Medical image segmentation method and apparatus for performing the same

Abstract

The present invention is a method performed by a processor of a medical image segmentation device, comprising the steps of: acquiring a medical image of a subject; inputting the medical image into a first prediction model trained to predict any one target region using the medical image as input, and determining a first region corresponding to any one target region; generating a plurality of sub-volume data using the first region; inputting the plurality of sub-volume data into a prediction model trained to predict any one target region using a 3D medical image as input, and determining a second region corresponding to the target region within the first region using the plurality of sub-volume data; and providing the second region corresponding to the target region within the medical image.

Description

[Technical Field]

[0001] The present invention relates to a method for segmenting a medical image and an apparatus for performing the same. [Background technology]

[0002] Generally, to determine the presence or absence of a disease in a target area of ​​a subject, medical images of the target area of ​​the subject taken through imaging medical examinations (e.g., X-ray, ultrasound, CT (Computer Tomography), Angiography, Positron Emission Computed Tomography (PET-CT), Single Photon Emission Computed Tomography (SPECT-CT), Magnetic Resonance Imaging (MRI), etc.) are used. Medical staff have performed diagnoses by visually checking the medical images to identify the target area and determining the presence or absence of a disease in the target area (e.g., the presence or absence of a tumor).

[0003] However, noise may exist in the medical image itself due to the performance of the imaging device, the movement of the patient, etc. If the target region (e.g., an organ or a tumor in an organ) in the medical image is identified based solely on the medical staff's findings, differences in opinion may occur depending on the medical staff's skills and experience, even for the same medical image.

[0004] Therefore, a method of predicting a target region from a medical image using an artificial neural network model that has been trained to predict a target region based on the medical image has been widely used. However, the artificial neural network model is configured to predict the area where the 3D target area is located using 2D slice-based medical images, and there is a problem that the accuracy and reliability of the prediction results are low for some slices.

[0005] The Background of the Invention has been prepared to facilitate a better understanding of the present invention and should not be construed as an admission that the matter described in the Background of the Invention exists as prior art. Summary of the Invention [Problem to be solved by the invention]

[0006] Previously, models have been disclosed for segmenting only target areas using 3D medical images, but the units of 3D learning data are small, and the information that can be learned is limited. Furthermore, when learning various types of target regions, the amount of data required for learning becomes excessively large, which limits the learning.

[0007] Therefore, there is a need for a method that can minimize the amount of calculations required for the artificial neural network model while improving the accuracy and reliability of the prediction results so as to eliminate the uncertainty of the artificial neural network model.

[0008] As a result, the inventors of the present invention have developed a prediction model that can quickly recognize the presence or absence and type of target area and accurately recognize the specific location for each of the multiple slices that make up a medical image, and a medical image segmentation method using the same.

[0009] The objects of the present invention are not limited to those mentioned above, and other objects not mentioned above will be clearly understood by those skilled in the art from the following description. [Means for solving the problem]

[0010] In order to solve the above-mentioned problems, a medical image segmentation method according to one embodiment of the present invention is provided, the method being performed by a processor of a medical image segmentation device and including the steps of: acquiring a medical image of a subject; inputting the medical image into a first prediction model trained to predict any one target region using the medical image as an input, thereby determining a first region corresponding to any one target region; generating a plurality of sub-volume data using the first region; inputting the plurality of sub-volume data into a prediction model trained to predict any one target region using a 3D medical image as an input, thereby determining a second region corresponding to the target region within the first region from the plurality of sub-volume data; and providing the second region corresponding to the target region within the medical image.

[0011] According to a feature of the present invention, the step of determining the second region may be a step of determining a plurality of second regions corresponding to two or more different target sites depending on the type of target site corresponding to the first region.

[0012] According to another aspect of the present invention, the sub-volume data may include data on a movement direction of voxels constituting the sub-volume data with respect to any one axis.

[0013] According to another aspect of the present invention, the step of generating the sub-volume data may include stacking a plurality of slices constituting the medical image to match a pre-stored height, and dividing the slices in a direction perpendicular to a plane of the slices to generate a plurality of sub-volume data.

[0014] According to another feature of the present invention, before generating the sub-volume data, the method may further include the steps of inputting the medical image into a classification model trained to classify the type of target region using the medical image as input, determining the type of target region included in each of a plurality of slices constituting the medical image, and grouping the plurality of slices by target region based on the classification result.

[0015] According to another feature of the present invention, the step of providing an area corresponding to the target region may be a step of providing a user interface screen including a segmentation model selection area for selecting the target region to be segmented in the medical image and a segmentation result display area for displaying the area corresponding to the target region.

[0016] According to another aspect of the present invention, the step of providing the region corresponding to the target region may further include a step of combining volume data in which a probability value corresponding to the target region in each of the plurality of sub-volume data is equal to or greater than a predetermined value using the prediction model, and displaying the combined volume data on the user interface screen.

[0017] According to another feature of the present invention, the step of providing a region corresponding to the target region may further include a step of displaying each of the plurality of sub-volume data on the user interface screen with different transparency levels depending on a probability value corresponding to the target region using the prediction model.

[0018] According to another feature of the present invention, the step of generating sub-volume data may be generated based on an axis including any one of a horizontal plane (Axial), a coronal plane (Coronal), and a sagittal plane, and the step of providing a region corresponding to the target region may further include the steps of acquiring, through the user interface screen, any one of a horizontal plane (Axial), a coronal plane (Coronal), and a sagittal plane for displaying the target region, and rendering the region corresponding to the target region based on the reference plane.

[0019] According to another aspect of the present invention, the method may further include, before the step of acquiring the medical image, acquiring training data having different slice thicknesses according to the type of medical image, generating a training dataset by performing bootstrapping a different number of times depending on the thickness of the training data, and generating the prediction model configured to predict any one target region based on the training dataset.

[0020] In order to solve the above-mentioned problems, another embodiment of the present invention provides a medical image segmentation device that acquires a medical image of a subject, inputs the medical image into a first prediction model trained to predict any one target region using the medical image as an input, determines a first region corresponding to the one target region, generates a plurality of sub-volume data using the first region, inputs the plurality of sub-volume data into a prediction model trained to predict any one target region using a 3D medical image as an input, determines a second region corresponding to the target region within the first region from the plurality of sub-volume data, and provides the second region corresponding to the target region within the medical image.

[0021] Further details of the embodiments are included in the detailed description and drawings. [Effects of the Invention]

[0022] By using multiple predictive models to predict target areas, the present invention can expand the range of information that can be learned compared to using a single predictive model that uses cube-shaped 3D medical images as input, while still improving learning efficiency by predicting target areas using only minimal learning data.

