit 55 acquires contribution information. The contribution information is information relating to the portion of the X-ray image data for which the feature quantity that satisfies a predetermined condition for the degree of contribution to the estimation of the X-ray age using the X-ray age estimation model is obtained. The predetermined condition may be a condition for classifying the degree of contribution into a plurality of stages. Further, the contribution information may be information representing the degree to which the feature amount obtained from each portion (for example, each pixel) of the X-ray image data contributed to the estimation of the X-ray age. When the Roentgen age estimation model is CNN, the contribution information obtaining unit 55 obtains, for example, a heat map generated by Grad-CAM as contribution information.

The output unit 56 outputs various data. The output may be displayed on a display, printed by a printer, written on a recording medium, or transmitted to another device connected to the X-ray age estimation device 50 via a network. The output unit 56 outputs the information on the X-ray age estimated by the estimation unit 54 and the contribution information acquired by the contribution information acquisition unit 55 .

Each of the Roentgen age estimation learning device 10 and the Roentgen age estimation device 50 is implemented by, as an example, a computer 80 shown in FIG. FIG. 4 is a diagram showing the hardware configuration of the computer 80. As shown in FIG. The computer 80 has a configuration in which a processor 81 , a storage unit 82 , a user interface 83 and a communication interface 84 are connected via a bus 85 . The processor 81 performs calculation and control. The processor 81 is, for example, a CPU (central processing unit) or a GPU (Graphics Processing Unit). The storage unit 82 stores a program for causing the computer 80 to function as the Roentgen age estimation learning device 10 or the Roentgen age estimation device 50 . The processor 81 reads the program from the storage unit 82 and executes it. Furthermore, the storage unit 82 has a work area and the like used when the processor 81 executes the program. The user interface 83 is an input device such as a keyboard, pointing device (mouse, tablet, etc.), buttons, touch panel, etc., and a display device such as a display. The communication interface 84 is for communicably connecting with other devices. Note that all or part of the functions of the X-ray age estimation learning device 10 and the functions of the X-ray age estimation device 50 are realized by hardware such as ASIC (Application Specific Integrated Circuit), PLD (Programmable Logic Device), and FPGA (Field Programmable Gate Array). hardware.

In addition, when the X-ray age estimation learning device 10 and the X-ray age estimation device 50 are integrated into the same device (hereinafter referred to as a learning estimation device), the model storage unit 11 and the model storage unit 52 may be the same. The estimation unit 16 and the estimation unit 54 may be the same, and the output unit 19 and the output unit 56 may be the same.

Further, the Roentgen age estimation learning device 10, the Roentgen age estimation device 50, and the learning estimation device may each be realized by a plurality of computer devices connected to a network. In this case, it is possible to arbitrarily choose which one of the plurality of computer devices realizes each functional unit of the Roentgen age estimation learning device 10, the Roentgen age estimation device 50, and the learning estimation device. Also, one functional unit may be realized by a plurality of computer devices.

For example, the data storage unit 12, the model storage unit 11, the conversion unit 13, the learning data generation unit 14, the learning unit 15, and the selection unit 18 of the X-ray age estimation learning device 10 may be realized by different computer devices. The storage unit 12 and the conversion unit 13, and the model storage unit 11, the learning data generation unit 14, the learning unit 15 and the selection unit 18 may be realized by different computer devices. The generation unit 14, the model storage unit 11, the learning unit 15, and the selection unit 18 may be implemented by different computer devices. Further, for example, the model storage unit 52, the conversion unit 53, the estimation unit 54, and the contribution information acquisition unit 55 of the Roentgen age estimation device 50 may be realized by different computer devices. The unit 56, the model storage unit 52, the estimation unit 54, and the contribution information acquisition unit 55 may be realized by different computer devices. 54 and the contribution information acquisition unit 55 may be implemented by different computer devices. A part or all of the data storage unit 12, the conversion unit 13, the learning data generation unit 14, and the learning unit 15 of the X-ray age estimation learning device 10 may be realized by a plurality of computer devices. When the learning unit 15 is implemented by a plurality of computer devices, each computer device may learn different types of X-ray age estimation models in parallel.

Next, learning processing of the X-ray age estimation model will be described. In the following, the case of using DNN as an X-ray age estimation model will be described as an example. The Roentgen age estimation learning device 10 learns the Roentgen age estimation model by transfer learning and fine tuning. As a result, it is possible to improve the DNN that has been learned in advance for other purposes and learn the Roentgen age estimation model in a shorter time and with less learning data than learning the DNN as the Roentgen age estimation model from the beginning. can.

FIG. 5 is a flowchart showing an example of processing of the X-ray age estimation learning device 10. As shown in FIG. The model storage unit 11 stores in advance an X-ray age estimation model in which an initial value is set for each parameter. This Roentgen age estimation model is generated by replacing the final layer of the learned DNN or the final layer and some layers for other purposes. A learned DNN receives feature values obtained from image data. Also, the output layer of the Roentgen age estimation model is one neuron whose output value is the estimated age, that is, the Roentgen age.

In this embodiment, the model storage unit 11 stores M types (M is an integer equal to or greater than 1) of X-ray age estimation models (1) to (M). Neural networks with different structures are used for the Roentgen age estimation models (1) to (M). An m-th (m is an integer of 1 or more and M or less) Roentgen age estimation model is described as an Roentgen age estimation model (m).

