KR102330340B1 - Medical image processing apparatus - Google Patents

Abstract

The present invention relates to a medical image processing apparatus. A medical image processing apparatus according to an embodiment of the present invention includes an image acquisition unit that acquires a medical image using X-rays, a display, and a control unit, wherein the control unit uses an artificial intelligence algorithm based on deep learning learned in advance. , detects at least one of a plurality of anatomical configurations included in the medical image, determines information on the detected at least one anatomical configuration, and displays the determined information together with the detected at least one anatomical configuration Indicated, the anatomical configuration may include at least one of an organ and a catheter. In addition, various embodiments are possible.

Description

Medical image processing apparatus {MEDICAL IMAGE PROCESSING APPARATUS}

The present invention relates to a medical image processing apparatus, and more particularly, to a medical image processing apparatus for processing a medical image.

A medical image processing apparatus is an apparatus for acquiring and processing an image of an inside of a body. A medical image processing apparatus is a non-invasive examination apparatus that provides images of internal organs of the body, fluids such as blood, medical instruments inserted into the body, etc. Based on the output medical image, it is possible to check the patient's condition and more accurately determine the presence or absence of a disease.

Recently, a Computer Aided Detection (CAD) system that analyzes objects included in a medical image and provides the analysis result to the user so that the user can more accurately read the medical image has been widely used. .

In addition, artificial intelligence (AI) is actively used in the image processing field. In particular, as the machine learns many images by itself through machine learning and deep learning, and the machine can analyze the images more accurately and quickly through the learned algorithm model, these artificial Research related to medical image processing using intelligence (AI) is being actively conducted in the CAD system field.

On the other hand, as in a medical image (hereinafter referred to as a medical image using X-rays) obtained by irradiating X-rays to an object and detecting X-rays passing through the object, internal organs of the body and medical instruments When displayed overlapping each other, even when using machine learning and deep learning, it is difficult to accurately learn the overlapping portion, so even if a conventional CAD system is used, it is difficult to accurately process and analyze the medical image.

SUMMARY OF THE INVENTION The present invention aims to solve the above and other problems.

Another object of the present invention is to provide a medical image processing apparatus capable of accurately detecting a plurality of anatomical components included in a medical image using X-rays.

In order to achieve the above object, a medical image processing apparatus according to an embodiment of the present invention includes an image acquisition unit for acquiring a medical image using X-rays, a display, and a control unit, wherein the control unit is based on pre-learned deep learning Detects at least one of a plurality of anatomical components included in the medical image using an artificial intelligence algorithm of It is displayed together with one anatomical component, and the anatomical component may include at least one of an organ and a catheter.

The effect of the image processing apparatus according to the present invention will be described as follows.

According to at least one embodiment of the present invention, even when regions corresponding to a plurality of anatomical configurations included in a medical image using X-rays overlap each other, it is possible to accurately detect a plurality of anatomical configurations. Image processing performance can be further improved.

Further scope of applicability of the present invention will become apparent from the following detailed description. However, it should be understood that the detailed description and specific embodiments such as preferred embodiments of the present invention are given by way of illustration only, since various changes and modifications within the spirit and scope of the present invention may be clearly understood by those skilled in the art.

1 is a diagram illustrating a medical image processing system using X-rays.
2 is an internal block diagram of a medical image processing apparatus according to an embodiment of the present invention.
3A to 3C are diagrams referenced for explanation of a learning operation for a medical image.
4 to 6 are diagrams referenced for the description of the operation of the medical image processing apparatus.
7A to 8 are diagrams referenced for explanation of a result of processing a medical image.
9 to 11 are flowcharts of a method of operating a medical image processing apparatus according to an embodiment of the present invention.

Hereinafter, the present invention will be described in detail with reference to the drawings. In the drawings, in order to clearly and briefly describe the present invention, the illustration of parts irrelevant to the description is omitted, and the same reference numerals are used for the same or extremely similar parts throughout the specification.

The suffixes "module" and "part" for the components used in the following description are given simply in consideration of the ease of writing the present specification, and do not give a particularly important meaning or role by themselves. Accordingly, the terms “module” and “unit” may be used interchangeably.

In the present application, terms such as "comprises" or "have" are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification exists, but one or more other features It is to be understood that it does not preclude the existence of, or addition of, elements, numbers, steps, operations, components, parts, or combinations thereof.

Also, in this specification, terms such as first and second may be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish one element from another.

1 is a diagram illustrating a medical image processing system using X-rays.

Referring to FIG. 1 , the medical image processing system 100 includes an X-ray irradiation unit 10 , X-ray detection units 20 and 30 , a connection frame unit 40 , a moving frame unit 50 , and/or a guide rail. (60) may be included.

The X-ray irradiation unit 10 can generate and irradiate X-rays. For example, the X-ray irradiator 10 may include an X-ray generator (not shown) that generates X-rays and a collimator (not shown) that adjusts the irradiation range of X-rays.

The X-ray detection units 20 and 30 may detect X-rays irradiated from the X-ray irradiation unit 10 . The X-ray detectors 20 and 30 may include a detector panel (not shown) that detects X-rays, a detector driving module (not shown) that controls driving of the detector panel and receives image data from the detector panel. have. In this case, the detector driving module may generate a medical image based on the image data received from the detector panel.

