Lung volume testing method and device

Abstract

To provide a lung volume diagnosis method and apparatus therefor.SOLUTION: The present invention relates to a lung volume diagnosis method and apparatus therefor. The lung volume diagnosis apparatus inputs a patient's two-dimensional medical treatment video and clinical information to an artificial intelligence model, grasps the patient's lung volume. The artificial intelligence model includes: a first nerve network for generating encoded video associated with two-dimensional medical treatment video; and a second nerve network including lung volume from data in which an output value of the first nerve network and the clinical information are coupled. The present invention is a technology developed through Seoul National University Hospital's "Severely Ill Patient Big Data Construction and AI Base CDSS Development" project (Theme Number: HI21C1074050021, Theme Specific Number: HI21C1074).SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a method and an apparatus for examining lung volume, and more particularly to a method and an apparatus for examining lung volume from medical images.

[0002] This technology was developed through the "Building big data for critically ill patients and developing AI-based CDSS" project run by Seoul National University Hospital (project number: HI21C1074050021, project specific number: HI21C1074). [Background technology]

[0003] Conventional lung volume is measured using a spirometer based on the amount of air a patient inhales and exhales. However, if the lungs become stiff due to disorders such as pulmonary fibrosis, the lungs lose elasticity, making it difficult to exhale and trapping more air in the lungs, which reduces the accuracy of lung volume measurements using conventional spirometers.

[0004] Lung volume can be calculated by isolating the lung area from a computed tomography (CT) image of the patient's chest. However, to calculate lung volume, a CT scan must be performed first, which is costly and time-consuming, and a complex process is required to separate and extract the lung area from the 3D medical image. Summary of the Invention [Problem to be solved by the invention]

[0005] The technical problem to be solved by the present invention is to provide a method and apparatus for easily determining a patient's lung volume from a two-dimensional medical image such as an X-ray image. [Means for solving the problem]

[0006] To achieve the above-mentioned technical objectives, an example of a lung volume testing method according to an embodiment of the present invention includes a step of inputting a two-dimensional medical image of a patient, a step of inputting clinical information of the patient, and a step of inputting the two-dimensional medical image and the clinical information into an artificial intelligence model to determine the lung volume of the patient.

[0007] In order to achieve the above-mentioned technical objectives, an example of a lung volume testing device according to an embodiment of the present invention includes an image input unit to which a two-dimensional medical image of a patient is input, a clinical information input unit to which clinical information of the patient is input, and a volume grasping unit that grasps the lung volume of the patient via an artificial intelligence model that inputs the two-dimensional medical image and the clinical information and outputs the lung volume. Effect of the Invention

[0008] According to an embodiment of the present invention, lung volume can be obtained from a two-dimensional medical image such as an X-ray image, and the accuracy of lung volume measurement can be improved by using clinical information together with the two-dimensional medical image. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 illustrates an example of a lung volume testing device according to an embodiment of the present invention. [Diagram 2] 1 is a flow chart illustrating an example of a lung volume testing method according to an embodiment of the present invention. [Diagram 3] FIG. 2 illustrates the structure of an example of an artificial intelligence model according to an embodiment of the present invention. [Figure 4] FIG. 2 is a diagram illustrating an example of a region division model according to an embodiment of the present invention. [Diagram 5] FIG. 2 is a diagram illustrating an example of a region division model according to an embodiment of the present invention. [Figure 6] 1 is a diagram illustrating the structure of various embodiments of an artificial intelligence model according to an embodiment of the present invention. [Figure 7] 1 is a diagram illustrating the structure of various embodiments of an artificial intelligence model according to an embodiment of the present invention. [Figure 8]1 is a diagram illustrating the structure of various embodiments of an artificial intelligence model according to an embodiment of the present invention. [Figure 9] FIG. 2 is a diagram illustrating an example of learning data of an artificial intelligence model according to an embodiment of the present invention. [Figure 10] 1 is a diagram illustrating a configuration of an example of a lung volume testing device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a lung volume testing method and device thereof according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

[0011] FIG. 1 is a diagram illustrating an example of a lung volume testing apparatus according to an embodiment of the present invention.