[0023] The present invention improves the prediction accuracy of a prediction model by inputting not only 2D images constituting a medical image but also before and after image information of sequentially captured 2D images into the prediction model for predicting a target region. In particular, the present invention improves prediction accuracy over the entire region of a 2D image by providing data on pixel changes (e.g., the movement direction of any one specified pixel) based on multiple pixels constituting the 2D images constituting a medical image.

[0024] The present invention linearly classifies medical images using a trained classification model before predicting the target area, thereby providing meaningful results that do not lean in one direction when using a prediction model.

[0025] The present invention can accurately predict the location of a target area in a medical image and visually display it to help medical staff diagnose the target area. The effects of the present invention are not limited to the above-mentioned examples, and various other effects are included within the scope of the present invention. [Brief explanation of the drawings]

[0026] [Figure 1] 1 is a block diagram showing the configuration of a medical image division system according to an embodiment of the present invention; [Figure 2] 1 is a block diagram showing the configuration of a medical staff device according to an embodiment of the present invention; [Figure 3] 1 is a block diagram showing the configuration of a medical image dividing device according to an embodiment of the present invention; [Figure 4] 1 is a schematic flowchart of a medical image segmentation method according to an embodiment of the present invention. [Figure 5] 1 is a schematic diagram illustrating a medical image segmentation method according to an embodiment of the present invention; [Figure 6a] 10 is a diagram illustrating an example of a user interface screen showing a medical image segmentation result according to an embodiment of the present invention. [Figure 6b]10 is a diagram illustrating an example of a user interface screen showing a medical image segmentation result according to an embodiment of the present invention. [Figure 7] 10 is an exemplary diagram of a user interface screen for using a prediction model for medical image segmentation according to an embodiment of the present invention; [Figure 8] FIG. 1 is a schematic diagram illustrating a method for learning a prediction model according to an embodiment of the present invention. [Figure 9] 1 is a schematic diagram illustrating an operation method of a prediction model for segmenting medical images according to an embodiment of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0027] The advantages and features of the present invention, and methods for achieving them, will become more apparent from the following detailed description of the embodiments in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, and may be embodied in various different forms. These embodiments are provided solely to ensure that the disclosure of the present invention is complete and to fully convey the scope of the invention to those skilled in the art to which the present invention pertains. The present invention is defined only by the scope of the claims. In connection with the description of the drawings, like reference numerals may be used to refer to like elements.

[0028] In this document, the terms "have," "can have," "include," or "can include" refer to the presence of a given feature (e.g., a value, function, operation, or component such as a part) and do not exclude the presence of additional features.

[0029] In this document, expressions such as "A or B," "at least one of A and / or B," or "one or more of A and / or B" include all possible combinations of the items listed together. For example, "A or B," "at least one of A and B," or "at least one of A or B" can refer to (1) at least one A, (2) at least one B, or (3) both at least one A and at least one B.

[0030] Terms such as "first," "second," "first," or "second" used herein may modify various components regardless of order and / or importance, and are used only to distinguish one component from another and do not limit the corresponding component. For example, a first user device and a second user device may refer to different user devices regardless of order or importance. For example, a first component may be referred to as a second component, and similarly, a second component may be referred to as a first component, without departing from the scope of the rights described herein.

[0031] When a component (e.g., a first component) is referred to as being "operatively or communicatively coupled with" or "connected to" another component (e.g., a second component), it should be understood that the component may be directly coupled to the other component or may be coupled through another component (e.g., a third component). In contrast, when a component (e.g., a first component) is referred to as being "directly coupled with" or "directly connected to" another component (e.g., a second component), it should be understood that there is no other component (e.g., a third component) between the component and the other component.

[0032] As used herein, the phrase "configured to" may be used in various ways, depending on the context, such as "suitable for," "having the capacity to," "designed to," "adapted to," "made to," or "capable of." The term "configured to" does not necessarily refer to hardware that is "specifically designed to." Instead, in some contexts, the phrase "apparatus configured to" may mean that the apparatus, together with other devices or components, is "capable of." For example, the phrase "a processor configured to perform A, B, and C" may refer to a dedicated processor (e.g., an embedded processor) for performing the operations, or to a general-purpose processor (e.g., a CPU or application processor) that can perform the operations by executing one or more software programs stored in a memory device.

[0033] The terms used in this document are merely used to describe particular embodiments and may not be intended to limit the scope of other embodiments. A singular expression may include a plural expression unless the context clearly dictates otherwise. Terms used herein, including technical or scientific terms, may have the same meaning as commonly understood by a person of ordinary skill in the art described in this document. Terms used in this document that are defined in a general dictionary may be interpreted to have the same or similar meaning as the meaning they have in the context of the relevant art, and unless explicitly defined in this document, they should not be interpreted in an idealized or overly formal sense. In some cases, even terms defined in this document may not be interpreted to exclude embodiments of this document.

[0034] The features of the various embodiments of the present invention may be partially or fully combined or combined with each other, and various technical interlocking and driving mechanisms are possible, as will be fully understood by those skilled in the art. Each embodiment may be implemented independently of the others, or may be implemented together in a related relationship.

[0035] For clarity of interpretation of this specification, the following defines terms used in this specification. The term "medical image" as used herein may refer to a two-dimensional image composed of multiple cuts (or slices) of a subject. Specifically, the medical image may be an enhanced or non-enhanced CT image generated in accordance with the DICOM standard. For example, the medical image may include a head and neck image including the entire area from the skull vertex to the lung apex, a chest image including the entire area from the thyroid to the liver dome, an abdominal image including the entire L1 spine at a position 3 cm away from the liver dome in the cranial direction, and a pelvic image including the entire ischium at a position 3 cm away from the L1 spine in the cranial direction.

[0036] The term "first prediction model" as used herein may refer to a model trained to predict a region corresponding to any one target region using a medical image as input. Specifically, the first prediction model in this specification may refer to a model trained to predict a first region corresponding to a target region in a slice constituting a medical image. For example, the first prediction model may be a model trained to output a region in each slice where a target region such as the head, neck, chest, abdomen, or pelvis is predicted to be located in a box form.

[0037] The term "second prediction model" as used herein may refer to a model trained to predict a region corresponding to one target region using a 3D medical image as input. Specifically, the second prediction model may be a model trained to predict whether each 3D slab or cube corresponds to a target region. For example, the second prediction model may be a model trained to predict whether each of a plurality of sub-volume data corresponds to a second region corresponding to a target region such as the head, neck, chest, abdomen, or pelvis. Meanwhile, the second region predicted by the second prediction model may be included in the first region, and the second region may include multiple regions depending on the type of target region predicted in the first region. As another example, the second prediction model may be a model trained to predict whether each of a plurality of sub-volume data corresponds to one region of a plurality of organs arranged sequentially in a part of a subject's body. As another example, the second prediction model may be a model trained to predict whether each of the plurality of volume data corresponds to a region of an organ, bone, or muscle that is suspected to be diseased, where the suspected region may be a region that is suspected to be a tumor, and the medical images may be medical images of the head and neck, chest, abdomen, or pelvis.

