KR102481564B1 - Medical image process apparatus and medical image learning method, and medical image process method - Google Patents

Abstract

A medical image processing method performed by a medical image processing apparatus according to an embodiment includes inputting a body X-ray image of a heart of a subject to be diagnosed to a pre-learned artificial neural network model, and outputting the body X-ray image as an output of the artificial neural network model. estimating the type of prosthetic valve in the heart and obtaining shape information related to the prosthetic valve; estimating the blood flow direction of the heart based on the obtained shape information; and determining the type of the prosthetic valve as a result of re-estimating the type of the prosthetic valve based on the estimated direction of blood flow.

Description

Medical image processing apparatus, medical image learning method, and medical image processing method

The present invention relates to a method and apparatus for learning medical images of the body, and a method and apparatus for processing medical images of the body.

A medical image refers to an image obtained using a medical imaging apparatus to include information about an internal structure of a body of a subject to be diagnosed. A medical imaging device used to obtain such a medical image is a non-invasive examination device, and displays structural details, internal tissues, and fluid flow in the body by photographing and processing the images to the user. A user such as a doctor may diagnose a patient's health condition and disease using a medical image output from a medical imaging device. For example, a medical imaging device includes an X-ray imaging device that radiates X-rays to an object and detects X-rays passing through the object to image an image, and a magnetic resonance image that provides a magnetic resonance image. There are a magnetic resonance imaging (MRI) device, a computed tomography (CT) device, and an ultrasound diagnosis device.

The prosthetic valve is implanted into the human or animal body and can be used as a passive, direct prosthetic valve, eg in or near a blood vessel. The prosthetic valve may be completely preformed and implanted as such, or it may be formed in situ using the artificial and/or natural components necessary to form a functional prosthetic valve.

These prosthetic valves can be largely divided into mechanical valves and tissue valves depending on the material, and depending on the location, mitral valves, aortic valves, tricuspid valves, and valves. It can be divided into pulmonary valve and the like. Medical images taken of humans or animals with artificial valves are photographed together with artificial valves, and the artificial valve included in the photographed medical images can be used for medical activities, etc. For this, it is necessary to distinguish the types.

According to the prior art, the type of prosthetic valve included in the medical image was classified depending on the doctor's visual observation of the medical image, but only a highly skilled doctor could accurately distinguish the type of the prosthetic valve, and the doctor's proficiency There is a problem in that the accuracy of classifying the type of prosthetic valve included in the medical image changes according to the method.

Korean Registered Patent No. 10-1517752 (registered on April 29, 2015)

According to an embodiment, a medical image processing method and apparatus capable of accurately distinguishing the type of an artificial valve included in a medical image using an artificial neural network model and a learning technique thereof are provided.

The problem to be solved by the present invention is not limited to those mentioned above, and another problem to be solved that is not mentioned will be clearly understood by those skilled in the art from the description below.

A medical image processing method performed by a medical image processing apparatus according to the first aspect includes inputting a body X-ray image of a heart of a diagnosis subject to a pre-learned artificial neural network model, and outputting the body as an output of the artificial neural network model. estimating the type of prosthetic valve in the heart of the X-ray image and obtaining shape information related to the prosthetic valve; estimating the blood flow direction of the heart based on the acquired shape information; and determining the type of the prosthetic valve as a result of re-estimating the type of the prosthetic valve based on the type of and the estimated direction of blood flow.

According to the second aspect, the computer program of the computer readable recording medium storing the computer program, when executed by the processor, provides instructions for causing the processor to perform the medical image processing method performed by the medical image processing apparatus. include

According to the third aspect, the computer program stored in the computer readable recording medium includes instructions for causing the processor to perform the medical image processing method performed by the medical image processing apparatus when executed by the processor.

A medical image learning method of a medical image processing apparatus according to a fourth aspect includes preparing body X-ray images for learning of a heart as inputs constituting a first training data set, and labels constituting the first training data set. preparing information on the type of artificial valve corresponding to each of the body X-ray images for learning and information on the shape related to the artificial valve, and a plurality of shape information and the plurality of shapes as input constituting a second learning data set. Preparing heart blood flow direction information corresponding to the information, and preparing type information of an artificial valve corresponding to the plurality of shape information and heart blood flow direction information as labels constituting the second learning data set; and training a first layer of the artificial neural network model using the first training data set and training a second layer of the artificial neural network model using the second training data set.