The conversion unit 13 performs predetermined data conversion on the X-ray image data of the learning data and verification data stored in the data storage unit 12 so as to obtain the feature amount to be input to the X-ray age estimation model (step S105). For example, the conversion unit 13 converts X-ray image data into image data having a predetermined data format and a predetermined data size. If data conversion is unnecessary, the X-ray age estimation learning device 10 does not have to perform the process of step S105.

The learning data generation unit 14 reads some or all of the learning data from the data storage unit 12 . The learning data generation unit 14 generates new learning data by replacing the image indicated by the X-ray image data included in the read learning data with the X-ray image data of an image randomly tilted within a predetermined angle. The learning data generation unit 14 may generate a plurality of new learning data with different tilt angles of the image from one learning data. The learning data generation unit 14 writes the generated learning data to the data storage unit 12 (step S110). By adding the learning data generated by the learning data generation unit 14 to learning, the X-ray age estimation model is generalized and overfitting is prevented.

The learning unit 15 sets the initial value 1 to the variable m (step S115). The learning unit 15 performs transfer learning of the Roentgen age estimation model (m) using the learning data (step S120). That is, the estimation unit 16 selects one piece of learning data and acquires a feature amount from the selected learning data. The estimating unit 16 inputs the acquired feature quantity to the Roentgen age estimation model (m) to obtain the result of roentgen age estimation. The updating unit 17 updates the values of some of the parameters of the Roentgen age estimation model (m) so that the difference between the obtained Roentgen age and the real age indicated by the selected learning data becomes small. do. Specifically, the updating unit 17 updates the values of parameters to be updated in transfer learning using an optimization algorithm so as to minimize the loss function calculated using the real age and the X-ray age. . The loss function is, for example, MSE (mean square error). Parameters to be updated in transfer learning are parameters not included in the learned CNN of the transfer source of the Roentgen age estimation model (m), but may include some parameters included in the CNN of the transfer source. The learning unit 15 repeats the above-described estimation of the X-ray age by the estimation unit 16 and updating of the parameter value to be updated by the update unit 17 for each piece of learning data.

Subsequently, the learning unit 15 fine-tunes the X-ray age estimation model (m) using the learning data (step S125). That is, the estimation unit 16 selects one piece of learning data and acquires a feature amount from the selected learning data. The estimating unit 16 inputs the acquired feature quantity to the Roentgen age estimation model (m) to obtain the result of roentgen age estimation. The updating unit 17 updates the values of all the parameters of the Roentgen age estimation model (m) so that the difference between the obtained Roentgen age and the real age indicated by the selected learning data becomes small. That is, the updating unit 17 updates the values of all parameters using an optimization algorithm so as to minimize the loss function calculated using the real age and the X-ray age. The loss function is, for example, MSE. The learning unit 15 repeats estimation by the estimating unit 16 and updating of all parameter values by the updating unit 17 for each piece of learning data.

The learning unit 15 determines whether or not the variable m has reached the model number M of the Roentgen age estimation model (step S130). When the learning unit 15 determines that the variable m has not reached the number of models M (step S130: NO), it adds 1 to the variable m (step S135). The learning unit 15 repeats the processing from step S120. When the learning unit 15 determines that the variable m has reached the number of models M (step S130: YES), the process proceeds to step S140.

The selection unit 18 inputs the feature amount obtained from the X-ray image data included in each verification data to each of the X-ray age estimation models (1) to (M), and obtains the X-ray age estimation result (step S140). . The selection unit 18 calculates the difference between the X-ray age obtained based on the X-ray image data included in the verification data and the actual age of the patient indicated by the verification data. The selection unit 18 calculates the value of the evaluation index using the calculated difference for each of the X-ray age estimation models (1) to (M). As evaluation indicators, for example, MSE, RMSE (root mean squared error), Pearson's correlation coefficient r between chronological age and X-ray age, MAE (mean absolute error), etc. can be used. can. The selection unit 18 selects an X-ray age estimation model (k) (k is an integer of 1 or more and M or less), which indicates that the value of the evaluation index has the highest estimation accuracy from among the X-ray age estimation models (1) to (M). is selected (step S145). The selection unit 18 may select the most accurate Roentgen age estimation model (k) by combining a plurality of evaluation indices. The output unit 19 outputs the Roentgen age estimation model estimated by the selection unit 18 .

Note that the Roentgen age estimation learning device 10 may learn the Roentgen age estimation model without enhancing the learning data by the learning data generator 14 . Alternatively, the Roentgen age estimation learning device 10 may learn the values of all the parameters of the Roentgen age estimation model based on learning data without performing transfer learning. Further, the Roentgen age estimation learning device 10 may perform learning processing of a plurality of types of Roentgen age estimation models in parallel. Further, when the number of models M=1, the Roentgen age estimation learning apparatus 10 does not have to perform the process of step S140. In this case, in step S145, the output unit 19 outputs the Roentgen age estimation model (1).

Subsequently, the process of estimating the X-ray age will be described. The Roentgen age estimation device 50 estimates the Roentgen age using the Roentgen age estimation model (k) learned by the Roentgen age estimation learning device 10 . FIG. 6 is a flowchart showing an example of processing of the X-ray age estimation device 50. As shown in FIG. The input unit 51 receives the Roentgen age estimation model (k) output from the Roentgen age estimation learning device 10 and writes it to the model storage unit 52 . The X-ray age estimation device 50 performs the processing in FIG. 6 for each X-ray image data for X-ray age estimation.