The X-ray detectors 20 and 30 may include a stand-type first X-ray detector 20 and/or a table-type second X-ray detector 30 .

For example, as shown in FIG. 1 , the X-ray irradiation unit 10 may irradiate X-rays in a direction from the back of the patient to the front, and the first X-ray detector 20 is from the back of the patient. X-rays transmitted through the front can be detected.

For example, unlike FIG. 1, when the patient is lying on the second X-ray detection unit 30, the X-ray irradiation unit 10 may irradiate X-rays in a direction from the front to the back of the patient, and the second X-ray The ray detector 30 may detect X-rays transmitted from the front to the back of the patient.

The X-ray irradiation unit 10 may be connected to the connection frame unit 40 . The connection frame part 40 may be formed in a shape whose length is adjustable. At this time, when the length of the connection frame unit 40 is adjusted, the height at which the X-ray irradiation unit 10 is located may also be adjusted.

The connection frame unit 40 may be installed on the guide rail 60 , and may be connected to the movable frame unit 50 movable along the guide rail 60 . At this time, when the position of the moving frame unit 50 is changed according to the guide rail 60 , the position of the X-ray irradiation unit 10 may also be changed.

The medical image processing system 100 may further include an input unit (not shown).

The input unit may receive various inputs for operations of components included in the medical image processing system 100 . For example, the input unit may receive a user input for controlling an X-ray irradiation time point, a position of the X-ray irradiation unit 10 , and the like.

The medical image processing system 100 may further include a display unit (not shown).

The display unit may output various screens related to the control of the medical image processing system 100 . For example, the output unit may include a control screen for controlling the medical image processing system 100 , a screen for guiding a user input, a screen on which medical images acquired through the X-ray detectors 20 and 30 are displayed, etc. can be printed out.

The medical image processing system 100 may further include a communication unit (not shown). The communication unit may include at least one communication module (not shown) for a wired/wireless communication connection.

The communication unit may be connected to an external device (eg, a mobile terminal, a notebook computer, etc.), and may transmit/receive data to/from the external device.

The communication unit may connect to an external server (not shown) through a network, and may transmit/receive data to and from the external server.

2 is an internal block diagram of a medical image processing apparatus according to an embodiment of the present invention.

The medical image processing apparatus 200 may include at least some of the components included in the medical image processing system 100 of FIG. 1 .

The medical image processing apparatus 200 may include an image acquisition unit 210 , a communication unit 220 , a storage unit 230 , an input unit 240 , an output unit 250 , and/or a control unit 260 .

The image acquisition unit 210 may include the X-ray irradiation unit 10 and the X-ray detection units 20 and 30 of FIG. 1 . For example, the image acquisition unit 210 may acquire a medical image based on X-rays irradiated from the X-ray irradiator 10 and detected through the X-ray detectors 20 and 30 , and the obtained medical The image may be transmitted to the controller 260 .

The medical image may include a plurality of anatomical components. Here, the anatomical configuration may include an organ such as bone or intestine and/or a catheter, which is a medical instrument inserted into the human body.

The communication unit 220 may transmit/receive a signal including data by wire/wireless. To this end, the communication unit 220 may include at least one communication module (not shown) for a wired/wireless communication connection.

In this case, the communication unit 220 may include at least a part of the communication unit of the medical image processing system 100 of FIG. 1 .

For example, the communication unit 220 may transmit/receive a signal using a wireless communication method such as Wi-Fi, Bluetooth, beacon, zigbee, and RFID (radio frequency identification). .

For example, the communication unit 220 may transmit/receive a signal to/from an external device (eg, a mobile terminal, a notebook computer, an external storage medium, etc.) connected through a wired communication method such as a universal serial bus (USB).

Meanwhile, the medical image processing apparatus 100 may receive a medical image from the outside through the communication unit 220 , and may transmit the received medical image to the controller 260 .

For example, the communication unit 220 may communicate with a server (not shown) through a network, and may receive a medical image from the server.

For example, the communication unit 220 may receive a medical image from an external device connected by wire.

The storage unit 230 may store a program for processing and controlling each signal in the control unit 260 , or may store a signal-processed image, audio, or data signal.

For example, the storage unit 230 stores application programs designed for the purpose of performing various tasks that can be processed by the control unit 260 , and upon request of the control unit 260 , selectively selects some of the stored application programs. can provide

The program stored in the storage unit 230 is not particularly limited as long as it can be executed by the control unit 260 .

Although the embodiment in which the storage unit 230 of FIG. 2 is provided separately from the control unit 260 is illustrated, the scope of the present invention is not limited thereto, and the storage unit 230 may be included in the control unit 260 .

The storage unit 230 may include a volatile memory (eg, DRAM, SRAM, SDRAM, etc.), a non-volatile memory (eg, a flash memory), a hard disk drive (HDD), or a solid state drive (Solid). -state drive (SSD), etc.) may be included.

The storage unit 230 may store a medical image acquired through the image acquisition unit 210 and/or the communication unit 220 .

The storage unit 230 may store pre-learned data, models, algorithms, etc. through machine learning such as deep learning. For example, the storage 230 may store data on a result of pre-learning a database of medical images using X-rays through machine learning.

Machine learning means that a computer learns from data without a human instructing the computer directly to logic, and allows the computer to solve a problem through this.