The computer program of the computer readable recording medium storing the computer program according to the fifth aspect, when executed by a processor, includes instructions for causing the processor to perform the medical image learning method of the medical image processing apparatus. .

According to the sixth aspect, the computer program stored in the computer readable recording medium includes instructions for causing the processor to perform the medical image learning method of the medical image processing apparatus when executed by the processor.

The medical image processing apparatus according to the seventh aspect includes: an input unit for receiving an X-ray body image of the heart of a patient to be diagnosed; An artificial neural network model unit that estimates the type of prosthetic valve in the image and obtains shape information related to the prosthetic valve, and estimates the blood flow direction of the heart based on the obtained shape information, and a processor configured to determine the type of the prosthetic valve as a result of re-estimating the type of the prosthetic valve based on the estimated direction of blood flow, and an output unit to output information about the type of the prosthetic valve determined by the processor. .

The medical image processing apparatus according to the eighth aspect includes, as input, body X-ray images for learning of the heart, and labels type information of an artificial valve corresponding to each of the X-ray images for learning and shape information related to the artificial valve. A type of prosthetic valve including a plurality of shape information and cardiac blood flow direction information corresponding to the plurality of shape information as inputs and corresponding to the plurality of shape information and the heart blood flow direction information; An input unit that receives a second training data set including information as labels, and the first layer of the artificial neural network model learns the first training data set received through the input unit, and the second training data set received through the input unit and an artificial neural network model unit through which the second layer of the artificial neural network model learns a training data set.

According to an embodiment, the type of prosthetic valve included in a medical image can be accurately classified using an artificial neural network model and a learning technique thereof. By providing information on the type of prosthetic valve classified using the artificial neural network model to the medical personnel, there is an effect of supporting the medical personnel to accurately and finally determine the end of the prosthetic valve included in the body X-ray image.

1 is a configuration diagram of a medical image processing apparatus according to an exemplary embodiment.
2 is a flowchart illustrating a process of learning an artificial neural network model of an artificial neural network model unit of a medical image processing apparatus according to a first embodiment of the present invention.
3 is a flowchart illustrating a process in which a medical image processing apparatus including an artificial neural network model learned according to the first embodiment of the present invention determines the type of an artificial valve in the heart of a subject to be diagnosed.
4 is a flowchart illustrating a process of learning an artificial neural network model of an artificial neural network model unit of a medical image processing apparatus according to a second embodiment of the present invention.
5 is a view showing shadow shapes for each type of prosthetic valve.

Advantages and features of the present invention, and methods of achieving them, will become clear with reference to the embodiments described below in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and only these embodiments make the disclosure of the present invention complete, and common knowledge in the art to which the present invention belongs. It is provided to completely inform the person who has the scope of the invention, and the present invention is only defined by the scope of the claims.

The terms used in this specification will be briefly described, and the present invention will be described in detail.

The terms used in the present invention have been selected from general terms that are currently widely used as much as possible while considering the functions in the present invention, but these may vary depending on the intention of a person skilled in the art or precedent, the emergence of new technologies, and the like. In addition, in a specific case, there is also a term arbitrarily selected by the applicant, and in this case, the meaning will be described in detail in the description of the invention. Therefore, the term used in the present invention should be defined based on the meaning of the term and the overall content of the present invention, not simply the name of the term.

When it is said that a certain part 'includes' a certain element in the entire specification, it means that other elements may be further included without excluding other elements unless otherwise stated.

In addition, the term 'unit' used in the specification means software or a hardware component such as FPGA or ASIC, and 'unit' performs certain roles. However, 'part' is not limited to software or hardware. A 'unit' may be configured to reside in an addressable storage medium and may be configured to reproduce one or more processors. Thus, as an example, 'unit' refers to components such as software components, object-oriented software components, class components and task components, processes, functions, properties, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays and variables. The functionality provided within the components and 'parts' may be combined into a smaller number of elements and 'parts' or further separated into additional elements and 'parts'.