The input unit 51 inputs X-ray image data and outputs it to the conversion unit 53 (step S205). The input X-ray image data is the X-ray image data of a subject whose X-ray age is to be estimated. The input unit 51 may further input information about the subject's age. The conversion unit 53 performs predetermined data conversion on the input X-ray image data so as to obtain the feature amount to be input to the X-ray age estimation model (k) (step S210). For example, the conversion unit 53 converts X-ray image data into image data having a predetermined data format and a predetermined data size. If data conversion is unnecessary, the X-ray age estimation device 50 does not have to perform the process of step S210.

The estimation unit 54 acquires feature amounts from the X-ray image data. The estimating unit 54 inputs the acquired feature amount to the Roentgen age estimation model (k) to obtain an estimation result of the Roentgen age (step S215). The contribution information acquiring unit 55 obtains a feature amount that satisfies a predetermined condition for the degree of contribution to the estimation of the X-ray age using the X-ray age estimation model among the feature amounts input to the X-ray age estimation model (k). Contribution information indicating portions of the image data is obtained (step S220). Specifically, the contribution information is a heat map generated using Grad-CAM. By Grad-CAM, the contribution information acquisition unit 55 obtains the position of the pixel from which the feature amount used when performing the estimation by the Roentgen age estimation model (k) and the degree to which the feature amount contributed to the estimation. get information. The contribution information acquisition unit 55 generates a heat map by superimposing a color representing the degree of contribution of each pixel to the estimation on the X-ray image data. Note that the contribution information acquiring unit 55 may use Guided Backpropagation to generate the heat map. The output unit 56 outputs the information of the X-ray age estimated by the estimation unit 54 and the heat map generated by the contribution information acquisition unit 55 as estimation results (step S225). When the information on the actual age is input, the output unit 56 may further output the information on the age difference, which is the difference between the X-ray age and the actual age.

Next, an example of learning an X-ray age estimation model and X-ray age estimation using the X-ray age estimation system 1 will be described. FIG. 7 is a diagram showing an outline of an embodiment using the X-ray age estimation system 1. In FIG. In this example, 11 types of CNNs, ResNet18, ResNet34, ResNet50, ResNet101, ResNet152, DenseNet121, DenseNet161, DenseNet169, DenseNet201, Inception-v4, and SENet154, were used for the Roentgen age estimation model. In order to perform transfer learning, each of these 11 types of pretrained CNN models M0 was obtained from ImageNet (URL: https://github.com/Cadene/pretrained-models.pytorch). The model M0 consists of a convolution part and a full connection part, and classifies image data into 1,000 categories. That is, the output layer of model M0 consists of 1,000 neurons corresponding to each category. The Roentgen age estimation model M1 used in the present embodiment uses the convolution part of the model M0 as it is, and replaces the full joint part of the model M0 with two new full joint parts. The two new fully-connected parts after replacement are a batch-normalization layer, a fully-connected layer consisting of 512 neurons using ReLU (Rectified Linear Unit) as the activation function, a batch-normalization layer, and It consists of the last fully connected layer. The final fully connected layer is connected to one neuron that outputs the chest radiograph age.

In addition, in this embodiment, the chest X-ray image data database A (https://cloud.google.com/healthcare/docs/resources/public-datasets/nih-chest) provided by the National Institutes of Health is learned. data, validation data and test data. The test data is data used to evaluate the accuracy of the Roentgen age estimation model trained by the Roentgen age estimation learning device 10 , and corresponds to input data input to the Roentgen age estimation device 50 . Database A is primarily US patient data. Of the 112,120 chest X-ray image data of 30,805 patients contained in database A, 63,328 are male. From the chest X-ray image data of this database A, 16 images of patients aged 100 years or older were removed, and the remaining chest X-ray image data were randomly divided into learning data, verification data and test data. As a result, 102,029 chest X-ray image data of 28,029 patients are used as training data D1, chest X-ray image data of 2,523 patients are used as verification data D2, and 613 chest X-ray images of 250 patients are used. The X-ray image data was divided into test data D3.

In addition, in order to evaluate the estimation accuracy in populations with different physiques, database B of chest X-ray image data provided by the Japan Society of Radiological Technology (JSRT) (http://db.jsrt.or.jp/eng. php) was also used for test data. Database B is data of Japanese patients. From the 247 chest X-ray image data of 247 patients included in database B, 2 images of unknown sex and age were excluded, and 245 image data of 245 patients were used as test data D4. Of the test data D4, 118 are for men.

Furthermore, database C containing chest X-ray image data of 1,562 patients hospitalized with heart disease was used as test data D5. Database C contains chest X-ray image data taken within two days of hospitalization. Of the data in the test data D5, 920 are for men. Chest X-ray image data in database A and database B is added with information on age and sex at the time the chest X-ray image data was captured, as well as finding information. The chest X-ray image data of database C is added with more detailed medical data than database A and database B in addition to age and sex. Clinical data includes information on clinical background and events such as re-hospitalization due to heart disease and death.

FIG. 8 is a diagram showing the age structure indicated by the data used by the Roentgen age estimation system. FIG. 8(a), FIG. 8(b), and FIG. 8(c) show age histograms of database A, database B, and database C, respectively. The age range for database A is 1 to 95 years, with a median of 49 years and an interquartile range of 35-39 years. The age range for database C is 18 to 98 years, with a median of 78 years and an interquartile range of 69-84 years.