[0038] In various embodiments, the first and second prediction models may be composed of at least one model or two or more ensemble models selected from the group consisting of VGG net, R, DenseNet based on CNN (Convolutional Neural Network), FCN (Fully Convolutional Network) having an encoder-decoder structure, DNN (deep neural network) such as SegNet, DeconvNet, DeepLAB V3+, U-net, SqueezeNet, Alexnet, ResNet18, MobileNet-v2, GoogLeNet, Resnet50, Resnet101, and Inception-v3.

[0039] In various embodiments, the second prediction model may be a model trained using training data having different slice thicknesses depending on the type of medical image. In the present invention, the slices may be resampled to unify the slice thicknesses based on a reference value, without performing a preprocessing process of increasing or decreasing the slice thickness. Specifically, the second prediction model uses 256x256x16 slab data or 96x96x96 cubic data as subvolume data. The second prediction model can be trained without image loss of the medical image by performing different numbers of bootstrappings depending on the slice thicknesses input to the training data. For example, if slices having a thickness of 3 mm are input in the initial training, the second prediction model may stack the slices to generate subvolume data in 48 mm units and perform training. If slices having a thickness of 5 mm are input in the next initial training, the second prediction model may stack the slices to generate subvolume data in 90 mm units and perform training. This type of training may be performed a specified number of times (epochs).

[0040] The term "classification model" as used herein may refer to a classification model trained to classify types of target regions using medical images as input. Specifically, the classification model may be a model trained to classify at least one target region among the head, neck, chest, abdomen, and pelvis, thereby grouping sequentially acquired slices by target region type (e.g., head, neck, chest, abdomen, pelvis).

[0041] In various embodiments, by using a classification model before predicting the presence or absence of a target region in volume data using the first and second prediction models, computational efficiency can be improved compared to using the first and second prediction models alone.

[0042] The present invention will now be described in detail by describing preferred embodiments of the present invention with reference to the accompanying drawings. FIG. 1 is a block diagram showing the configuration of a medical image division system according to an embodiment of the present invention.

[0043] 1, the medical image segmentation system 1000 may be a system configured to segment only a region of interest in a medical image. Here, segmenting a region of interest can be understood as displaying a region corresponding to the region of interest in a medical image. To this end, the medical image segmentation system 1000 may include an image capturing device 100 that captures a part of a subject's body, a medical staff device 200 that identifies the region of interest and diagnoses the subject, and a medical image segmentation device 300 that segments the region of interest in the medical image.

[0044] The imaging device 100 is a device capable of obtaining medical images of a target region of a subject 10, such as a human or animal, and may include a cylindrical bore 110 into which the subject 10 is carried and a transport device 130 on which the subject 10 is seated and which carries the subject 10 inside. Here, the target region may include various organs, bones, muscles, etc. of the subject 10, such as the brain, neck, chest, abdomen, pelvis, etc. The imaging device 100 can obtain medical images of the subject 10 by irradiating the subject with X-rays capable of projecting the subject 10.

[0045] In various embodiments, the imaging device 100 can acquire a gray scale or RGB two-dimensional image, a single still image, a video consisting of multiple cuts, etc. as a medical image for predicting the area corresponding to the target area.

[0046] The medical staff device 200 may be a device that can send a medical image division request to the medical image division device 300 and display the medical image division result. For example, the medical staff device 200 may include a smartphone, a tablet PC (Personal Computer), a notebook computer, a PC, etc.

[0047] In various embodiments, the medical staff device 200 may display, through a user interface screen, the medical image segmentation results provided by the medical image segmentation device 300. For example, the medical image segmentation results may be provided in a manner in which regions corresponding to target regions in each of a plurality of slices based on any one plane are highlighted in different colors.

[0048] In various embodiments, the medical staff device 200 is configured to be able to install or execute a web or mobile application or program provided by the medical image splitting device 300, and can perform a series of medical image splitting methods performed by the medical image splitting device 300 described below.

[0049] The medical image segmentation device 300 can acquire a medical image of a subject from the image capture device 100 and predict a region corresponding to a target region based on the acquired medical image. To this end, the medical image segmentation device 300 may include a general-purpose computer, laptop, data server, etc. capable of performing deep learning on and analyzing medical images.

[0050] In various embodiments, the medical image segmentation device 300 may input a plurality of slices constituting the medical image to a first prediction model trained to predict any one target region using the medical image as an input, and determine a first region corresponding to the target region in each of the plurality of slices. For example, the medical image segmentation device 300 may output a first region corresponding to a target region such as the brain, neck, chest, abdomen, or pelvis in a box form for each slice.

[0051] In the present invention, the medical image segmentation device 300 can predict a region corresponding to the target region in each of a plurality of slices constituting a medical image, as well as generate a plurality of subvolume data using the plurality of slices, and predict a region corresponding to the target region based on the subvolume data. Here, the subvolume data can be generated by stacking a plurality of slices constituting a medical image to match the pre-stored height and dividing the slices in a direction perpendicular to the plane of the slices. For example, the subvolume data can be slab data of 256×256×16 size, or as another example, cubic data of 96×96×96 size. In addition, voxels constituting the subvolume data each have a size of 2×2×3 mm. 3 , 0.7×0.7×1.0mm 3 The size of the voxel can be adjusted to 0.7 x 0.7 x 1.0 mm. 3 From 2 x 2 x 3 mm 3 The size can vary within the range of

[0052] The medical image segmentation device 300 can sequentially utilize two types of prediction models, one using a 2D image as input and the other using a 3D image as input. The medical image segmentation device 300 of the present invention can improve prediction accuracy compared to predicting a target region using only a prediction model using slices as input, and can expand the diversity of prediction results compared to predicting a target region using only a prediction model using 3D volume data as input.

[0053] In the present invention, the plurality of subvolume data includes data on the movement direction of voxels constituting the subvolume data based on any one axis, thereby improving the accuracy of the prediction result compared to predicting the target region only in slice units. For example, if the subvolume data is slab data, each subvolume data may include data on the movement direction of voxels in a direction perpendicular to a horizontal plane (axial). As another example, if the subvolume data is cubic data, each subvolume data may include data on the movement direction of voxels in at least one of a horizontal plane (axial), a coronal plane, and a sagittal plane.