[0012] 1, a lung volume testing apparatus 100 receives a two-dimensional medical image 110 and clinical information 120, and then determines and outputs a lung volume 130. In one embodiment, the lung volume testing apparatus 100 may receive the two-dimensional medical image 110 and the clinical information 120 from a picture archiving and communication system (PACS) and an electronic medical record (EMR) system.

[0013] The two-dimensional medical image 110 may be a chest X-ray image. Alternatively, the two-dimensional medical image 110 may be a segmented image obtained by segmenting only the lung region from the chest X-ray image. An example of a method for generating a segmented image of the lung region from an X-ray image is illustrated in FIG. 4 and FIG. 5.

[0014] Since the lung volume 130 has a significant correlation with the age and / or sex of the patient, the clinical information 120 may also include the age and / or sex. In another embodiment, the clinical information 120 may further include quantitative data such as the area of ​​the lung region in addition to the age and sex. In addition, various information having a correlation with the lung volume may be included in the clinical information 120, but is not limited to this embodiment. However, in the following, the case where the clinical information 120 includes the age and / or sex will be mainly described as an example.

[0015] FIG. 2 is a flow chart illustrating an example of a lung volume testing method according to an embodiment of the present invention.

[0016] 2, a lung volume testing device 100 (hereinafter referred to as the "device") receives a two-dimensional medical image and clinical information (S200). The two-dimensional medical image may be a chest X-ray image or a segmented image obtained by segmenting a lung region from the chest X-ray image. The clinical information may include age, sex, etc.

[0017] The device 100 inputs the two-dimensional medical image and the clinical information into an artificial intelligence model to determine the lung volume (S210). For example, the artificial intelligence model may be implemented as an artificial neural network that predicts and outputs the lung volume when the two-dimensional medical image and the clinical information are input. The two-dimensional medical image and the clinical information are different types of data. This embodiment provides an artificial intelligence model including a first neural network and a second neural network to accurately predict the lung volume based on two different types of data. An example of the structure of an artificial intelligence model according to an embodiment of the present invention is illustrated in FIG. 3.

[0018] FIG. 3 is a diagram illustrating an example structure of an artificial intelligence model according to an embodiment of the present invention.

[0019] Referring to FIG. 3, the artificial intelligence model 300 includes a first neural network 310 and a second neural network 320. The first neural network 310 and the second neural network 320 may be the same type of artificial neural network or different types of artificial neural networks. For example, the first neural network 310 is a U-Net and the second neural network 320 is a CNN (convolutional neural network). In addition, the first neural network 310 and the second neural network 320 may be embodied by various types of conventional artificial neural networks, but are not limited to a specific type. However, for convenience of explanation, the following description will be mainly given of the case where the first neural network 310 is a U-Net.

[0020] The first neural network 310 receives the two-dimensional medical image 110. The first neural network 310 may directly use various conventional artificial neural networks such as U-Net, or may be modified based on the same. In one embodiment, the first neural network 310 is a model that outputs an encoded image (or a feature map) by encoding (or downsampling) the two-dimensional medical image 110 (see FIG. 6). In another embodiment, the first neural network 310 is a model that, when a two-dimensional medical image is input, performs an encoding process (i.e., downsampling) and a decoding process (i.e., upsampling) and outputs a restored image (see FIGS. 7 and 8).

[0021] The second neural network 320 is a model that outputs lung volume when data obtained by concatenating the output value of the first neural network 310 with clinical information 120 is input. The output value of the first neural network 310 is also feature information output from a specific layer among a plurality of layers that constitute the first neural network 310. Examples of various connection structures between the first neural network 310 and the second neural network 320 are shown in Figures 6 to 8.

[0022] 4 and 5 are diagrams illustrating an example of a region division model according to an embodiment of the present invention.

[0023] 4, the region segmentation model 400 is an artificial intelligence model that, when an X-ray image 410 is input, generates a segmented image 420 of a lung region from the X-ray image 410. An example of a segmented image 510 obtained by segmenting a lung region from a chest X-ray image 500 is shown in FIG. 5. In the X-ray image 500, various tissues (lungs, bones, muscles, organs, etc.) are displayed superimposed on one plane, but in the segmented image 510, only lung tissue is displayed. When the segmented image 510 of the lung region is used in a two-dimensional medical image, the accuracy of lung volume measurement can be improved.