Hereinafter, with reference to the accompanying drawings, embodiments of the present invention will be described in detail so that those skilled in the art can easily carry out the present invention. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted.

In the present specification, a subject to be diagnosed or a patient may include a human or an animal, or a part of a human or animal.

In addition, 'image' in this specification may refer to multi-dimensional data composed of discrete image elements (eg, pixels in a 2D image and voxels in a 3D image). there is.

The medical image processing apparatus, the medical image learning method, and the medical image processing method of the present invention can be implemented in various embodiments, and in the following description, the first and second embodiments are representatively described. do it with 1 is a configuration diagram of a medical image processing apparatus 100 according to an embodiment of the present invention, and the illustrated medical image processing apparatus 100 performs a medical image learning method and a medical image processing method implemented in various forms. However, the description of the medical image processing apparatus 100 will be divided into the first embodiment and the second embodiment.

<First Embodiment>

FIG. 1 illustrates a configuration of a medical image processing apparatus 100 according to an exemplary embodiment, but what is illustrated in FIG. 1 is merely illustrative. The medical image processing apparatus 100 may be implemented in or include a PC or server.

Referring to FIG. 1 , the medical image processing apparatus 100 includes an input unit 110, an artificial neutral network model unit 120, a processing unit 130, and an output unit 140.

The input unit 110 receives body X-ray images for heart learning as an input constituting a learning data set. In addition, the input unit 110 receives type information of the artificial valve in the heart of the body X-ray image and shape information related to the artificial valve as labels constituting the learning data set. In addition, the input unit 110 receives an X-ray image of the body of the patient's heart. Here, the learning data set is to be used when learning the artificial neural network model unit 120, and the body X-ray image of the heart of the subject to be diagnosed is input to the artificial neural network model unit 120 that has been previously learned, and the type of artificial valve is to determine

The artificial neural network model unit 120 learns the artificial neural network model from the learning data set received through the input unit 110 . In addition, the learned artificial neural network model unit 120 estimates the type of artificial valve in the body X-ray image of the subject from the body X-ray image of the subject to be diagnosed received through the input unit 110, and the type related to the artificial valve. Acquire information. For example, the artificial neural network model unit 120 may obtain information about the shape of a shadow of the artificial valve as shape information related to the artificial valve. For example, the artificial neural network model may be a Mask R-CNN model. The Mask R-CNN model can use body X-ray images for learning of the heart as training images of the Mask R-CNN model, and information on the type of artificial valve in the heart of the body X-ray images corresponding to the body X-ray images for learning, respectively. And shape information related to the prosthetic valve can be used as a mask for the Mask R-CNN model. For example, the artificial neural network model unit 120 may include a memory storing instructions programmed to perform functions as an artificial neural network model and a microprocessor executing these instructions.

The processing unit 130 estimates the blood flow direction of the heart based on the shape information related to the prosthetic valve obtained by the artificial neural network model unit 120, and estimates the type and type of the prosthetic valve estimated by the artificial neural network model unit 120. The type of prosthetic valve is determined as a result of re-estimating the type of prosthetic valve based on the direction of blood flow. For example, when estimating the direction of blood flow, the processing unit 130 may estimate the direction of blood flow of the heart based on a result of comparing the obtained shape information with the preset shape information. For example, the preset shape information may be a plurality of crown shapes, and when estimating a blood flow direction, the processing unit 130 may estimate a blood flow direction for any one of the tissue valves respectively corresponding to the plurality of crown shapes. For example, the processing unit 130 performs a function of determining the type of artificial valve as a result of re-estimating the type of artificial valve based on the type of artificial valve estimated by the artificial neural network model unit 120 and the estimated direction of blood flow. It may include a memory that stores instructions programmed to do so, and a microprocessor that executes these instructions.

The output unit 140 outputs information about the type of prosthetic valve determined by the processing unit 130 . For example, the output unit 140 may output information on the type of prosthetic valve determined by the processing unit 130 to be recognized from the outside. The output unit 140 may include a port for outputting information on the type of prosthetic valve, a wired communication module, or a wireless communication module. Alternatively, the output unit 140 may include an image display device capable of outputting information about the type of prosthetic valve in the form of an image.