In this embodiment, the conversion unit 13 and the conversion unit 53 convert the chest X-ray image data into png format image data of 320×320 pixels (step S105 in FIG. 5, step S210 in FIG. 6). Further, the learning data generation unit 14 performed random padding and rotation on the image indicated by the chest X-ray image data included in the learning data obtained from the database A to generate additional learning data (step S110 in FIG. 5). . Rotation was randomized between -20 and 20 degrees, and no image reversal was performed. 5, the update unit 17 updates the parameter values using MSE (mean square error) shown in the following equation (1) as a loss function. Note that n is the number of learning data. _yi is the correct age (actual age) indicated by the learning data, and y ^{^} _i is the chest X-ray age estimated from the chest X-ray image data included in the learning data.

The update unit 17 uses Adam (Adaptive Moment Estimation) and CLR (Cyclical Learning Rate) as optimization algorithms in steps S120 and S125 of FIG.

FIG. 9 is a diagram showing the estimation accuracy of each Roentgen age estimation model. FIG. 9 is obtained by performing the processing of step S140 in FIG. indicates the value of the evaluation index. Evaluation indices are MSE, RMSE, Pearson's correlation coefficient r(R) between chronological age and chest radiograph age, and MAE. It can be seen from FIG. 9 that SENet 154 has the highest estimation accuracy.

FIG. 10 is a diagram showing the correspondence between chest X-ray age and actual age obtained using verification data D2. FIG. 10 shows the chest X-ray age obtained by the X-ray age estimation learning device 10 by inputting the feature amount obtained from the chest X-ray image data included in the verification data D2 into the trained X-ray age estimation model using SENet 154. , and the actual age indicated by the verification data D2 are plotted. Pearson's correlation coefficient r-value (R) was 0.9516 and p-value (P) was less than 2.2×10 ⁻³²³ .

FIG. 11 is a diagram showing the correspondence between chest X-ray age and actual age obtained using test data D3 and D4. FIG. 11(a) plots the correspondence between the chest X-ray age obtained by the X-ray age estimating device 50 using the test data D3 as an input and the actual age indicated by the test data D3. FIG. 11(b) plots the correspondence between the chest X-ray age obtained by the X-ray age estimating device 50 using the test data D4 as an input and the actual age indicated by the test data D4. In FIGS. 11(a) and 11(b), a trained X-ray age estimation model using SENet154 was used to estimate the chest X-ray age.

As shown in FIGS. 11(a) and 11(b), there is a very strong correlation between the chest X-ray age estimated by the X-ray age estimation device 50 and the actual age. For test data D3, the r-value (R) of Pearson's correlation coefficient is 0.961, the p-value (P) is less than 2.2×10 ⁻³²³ , and the MAE between chest radiograph age and chronological age is 3. was 79 years old. In the case of test data D4, the r value (R) of Pearson's correlation coefficient is 0.916, p value (P) = 1.51 × 10 ^-98 , and the MAE between the chest X-ray age and the actual age is was 4.56 years old. FIG. 11(c) plots the average ages estimated by four experienced doctors based on chest X-ray images and their actual ages. Although there is a relationship between the age estimated by the doctor and the actual age, it can be seen that the accuracy of the estimation result by the X-ray age estimation device 50 is better.

FIG. 12 is a diagram showing a heat map output by the X-ray age estimation device 50. As shown in FIG. FIG. 12( a ) is a heat map when Grad-CAM is used for the contribution information acquisition unit 55 of the X-ray age estimation device 50 . FIG. 12(b) is a heat map when Guided Grad-CAM is used for the contribution information acquisition unit 55 of the X-ray age estimation device 50. FIG. According to FIGS. 12(a) and 12(b), the vicinity of the upper part of the mediastinum and the circumference of the thorax contribute greatly to the estimation, and the shape and calcification of the aorta affect the estimation of chest X-ray age. can be considered.

FIG. 13 is a diagram showing the relationship between the age difference and the odds ratio with findings. The age difference is the difference between the chest X-ray age obtained by the X-ray age estimation device 50 using the test data D3 and D4 and the actual age. The X-ray age estimation device 50 uses a trained X-ray age estimation model using SENet 154 to estimate the chest X-ray age. FIG. 13 also shows 95% confidence intervals on a log scale. FIG. 13 shows that the greater the age difference, the higher the odds ratio for those with findings.

FIG. 14 is a diagram showing the relationship between age differences and heart failure outcomes. The age difference is the difference between the chest X-ray age obtained by the X-ray age estimating device 50 using the test data D5 and the actual age. The X-ray age estimation device 50 uses a trained X-ray age estimation model using SENet 154 to estimate the chest X-ray age. The accuracy of chest radiographic age estimates obtained using test data D5 was reduced. However, Pearson's correlation coefficient r value was 0.796 and p value was 4.6×10 ⁻²⁹¹ , confirming a strong correlation.

Furthermore, based on the age difference, the test data D5 is divided into three groups: a group G1 whose chest X-ray age is younger, a group G2 whose chest X-ray age is almost the same as the actual age, and a group G3 whose chest X-ray age is higher. separated into two groups. Group G1 and group G3 are each 20% of the total, and group G2 is 60% of the total. Then, the event-free survival rate was calculated for each group based on the clinical data added to the test data D5. Events here were a composite endpoint of readmission for heart failure, heart transplant, and death from any cause. FIG. 14 shows changes over time in event-free survival rates by group. It can be seen from FIG. 14 that the event-free survival rate is higher when the chronological age is younger than the chest X-ray age, and the event-free survival rate is lower when the chronological age is higher than the chest X-ray age. Thus, in patients with heart failure, prognosis (presence or absence of occurrence of cardiovascular events) can be stratified using chest X-ray age. Therefore, the age difference between chest X-ray age and chronological age is useful for preventing overlooking of chest X-ray findings in healthy subjects. Furthermore, age on chest radiograph may help guide disease treatment decisions, as age differences reflect disease severity.