[0024] The region segmentation model 400 may also be realized by various artificial neural networks such as CNN. The region segmentation model 400 may be generated by training it using a supervised learning method. For example, the region segmentation model 400 may be trained using learning data including an X-ray image 410 and a two-dimensional image (answer paper) of the lung region as a data set. That is, the region segmentation model 400 may be trained so that when the X-ray image 410 of the learning data is input, the region segmentation model 400 outputs a segmented image 420, and the difference between the segmented image 420 and the two-dimensional image (answer paper) of the lung region of the learning data is reduced. The method of training the region segmentation model 400 using the learning data and a supervised learning method is a well-known configuration, and therefore further description thereof will be omitted.

[0025] In one embodiment, the learning data can be generated based on a CT (computed tomography) image. For example, a three-dimensional lung region is divided into a chest CT image using various conventional division algorithms. Then, the two-dimensional image generated by projecting the chest CT image onto a two-dimensional plane can be used as an X-ray image of the learning data, and the two-dimensional lung image generated by projecting the three-dimensional lung region onto the two-dimensional plane can be used as a division image. In addition, the learning data for the region division model 400 can be generated by various methods, but is not limited to this embodiment.

[0026] The device 100 can obtain quantitative data 430, such as the area of ​​the lung region, from the segmented image 420 obtained via the region segmentation model 400. The device 100 can use the area of ​​the lung region, etc., as clinical information.

[0027] FIG. 6 is a diagram illustrating a structure of a first embodiment of an artificial intelligence model according to an embodiment of the present invention.

[0028] Referring to FIG. 6, the artificial intelligence model includes a first neural network 600 that performs an encoding process (eg, downsampling) and a second neural network 610 that predicts lung volume.

[0029] The first neural network 600 outputs an encoded image obtained by encoding a 2D medical image when a 2D medical image is input. The first neural network 600 may be configured by omitting the decoder portion from various types of conventional artificial neural networks including an encoder and a decoder. For example, the first neural network 600 may be configured by a layer (i.e., an encoder) that performs an encoding process in a U-Net.

[0030] The second neural network 610 outputs lung volume when the output value (i.e., the coded image) of the first neural network 600 and clinical information are input. The second neural network 610 can be realized by various types of artificial neural networks such as CNN.

[0031] The device 100 can train an entire artificial intelligence model in which the first neural network 600 and the second neural network 610 are connected. For example, the artificial intelligence model can be trained and generated through a regression estimation method. In another embodiment, the device 100 can train the first neural network 600 and the second neural network 610 separately, or train only the second neural network 620. A method of training an artificial intelligence model using learning data will be further described with reference to FIG. 9.

[0032] FIG. 7 is a diagram illustrating the structure of a second embodiment of an artificial intelligence model according to an embodiment of the present invention.

[0033] Referring to FIG. 7, the artificial intelligence model includes a first neural network 700 that performs encoding and decoding processes, and a second neural network 710 that predicts lung volume.

[0034] The first neural network 700 includes an encoder that encodes the two-dimensional medical image when the two-dimensional medical image is input, and a decoder that decodes the encoded data to generate a reconstructed image. For example, the first neural network 700 may be embodied by a U-Net. The first neural network 700 may be trained to reduce a difference between the two-dimensional medical image, which is input data, and the reconstructed image, which is output data. For example, the first neural network 700 may be trained by a self-supervised learning method.

[0035] The second neural network 710 outputs lung volume when data combining feature information (e.g., encoded image) and clinical information output from the encoder of the first neural network 700 is input. The second neural network 710 may be implemented by the same artificial neural network as the second neural network 610 described in FIG. 6 or may be implemented by a different artificial neural network.

[0036] The first neural network 700 and the second neural network 710 of this embodiment are also generated through separate learning processes. The learning methods of the first neural network 700 and the second neural network 710 will be further described in FIG.

[0037] FIG. 8 is a diagram illustrating the structure of a third embodiment of an artificial intelligence model according to an embodiment of the present invention.

[0038] Referring to FIG. 8, the artificial intelligence model includes a first neural network 800 that performs encoding and decoding processes, and a second neural network 810 that predicts lung volume.

[0039] The first neural network 800 includes an encoder for encoding the two-dimensional medical image when the two-dimensional medical image is input, and a decoder for decoding the encoded data and restoring the image. For example, the first neural network 800 may be implemented by a U-Net.