2 is a flowchart illustrating a process of learning an artificial neural network model of the artificial neural network model unit 120 of the medical image processing apparatus 100 according to the first embodiment of the present invention, and FIG. 3 is a flowchart illustrating the first embodiment of the present invention. A flowchart illustrating a process in which the medical image processing apparatus 100 including the artificial neural network model trained according to is determined the type of artificial valve in the heart of a subject to be diagnosed.

Hereinafter, with reference to FIGS. 1 to 3, a process of learning the artificial neural network model according to the first embodiment of the present invention and a process of determining the type of artificial valve using the learned artificial neural network model will be described in detail. .

First, an operator of the medical image processing apparatus 100 may prepare body X-ray images for heart learning as input constituting a learning data set (S210), and body X-ray images as labels constituting the learning data set. Type information of the prosthetic valve in the heart of the patient and shape information related to the prosthetic valve may be prepared (S220). Here, the shape information related to the prosthetic valve included in the training data set as a label may be information about the shape of the shadow of the prosthetic valve.

When the prepared training data set is input through the input unit 110, the artificial neural network model of the artificial neural network model unit 120 learns the training data set (S230). In this way, the number of times the artificial neural network model unit 120 learns the training data set is not particularly limited, but as one of the embodiments, learning may be performed within a range of appropriate numbers for the best learning effect.

After learning of the artificial neural network model unit 120 of the medical image processing apparatus 100, the operator of the medical image processing apparatus 100 prepares an X-ray image of the body of the subject to be diagnosed and inputs the image through the input unit 110. can do.

Then, the artificial neural network model unit 120 learned through steps S210 to S230 estimates the type of artificial valve in the body X-ray image from the artificial neural network model from the body X-ray image of the subject to be diagnosed received through the input unit 110. and obtain shape information related to the prosthetic valve. Here, the artificial neural network model unit 120 may acquire information about the shape of the shadow of the artificial valve as shape information related to the artificial valve (S310).

Here, the type of artificial valve estimated by the artificial neural network model unit 120 through step S310 may or may not match the actual type of artificial valve existing in the body of the subject to be diagnosed. Several prosthetic valves existing in the body of the diagnosis subject may be included in the X-ray image in a similar position, or in the form of overlapping a plurality of prosthetic valves in a certain direction due to an enlarged heart or a change in body shape. In these cases, step S310 The possibility that the type of prosthetic valve estimated by the method is inconsistent with the actual type of prosthetic valve existing in the body is relatively higher than in other cases.

Accordingly, the processing unit 130 of the medical image processing apparatus 100 estimates the blood flow direction of the heart based on the shape information related to the prosthetic valve acquired in step S310. Here, the reason why the processing unit 130 estimates the blood flow direction of the heart is to reflect the estimated blood flow direction of the heart to more accurately classify the type of artificial valve. is likely to deteriorate.

The processing unit 130 compares the shape information related to the prosthetic valve acquired by the artificial neural network model unit 120 with the preset shape information in step S310 to determine whether the obtained shape information is included in the preset shape information. For example, the preset shape information may be a plurality of crown shapes, that is, various crown shapes. 5 is a view showing shadow shapes for each type of prosthetic valve. Referring to FIG. 5 , among tissue valves, it can be seen that aortic valves (Aortic Bio), pulmonary artery valves (Pulmonary Bio), mitral valves (Mitral Bio), and tricuspid valves (Tricuspid Bio) have crown-shaped shadows.

Subsequently, the processing unit 130 estimates the blood flow direction of the heart based on a result of comparing the shape information related to the prosthetic valve acquired by the artificial neural network model unit 120 with the preset shape information in step S310. For example, it is possible to estimate whether blood flow flows from the lower part of the heart to the upper right side or from the right side of the heart to the left side depending on which of the plurality of crown shapes it is in (S320).