Deep learning is a method of teaching a human way of thinking to a computer based on Artificial Neural Networks (ANN). The artificial neural network (ANN) may be implemented in the form of software or in the form of hardware such as a chip.

For example, an artificial neural network (ANN), a deep neural network (DNN), a convolutional neural network (CNN), a recurrent neural network (RNN), a deep belief network; DBN), etc., may include various types of algorithms.

Meanwhile, the deep learning-based pre-learned artificial intelligence algorithm model may be a segmentation model.

In FIG. 3A, the segmentation model processes the obtained medical image 310 through an encoder network and a decoder network having a plurality of convolution layers, and an image-processed medical image ( 320), but the present invention is not limited thereto, and may be implemented with models such as FCN (Fully Convolutional Network), U-net, DeepLab, and the like.

Meanwhile, in the case of a medical image using X-rays, a plurality of anatomical components included in the medical image may be displayed overlapping each other due to characteristics of X-rays passing through an object. For example, when the chest among body parts is photographed using X-rays, organs such as the heart, lungs, and ribs may be displayed overlapping each other.

At this time, when a medical image using X-rays is simply learned through a segmentation model, it is difficult to accurately perform learning of overlapping organs, so there is a problem in that it is difficult to accurately detect a plurality of anatomical configurations included in the medical image. .

In consideration of this, according to various embodiments of the present disclosure, a model pre-trained through machine learning stored in the medical image processing apparatus 200 is mapped to any one of a plurality of classes. A medical image including a region (eg, a pixel) may be a model based on a result of learning through machine learning.

Here, the plurality of classes may refer to a set including at least one of a plurality of anatomical configurations. In this case, the plurality of classes may represent a set including any one of the plurality of anatomical configurations to a set including all of the plurality of anatomical configurations.

For example, a plurality of classes for the chest may be determined as shown in Table 1 below.

class Chest 10 lung 11 lung + heart 12 lungs + ribs 13 lung + keel 14 lungs + heart + ribs 15 lung + heart + keel 16 lungs + heart + ribs + keel 20 heart 21 heart + ribs 22 heart + keel 23 heart + ribs + keel 30 rib 31 rib + keel 40 keel

In this case, when each of the detailed regions constituting the medical image of the chest shown in FIG. 3B is mapped to any one of a plurality of classes for the chest, the medical image shown in FIG. 3B is the same as the medical image shown in FIG. 3C . can be reconstructed together.

Referring to FIG. 3C , the detailed region 330 corresponding to the lungs and ribs is class 12, the detailed region 335 corresponding to the lungs, heart, and ribs is class 14, and the detailed region 340 corresponding to the lungs is class 10 , the detailed region 350 corresponding to the lungs and the heart may be mapped to class 11, and the detailed region 360 corresponding to the rib may be mapped to class 30, respectively.

That is, the model previously learned through machine learning stored in the storage unit 230 of the medical image processing apparatus 200 is based on the result of learning the medical image reconstructed according to a plurality of classes through machine learning as shown in FIG. 3C . It may be a model based

Meanwhile, the storage 230 may store images of a plurality of anatomical components included in the medical image.

For example, the storage unit 230 may store a database for images corresponding to a portion of various catheters. In this case, the images included in the database may be various images in which the catheter is disposed in the center, as shown in FIG. 4 .

For example, the storage unit 230 may store a database for images corresponding to a portion of various bones or joints.

The input unit 240 may receive various user commands related to the operation of the medical image processing apparatus 200 , and may transmit a control signal corresponding to the input command to the control unit 260 .

The input unit 240 may include a touch screen, a keyboard, a mouse, a microphone, and the like.

The input unit 240 may include at least a portion of the input unit of the medical image processing system 100 .

The output unit 250 may include a display device such as a display (not shown) and a light emitting diode (LED), and may display a screen related to an operation of the medical image processing apparatus 200 .

For example, the output unit 250 may output an image-processed medical image through a display.

The output unit 250 may include an audio device such as a speaker or a buzzer. For example, the output unit 250 may output a sound effect for the state of the medical image processing apparatus 200 , and may output a predetermined warning sound when an error occurs.

The controller 260 may include at least one processor, and may control overall operations of the medical image processing apparatus 200 using the processor included therein. Here, the processor may be a general processor such as a central processing unit (CPU). Of course, the processor may be a dedicated device such as an ASIC or other hardware-based processor.

When the controller 260 includes a plurality of processors, the plurality of processors may be integrated on one chip or may be physically separated from each other.

The controller 260 may process a medical image acquired through the image acquisition unit 210 and/or the communication unit 220 .

The controller 260 may detect at least one of a plurality of anatomical components included in a medical image using X-rays acquired through the image acquisition unit 210 and/or the communication unit 220 .

In this case, the controller 260 may detect at least one of a plurality of anatomical components included in the medical image by using a deep learning-based artificial intelligence algorithm previously learned through machine learning. Here, the deep learning-based artificial intelligence algorithm previously learned through machine learning may be a model stored in the storage unit 230 .

The control unit 260 includes a data learning unit 261 for learning a medical image through machine learning such as deep learning and a data processing unit for processing a medical image using an artificial intelligence algorithm based on deep learning previously learned through machine learning ( 262) may be included.