[0054] In various embodiments, the medical image segmentation device 300 may input a plurality of subvolume data to a second prediction model trained to predict any one target region using a 3D medical image as input, and determine a second region corresponding to the target region in the plurality of subvolume data. In other words, the medical image segmentation device 300 may predict whether each of the plurality of subvolume data corresponds to the target region. For example, the medical image segmentation device 300 may predict whether each of the plurality of subvolume data corresponds to the target region, such as the brain, neck, chest, abdomen, or pelvis.

[0055] To this end, the medical image segmentation device 300 may include multiple second prediction models capable of predicting specific regions of the head, neck, chest, abdomen, and pelvis. The medical image segmentation device 300 may use any one of the multiple second prediction models to match the region predicted through the first prediction model. That is, the medical image segmentation device 300 may determine a first region representing a general outline of the target region through the first prediction model and determine a second region representing the specific shape of the target region through the second prediction model. As a result, the first and second regions determined by the first and second prediction models may correspond to the same target region. However, depending on the location of the organ within the body, the first and second regions may correspond to different target regions, and multiple second regions may be included within the first region. For example, the first region may be the lung, and the multiple second regions may be the left and right lungs.

[0056] In various embodiments, the medical image segmentation device 300 can predict target regions within the neck, such as the thyroid gland, oral cavity, mandible, left submandibular gland, right submandibular gland, pharynx, larynx, left parotid gland, right parotid gland, left temporomandibular joint, right temporomandibular joint, left brachial plexus, and right brachial plexus, through the first and second prediction models. As another example, the medical image segmentation device 300 can predict target regions such as the thyroid gland, left lung, right lung, heart, left humerus, right humerus, trachea, left bronchus, right bronchus, left brachial plexus, and right brachial plexus within the chest through the first and second prediction models. As another example, the medical image segmentation device 300 can predict target regions such as the liver, pancreas, gallbladder, spleen, left kidney, and right kidney within the abdomen through the first and second prediction models. As another example, the medical image segmentation device 300 can predict target regions corresponding to the spinal cord, esophagus, stomach, duodenum, cauda equina, and bowel within consecutive organs through the first and second prediction models.

[0057] In various embodiments, the medical image segmentation device 300 may input a medical image of a subject to a classification model trained to classify the type of target region from the medical image, and determine the type of target region included in each of a plurality of slices constituting the medical image. For example, the classification model may determine whether a slice includes one of the target regions, including the head, neck, chest, abdomen, pelvis, optics, heart, breast, and bowels. The medical image segmentation device 300 may group the plurality of slices by target region based on the classification result.

[0058] In various embodiments, the medical image segmentation device 300 may combine predicted results based on multiple subvolume data, display a region corresponding to the target region in the medical image, and provide the display to the medical staff device 200. For example, the medical image segmentation device 300 may provide the medical staff device 200 with a user interface screen including a segmentation result display area displaying the region corresponding to the target region. Here, the user interface screen may include a segmentation model selection area that allows the medical staff to select the target region to be segmented in the medical image, thereby highlighting only the specific target region that the medical staff wants to confirm in the medical image. In addition, while the medical image segmentation device 300 predicts the region corresponding to the target region based on multiple subvolume data, the medical image segmentation device 300 may display and provide the region of the target region captured in one of the reference planes in the axial, coronal, and sagittal directions, depending on the medical staff's selection.

[0059] So far, a medical image splitting system 1000 according to one embodiment of the present invention has been described. According to the present invention, the medical image splitting system 1000 accurately predicts the location of a target area in a medical image and visually displays it, thereby assisting medical staff in diagnosing the target area.

[0060] In the following, a medical staff device 200 that utilizes the medical image segmentation results will be described with reference to FIG. FIG. 2 is a block diagram showing the configuration of a medical staff device according to an embodiment of the present invention.

[0061] 2, medical staff device 200 may include a memory interface 210, one or more processors 220, and a peripherals interface 230. The various components within medical staff device 200 may be coupled by one or more communication buses or signal lines.

[0062] The memory interface 210 is connected to the memory 250 and can transmit various data to the processor 220. Here, the memory 250 can include at least one type of storage medium among a flash memory type, a hard disk type, a multimedia card micro type, a card-type memory (e.g., SD or XD memory), RAM, SRAM, ROM, EEPROM, PROM, network storage, cloud, and a blockchain database.

[0063] In various embodiments, the memory 250 may store a web / app application or program for segmenting only a region corresponding to a target region in a medical image. The memory 250 may also store identification information of the subject (e.g., age, gender, presence or absence of disease), the medical image of the subject, a region corresponding to the target region in the medical image, etc.

[0064] In various embodiments, the memory 250 may store at least one of an operating system 251, a communications module 252, a graphic user interface module (GUI) 253, a sensor processing module 254, a telephony module 255, and an application module 256. Specifically, the operating system 251 may include instructions for processing basic system services and instructions for performing hardware operations. The communications module 252 may communicate with at least one of another device, computer, and server. The graphic user interface module (GUI) 253 may process a graphic user interface. The sensor processing module 254 may process sensor-related functions (e.g., processing audio input received through one or more microphones 292). The telephony module 255 may process telephony-related functions. The application module 256 may perform various functions of user applications, such as electronic messaging, web browsing, media processing, searching, imaging, and other processing functions. Additionally, the medical staff device 200 may store in memory 250 one or more software applications 256-1, 256-2 associated with one type of service (eg, a medical image segmentation application).

[0065] In various embodiments, memory 250 can store a digital assistant client module 257 (hereinafter, DA client module), which can store commands for performing client-side functions of the digital assistant and various user data 258 (e.g., user-customized vocabulary data, preference data, user electronic address book, etc.).

[0066] Meanwhile, the DA client module 257 can obtain user voice input, text input, touch input, and / or gesture input through various user interfaces (e.g., the I / O subsystem 240) provided on the medical staff device 200.

[0067] The DA client module 257 can also output audiovisual and tactile data. For example, the DA client module 257 can output data consisting of a combination of at least two or more of voice, sound, notification, text message, menu, graphic, video, animation, and vibration. In addition, the DA client module 257 can communicate with a digital assistant server (not shown) using the communication subsystem 280.

[0068] In various embodiments, the DA client module 257 can collect additional information about the surrounding environment of the medical staff device 200 from various sensors, subsystems, and peripheral devices to form a context associated with the user input. For example, the DA client module 257 can provide context information along with the user input to the digital assistant server to infer the user's intent. Here, context information that may accompany the user input can include sensor information such as light, ambient noise, ambient temperature, images of the surrounding environment, video, etc. As another example, the context information can include the physical state of the medical staff device 200 (e.g., device orientation, device location, device temperature, power level, speed, acceleration, motion pattern, cellular signal strength, etc.). As yet another example, the context information can include information related to the software state of the medical staff device 200 (e.g., processes running on the medical staff device 200, installed programs, past and present network activity, background services, error logs, resource usage, etc.).