[0040] The second neural network 810 is located at the output end of the first neural network 800. When the second neural network 810 receives data obtained by combining the restored image output from the first neural network 800 with clinical information, the second neural network 810 outputs lung volume.

[0041] The device 100 can train the entire artificial intelligence model in which the first neural network 800 and the second neural network 810 are connected. In another embodiment, the device 100 can train the first neural network 800 and the second neural network 810 separately, or train only the second neural network. The method of training the artificial intelligence model using the learning data will be further described with reference to FIG. 9.

[0042] FIG. 9 is a diagram illustrating an example of learning data of an artificial intelligence model according to an embodiment of the present invention.

[0043] Referring to FIG. 9, the training data 900 of the artificial intelligence model includes a data set of a two-dimensional medical image 910, clinical information 920, and an actual lung volume 930. The two-dimensional medical image 910 may be a chest X-ray image or the segmented image described in FIG. 4 and FIG. 5. The clinical information 920 may include the age and sex of the patient. In another embodiment, the clinical information 920 may further include the area of ​​the lung region obtained from the segmented image of FIG. 4. The actual lung volume 930 may be obtained through a three-dimensional segmentation model 940 that segments a three-dimensional lung region from a CT image 950. The three-dimensional segmentation model 940 may be any of various conventional models implemented by the artificial intelligence model. In addition, the actual lung volume of the training data may be obtained through various conventional methods.

[0044] The apparatus 100 can use the training data 900 to train the various artificial intelligence models of FIGS.

[0045] First, the learning method of the artificial intelligence model in FIG. 6 will be described.

[0046] The device 100 inputs a two-dimensional medical image 910 of the training data 900 to the first neural network 600. The device 100 inputs data obtained by combining the coded image output from the first neural network 600 and clinical information 920 of the training data 900 to the second neural network 610 to predict lung volume. Hereinafter, the lung volume predicted and output by the second neural network 610 is referred to as a "predicted lung volume." The device 100 trains both the first neural network 600 and the second neural network 610 based on a loss function indicating the difference between the predicted lung volume and an actual lung volume 930 of the training data 900. The loss function is a loss function related to the entire first neural network 600 and the second neural network 610. The device 100 performs a training process to adjust parameter values ​​of the first neural network 600 and the second neural network 610 so that the value of the loss function decreases.

[0047] Next, a learning method for the artificial intelligence model in FIG. 7 will be described.

[0048] The device 100 inputs the two-dimensional medical image 910 of the training data 900 to the first neural network 700. The first neural network 700 outputs a reconstructed image through encoding and decoding processes. The first neural network 700 performs a first training process to adjust parameter values ​​of the first neural network 700 so that a value of a loss function indicating a difference between the two-dimensional medical image 910 of the training data 900 and the reconstructed image is reduced. That is, the first neural network 700 is generated by self-supervised learning.

[0049] When the first training process is completed, the device 100 inputs the two-dimensional medical image 910 of the learning data 900 to the first neural network 700 that has completed training. The device 100 inputs data combining feature information of the encoder of the first neural network 700 (i.e., the encoded image) and clinical information 920 of the learning data 900 to the second neural network 710 to obtain the predicted lung volume. The device 100 performs a second training process to adjust the parameter values ​​of the second neural network 710 so that the value of a loss function indicating the difference between the predicted lung volume and the actual lung volume 930 of the learning data 900 decreases. Since the first neural network 700 has completed learning in advance, the learning state of the first neural network 700 is fixed. That is, in the second training process, the second neural network 710 excluding the first neural network 700 is trained.

[0050] Next, a learning method for the artificial intelligence model in FIG. 8 will be described.

[0051] The device 100 inputs the two-dimensional medical image 910 of the training data 900 to the first neural network 800 to obtain a reconstructed image. The device 100 inputs data combining the reconstructed image and clinical information output from the first neural network 800 to the second neural network 810 to obtain a predicted lung volume. The device 100 performs a training process to adjust parameter values ​​of the first neural network 800 and the second neural network 810 so that a value of a loss function indicating a difference between the predicted lung volume and an actual lung volume 930 of the training data 900 decreases. In the artificial intelligence model of FIG. 7, the first neural network 700 and the second neural network 710 are trained separately, whereas in the artificial intelligence model of FIG. 8, the first neural network 800 and the second neural network 810 are trained together at one time.