When detecting an organ from among a plurality of anatomical components included in a medical image, the controller 260 may determine a plurality of classes corresponding to at least one of a plurality of organs included in the corresponding medical image.

In this case, the controller 260 may determine a body part corresponding to the medical image, and may determine a plurality of organs included in the corresponding medical image according to the determined body part.

For example, when the medical image is a medical image of the chest shown in FIG. 5A , the controller 260 may determine a body part corresponding to the medical image as the chest, and remove a plurality of organs included in the medical image. (lung), heart (heart), rib (lib) and keel (carina) can be judged.

In this case, the controller 260 may determine the body part corresponding to the medical image as the chest, based on the data on the body part stored in the storage 230 .

The controller 260 may determine a plurality of classes as shown in Table 1 in relation to a medical image of the chest.

Meanwhile, the controller 260 may map the detailed region of the medical image to any one of a plurality of classes by using a deep learning-based AI algorithm previously learned through machine learning.

Referring to FIG. 5B , the controller 260 processes the medical image of the chest of FIG. 5A through an artificial intelligence algorithm based on deep learning previously learned for the chest, a detailed region 510 corresponding to the rib, the lung The detailed region 520 corresponding to , and the detailed region 530 corresponding to the lungs and ribs may be mapped to different classes.

When detecting a specific organ, the controller 260 may determine a class corresponding to a specific organ to be detected from among a plurality of classes.

For example, based on Table 1, when the controller 260 detects a lung, classes 10 to 17 among a plurality of classes may be determined as a class corresponding to the lung.

In this case, the controller 260 may determine a detailed region mapped to classes 10 to 17 corresponding to the lung, respectively, from among the plurality of detailed regions constituting the medical image, and detect the lung based on the determined detailed region .

That is, when all pixels mapped to classes 10 to 17 corresponding to the lung are combined among a plurality of pixels constituting the medical image, the lung may be detected from the medical image.

On the other hand, when detecting a catheter among a plurality of anatomical components included in the medical image, the controller 260 uses an artificial intelligence algorithm based on deep learning previously learned through machine learning to correspond to the catheter in the medical image. A region of interest (ROI) including the object may be determined, and the catheter may be detected based on the determined region of interest.

For example, the controller 260 may determine a portion of the medical image corresponding to an image corresponding to a portion of the catheter stored in the storage 230 as the region of interest.

The controller 260 may check the direction of the object corresponding to the catheter included in the ROI.

For example, when a portion of the medical image corresponding to the first image 401 of FIG. 4 is determined as the region of interest, the controller 260 may change the direction of the object corresponding to the catheter included in the region of interest to the left. You can decide between the top and the bottom right.

For example, when the portion corresponding to the fifth image 405 of FIG. 4 among a portion of the medical image is determined as the region of interest, the controller 260 may change the direction of the object corresponding to the catheter included in the region of interest to the left. You can decide between the bottom and the top right.

The controller 260 may determine another ROI including the object corresponding to the catheter based on the direction of the object corresponding to the catheter included in the ROI.

For example, as shown in FIG. 6 , when the first portion 601 of the medical image is determined as the region of interest corresponding to the first catheter, the controller 260 is included in the first portion 601 determined as the region of interest, The direction of the object corresponding to the catheter may be determined to the lower left.

At this time, the controller 260 determines whether there is a portion corresponding to the image corresponding to a portion of the catheter stored in the storage 230 among the medical images according to the direction of the object corresponding to the catheter included in the region of interest. can be checked repeatedly.

For example, as shown in FIG. 6 , when the first portion 601 of the medical image is determined as the region of interest corresponding to the first catheter, the second portion ( ) of the medical image according to the direction of the object corresponding to the first catheter Each portion extending from 602 to the third portion 603 may be determined as a region of interest corresponding to the first catheter.

Meanwhile, each of the plurality of ROIs determined to include the object corresponding to the catheter may include the object corresponding to the catheter in the center.

In addition, each of the plurality of regions of interest determined to include the object corresponding to the catheter may at least partially overlap each other based on the object corresponding to the catheter.

For example, as shown in FIG. 6 , in the first portion 601 and the second portion 602 of the medical image, regions corresponding to the first catheter may overlap each other.

Meanwhile, when the region of interest includes an object corresponding to the edge of the catheter or another region of interest is no longer identified according to the direction of the object corresponding to the catheter, the controller 260 determines the region of interest as the distal region. can

For example, as in the first portion 601 of FIG. 6 , the region of interest corresponding to the image corresponding to the edge of the catheter may be determined as the distal region.

Meanwhile, the controller 260 may detect the catheter based on the plurality of regions of interest. In this case, the controller 260 may detect the catheter in consideration of whether the plurality of regions of interest overlap each other.

For example, as shown in FIG. 6 , the controller 260 controls the first catheter based on a plurality of regions of interest that at least partially overlap each other from the first portion 601 to the third portion 603 of the medical image. can be detected. In this case, the controller 260 may connect the centers of a plurality of regions of interest corresponding to the first catheter to each other to detect the first catheter.

Meanwhile, when a distal region exists among the plurality of regions of interest, the controller 260 may determine whether another distal region for the same catheter exists within a predetermined distance from the corresponding distal region.