[0069] In various embodiments, memory 250 may contain additional or deleted instructions. Furthermore, medical staff device 200 may include additional components or may exclude some components from those shown in FIG.

[0070] The processor 220 can control the overall operation of the medical staff device 200 and can execute various commands to run applications or programs stored in the memory 250 to display an area corresponding to a target site in a medical image to the medical staff or to implement a user interface that allows the medical staff to check the prediction results provided through the first and second prediction models.

[0071] The processor 220 may correspond to a computing device such as a central processing unit (CPU) or an application processor (AP), and may be implemented in the form of an integrated chip (IC) such as a system on chip (SoC) that integrates various computing devices that perform machine learning, such as a neural processing unit (NPU).

[0072] In various embodiments, the processor 220 can display the target area in the medical image through a user interface screen via a medical image segmentation application or program provided by the medical image segmentation device 300, or can request segmentation of the target area and display the results.

[0073] In various embodiments, the processor 220 may acquire a medical image from the imaging device 100 and generate multiple sub-volume data using the medical image. Alternatively, the processor 220 may generate sub-volume data only for a region of a target region predicted through a first prediction model, rather than for the entire region of the medical image. Specifically, the processor 220 may input the medical image to a first prediction model trained to predict a target region using the medical image as input, and determine a first region corresponding to the target region. Furthermore, the processor 220 may input the sub-volume data generated for only the first region to a second prediction model trained to predict a target region using the 3D medical image as input, and determine a second region corresponding to the target region from the multiple sub-volume data. The processor 220 may then provide the region corresponding to the target region within the medical image through a user interface screen.

[0074] The peripheral interface 230 can be connected to various sensors, subsystems, and peripheral devices to provide data to enable the medical staff device 200 to perform various functions. Here, when the medical staff device 200 performs a certain function, it can be understood that the function is performed by the processor 220.

[0075] The peripheral interface 230 may receive data from a motion sensor 260, an illumination sensor (light sensor) 261, and a proximity sensor 262, through which the medical staff device 200 may perform orientation, light, and proximity sensing functions, etc. As another example, the peripheral interface 230 may receive data from other sensors 263 (positioning system - GPS receiver, temperature sensor, biometric recognition sensor), through which the medical staff device 200 may perform functions related to the other sensors 263.

[0076] In various embodiments, the medical staff device 200 may include a camera subsystem 270 connected to the peripheral interface 230 and an optical sensor 271 connected thereto, through which the medical staff device 200 can perform various imaging functions such as taking photographs and recording video clips.

[0077] In various embodiments, medical staff device 200 can include a communications subsystem 280 coupled to peripheral interface 230. Communications subsystem 280 can be configured with one or more wired / wireless networks and can include various communications ports, radio frequency transceivers, and optical transceivers.

[0078] In various embodiments, the medical staff device 200 includes an audio subsystem 290 coupled to the peripheral interface 230, which may include one or more speakers 291 and one or more microphones 292, enabling the medical staff device 200 to perform voice-activated functions, such as voice recognition, voice duplication, digital recording, and telephone functions.

[0079] In various embodiments, the medical staff device 200 can include an I / O subsystem 240 coupled to the peripheral interface 230. For example, the I / O subsystem 240 can control a touchscreen 243 included in the medical staff device 200 through a touchscreen controller 241.

[0080] For example, the touchscreen controller 241 may detect a user's touch and movement or the cessation of touch and movement using any one of a number of touch sensing technologies, such as capacitive, resistive, infrared, surface acoustic wave technology, a proximity sensor array, etc. As another example, the I / O subsystem 240 may control other input / control devices 244 included in the medical staff device 200 through an other input controller 242. As one example, the other input controller 242 may control one or more buttons, rocker switches, thumbwheels, an infrared port, a USB port, and a pointer device such as a stylus.

[0081] So far, the medical staff device 200 according to one embodiment of the present invention has been described. According to the present invention, the medical staff device 200 can be used to accurately identify the location and size of a target region to be diagnosed in a medical image, such as an organ, bone, or tumor, thereby accurately diagnosing the health condition of a subject.

[0082] Hereinafter, a medical image segmentation device 300 that provides a segmented result of a target region in a medical image will be described with reference to FIG. FIG. 3 is a block diagram showing the configuration of a medical image dividing device according to an embodiment of the present invention.

[0083] Referring to FIG. 3, a medical image splitter 300 may include a communication interface 310, a memory 320, an I / O interface 330, and a processor 340, each of which may communicate with each other through one or more communication buses or signal lines.

[0084] The communication interface 310 can be connected to and exchange data with the image capturing device 100 and the medical staff device 200 through a wired / wireless communication network. For example, the communication interface 310 can receive a medical image of a subject from the image capturing device 100 or the medical staff device 200. As another example, the communication interface 310 can transmit a predicted result of a region corresponding to a target region in a medical image to the medical staff device 200 and provide a user interface screen for visually displaying the predicted result.

[0085] Meanwhile, the communication interface 310 that enables the transmission and reception of such data includes a wired communication port 311 and wireless circuitry 312, where the wired communication port 311 can include one or more wired interfaces, such as Ethernet, Universal Serial Bus (USB), Firewire, etc. The wireless circuitry 312 can transmit and receive data to and from external devices through RF signals or optical signals. In addition, the wireless communication can use at least one of a number of communication standards, protocols, and technologies, such as GSM, EDGE, CDMA, TDMA, Bluetooth, Wi-Fi, VoIP, Wi-MAX, or any other suitable communication protocol.

[0086] The memory 320 may store various data used in the medical image segmentation device 300. For example, the memory 320 may store identification information of the image capture device 100 and the medical staff device 200, first and second prediction models trained to predict a region corresponding to any one target region using a 3D medical image as input, and configuration and training data of a classification model trained to classify types of target regions using a medical image as input.

[0087] In various embodiments, the memory 320 may include a volatile or non-volatile storage medium capable of storing various data, instructions, and information. For example, the memory 320 may include at least one of the following types of storage media: flash memory, hard disk, multimedia card micro, card-type memory (e.g., SD or XD memory), RAM, SRAM, ROM, EEPROM, PROM, network storage, cloud, and blockchain database.

[0088] In various embodiments, the memory 320 may store configurations for at least one of an operating system 321 , a communications module 322 , a user interface module 323 , and one or more applications 324 .

[0089] The operating system 321 (e.g., an embedded operating system such as LINUX, UNIX, MAC OS, WINDOWS, VxWorks, etc.) may include various software components and drivers for controlling and managing general system operations (e.g., memory management, storage device control, power management, etc.) and may support communication between various hardware, firmware, and software components.

[0090] The communications module 322 can facilitate communication with other devices through the communications interface 310. The communications module 322 can include various software components for processing data received by the wired communications port 311 or wireless circuitry 312 of the communications interface 310.