[0052] FIG. 10 is a diagram illustrating an example of a lung volume testing device according to an embodiment of the present invention.

[0053] 10, the lung volume testing apparatus 100 includes a region segmentation model 1000, an image input unit 1010, a clinical information input unit 1020, a volume assessment unit 1030, an artificial intelligence model 1040, and a learning unit 1050. In one embodiment, the region segmentation model 1000 may be omitted. In another embodiment, if the artificial intelligence model 1040 has completed learning in advance, the learning unit 1050 may be omitted. In yet another embodiment, the lung volume testing apparatus 100 may be embodied by a computing device including a memory, a processor, and an input / output device. In this case, each component may be embodied by software and executed by the processor after being loaded into the memory.

[0054] The image input unit 1010 receives a two-dimensional medical image of a patient. In one embodiment, the image input unit 1010 may receive a chest X-ray image as the two-dimensional medical image. In another embodiment, the image input unit 1010 may receive a segmented image formed by a lung region segmented from an X-ray image via the region segmentation model 1000 as the two-dimensional medical image.

[0055] The clinical information input unit 1020 receives clinical information of the patient. In one embodiment, the clinical information input unit 1020 may receive the age and sex of the patient as clinical information. In another embodiment, the clinical information input unit 1020 may receive quantitative data such as the area of ​​the lung region grasped via the region segmentation model 1000, and the age and sex of the patient as clinical information.

[0056] The volume determination unit 1030 receives two-dimensional medical images and clinical information as input, and determines the lung volume of the patient via an artificial intelligence model 1040 that outputs the lung volume.

[0057] The first embodiment of the artificial intelligence model 1040 also includes a first neural network for encoding an image and a second neural network for predicting lung volume. In this case, the volume grasping unit 1040 generates an encoded image related to a two-dimensional medical image through the first neural network, and inputs data combining the encoded image and clinical information to the second neural network to grasp the lung volume. The first embodiment of the artificial intelligence model 1040 is illustrated in FIG. 6.

[0058] The second embodiment of the artificial intelligence model 1040 includes a first neural network that compresses and restores an image through an encoder and a decoder, and a second neural network that predicts a lung volume. In this case, the volume grasping unit 1040 can grasp the lung volume by inputting data that combines encoder feature information and clinical information related to the two-dimensional medical image to the second neural network. The second embodiment of the artificial intelligence model 1040 is illustrated in FIG. 7.

[0059] The third embodiment of the artificial intelligence model 1040 includes a first neural network that compresses and restores an image through an encoder and a decoder, and a second neural network that predicts a lung volume. In this case, the volume determining unit 1040 obtains a restored image related to a two-dimensional medical image through the first neural network, and inputs data combining the restored image and clinical information to the second neural network to determine the lung volume. The third embodiment of the artificial intelligence model 1040 is illustrated in FIG. 8.

[0060] The learning unit 1050 uses the learning data to train the artificial intelligence model 1040 including the first neural network and the second neural network in a supervised learning method. In the case of the first artificial intelligence model of Fig. 6, the learning unit 1050 inputs a two-dimensional medical image of the learning data to the first neural network that performs image coding, inputs data obtained by combining the output value of the first neural network with clinical information of the learning data to the second neural network to predict lung volume, and trains both the first neural network and the second neural network so that the difference between the predicted lung volume and the actual lung volume of the learning data is reduced.

[0061] In the case of the second artificial intelligence model of Fig. 7, the learning unit 1050 also includes a first learning unit and a second learning unit. The first learning unit can input the two-dimensional medical image of the learning data to the first neural network performing image encoding and image decoding, and train the first neural network by a self-guided learning method. The second learning unit can grasp feature information related to the two-dimensional medical image of the learning data through the first neural network that has completed learning, input data combining the feature information and clinical information of the learning data to the second neural network to predict lung volume, and train the second neural network so that the difference between the predicted lung volume and the actual lung volume of the learning data is minimized.

[0062] In the case of the third artificial intelligence model of Figure 8, the learning unit 1050 inputs a two-dimensional medical image of the learning data to a first neural network that performs image encoding and image decoding, generates a reconstructed image, and inputs data combining the reconstructed image and clinical information of the learning data to a second neural network to predict lung volume, and can train both the first and second neural networks so that the difference between the predicted lung volume and the actual lung volume of the learning data is reduced.