In this case, if there is no other end region within a predetermined distance from the corresponding end region, the controller 260 may determine the corresponding end region as the edge of the catheter.

For example, as shown in FIG. 6 , among the plurality of regions of interest corresponding to the first catheter, in the case of the first portion 601 that is the distal region, there is no other distal region for the first catheter within a predetermined distance. The first portion 601 may be determined as the edge of the first catheter.

Meanwhile, referring to region 610 of FIG. 6 , among a plurality of regions of interest corresponding to the first catheter, in the case of a third part 603 that is an end region, a fourth part that is another end region with respect to the first catheter within a predetermined distance. Since 604 is present, the controller 260 can determine whether the two end regions 603 and 604 are connected.

In this case, the controller 260 may determine whether to connect the two end regions based on the direction of the object corresponding to the catheter included in the two end regions.

As in the area 610 of FIG. 6 , the lower left corner, which is the direction of the object corresponding to the first catheter included in the third part 603 , and the direction of the object corresponding to the first catheter included in the fourth part 604 . When the right upper ends correspond to each other and the first catheters extending in each direction are connected to each other, the controller 260 may connect the two end regions 603 and 604 to each other. In this case, it may be determined that the two distal regions 603 and 604 connected to each other are not edges of the first catheter.

Also, the controller 260 may more accurately detect the first catheter based on the connection relationship between the two distal regions 603 and 604 . For example, when outputting the result of detecting the first catheter through the output unit 250 , the controller 260 may display the centers of the two end regions 603 and 604 by connecting them to each other.

Similarly, the controller 260 may determine the fifth portion 605 of the medical image as the region of interest for the second catheter and the sixth portion 606 of the medical image as the region of interest for the third catheter. can decide Also, the controller 260 may detect the second catheter based on the plurality of regions of interest for the second catheter, and may detect the third catheter based on the plurality of regions of interest for the third catheter.

Meanwhile, when detecting a bone or joint among a plurality of anatomical components included in the medical image, the controller 260 determines a region of interest (ROI) including an object corresponding to the bone or joint in the medical image. Also, a bone or a joint may be detected based on the determined region of interest.

For example, the controller 260 may determine a portion of the medical image corresponding to an image corresponding to a portion of a bone or joint stored in the storage 230 as the region of interest.

The controller 260 may determine information on the detected anatomical configuration from among a plurality of anatomical configurations included in the medical image using X-rays.

When an organ is detected from among a plurality of anatomical components, the controller 260 may determine information on the organ based on the detected shape of the organ.

For example, the controller 260 may calculate a ratio between the horizontal length of the lung and the horizontal length of the heart as information about the heart, based on the shape of the lung and the heart.

For example, the controller 260 may calculate a difference between positions of both diaphragms as information about the diaphragm, based on the shape of the lungs.

For example, the controller 260 may determine the corners and centers of each vertebra based on the shape of the vertebrae, and based on the corners and centers of each vertebra. It is possible to determine information about the vertebrae.

When a catheter is detected among a plurality of anatomical components, the controller 260 may determine information about the catheter based on the detected shape of the catheter.

For example, the controller 260 may determine the position of the edge of the catheter based on the shape of the catheter as information about the catheter.

The controller 260 may display a medical image through the display of the output unit 250 .

At this time, when a user input for selecting any one of the anatomical components included in the medical image displayed through the output unit 250 is received through the input unit 240 , the controller 260 receives information on the selected anatomical component. , may be output through the output unit 250 . In this regard, it will be described with reference to FIGS. 7A to 7F and 8 .

Referring to FIG. 7A , according to a user input received through the input unit 240 , the position of the pointer 700 displayed on the display may be located in the lung.

In this case, the controller 260 may determine the lung as an anatomical configuration selected according to a user input from among the plurality of anatomical configurations, and display the lungs to be distinguished from other anatomical configurations.

For example, as shown in FIG. 7A , a separate dotted line 701 may be displayed according to the shape of the lung, which is an anatomical configuration selected according to a user input.

Also, the controller 260 may display an indicator 702 indicating a costophrenic angle (CPA) as information about the lung, which is an anatomical configuration selected according to a user input.

At this time, when the costal diaphragm angle (CPA) does not fall within the preset normal range, the control unit 260 may output a warning message through the output unit 250 .

For example, when the costal diaphragm angle (CPA) does not fall within a preset normal range, the controller 260 may output a warning message warning that there is a possibility that the pleural fluid may be filled with more than a predetermined level between the pleural spaces. .

Referring to FIG. 7B , according to a user input received through the input unit 240 , the position of the pointer 700 displayed on the display may be located on the keel.

In this case, the controller 260 may determine the keel as an anatomical configuration selected according to a user input, among a plurality of anatomical configurations, and display the keel to be distinguished from other anatomical configurations.

Also, the controller 260 may display indicators 703 and 704 indicating a tracheal deviation of the airway as information about the keel, which is an anatomical configuration selected according to a user input. In this case, when the degree of airway bias does not fall within the preset normal range, the controller 260 may output a warning message through the output unit 250 .

For example, when the degree of airway bias does not fall within a preset normal range, the controller 260 may output a warning message warning of the possibility of compression by a tension pneumothorax or a tumor. .

Referring to FIG. 7C , according to a user input received through the input unit 240 , the position of the pointer 700 displayed on the display may be located on the diaphragm.