[0091] The user interface module 323 can receive user requests or inputs from a keyboard, touch screen, keyboard, mouse, microphone, etc. through the I / O interface 330 and provide a user interface on a display.

[0092] The application 324 may include programs or modules configured to be executed by one or more processors 340. Here, the application for medical image classification and segmentation may be implemented on a server farm.

[0093] The I / O interface 330 can connect at least one of an input / output device (not shown) of the medical image splitter 300, such as a display, a keyboard, a touch screen, and a microphone, to the user interface module 323. The I / O interface 330 can receive user input (e.g., voice input, keyboard input, touch input, etc.) together with the user interface module 323 and process commands according to the received input.

[0094] The processor 340 is connected to the communication interface 310, the memory 320, and the I / O interface 330 to control the overall operation of the medical image segmentation device 300, and can learn the first and second prediction models and classification models through applications or programs stored in the memory 320, and can execute various commands to segment target areas in the medical image when a new medical image is input.

[0095] The processor 340 may correspond to a computing device such as a central processing unit (CPU) or an application processor (AP). The processor 340 may also be implemented in the form of an integrated chip (IC) such as a system on chip (SoC) in which various computing devices are integrated. Alternatively, the processor 340 may include a module for calculating an artificial neural network model, such as a neural processing unit (NPU).

[0096] The method by which the processor 340 of the medical image segmentation device 300 segments a medical image will be described below with reference to FIGS. FIG. 4 is a schematic flowchart of a medical image segmentation method according to an embodiment of the present invention.

[0097] 4, the processor 340 may acquire medical images of the subject (S110). Here, the medical images are 2D images composed of a plurality of cuts (or slices) and may be enhanced or non-enhanced CT images. For example, the medical images may include a head and neck image including the entire area from the skull vertex to the lung apex, a chest image including the entire area from the thyroid to the liver dome, an abdominal image including the entire L1 spine at a position 3 cm away from the liver dome in the cranial direction, and a pelvic image including the entire ischium at a position 3 cm away from the L1 spine in the cranial direction.

[0098] In various embodiments, the processor 340 may classify and group the plurality of slices constituting the medical image according to the target region, thereby improving the computational efficiency of the second prediction model. Specifically, the processor 340 may input the medical image of the subject to a classification model trained to classify the type of target region using the medical image as input, and determine the type of target region included in each of the plurality of slices constituting the medical image. For example, the classification model may determine whether the slice includes one of the target regions, namely, the brain, neck, chest, abdomen, and pelvis. The medical image segmentation device 300 may group the plurality of slices according to the target region based on the classification result.

[0099] After step S110, the processor 340 may input the medical image to a first prediction model trained to predict any one target region using the medical image as input, and determine a first region corresponding to any one target region (S120). Specifically, the first prediction model may be a model trained to predict a first region corresponding to a target region in a slice constituting the medical image. For example, the processor 340 may output a region in each slice in which a target region, such as the head, neck, chest, abdomen, or pelvis, is predicted to be located in a box form through the first prediction model, and determine this as the first region.

[0100] After step S120, the processor 340 can generate a plurality of subvolume data using the first region (S130). Specifically, the subvolume data can be generated by stacking a plurality of slices constituting the medical image to match the pre-stored height and dividing the slices in a direction perpendicular to the plane of the slices. For example, the subvolume data can be slab data of 256x256x16 size, or as another example, the subvolume data can be cubic data of 96x96x96 size. In addition, the voxels constituting the subvolume data can each be 2x2x3mm. 3 , 0.7×0.7×1.0mm 3 The size of the voxel can be adjusted to 0.7 x 0.7 x 1.0 mm. 3 From 2 x 2 x 3 mm 3 The processor 340 can generate a plurality of sub-volume data based on an axis including any one of a horizontal plane (Axial), a coronal plane (Coronal), and a sagittal plane (Sagittal).

[0101] In various embodiments, the plurality of subvolume data may include data on the movement direction of voxels constituting the subvolume data based on any one axis, thereby improving the accuracy of prediction results compared to predicting the target region in slice units. For example, if the subvolume data is slab data, each subvolume data may include data on the movement direction of voxels in a direction perpendicular to a horizontal plane (axial). As another example, if the subvolume data is cubic data, each subvolume data may include data on the movement direction of voxels in at least one of a horizontal plane (axial), a coronal plane, and a sagittal plane.

[0102] After step S130, the processor 340 inputs multiple subvolume data into a second prediction model trained to predict any one target region using the 3D medical image as input, and can determine a second region corresponding to the target region within the first region using the multiple subvolume data (S140).

[0103] In relation to this, FIG. 5 is a schematic diagram illustrating a medical image segmentation method according to an embodiment of the present invention. 5, a processor 340 may input a medical image 11 consisting of a plurality of slices into a first prediction model 12 to determine a first region 13 corresponding to a target region. Here, the medical image 11 may be a medical image classified as including one of the target regions, such as the head, neck, chest, abdomen, and pelvis. The processor 340 may stack and divide only the first region 13 corresponding to the target region to generate a plurality of sub-volume data. The processor 340 may use the medical image 11 to generate a plurality of sub-volume data defined as slab data 14a having a size of 256×256×16 or cubic data 14b having a size of 96×96×96.

[0104] In various embodiments, the processor 340 may input the slab data 14a to a second prediction model 15a for the 2-1 target region or input the cubic data 14b to a second prediction model 15b for the 2-2 target region depending on the type of generated subvolume data, and output a prediction result 16 indicating whether each subvolume data corresponds to a region of interest. Here, the target region may include organs such as the brain, neck, chest, abdomen, or pelvis, bones, or muscles, or tumors suspected of being diseases formed around organs, bones, or muscles.

[0105] In this manner, processor 340 can determine a first region representing the general shape of the target region through the first prediction model and a second region representing the specific shape of the target region through the second prediction model. Thus, the first and second regions determined by the first and second prediction models may correspond to the same target region. However, depending on the location of the organ within the body, the first and second regions may correspond to different target regions, and multiple second regions may be included within the first region. In other words, processor 340 can determine multiple second regions corresponding to two or more different target regions depending on the type of target region corresponding to the first region. For example, the first region may be the lung, and the multiple second regions may be the left and right lungs.

[0106] Meanwhile, to obtain such prediction results, the second prediction model may be trained using training data having different slice thicknesses depending on the type of medical image. Specifically, the second prediction model uses 256x256x16 slab data or 96x96x96 cubic data as subvolume data. The processor 340 performs different numbers of bootstrapping depending on the slice thickness input to the training data, thereby training the second prediction model without image loss of the medical image. For example, if slices having a thickness of 3 mm are input at the beginning of training, the processor 340 may stack the slices to generate subvolume data in 48 mm units to train the second prediction model. If slices having a thickness of 5 mm are input at the next beginning of training, the processor 340 may stack the slices to generate subvolume data in 90 mm units to train the second prediction model. The processor 340 may perform such training for the number of epochs specified in the memory 320.