Next, the processor 130 re-estimates the type of artificial valve in the heart of the body X-ray image input in step S310 based on the type of artificial valve estimated in step S310 and the direction of blood flow estimated in step S320. As a result, the type of prosthetic valve is determined. If it is confirmed in step S320 that the shape information related to the prosthetic valve is not included in the preset shape information, the estimated result in step S310 is finally determined as the type of the prosthetic valve. However, if it is confirmed in step S320 that the shape information related to the prosthetic valve is included in the preset shape information, the type of artificial valve may be corrected by reflecting the estimated blood flow direction instead of using the result estimated in step S310 as it is. . For example, in the case where blood flow is estimated to flow from the lower side of the heart to the right side, regardless of the estimation result of step S310, the type of artificial valve can be finally determined as the aortic valve, and the blood flow estimated from the right side to the left side of the heart is estimated. In this case, the type of artificial valve may be finally determined as a tricuspid valve regardless of the estimation result of step S310 (S330).

Thereafter, the output unit 140 outputs information about the type of prosthetic valve determined by the processing unit 130 . For example, the output unit 140 outputs information on the type of prosthetic valve determined by the processing unit 130 in the form of an image recognizable from the outside or as a separate electronic device through a wired communication module or a wireless communication module. can be sent

<Second Embodiment>

The second embodiment can be said to be a case where the artificial neural network model unit 120 performs up to the point where the processing unit 130 re-estimates the type of prosthetic valve in the heart in the first embodiment.

FIG. 1 illustrates a configuration of a medical image processing apparatus 100 according to an exemplary embodiment, but what is illustrated in FIG. 1 is merely illustrative. The medical image processing apparatus 100 may be implemented in or include a PC or server.

Referring to FIG. 1 , the medical image processing apparatus 100 includes an input unit 110, an artificial neural network model unit 120, a processing unit 130, and an output unit 140.

The input unit 110 receives body X-ray images for learning of the heart as input constituting the first learning data set. In addition, the input unit 110 receives type information of the artificial valve in the heart of the body X-ray image and shape information related to the artificial valve as labels constituting the first learning data set. In addition, the input unit 110 receives a plurality of shape information and heart blood flow direction information corresponding to the plurality of shape information as input constituting the second learning data set. In addition, the input unit 110 receives information about the type of prosthetic valve corresponding to a plurality of shape information and heart blood flow direction information as labels constituting the second learning data set. For example, the shape information related to the prosthetic valve included as a label of the first learning data set may be information about the shape of a shadow of the prosthetic valve. Also, the plurality of shape information may be a plurality of crown shapes, and the heart blood flow direction information may be a blood flow direction for any one of the tissue valves respectively corresponding to the plurality of crown shapes. In addition, the input unit 110 receives an X-ray image of the body of the patient's heart. Here, the first learning data set and the second learning data set are used for learning the artificial neural network model unit 120, and the body X-ray image of the heart of the subject to be diagnosed is pre-learned artificial neural network model unit 120. ) to determine the type of prosthetic valve.

In the artificial neural network model unit 120, the first layer of the artificial neural network model learns the first training data set received through the input unit 110, and the second training data set input through the input unit 110 is used as the artificial neural network model. The second layer of learns. In addition, the artificial neural network model unit 120 including the learned artificial neural network model determines that the first layer of the artificial neural network model from the body X-ray image of the subject to be diagnosed received through the input unit 110 is the artificial valve in the body X-ray image. The type is estimated, and shape information related to the prosthetic valve is acquired. In addition, the second layer of the artificial neural network model estimates the blood flow direction of the heart based on the acquired shape information related to the prosthetic valve, and based on the type of prosthetic valve estimated by the first layer and the estimated blood flow direction, the prosthetic valve is estimated. As a result of re-estimating the type, the type of prosthetic valve is determined. For example, the artificial neural network model may be a Mask R-CNN model. For example, the artificial neural network model unit 120 may include a memory storing instructions programmed to perform functions as an artificial neural network model and a microprocessor executing these instructions.

The processing unit 130 controls the output unit 140 to output the type of prosthetic valve determined by the artificial neural network model unit 120 . For example, processing unit 130 may include a microprocessor that executes various instructions.

The output unit 140 outputs information about the type of prosthetic valve under the control of the processing unit 130 . For example, the output unit 140 may output information on the type of prosthetic valve determined by the processing unit 130 to be recognized from the outside. The output unit 140 may include a port for outputting information on the type of prosthetic valve, a wired communication module, or a wireless communication module. Alternatively, the output unit 140 may include an image display device capable of outputting information about the type of prosthetic valve in the form of an image.