[0063] The present invention may also be embodied as computer readable program code on a computer readable recording medium. The computer readable recording medium includes all types of recording devices in which data readable by a computer system is stored. Examples of the computer readable recording medium include a read only memory (ROM), a random access memory (RAM), a compact disc read only memory (CD-ROM), a magnetic tape, a floppy disk, and an optical data storage device. The computer readable recording medium may also be distributed among computer systems coupled to a network, so that the computer readable code may be stored and executed in a distributed fashion.

[0064] The present invention has been described above with reference to its preferred embodiment. Those skilled in the art will understand that the present invention can be embodied in modified forms without departing from the essential characteristics of the present invention. Therefore, the disclosed embodiments should be considered from an illustrative rather than a restrictive perspective. The scope of the present invention is defined by the claims, not the above description, and all differences within the scope of the equivalents should be interpreted as being included in the present invention. [Explanation of symbols]

[0065] 100 Lung volume test device 110,910 2D medical images 120,920 Clinical information 130 lung volume 300,1040 Artificial Intelligence Models 310 First Neural Network 320 Second Neural Network 400,1000 domain division model 410 X-ray images 420 split video' 430 Quantitative data such as area 900 training data 930 Actual Lung Volume 940 3D division model 950 CT image 1010 Video input section 1020 Clinical information input unit 1030 Volume grasping part 1050 Learning Department

Claims

1. receiving a two-dimensional medical image of a patient; receiving clinical information of the patient; inputting the two-dimensional medical image and the clinical information into an artificial intelligence model to determine the lung volume of the patient; The step of determining the lung volume includes: generating an encoded image relating to the two-dimensional medical image through a first neural network; and inputting the combined data of the coded image and the clinical information into a second neural network to predict lung volume.

2. 2. The method of claim 1, wherein the two-dimensional medical image is a chest X-ray image.

3. The method of claim 1 , wherein the clinical information includes the age and / or sex of the patient.

4. 2. The method of claim 1, wherein the two-dimensional medical image is a segmented image obtained by segmenting a lung region from an X-ray image.

5. 5. The method of claim 4, further comprising: generating a segmented image of the lung region from the two-dimensional medical image by using a region segmentation model trained using learning data including an X-ray image and a segmented image of the lung region.

6. A step of inputting a two-dimensional medical image of a patient; receiving clinical information of the patient; inputting the two-dimensional medical image and clinical information into an artificial intelligence model to estimate the lung volume of the patient; the two-dimensional medical image is a segmented image obtained by segmenting a lung region from an X-ray image; The method for examining lung volume, wherein the clinical information includes an area of ​​the lung region determined based on the segmented images.

7. 2. The method of claim 1, further comprising the step of training the first neural network and the second neural network by a teaching-learning method using learning data including two-dimensional medical images, clinical information, and actual lung volumes.

8. A step of inputting a two-dimensional medical image of a patient; receiving clinical information of the patient; inputting the two-dimensional medical image and clinical information into an artificial intelligence model to estimate the lung volume of the patient; The step of determining the lung volume includes: inputting the 2D medical image into a first neural network including an encoder and a decoder; and inputting data obtained by combining the feature information output from the encoder and the clinical information into a second neural network to predict lung volume.

9. training the artificial intelligence model including the first neural network and the second neural network; The step of training the first neural network comprises: inputting a two-dimensional medical image into the first neural network; training the first neural network so that a difference between a restored image output by the first neural network and the two-dimensional medical image is reduced; The step of training the second neural network comprises: inputting a 2D medical image into the first neural network that has been trained; The lung volume testing method of claim 8, further comprising the step of inputting data combining feature information and clinical information output from the encoder of the first neural network into the second neural network, and training the second neural network so as to reduce the difference between the obtained predicted lung volume and the actual lung volume.

10. A step of receiving a two-dimensional medical image of a patient; receiving clinical information of the patient; inputting the two-dimensional medical image and clinical information into an artificial intelligence model to estimate the lung volume of the patient; The step of determining the lung volume includes: inputting the 2D medical image into a first neural network including an encoder and a decoder to generate a reconstructed image; and inputting the combined data of the reconstructed image and the clinical information into a second neural network to predict lung volume.