An example of a system to which the X-ray age estimation device 50 described above is applied will be described.
FIG. 15 is a diagram showing the configuration of the X-ray imaging system 100. As shown in FIG. The X-ray imaging system 100 has an input unit 110 , an X-ray imaging device 120 , an X-ray age estimation device 130 and an output unit 140 . The Roentgen age estimation device 130 is the Roentgen age estimation device 50 described above.

The input unit 110 is a user interface for inputting instructions to the X-ray imaging system 100 and inputting various kinds of information. The input unit 110 is, for example, a button, touch panel, keyboard, barcode reader, or the like. The input unit 110 may receive information from another device connected to the X-ray imaging system 100 via a network, or may read information from a recording medium. Various information input to the input unit 110 includes subject information. Subject information includes, for example, subject identification information, name, date of birth, actual age, and other information.

The X-ray imaging device 120 is a conventional device for X-raying a subject. The X-ray imaging device 120 performs X-ray imaging in accordance with instructions input from the input unit 110 . The X-ray imaging device 120 outputs X-ray image data of the subject obtained by X-ray photography to the X-ray age estimation device 130 . Information on the photographing date and time is added to the X-ray image data.

The Roentgen age estimation device 130 is the Roentgen age estimation device 50 described above. When the input unit 110 also serves as the input unit 51 of the Roentgen age estimation device 50 , the Roentgen age estimation device 130 may have a configuration in which the input unit 51 is removed from the Roentgen age estimation device 50 . If the output unit 140 also serves as the output unit 56 of the Roentgen age estimation device 50 , the Roentgen age estimation device 130 may have a configuration in which the output unit 56 is removed from the Roentgen age estimation device 50 . The X-ray age estimating device 130 receives the X-ray image data output from the X-ray imaging device 120 and the actual age information included in the subject information input by the input unit 110, and performs the X-ray age estimation process shown in FIG. I do. Note that the X-ray age estimation device 130 may calculate the actual age information using the information on the date of birth included in the subject information and the information on the photographing date and time added to the X-ray image data. Roentgen age estimation device 130 outputs the information on Roentgen age, the information on the age difference, and the heat map to output unit 140 in step S225 of FIG.

The output unit 140 outputs various data. The output is, for example, display on a display, printing by a printer, writing on a recording medium, or transmission to another device connected to the X-ray imaging system 100 via a network. The output unit 140 outputs the X-ray image data output from the X-ray imaging device 120, the subject information input from the input unit 110, the X-ray age information, the age difference information and the heat map output from the X-ray age estimation device 130. and output the inspection result data including. For example, every time an X-ray image is captured by the X-ray imaging device 120, the output unit 140 displays all of the test result data, or a part of the test result data including at least X-ray age information or age difference information, on the display. indicate. Furthermore, the output unit 140 writes the inspection result data to a recording medium or transmits it to another device.

In addition, it is not necessary to input the information of the real age to the X-ray age estimation device 130 . In this case, the age difference information is not output from the X-ray age estimation device 130 . The output unit 140 calculates age difference information from the difference between the actual age information indicated by the subject information and the X-ray age information output from the X-ray age estimation device 130 .

FIG. 16 is a diagram showing the configuration of the evaluation system 200. As shown in FIG. The evaluation system 200 has an X-ray age estimation device 210 and an evaluation device 220 . The Roentgen age estimation device 210 is the Roentgen age estimation device 50 of the embodiment described above. When the input unit 230 of the evaluation device 220 (to be described later) also serves as the input unit 51 of the Roentgen age estimation device 50 , the Roentgen age estimation device 210 may have a configuration in which the input unit 51 is removed from the Roentgen age estimation device 50 . When the output unit 250 of the evaluation device 220 (to be described later) also serves as the output unit 56 of the Roentgen age estimation device 50 , the Roentgen age estimation device 210 may have a configuration in which the output unit 56 is removed from the Roentgen age estimation device 50 . The Roentgen age estimating device 210 outputs information on the Roentgen age of the subject obtained by performing the processing of FIG. 6 to the evaluation device 220 . The roentgen age estimation device 210 may output to the evaluation device 220 information on the age difference between the actual age of the subject and the roentgen age instead of or in addition to the roentgen age.

The evaluation device 220 performs a predetermined calculation using the roentgen age of the subject estimated by the roentgen age estimating device 210 and the information on the biological body or lifestyle of the subject or a numerical value representing the information. Evaluate for a disease or condition. The evaluation device 220 may perform evaluation using information on the age difference of the subjects in place of or in addition to the X-ray age. The evaluation device 220 is realized by the computer 80 shown in FIG. 4 as an example. In this case, the storage unit 82 of the computer 80 stores a program for causing the computer 80 to function as the evaluation device 220. FIG. All or part of the functions of the evaluation device 220 may be realized using hardware such as ASIC and PLD.