4 is a flowchart illustrating a process of learning an artificial neural network model of the artificial neural network model unit 120 of the medical image processing apparatus 100 according to the second embodiment of the present invention.

Hereinafter, with reference to FIGS. 1 and 4, a process of learning the artificial neural network model according to the second embodiment of the present invention and a process of determining the type of artificial valve using the learned artificial neural network model will be described in detail. .

First, an operator of the medical image processing apparatus 100 may prepare body X-ray images for learning of the heart as inputs constituting the first learning data set (S410), and as labels constituting the first learning data set. Information on the type of artificial valve in the heart of the body X-ray image and information on the shape of the artificial valve may be prepared (S420). Here, the shape information related to the prosthetic valve included in the training data set as a label may be information about the shape of the shadow of the prosthetic valve.

In addition, the operator of the medical image processing apparatus 100 may prepare a plurality of shape information and heart blood flow direction information corresponding to the plurality of shape information as inputs constituting the second learning data set (S430). As a label constituting the learning data set, information on the type of prosthetic valve corresponding to a plurality of shape information and heart blood flow direction information may be prepared (S440).

When the first training data set and the second training data set thus prepared are input through the input unit 110, the first training data set is learned by the first layer of the artificial neural network model, and the second training data set is used by the artificial neural network model. Layer 2 learns. In this way, the number of times that the artificial neural network model unit 120 learns the learning data set is not particularly limited, but as one of the embodiments, learning may be performed within an appropriate number of times for the best learning effect (S450).

After learning of the artificial neural network model unit 120 of the medical image processing apparatus 100, the operator of the medical image processing apparatus 100 prepares an X-ray image of the body of the subject to be diagnosed and inputs the image through the input unit 110. can do.

Then, in the artificial neural network model unit 120 learned through steps S410 to S450, the artificial neural network model determines the type of artificial valve in the body X-ray image from the body X-ray image of the subject to be diagnosed received through the input unit 110. do. The process of determining the type of artificial valve in the body X-ray image by the artificial neural network model unit 120 is that the first layer and the second layer of the artificial neural network model sequentially perform steps S310 to S330 of the first embodiment. This can be done, and since it can be inferred from the above descriptions, the detailed description thereof will be omitted.

Thereafter, the output unit 140 outputs information about the type of artificial valve determined by the artificial neural network model unit 120 according to the control of the processing unit 130 . For example, the output unit 140 outputs information on the type of prosthetic valve determined by the processing unit 130 in the form of an image recognizable from the outside or as a separate electronic device through a wired communication module or a wireless communication module. can be sent

On the other hand, each step included in the medical image learning method and medical image processing method according to the above-described first and second embodiments is carried out in a computer readable recording medium recording a computer program programmed to perform these steps. can be implemented

In addition, each step included in the medical image learning method and medical image processing method according to the above-described first and second embodiments is a computer program stored in a computer readable recording medium programmed to perform these steps. can be implemented in the form

Combinations of each step in each flowchart attached to the present invention may be performed by computer program instructions. Since these computer program instructions may be loaded into a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing equipment, the instructions executed by the processor of the computer or other programmable data processing equipment function as described in each step of the flowchart. create a means to do them. These computer program instructions can also be stored on a computer usable or computer readable medium that can be directed to a computer or other programmable data processing equipment to implement functions in a particular way, so that the computer usable or computer readable It is also possible that the instructions stored on the recording medium produce an article of manufacture containing instruction means for performing the functions described in each step of the flowchart. The computer program instructions can also be loaded on a computer or other programmable data processing equipment, so that a series of operational steps are performed on the computer or other programmable data processing equipment to create a computer-executed process to generate computer or other programmable data processing equipment. Instructions for performing the processing equipment may also provide steps for executing the functions described at each step in the flowchart.

Further, each step may represent a module, segment or portion of code that includes one or more executable instructions for executing the specified logical function(s). It should also be noted that in some alternative embodiments it is possible for the functions mentioned in the steps to occur out of order. For example, two steps shown in succession may in fact be performed substantially concurrently, or the steps may sometimes be performed in reverse order depending on the function in question.