11. The lung volume testing method of claim 10, further comprising the step of training the first neural network and the second neural network by a teaching-learning method using learning data including two-dimensional medical images, clinical information, and actual lung volume.

12. An image input unit to which a two-dimensional medical image of a patient is input; a clinical information input unit to which clinical information of the patient is input; and a volume estimation unit for estimating the lung volume of the patient by inputting the two-dimensional medical image and clinical information into an artificial intelligence model, The artificial intelligence model is a first neural network for encoding an image; a second neural network for predicting lung volume; The lung volume examination device is characterized in that the volume grasping unit generates an encoded image related to the two-dimensional medical image via the first neural network, and inputs data combining the encoded image and the clinical information to the second neural network to grasp the lung volume.

13. An image input unit to which a two-dimensional medical image of a patient is input; a clinical information input unit to which clinical information of the patient is input; and a volume estimation unit for estimating the lung volume of the patient by inputting the two-dimensional medical image and clinical information into an artificial intelligence model, The artificial intelligence model is a first neural network for encoding and decoding video via an encoder and a decoder; a second neural network for predicting lung volume; The lung volume examination device is characterized in that the volume grasping unit grasps the lung volume by inputting data combining feature information related to the two-dimensional medical image grasped via the encoder and the clinical information into the second neural network.

14. An image input unit to which a two-dimensional medical image of a patient is input; a clinical information input unit to which clinical information of the patient is input; and a volume estimation unit for estimating the lung volume of the patient by inputting the two-dimensional medical image and clinical information into an artificial intelligence model, The artificial intelligence model is a first neural network for encoding and decoding video via an encoder and a decoder; a second neural network for predicting lung volume; The lung volume examination device is characterized in that the volume grasping unit obtains a restored image related to the two-dimensional medical image via the first neural network, and inputs data combining the restored image and the clinical information to the second neural network to grasp the lung volume.

15. An image input unit to which a two-dimensional medical image of a patient is input; a clinical information input unit to which clinical information of the patient is input; and a volume estimation unit for estimating the lung volume of the patient by inputting the two-dimensional medical image and clinical information into an artificial intelligence model, a learning unit for training the artificial intelligence model including the first neural network and the second neural network by a teaching-learning method using learning data; The learning unit is inputting the two-dimensional medical image of the training data into the first neural network that performs image encoding; inputting data obtained by combining the output value of the first neural network with clinical information of the learning data into the second neural network to predict lung volume; A lung volume testing apparatus, comprising: training said first neural network and said second neural network so that a difference between a predicted lung volume and an actual lung volume of said learning data is minimized.

16. An image input unit to which a two-dimensional medical image of a patient is input; a clinical information input unit to which clinical information of the patient is input; and a volume estimation unit for estimating the lung volume of the patient by inputting the two-dimensional medical image and clinical information into an artificial intelligence model, a learning unit for training the artificial intelligence model including the first neural network and the second neural network by a teaching-learning method using learning data; The learning unit is a first learning unit for inputting the two-dimensional medical image as the learning data to the first neural network, the first neural network including an encoding process and a decoding process, and training the first neural network by a self-guided learning method; a second learning unit that grasps feature information related to the two-dimensional medical image of the learning data through the first neural network that has completed learning, inputs data combining the feature information and clinical information of the learning data into the second neural network to predict lung volume, and trains the second neural network so that a difference between the predicted lung volume and an actual lung volume of the learning data is reduced.

17. An image input unit to which a two-dimensional medical image of a patient is input; a clinical information input unit to which clinical information of the patient is input; and a volume estimation unit for estimating the lung volume of the patient by inputting the two-dimensional medical image and clinical information into an artificial intelligence model, a learning unit for training the artificial intelligence model including the first neural network and the second neural network in a teaching-learning manner using learning data; The learning unit is inputting the two-dimensional medical image of the learning data into the first neural network including an encoding process and a decoding process to generate a restored image; inputting information obtained by combining the restored image and the clinical information of the training data into the second neural network to predict lung volume; A lung volume testing apparatus, comprising: training said first neural network and said second neural network so that a difference between a predicted lung volume and an actual lung volume of said learning data is reduced.

18. A computer readable recording medium having recorded thereon a computer program for carrying out the method according to any one of claims 1 to 11.

Images