Here, an example of conventional evaluation that does not use information on radiographic age and age difference will be described. One example is the Suita score, which predicts future onset of angina pectoris, myocardial infarction, etc. (see, for example, http://www.ncvc.go.jp/pr/release/006484.html). The Suita score uses, as risk factors, biological information and lifestyle habits of subjects such as age, sex, current smoking, diabetes, blood pressure, LDL cholesterol, HDL cholesterol, and CKD (chronic kidney disease). A subject is given a score for each type of these risk factors, and based on the total score, the subject's 10-year probability of developing coronary artery disease is obtained. A scoring method is determined according to the type of each risk factor. For example, in the case of age, discrete scores are determined according to which one of the categorized stepwise numerical ranges such as 35 to 44 years old, 45 to 54 years old, and so on. In the case of blood pressure, discrete scores such as optimal blood pressure, normal blood pressure, Stage 1 hypertension, and Stage 2 or higher hypertension are determined according to blood pressure conditions. In addition, in the case of CKD, it is categorized based on the state according to the value of eGFR (estimated glomerular filtration rate), such as Stage 1 or 2 (eGFR ≥ 60), Stage 3 (eGFR 30 ~ < 60), Stage 4 or 5 (eGFR < 30). A discrete score is determined according to which of the categories listed above belongs to. Thus, for some types of risk factors, the score is determined by the category to which the risk factor value or condition belongs. Also, for some types of risk factors, the score differs depending on whether or not the risk factor applies. For example, if the smoker is currently smoking, a predetermined score is given, and if the smoker is not currently smoking, no score is given, in other words, 0 points are given.

In addition, as other examples, CHADS ₂ score, CHA ₂ DS ₂ -VASc score used for risk assessment of cerebral infarction in non-valvular atrial fibrillation (e.g., https://www.jhf.or.jp/pro /hint/c3/hint005.html). The CHADS ₂ score is a risk evaluation value obtained by totaling scores according to the presence or absence of risk factors for heart failure, hypertension, age 75 or older, diabetes, and cerebral infarction/transient ischemic attack. The CHA ₂ DS ₂ -VASc score is a risk assessment value obtained by adding the age of 65 to 74 years old, vascular disease complications, and gender to the risk factors, and summing the scores according to the presence or absence of each risk factor. .

The evaluation value in the above example is calculated by substituting the value of each evaluation parameter into a predetermined calculation formula for adding the evaluation parameter corresponding to each risk factor related to the subject to be evaluated. The value of each evaluation parameter is determined based on the risk factor corresponding to the evaluation parameter, that is, information on the subject's biological body or lifestyle.

The evaluation device 220 of the present embodiment uses an evaluation parameter corresponding to the roentgen age of the subject estimated by the roentgen age estimation device 210 and an evaluation parameter corresponding to information on the biological body or lifestyle of the subject. An evaluation value for evaluating a disease or pathological condition is calculated by calculation. Instead of or in addition to the X-ray age, an evaluation parameter according to the age difference between the X-ray age and the actual age may be used to calculate the evaluation value.

The evaluation device 220 has an input section 230 , an evaluation section 240 and an output section 250 . The input unit 230 inputs data. Input may be made using a user interface such as a keyboard, mouse, or touch panel, may be received from another device connected via a network, or may be read from a recording medium. The input unit 230 inputs information about the living body or lifestyle according to the evaluation parameters of the subject, and further inputs either or both of the information on the X-ray age and the information on the age difference output by the X-ray age estimation device 210 . The input unit 230 outputs the input value to the evaluation unit 240 .

When the information on the age by X-rays is input from the age estimation device 210 for X-rays, the evaluation unit 240 sets the evaluation parameter corresponding to the age by X-rays to a value corresponding to the age by X-rays. Similarly, when the age difference information is input from the X-ray age estimation device 210, the evaluation unit 240 sets the evaluation parameter corresponding to the age difference to a value corresponding to the age difference. Note that the evaluation unit 240 may calculate the age difference from the difference between the X-ray age input from the X-ray age estimation device 210 and the actual age input from the input unit 230 . Thus, the age difference is also information obtained by performing a predetermined calculation on the X-ray age. Furthermore, the evaluation unit 240 sets values corresponding to the information to each evaluation parameter corresponding to each information about the biological body or lifestyle of the subject. Examples of biological information include information such as age, sex, test results of various clinical tests, conditions classified based on test results of various clinical tests, various past diseases, and various current diseases. Not limited. In addition, examples of information related to life include, but are not limited to, information on smoking, sleeping hours, meals, and the like.

The value set in the evaluation parameter is, for example, the value indicated by the information corresponding to the evaluation parameter, the value indicated by the information corresponding to the evaluation parameter, the value indicated by the information corresponding to the evaluation parameter, or the category to which the state belongs. a value obtained by performing a predetermined operation on a value indicated by information corresponding to the evaluation parameter; and a value corresponding to whether or not the information corresponds to the evaluation parameter. The evaluation unit 240 may use the value of the X-ray age as the evaluation parameter value as it is, and the range in which the X-ray age is categorized (for example, under a1, a1 to a2, a2 to a3, etc.) It may be an evaluation parameter value according to whether it belongs to. Further, for example, the evaluation unit 240 may use the value of the age difference as the evaluation parameter value as it is, and the range in which the age difference is categorized (for example, age b or older, age b to -b, age -b or younger, etc.) It may be an evaluation parameter value according to which one of them it belongs to.