[0107] Alternatively, the processor 340 may pre-process a plurality of slices input to the second prediction model, i.e., a slice including only the first region, and use the pre-processed slices as training data. For example, the processor 340 may copy one slice and stack it to match the reference thickness of the training data, or may perform linear interpolation to resample the slice thickness to match the reference thickness.

[0108] 4, the processor 340 may provide a second region corresponding to the target region within the medical image (S150). Specifically, the processor 340 may determine whether each of the plurality of sub-volume data corresponds to the target region through the second prediction model, and may segment each of the plurality of sub-volume data according to the result. The processor 340 may combine the segmentation results to display the second region corresponding to the target region in a plurality of slices constituting the medical image.

[0109] In this regard, FIGS. 6a and 6b are exemplary diagrams of user interface screens showing medical image segmentation results according to an embodiment of the present invention. 6a and 6b, the processor 340 may display a second region corresponding to the target region in the medical image and provide it to the medical staff device 200. Specifically, the processor 340 may combine volume data in which a probability value corresponding to the target region in each of a plurality of sub-volume data is equal to or greater than a preset value using the second prediction model, and display the combined volume data on the user interface screen. For example, as shown in FIG. 6a, the processor 340 may highlight and display only the edge region 17 of the target region in which a probability value corresponding to the target region is equal to or greater than a preset value. In addition, the processor 340 may display regions predicted to be the target region in the medical image in different colors depending on the type of region.

[0110] Furthermore, the processor 340 may use the second prediction model to display each of the plurality of sub-volume data on the user interface screen at different transparency levels depending on the probability value corresponding to the target region. For example, as shown in FIG. 6b, the processor 340 may highlight a region 17' predicted to be the target region in the medical image at different transparency levels depending on the probability value. The processor 340 may control the region 17' to be highlighted in the medical image by lowering the transparency as the probability value increases.

[0111] In various embodiments, the processor 340 may provide a user interface screen including a segmentation model selection area for selecting a target region to be segmented within a medical image and a segmentation result display area 15 for displaying the region corresponding to the target region.

[0112] In a related regard, FIG. 7 is an example diagram of a user interface screen for utilizing a predictive model for medical image segmentation according to one embodiment of the present invention. Referring to FIG. 7 , the processor 340 may train the first and second prediction models and provide the medical staff device 200 with a user interface screen capable of displaying prediction results obtained through the first and second prediction models. The user interface screen may include an area 21 in which learning parameters for the second prediction model can be input and adjusted. For example, batch size, iteration, and epoch parameters may be set through the user interface screen. The processor 340 may train the second prediction model based on the parameters input through the user interface screen, and the user interface screen may include an area 22 displaying the learning progress and an area 23 displaying the learning dataset used for learning. The user interface screen may also include a graphic object 24 for training the second prediction model and an area 25 displaying a list of second prediction models undergoing learning when the corresponding graphic object 24 is selected. The area 25 displaying the list of second prediction models undergoing learning may also display the loss function and measure used for learning, and may include a graphic object allowing selection of whether to visualize and display prediction results. The train value is updated each time a batch is completed, and the processor 340 may update the valid value once the entire learning data has been learned, i.e., each time an epoch is completed. The user interface screen may also include an area 26 for graphically displaying the train value and the valid value for each second prediction model through learning of the second prediction model. For example, the train value may be displayed as a solid line, and the valid value may be displayed as a dotted line. The user interface screen may also include an area 30 for displaying a prediction result of the target region using the second prediction model within the medical image. Here, the user interface screen may include a graphic object 27 for selecting a cross-section based on which the region corresponding to the target region is to be displayed, a graphic object 28 for selecting a display format for the region corresponding to the target region, and a graphic object 29 for setting the opacity for displaying the region corresponding to the target region within the medical image.Here, the user interface screen can highlight and display only the region corresponding to the target region in the medical image, and can also display the prediction results for each of the plurality of sub-volume data.

[0113] The processor 340 can acquire one of the reference planes, axial, coronal, and sagittal, for displaying the target area through a user interface screen, depending on the selection of the medical staff, and can render the area corresponding to the target area based on the reference plane to provide the area corresponding to the target area.

[0114] So far, we have described a medical image segmentation method performed by the processor 340 of the medical image segmentation device 300 according to an embodiment of the present invention. According to the present invention, the medical image segmentation device 300 goes beyond inputting only 2D images constituting a medical image to the first and second prediction models that predict a target region, and instead provides before and after image information of 2D images captured sequentially through volume data, thereby improving the prediction accuracy of the model and expanding the diversity of prediction results through two types of prediction models compared to predicting a target region using a single prediction model.

[0115] The overall flow of the prediction model learning process and prediction process performed by the medical image segmentation device 300 will be described below. FIG. 8 is a schematic diagram illustrating a method for training a prediction model according to an embodiment of the present invention.

[0116] 8, the medical image segmentation device 300 may preprocess a plurality of slices (learning data sets) used for training the second prediction model and generate a plurality of subvolume data based on the preprocessed slices. The medical image segmentation device 300 may input a plurality of subvolume data sets to perform training and may provide an operator that can correct a difference between the training result and the actual result.

[0117] As described above, the medical image segmentation device 300 can learn the second prediction model, and when a prediction request for a medical image is received at the request of an administrator or medical staff, the second prediction model can be loaded to segment an area corresponding to a target site within the medical image.

[0118] In this regard, FIG. 9 is a schematic diagram illustrating the operation of a prediction model for segmenting medical images according to an embodiment of the present invention. 9, the medical image segmentation device 300 can acquire enhanced and non-enhanced CT images generated according to the DICOM standard and check whether to perform auto-segmentation. For example, the medical staff device 200 can directly segment the target area in the medical image or request auto-segmentation through a web or app application provided by the medical image segmentation device 300.

[0119] If automatic segmentation is requested, the medical image segmentation device 300 can use a classification model to detect which target region is imaged in each of the slices constituting the medical image. That is, the medical image segmentation device 300 can classify (or detect) each of the slices into the brain, neck, chest, abdomen, pelvis, and other organs using the classification model, and can apply additional first and second prediction models depending on the type of target region or the type of subject. For example, for a slice classified as the brain, the medical image segmentation device 300 can further use first and second prediction models to predict the optic region. As another example, for a slice classified as the chest in a medical image of a female subject, the medical image segmentation device 300 can further use first and second prediction models to predict the breast region. As another example, in a medical image of a subject with cardiac disease, for a slice classified as a chest, the medical image segmentation device 300 may further use a prediction model for predicting a heart region. As another example, in a slice classified as an abdomen and a pelvis, the medical image segmentation device 300 may further use first and second prediction models for predicting a bowel region.