In this case, the controller 260 may determine the diaphragm as an anatomical configuration selected according to a user input among a plurality of anatomical configurations, and display the diaphragm to be distinguished from other anatomical configurations.

For example, as shown in FIG. 7C , separate dotted lines 705 and 706 may be displayed according to the shape of the diaphragm, which is an anatomical configuration selected according to a user input.

Also, the controller 260 may display an indicator (not shown) indicating a difference between positions of the diaphragms on both sides as information on the diaphragm, which is an anatomical configuration selected according to a user input. In this case, when the left diaphragm is located above the right diaphragm, the controller 260 may output a warning message through the output unit 250 .

For example, when the left diaphragm is located above the right diaphragm, the control unit 260 may output a warning message warning the possibility of gas (gas) filling the stomach or paralysis of the pleural nerve. can

Referring to FIG. 7D , according to a user input received through the input unit 240 , the position of the pointer 700 displayed on the display may be located in the heart.

In this case, the controller 260 may determine the heart from among the plurality of anatomical components as an anatomical component selected according to a user input, and display the heart to be distinguished from other anatomical components.

For example, as shown in FIG. 7D , the color of the heart, which is an anatomical component selected according to a user input, may be displayed differently from the colors of other anatomical components.

Also, the controller 260 may display indicators 707 and 708 indicating a difference between the horizontal length of the lung and the horizontal length of the heart as information about the heart, which is an anatomical configuration selected according to a user input. In this case, when the horizontal length of the heart is more than half the horizontal length of the lungs, the controller 260 may output a warning message through the output unit 250 .

For example, when the horizontal length of the heart is more than half the horizontal length of the lungs, the controller 260 may output a warning message warning of the possibility of cardiac hypertrophy (cardiomegaly).

Referring to FIG. 7E , according to a user input received through the input unit 240 , the position of the pointer 700 displayed on the display may be located in the vertebrae.

In this case, the control unit 260 may determine a vertebra among a plurality of anatomical components as an anatomical component selected according to a user input, and display the vertebrae to be distinguished from other anatomical components.

For example, as shown in FIG. 7E , a separate solid line 709 may be displayed according to the shape of the vertebra, which is an anatomical configuration selected according to a user input.

Also, the controller 260 may display an indicator 710 indicating the center, inclination, etc. of each of the vertebrae as information about the vertebra, which is an anatomical configuration selected according to a user input. The controller 260 may display an indicator (not shown) indicating a disc between vertebrae.

In this case, the controller 260 may output a warning message through the output unit 250 when the position difference between the centers of the vertebrae is equal to or greater than a predetermined criterion.

For example, when the position difference between the centers of the vertebrae is greater than or equal to a predetermined criterion, the controller 260 may output a warning message warning of the possibility of scoliosis.

Referring to FIG. 7F , according to a user input received through the input unit 240 , the position of the pointer 700 displayed on the display may be located in the joint.

In this case, the controller 260 may determine a joint from among the plurality of anatomical configurations as an anatomical configuration selected according to a user input, and display the joint to be distinguished from other anatomical configurations.

Also, the controller 260 may display an indicator 711 indicating a keypoint of the joint as information about a joint, which is an anatomical configuration, selected according to a user input. The control unit 260 may display information about a distance between feature points of a joint, an average of distances, a deviation, and the like.

In this case, the controller 260 may output a warning message through the output unit 250 when the average of the distances between feature points of the joint is less than a predetermined standard.

For example, when the average of the distances between feature points of the joints is less than a predetermined criterion, the controller 260 may output a warning message warning the possibility of arthritis.

Meanwhile, referring to FIG. 8 , according to a user input received through the input unit 240 , the position of the pointer 700 displayed on the display may be located in any one of the catheters.

In this case, the controller 260 may determine the catheter as an anatomical configuration selected according to a user input, among a plurality of anatomical configurations, and display the catheter to be distinguished from other anatomical configurations.

In addition, the control unit 260 may display an indicator that emphasizes the edge of the specific catheter as information about the specific catheter selected according to the user input.

For example, as shown in FIG. 8 , when a plurality of catheters 810 , 820 , and 830 are included in a medical image, colors of the plurality of catheters 810 , 820 , and 830 may be displayed differently from each other. Also, when the pointer 700 is positioned at a peripherally inserted central catheter (PICC) 820 , an indicator 825 indicating an edge of the PICC 820 may be displayed.

As described above, according to various embodiments of the present disclosure, even when regions corresponding to a plurality of anatomical configurations included in a medical image overlap each other, it is possible to accurately detect a plurality of anatomical configurations, processing performance can be improved.

9 to 11 are flowcharts of a method of operating a medical image processing apparatus according to an embodiment of the present invention.

Referring to FIG. 9 , in operation S901 , the medical image processing apparatus 200 may acquire a medical image and check the acquired medical image.

For example, the medical image processing apparatus 200 may acquire a medical image using X-rays through the image acquisition unit 210 and/or the communication unit 220 .

For example, the medical image processing apparatus 200 may determine a body position corresponding to the acquired medical image, and may determine a plurality of organs or catheters included in the corresponding medical image according to the determined body part. .

The medical image processing apparatus 200 may determine whether to detect an organ among a plurality of anatomical components included in the medical image in operation S902 .