The evaluation unit 240 calculates the evaluation value by substituting the value of each evaluation parameter into the calculation formula for calculating the evaluation value representing the evaluation of the evaluation target with a quantitative value using each evaluation parameter. An example of the calculation formula is addition of evaluation parameters, but is not limited to this, and any calculation formula is used. When the calculation formula is a weighted addition of evaluation parameters, weights for other evaluation parameters may be determined according to the value of the evaluation parameter corresponding to the X-ray age or age difference. An evaluation target is information about a disease or pathological condition, such as severity of the disease or pathological condition, prognosis prediction, or the like. In addition, the evaluation unit 240 stores in advance the correspondence between the evaluation value and the evaluation (for example, low/medium/high risk, x1%/x2%/, etc.), and obtains the evaluation corresponding to the calculated evaluation value. may The output unit 250 outputs one or both of the evaluation value calculated by the evaluation unit 240 and the evaluation corresponding to the evaluation value to the output unit 250 as evaluation results.

Artificial intelligence may be used to calculate the evaluation value using each evaluation parameter. In this case, the evaluation value calculation formula corresponds to, for example, DNN. When using the DNN, the input of the DNN is the value of each evaluation parameter, and the output is the evaluation value. For the values of the evaluation parameters corresponding to the X-ray age and age difference, the X-ray age and age difference values may be used as they are, or discrete values corresponding to the numerical ranges to which the X-ray age and age difference values belong may be used. Alternatively, continuous values normalized to X-ray age or age differences may be used.

Note that the types of evaluation parameters used to calculate the evaluation value and the calculation (calculation formula) for calculating the evaluation value are statistically determined by learning using learning data according to the evaluation target.

The output unit 250 outputs the evaluation result received from the evaluation unit 240 . The output includes, for example, one or more of display on a display, printing by a printer, writing on a recording medium, and transmission to another device connected to the evaluation device 220 via a network.

According to the embodiments described above, artificial intelligence can be used to estimate the age of a subject from chest X-ray image data. The new concepts of X-ray age and chest X-ray age are intuitively easy for patients to understand and may be widely used. In addition, the age difference between the chest X-ray age and the actual age can support the determination of whether or not there is an abnormal finding. Therefore, it is expected that abnormal findings will be overlooked less in general medical examinations and medical treatments. In addition, since it is possible to highlight and indicate the location where anomalies are detected, it becomes possible to discuss in more depth what the cause is even for the same outcome. In addition, the age difference between the chest X-ray age and the chronological age reflects the severity of the disease. This age difference stratifies the prognosis of heart failure patients after chest X-ray image data is taken. Therefore, it is expected that it will be useful in determining the treatment policy for diseased patients, and there is a possibility that the X-ray age and the chest X-ray age will be widely used by medical personnel. In addition, chest X-ray images have been difficult to utilize in digital medical care because of variations in findings depending on the reader. In this embodiment, the chest X-ray findings can be converted into a unique quantitative value of chest X-ray age, so it is considered possible to contribute to the promotion of digital medicine.

Furthermore, when X-rays are taken of a subject in health checkups and medical treatments, it is possible to provide information on X-ray ages and age differences. Therefore, clinically useful information can be rapidly provided. Furthermore, in the evaluation of the severity of diseases and pathological conditions and in prognostic prediction research, it is possible to incorporate information on X-ray age and age difference as one of the evaluation parameters. Therefore, Roentgen age can be used as a new research tool.

In the above-described embodiment, the case of using X-ray image data of the chest as the data of the image obtained by transmissive imaging of the inside of the human body has been described as an example. An image captured by another method, or an image of another part such as the brain may be used.

According to the above-described embodiment, the Roentgen age estimation learning device comprises a learning unit. The learning unit is a model that shows the correspondence between the feature amount obtained from the X-ray image data and the subject's age, using the X-ray image data of the subject and the age of the subject when the X-ray image was taken. Train the Roentgen age estimation model.

The X-ray age estimating device includes an estimating unit. The estimation unit inputs the feature amount obtained from the data of the X-ray image of the target subject for age estimation, which is the X-ray image of the subject whose age is to be estimated, to the X-ray age estimation model learned by the X-ray age estimation learning device. Get the age estimation result. The X-ray age estimation device may further include a contribution information acquisition unit. The contribution information acquisition unit acquires contribution information, which is information relating to a portion of the estimation target X-ray image from which a feature amount that satisfies a predetermined condition for a degree of contribution to the age estimation of the age estimation target subject is obtained. The Roentgen age estimation device may further include an output unit. The output unit outputs one or more of information on the age of the result of estimation by the estimation unit, information on the difference between the actual age of the subject whose age is to be estimated and the age of the result of estimation by the estimation unit, and contribution information.

For example, the X-ray image is an X-ray image of the chest of the subject, and the estimation target X-ray image is an X-ray image of the chest of the subject whose age is to be estimated.

Also, the X-ray imaging system may be equipped with an X-ray age estimator. An X-ray apparatus included in an X-ray imaging system inputs X-ray image data obtained by X-ray imaging of a subject to an X-ray age estimation apparatus as estimation target X-ray image data.

The evaluation device also includes an evaluation unit. The evaluation unit determines whether the subject's disease or Evaluate pathology.

The image age estimation learning device includes a learning unit. The image age estimation learning device corresponds to, for example, the X-ray age estimation learning device 10 of the embodiment. The learning unit obtains from the transmission image data using the data of the transmission image, which is an image obtained by transmission photography of the human body of the subject, and the age information of the subject when the transmission image was captured. It learns an age estimation model that shows the correspondence between the features obtained and the age of the subject. The age estimation model corresponds to, for example, the Roentgen age estimation model of the embodiment. Transmitted images include, for example, images obtained by X-ray imaging and images obtained by MRI. A feature amount is, for example, a pixel value.