The above description is merely an example of the technical idea of the present invention, and various modifications and variations can be made to those skilled in the art without departing from the essential qualities of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, but to explain, and the scope of the technical idea of the present invention is not limited by these embodiments. The protection scope of the present invention should be construed according to the claims below, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

100: medical image processing device
110: input unit
120: artificial neural network model unit
130: processing unit
140: output unit

Claims (13)

A medical image processing method performed by a medical image processing apparatus,
A body X-ray image of the heart of the patient to be diagnosed is input to a pre-learned artificial neural network model, and as an output of the artificial neural network model, the type of artificial valve in the heart of the body X-ray image is estimated, and the artificial valve and obtaining related shape information;
estimating a blood flow direction of the heart based on the acquired shape information;
Determining the type of the artificial valve as a result of re-estimating the type of the artificial valve based on the estimated type of the artificial valve and the estimated blood flow direction
Medical image processing method. According to claim 1,
The obtaining of the shape information includes obtaining information on the shape of the shadow of the prosthetic valve.
Medical image processing method. According to claim 1,
The estimating of the blood flow direction may include comparing the acquired shape information with preset shape information;
Estimating the blood flow direction of the heart based on the result of the comparison
Medical image processing method. According to claim 3,
The preset shape information is a plurality of crown shapes,
The estimating of the blood flow direction may include estimating a blood flow direction for any one of the tissue valves corresponding to the plurality of crown shapes.
Medical image processing method. A computer-readable recording medium storing a computer program,
When the computer program is executed by a processor,
A computer-readable recording medium comprising instructions for causing the processor to perform the method according to any one of claims 1 to 4. As a computer program stored on a computer-readable recording medium,
When the computer program is executed by a processor,
A computer program comprising instructions for causing the processor to perform the method according to any one of claims 1 to 4. As a medical image learning method of a medical image processing device,
Preparing body X-ray images for heart learning as input constituting a first learning data set;
preparing information on the type of artificial valve corresponding to each of the X-ray images for learning and shape information related to the artificial valve as labels constituting the first learning data set;
preparing a plurality of shape information and heart blood flow direction information corresponding to the plurality of shape information as input constituting a second learning data set;
Preparing type information of an artificial valve corresponding to the plurality of shape information and the heart blood flow direction information as labels constituting the second learning data set;
Learning a first layer of an artificial neural network model using the first training data set and learning a second layer of the artificial neural network model using the second training data set
Medical image learning method. According to claim 7,
The shape information related to the prosthetic valve is information about the shape of the shadow of the prosthetic valve.
Medical image learning method. According to claim 7,
The plurality of shape information used as the input of the second learning data set is a plurality of crown shapes, and the prosthetic valve type information used as the label of the second learning data set is tissue valve type information.
Medical image learning method. A computer-readable recording medium storing a computer program,
When the computer program is executed by a processor,
A computer-readable recording medium comprising instructions for causing the processor to perform the method according to any one of claims 7 to 9. As a computer program stored on a computer-readable recording medium,
When the computer program is executed by a processor,
A computer program comprising instructions for causing the processor to perform the method according to any one of claims 7 to 9. An input unit for receiving an X-ray image of the heart of a subject to be diagnosed;
An artificial neural network model unit for estimating the type of an artificial valve in the body X-ray image of the diagnosis subject from the body X-ray image of the diagnosis subject input through the input unit and acquiring shape information related to the artificial valve;
The blood flow direction of the heart is estimated based on the obtained shape information, and the type of the prosthetic valve is determined as a result of re-estimating the type of the prosthetic valve based on the estimated type of prosthetic valve and the estimated blood flow direction. a processing unit that
And an output unit for outputting information about the type of the prosthetic valve determined by the processing unit.
Medical image processing device. A first learning data set including, as input, body X-ray images for learning of the heart, and including type information of an artificial valve corresponding to each of the body X-ray images for learning and shape information related to the artificial valve as a label and a plurality of shape information and cardiac blood flow direction information corresponding to the plurality of shape information as inputs, and a label containing type information of an artificial valve corresponding to the plurality of shape information and the heart blood flow direction information as a label. an input unit for receiving a training data set;
An artificial neural network in which the first layer of the artificial neural network model learns the first training data set received through the input unit, and the second layer of the artificial neural network model learns the second training data set input through the input unit. including the model
Medical image processing device.

Images