For example, the medical image processing apparatus 200 may output, through the output unit 250 , a first guide screen guiding to determine whether an organ is detected and at least one organ to be detected, and the input unit According to a user input received through 240 , an institution to be detected may be determined.

When it is determined not to detect an organ in operation S903 , the medical image processing apparatus 200 may determine whether to detect a catheter.

For example, the medical image processing apparatus 200 may output, through the output unit 250 , a second guide screen guiding whether a catheter is detected and at least one catheter to be detected, and the input unit 240 . A catheter to be detected may be determined according to a user input received through the .

In this case, the medical image processing apparatus 200 may display the first guide screen and the second guide screen simultaneously or sequentially.

When it is determined that neither the organ nor the catheter are detected in operation S904 , the medical image processing apparatus 200 may output the acquired medical image as it is through the output unit 250 .

Meanwhile, in operation S905 , when it is determined to detect at least one organ among a plurality of anatomical components included in the medical image, the medical image processing apparatus 200 may detect at least one of the plurality of organs included in the medical image to be detected. of organs can be detected, and information about the detected organs can be determined. In this regard, it will be described with reference to FIG. 10 .

Referring to FIG. 10 , in operation S1010 , the medical image processing apparatus 200 may determine a plurality of classes corresponding to at least one of organs included in the medical image.

For example, among the plurality of classes, a first class may correspond to a first institution, a second class may correspond to a second institution, and a third class may correspond to both the first institution and the second institution.

In operation S1020 , the medical image processing apparatus 200 may map each detailed region of the medical image to any one of a plurality of classes by using a deep learning-based artificial intelligence algorithm previously learned through machine learning. .

For example, the medical image processing apparatus 200 maps a detailed region corresponding only to a first institution among a plurality of detailed regions of a medical image to a first class, and a detailed region corresponding only to a second institution is a second class , and a subregion corresponding to both the first organization and the second organization may be mapped to the third class.

In operation S1030 , the medical image processing apparatus 200 may determine at least one class corresponding to a specific organ to be detected from among a plurality of classes.

In operation S1040 , the medical image processing apparatus 200 may determine a detailed region mapped to at least one class determined in operation S1030 from among a plurality of detailed regions constituting a medical image.

The medical image processing apparatus 200 may detect an organ to be detected according to the detailed area determined in operation S1040 in operation S1050.

For example, the medical image processing apparatus 200 may detect an organ to be detected from the medical image by combining all pixels mapped to a class corresponding to an organ to be detected among a plurality of pixels constituting a medical image. have.

The medical image processing apparatus 200 may determine information on the detected organ in operation S1060 .

For example, the medical image processing apparatus 200 may determine information about a corresponding organ based on the detected shape of the organ.

Referring back to FIG. 9 , the medical image processing apparatus 200 may determine whether to detect a catheter in operation S906 .

When it is determined in operation S907 to detect the catheter in operation S903 or S906, the medical image processing apparatus 200 may detect a catheter included in the medical image and determine information on the detected catheter. In this regard, it will be described with reference to FIG.

Referring to FIG. 11 , in operation S1110 , the medical image processing apparatus 200 uses an artificial intelligence algorithm based on deep learning previously learned through machine learning, the region of interest including the object corresponding to the catheter in the medical image. It can be determined whether this exists or not.

For example, the medical image processing apparatus 200 may detect a region of interest including an object corresponding to a catheter in a medical image based on a database of images corresponding to a portion of various catheters stored in the storage 230 . It can be determined whether it exists or not.

In operation S1120 , the medical image processing apparatus 200 may determine whether the ROI corresponds to the distal region when the ROI exists.

For example, when the ROI includes an object corresponding to the edge of the catheter or another ROI is no longer identified according to the direction of the object corresponding to the catheter, the medical image processing apparatus 200 may detect the ROI. can be determined as the terminal region.

In operation S1130 , when the ROI does not correspond to the distal region, the medical image processing apparatus 200 performs another operation including the object corresponding to the catheter based on the direction of the object corresponding to the catheter included in the ROI. A region of interest can be determined.

Meanwhile, in operation S1140 , when the ROI corresponds to the distal region, the medical image processing apparatus 200 may determine whether another distal region for the same catheter exists within a predetermined distance from the ROI.

In operation S1150 , when another distal region does not exist within a predetermined distance from the ROI, the medical image processing apparatus 200 may determine the ROI as the edge of the catheter.

Meanwhile, in operation S1160 , when another end region exists within a predetermined distance from the corresponding region of interest, the medical image processing apparatus 200 may determine whether the other end region is a connectable target with the corresponding region of interest.

For example, the medical image processing apparatus 200 may determine whether to connect the two distal regions based on a direction of an object corresponding to a catheter included in the two distal regions.

In operation S1150 , if the other end region is not an object connectable to the region of interest, the medical image processing apparatus 200 may determine the region of interest as the edge of the catheter and branch to operation S1110 to determine the presence of the region of interest. It can be repeated to check whether or not

Meanwhile, in operation S1160 , if the other end region is a connectable target with the corresponding region of interest, the medical image processing apparatus 200 may connect the corresponding region of interest and the other end region to each other, and branch in operation S1110 to The check for existence can be repeated.