The image age estimating device includes an estimating unit. The image age estimation device corresponds to, for example, the X-ray age estimation device 50 of the embodiment. The estimating unit inputs, to the age estimation model learned by the image age estimation learning device, the feature amount obtained from the data of the estimation target image, which is the transmission image of the subject for age estimation, and calculates the age of the subject for age estimation. We obtain the estimation result of

Although the embodiments of the present invention have been described above with reference to the drawings, it is clear that the above embodiments are merely examples of the present invention and that the present invention is not limited to the above embodiments. be. Therefore, additions, omissions, substitutions, and other modifications of components may be made without departing from the technical idea and scope of the present invention.

Reference Signs List 1... Roentgen age estimation system, 10... Roentgen age estimation learning device, 11... Model storage unit, 12... Data storage unit, 13... Conversion unit, 14... Learning data generation unit, 15... Learning unit, 16... Estimation unit, 17 ... update unit 18 ... selection unit 19 ... output unit 50 ... X-ray age estimation device 51 ... input unit 52 ... model storage unit 53 ... conversion unit 54 ... estimation unit 55 ... contribution information acquisition unit 56 ... output section 80 ... computer 81 ... processor 82 ... storage section 83 ... user interface 84 ... communication interface 85 ... bus 100 ... X-ray imaging system 110 ... input section 120 ... X-ray imaging apparatus 130 ... roentgen age estimation device 140 output unit 200 evaluation system 210 roentgen age estimation device 220 evaluation device 230 input unit 240 evaluation unit 250 output unit

Claims (14)

X-ray image data of a subject and information on the age of the subject when the X-ray image was taken, X-ray age indicating the correspondence between the feature value obtained from the X-ray image data and the age of the subject a learning unit that learns the estimation model;
Roentgen age estimation learning device. The X-ray image is a chest X-ray image,
The X-ray age estimation learning device according to claim 1. An X-ray age estimation model trained using X-ray image data of a subject and information about the age of the subject when the X-ray image was taken, wherein the feature amount obtained from the X-ray image data and the A feature amount obtained from the data of an estimated target X-ray image, which is an X-ray image of a subject whose age is to be estimated, is input into the X-ray age estimation model that indicates the correspondence with the subject's age, and the age of the subject whose age is to be estimated is calculated. an estimation unit that obtains an estimation result;
Roentgen age estimator. A contribution information acquiring unit that acquires contribution information that is information about a portion of the X-ray image of the estimation target from which the feature quantity satisfying a predetermined condition for the degree of contribution to age estimation of the subject for age estimation satisfies a predetermined condition;
4. The radiographic age estimation device of claim 3, further comprising: Output for outputting one or more of information on the age of the estimation result by the estimation unit, information on the difference between the actual age of the subject whose age is to be estimated and the age of the estimation result by the estimation unit, and the contribution information further comprising a part,
The X-ray age estimating device according to claim 4. The X-ray image and the estimated target X-ray image are chest X-ray images,
The Roentgen age estimating device according to any one of claims 3 to 5. obtained from the transmission image data by using the data of the transmission image, which is an image obtained by transmission photography of the human body of the subject, and the age information of the subject when the transmission image was captured a learning unit that learns an age estimation model that indicates the correspondence between the feature amount and the subject's age;
An image age estimation learning device comprising: An age estimation model learned using transmission image data obtained by transmission imaging of a subject's human body and age information of the subject when the transmission imaging image was captured, The feature amount obtained from the data of the estimation target image, which is the transmission image of the subject whose age is to be estimated, is added to the age estimation model showing the correspondence between the feature amount obtained from the data of the transmission image and the age of the subject. and an estimation unit that obtains a result of estimating the age of the subject whose age is to be estimated,
An image age estimation device comprising: The X-ray age estimation device according to any one of claims 3 to 6;
an X-ray imaging device that performs X-ray photography of a subject and inputs data of the X-ray image of the subject obtained by the X-ray photography into the X-ray age estimation device as data of the X-ray image to be estimated;
X-ray system with. Predetermined calculation using the estimated age of the subject by the X-ray age estimation apparatus according to any one of claims 3 to 6 and information on the biological body or lifestyle of the subject or a numerical value representing the information An evaluation unit that evaluates the subject's disease or condition based on the results of the
An evaluation device comprising: X-ray image data of a subject and information on the age of the subject when the X-ray image was taken, X-ray age indicating the correspondence between the feature amount obtained from the X-ray image data and the age of the subject a learning step for learning the estimation model;
Roentgen age estimation learning method comprising: An X-ray age estimation model trained using X-ray image data of a subject and information about the age of the subject when the X-ray image was taken, wherein the feature amount obtained from the X-ray image data and the A feature amount obtained from the data of an estimated target X-ray image, which is an X-ray image of a subject whose age is to be estimated, is input into the X-ray age estimation model that indicates the correspondence with the subject's age, and the age of the subject whose age is to be estimated is calculated. an estimation step for obtaining an estimation result;
Roentgen age estimation method comprising: the computer,
A program for operating as the X-ray age estimation learning device according to claim 1 or 2. the computer,
A program for operating as the X-ray age estimating device according to any one of claims 3 to 6.

Images