[0120] As described above, the medical image segmentation device 300 can classify and group slices using a classification model, divide the grouped slices into a plurality of sub-volume data, and then predict whether each of the plurality of sub-volume data is a region corresponding to a target region using first and second prediction models based on the grouped slices.

[0121] Although one embodiment of the present invention has been described in detail above with reference to the accompanying drawings, the present invention is not necessarily limited to such an embodiment and may be variously modified within the scope of the technical concept of the present invention. Therefore, the disclosed embodiments of the present invention are intended to be illustrative rather than limiting the technical concept of the present invention, and the scope of the technical concept of the present invention is not limited by such embodiments. Therefore, the above-described embodiments should be understood to be illustrative in all respects and not limiting. The scope of protection of the present invention should be interpreted by the scope of the following claims, and all technical concepts within the scope equivalent thereto should be interpreted as being included in the scope of the present invention.

Claims

1. 1. A method performed by a processor of a medical image segmentation device, comprising: acquiring a medical image of the subject; inputting the medical image into a first prediction model that has been trained to predict any one target region using the medical image as an input, and determining a first region corresponding to any one target region; generating a plurality of sub-volume data using the first region; inputting the plurality of sub-volume data into a prediction model trained to predict any one target region using a 3D medical image as an input, and determining a second region corresponding to the target region within the first region from the plurality of sub-volume data; and providing a second region within the medical image corresponding to the region of interest.

2. The step of determining the second region includes: The medical image segmentation method according to claim 1 , further comprising the step of determining a plurality of second regions corresponding to two or more different target regions according to the type of target region corresponding to the first region.

3. The sub-volume data is The medical image segmentation method according to claim 1 , further comprising data on a movement direction of voxels constituting the sub-volume data with respect to any one of the axes.

4. The step of generating sub-volume data includes:

2. The medical image segmentation method of claim 1, further comprising the step of stacking a plurality of slices constituting the medical image to match the height of the pre-stored slices, and dividing the slices in a direction perpendicular to the plane of the slices to generate a plurality of sub-volume data.

5. Before generating the sub-volume data, inputting a medical image into a classification model trained to classify the type of target region using the medical image; Determining the type of target region included in each of the plurality of slices constituting the medical image; and The medical image segmentation method according to claim 1 , further comprising: grouping the plurality of slices by target region according to a classification result.

6. The step of providing a region corresponding to the target site comprises:

2. The medical image segmentation method of claim 1, further comprising providing a user interface screen including a segmentation model selection area for selecting a target region to be segmented in the medical image and a segmentation result display area for displaying a region corresponding to the target region.

7. The step of providing a region corresponding to the target site comprises:

7. The medical image segmentation method of claim 6, further comprising: combining volume data in which a probability value corresponding to the target region in each of the plurality of sub-volume data is equal to or greater than a predetermined value using the prediction model, and displaying the combined volume data on the user interface screen.

8. The step of providing a region corresponding to the target site comprises: The medical image segmentation method of claim 6 , further comprising: displaying each of the plurality of sub-volume data on the user interface screen with different transparency levels according to a probability value corresponding to the target region using the prediction model.

9. The step of generating sub-volume data includes: It is generated based on an axis including one of the horizontal plane (Axial), coronal plane (Coronal), and sagittal plane (Sagittal), The step of providing a region corresponding to the target site comprises: acquiring, through the user interface screen, any one of an axial plane, a coronal plane, and a sagittal plane for displaying the target region; and The medical image segmentation method according to claim 6 , further comprising the step of: rendering a region corresponding to the target region based on the reference plane.

10. Before the step of acquiring the medical image, acquiring training data having different slice thicknesses according to types of medical images; generating a training data set by performing bootstrapping a different number of times depending on the thickness of the training data; and The medical image segmentation method of claim 1 , further comprising: generating the prediction model configured to predict any one target region based on the training dataset.

11. communication interface; memory; and a processor operatively coupled to the communication interface and the memory; The processor: a first prediction model trained to predict one target region using the medical image as an input, to determine a first region corresponding to the target region; a plurality of sub-volume data sets are generated using the first region; a prediction model trained to predict one target region using the 3D medical image as an input, to determine a second region corresponding to the target region within the first region using the plurality of sub-volume data sets; and a medical image segmentation device configured to provide the second region corresponding to the target region within the medical image.

12. The processor: The medical image segmentation device according to claim 11 , configured to determine a plurality of second regions corresponding to two or more different target regions according to the type of target region corresponding to the first region.

13. The subvolume data is The medical image segmentation device according to claim 11, further comprising data on a movement direction of voxels constituting the sub-volume data with respect to any one of the axes.

14. The processor: The medical image segmentation device according to claim 11, configured to stack a plurality of slices constituting the medical image to match a pre-stored height, and to generate a plurality of sub-volume data by segmenting in a direction perpendicular to a plane of the slices.

15. The processor:

12. The medical image segmentation device of claim 11, further configured to: input the medical image into a classification model that has been trained to classify types of target regions using the medical image as input, determine the type of target region included in each of a plurality of slices constituting the medical image, and group the plurality of slices by target region according to the classification result, before generating the sub-volume data.

16. The processor: The medical image segmentation device of claim 11, configured to provide a user interface screen including a segmentation model selection area for selecting a target region to be segmented in a medical image and a segmentation result display area for displaying a region corresponding to the target region.

17. The processor:

17. The medical image segmentation device of claim 16, further configured to combine volume data in which a probability value corresponding to the target region in each of the plurality of sub-volume data is equal to or greater than a preset value using the prediction model, and display the combined volume data on the user interface screen.

18. The processor: The medical image segmentation device according to claim 16, further configured to use the prediction model to display each of the plurality of sub-volume data on the user interface screen with different transparency depending on a probability value corresponding to the target region.

19. The processor: The sub-volume data is generated based on an axis including any one of an axial plane, a coronal plane, and a sagittal plane; 17. The medical image segmentation device of claim 16, further configured to: acquire, through the user interface screen, one of an axial plane, a coronal plane, and a sagittal plane for displaying the target region in order to provide the region corresponding to the target region; and render the region corresponding to the target region based on the reference plane.

20. The processor:

12. The medical image segmentation device of claim 11, further comprising: before acquiring the medical image, acquiring training data having different slice thicknesses according to a type of the medical image, performing bootstrapping a different number of times depending on the thickness of the training data to generate a training dataset; and generating the prediction model configured to predict any one target region based on the training dataset.

Images