In this case, the region of interest, which is the two distal regions connected to each other, may be determined not to be the edge of the catheter.

Meanwhile, the medical image processing apparatus 200 may determine the position of the edge of the catheter as information about the catheter based on the shape of the catheter detected in operations S1110 to 1160 .

Referring again to FIG. 9 , the medical image processing apparatus 200 may output a medical image through the output unit 250 in operation S908 .

In this case, the medical image processing apparatus 200 may output information on the anatomical configuration detected in operation S905 and/or operation S907 (eg, a list of detected organs, a list of catheters, etc.). may be

In operation S909 , the medical image processing apparatus 200 may determine whether a user input for selecting one of organs and/or catheters included in the displayed medical image output through the output unit 250 is received.

In operation S910 , the medical image processing apparatus 200 may output information on an anatomical configuration selected according to a user input through the output unit 250 .

In operation S911 , the medical image processing apparatus 200 may determine whether the display of the medical image is terminated.

For example, when a user input terminating the display of the medical image is received through the input unit 240 , the medical image processing apparatus 200 may end the display of the medical image.

The accompanying drawings are only for easy understanding of the embodiments disclosed in the present specification, and the technical spirit disclosed herein is not limited by the accompanying drawings, and all changes and equivalents included in the spirit and scope of the present invention It should be understood to include water or substitutes.

Meanwhile, the method of operating an image processing apparatus of the present invention can be implemented as processor-readable codes on a processor-readable recording medium provided in the image processing apparatus. The processor-readable recording medium includes all types of recording devices in which data readable by the processor is stored. Examples of the processor-readable recording medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc., and also includes those implemented in the form of carrier waves such as transmission over the Internet . In addition, the processor-readable recording medium is distributed in a computer system connected through a network, so that the processor-readable code can be stored and executed in a distributed manner.

In addition, although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above, and the technical field to which the present invention belongs without departing from the gist of the present invention as claimed in the claims In addition, various modifications may be made by those of ordinary skill in the art, and these modifications should not be individually understood from the technical spirit or perspective of the present invention.

Claims (10)

A medical image processing apparatus comprising:
an image acquisition unit for acquiring a medical image using X-rays;
display; and
including a control unit;
The control unit is
Each of a plurality of detailed regions constituting the medical image is mapped to any one of a plurality of classes corresponding to at least one of a plurality of organs by using a pre-trained deep learning-based artificial intelligence algorithm (mapping),
detecting an organ to be detected from among a plurality of organs included in the medical image based on the mapping result;
Through the display, the detection target organ is displayed,
Among the plurality of classes, a first class corresponds to a first institution, a second class corresponds to a second institution, and a third class corresponds to both the first institution and the second institution. image processing device. According to claim 1,
further comprising an input unit,
The control unit is
When a user input for selecting any one of the plurality of organs as the detection target is received through the input unit, the organ selected as the detection target according to the user input is displayed separately from other organs,
The medical image processing apparatus of claim 1, wherein information on the organ selected as the detection target according to the user input is displayed on the display. delete According to claim 1,
The control unit is
determining a body part corresponding to the medical image,
and determining a plurality of organs included in the medical image according to the determined body part. According to claim 1,
The control unit is
Among the plurality of sub-regions, a sub-region corresponding only to a first institution is mapped to the first class, a sub-region corresponding only to a second institution is mapped to the second class, and the first institution and the second institution The medical image processing apparatus according to claim 1, wherein the sub-regions corresponding to all are mapped to the third class. According to claim 1,
The control unit is
Determining at least one class corresponding to the organ to be detected from among the plurality of classes,
determining at least one subregion mapped to the determined class from among the plurality of subregions;
The medical image processing apparatus of claim 1, wherein the organ to be detected is detected based on the determined detailed area. A medical image processing apparatus comprising:
an image acquisition unit for acquiring a medical image using X-rays;
display; and
including a control unit,
The control unit is
Determining a region of interest (ROI) including an object corresponding to a catheter in the medical image by using a pre-trained deep learning-based artificial intelligence algorithm,
Based on the direction of the object corresponding to the catheter included in the determined region of interest, determining another region of interest including the object corresponding to the catheter,
If there is a first distal region including an object corresponding to the edge of the catheter among the plurality of regions of interest, a second distal region including the object corresponding to the edge of the catheter within a predetermined distance from the first distal region to determine whether it exists,
determining a region of interest corresponding to an edge of the catheter among the plurality of regions of interest based on the determination result;
and detecting the catheter based on a plurality of ROIs determined to include an object corresponding to the catheter. 8. The method of claim 7,
Each of the plurality of regions of interest,
Containing an object corresponding to the catheter in the center,
Based on the object corresponding to the catheter, at least partially overlapping each other,
The control unit is
The medical image processing apparatus of claim 1, wherein the catheter is detected in consideration of whether the plurality of regions of interest overlap each other. 8. The method of claim 7,
The control unit is
and when the second end region does not exist within the predetermined distance from the first end region, the first end region is determined as a region of interest corresponding to an edge of the catheter. 8. The method of claim 7,
The control unit is
When the second end region is within the predetermined distance from the first end region, based on the orientation of the object corresponding to the edge of the catheter included in the first end region and the second end region, respectively, the and determining whether the first end region and the second end region are connected.

Images