KR102606249B1 - Lesion analysis method using cerebrovascular modeling and apparatus thereof - Google Patents

Abstract

The present invention relates to a lesion analysis method and device using cerebrovascular modeling. One embodiment of the present invention includes the steps of the lesion analysis device receiving a cerebrovascular image, the lesion analysis device receiving a cerebrovascular image from the cerebrovascular image. Extracting a vascular region, generating a cerebrovascular model in continuous space based on the extracted cerebrovascular region by the lesion analysis device, generating a plurality of brain vessels from the cerebrovascular model by the lesion analysis device. It provides a lesion analysis method using cerebrovascular modeling, including calculating vascular characteristics and deriving a disease state and prognosis based on the cerebrovascular characteristics using an artificial intelligence algorithm by the lesion analysis device.

Description

Lesion analysis method using cerebrovascular modeling and apparatus thereof}

The present invention relates to a lesion analysis method and device using cerebrovascular modeling, and more specifically, to calculating cerebrovascular characteristics in continuous space using cerebrovascular modeling, and based on this, the state and prognosis of the disease. It relates to a lesion analysis method and device for deriving .

The cerebrovascular structure is used as an important factor in pre-diagnosing brain diseases or tracking the progress of brain diseases.

However, conventional medical imaging techniques such as Magnetic Resonance Angiography (MRA) convert and display the anatomical structure of the brain in discrete space, so they are used in quantifying fine structures such as cerebral blood vessels. This can cause large errors. Therefore, there is a need for the development of technology that can quantify the brain vascular structure by modeling it in continuous space.

Korean Patent Publication No. 10-2095731 (registered on March 26, 2020) relates to an MRA image learning method and a vascular lesion auxiliary diagnosis method of a deep learning-based auxiliary diagnosis system, including blood vessel extraction in which the vascular region is extracted from the MRA image. Steps and; a cube extraction step in which a plurality of three-dimensional cube regions of a preset size are extracted along the blood vessel region by the auxiliary diagnosis system; a learning data extraction step of classifying the plurality of 3D cube regions into positive cubes and negative cubes based on the labels of the MRA images by the auxiliary diagnosis system; and a model creation step in which a learning model is generated by learning the positive cube and the negative cube as learning data by the auxiliary diagnosis system.

Since this method extracts blood vessels by segmenting the anatomical structure of the brain transformed into a discrete space, large errors may occur when applied to the diagnosis of very fine structures such as cerebral blood vessels.

1. Korean Patent Registration No. 10-2095731 (registered on March 26, 2020)

The technical problem to be solved by the present invention is to provide a lesion analysis method and device for calculating cerebrovascular characteristics in continuous space using cerebrovascular modeling and deriving disease status and prognosis based on this.

In order to solve the above-described technical problem, an embodiment of the present invention provides a lesion analysis method performed in a lesion analysis device using cerebrovascular modeling.

The lesion analysis method using cerebrovascular modeling includes the steps of receiving a cerebrovascular image by the lesion analysis device, extracting a cerebrovascular region from the cerebrovascular image by the lesion analysis device, and extracting a cerebrovascular region from the cerebrovascular image by the lesion analysis device. Generating a cerebrovascular model in continuous space based on the vascular region, calculating a plurality of cerebrovascular characteristics from the cerebrovascular model by the lesion analysis device, and performing an artificial intelligence algorithm by the lesion analysis device. It may include deriving a disease state and prognosis based on the cerebrovascular characteristics.

According to an embodiment of the present invention, the step of generating the brain blood vessel model by the lesion analysis device includes dividing the surface of the extracted brain blood vessel region into isosurfaces to generate the brain blood vessel surface model. , dividing the three-dimensional continuous space equally into a plurality of cells based on each vertex of the isoform surface, and extracting the center line of the brain blood vessel from the boundary surface of each cell.

According to an embodiment of the present invention, the step of equally dividing into a plurality of cells may include setting a set of spatial coordinates in which the closest vertex among the vertices of the isosurface is the same as one cell.

According to one embodiment of the present invention, the step of extracting the center line of the brain blood vessel includes determining a starting point of the center line from a lower slice of the brain blood vessel area, the skeleton from the brain blood vessel area, or the brain blood vessel surface model. It may include generating a structure, refining the skeletal structure to determine the end points of the center line, and extracting the center line by tracking the boundary surface of each cell connecting the start point and the end point.

According to one embodiment of the present invention, the step of determining the end points of the center line includes generating a skeletal structure by skeletonizing the brain blood vessel region or the brain blood vessel surface model, and branching from the skeletal structure to a certain length or less. pruning and refining the nodes, generating a linked list of a tree structure based on the refined skeleton structure, and specifying leaf nodes from the linked list. It may include determining endpoints.

The step of calculating the plurality of cerebral blood vessel characteristics by the lesion analysis device according to an embodiment of the present invention includes grouping the cerebral blood vessels into a plurality of groups based on a branch point of the center line of the cerebral blood vessel model, and It may include calculating the plurality of cerebrovascular characteristics for each of the groups.

According to one embodiment of the present invention, the plurality of cerebrovascular characteristics include cerebrovascular cross-sectional area, maximum inscribed hole radius, maximum radius, minimum radius, maximum-minimum radius ratio, surface perimeter, tortuosity, curvature, circularity, It may include at least one of branch angles and calculation values calculated based on them.

According to an embodiment of the present invention, the step of deriving the state and prognosis of the disease includes extracting disease-linked attributes through feature extraction from the cerebrovascular characteristics by the artificial intelligence algorithm, and The artificial intelligence algorithm may include a step of deriving the disease status and prognosis of the test subject based on the extracted disease-linked attributes.

According to one embodiment of the present invention, the artificial intelligence algorithm can derive the condition and prognosis of the disease using the Cox-Lasso model.

According to an embodiment of the present invention, the artificial intelligence algorithm repeatedly learns a plurality of learning data sets containing the cerebrovascular characteristics, extracts disease-linked properties, and learns based on the extracted disease-linked properties. The disease status and prognosis of each data set may be derived, and the quality of the derived results may be compared and evaluated with the correct answer data included in the learning data set, so that the quality is higher than a preset standard value.

One embodiment of the present invention includes cerebrovascular images, software for cerebrovascular modeling and lesion analysis, and a memory for storing data, and extracting cerebrovascular regions from cerebrovascular images, based on the extracted cerebrovascular regions. At least one processor configured to generate a cerebrovascular model in continuous space, calculate a plurality of cerebrovascular characteristics from the cerebrovascular model, and derive a disease state and prognosis based on the cerebrovascular characteristics. , provides a lesion analysis device using cerebrovascular modeling.

One embodiment of the present invention provides a computer program stored in a medium for executing any one of the above-described methods using a computing device.

The present invention can derive the geometric characteristics of cerebral blood vessels from a brain blood vessel model generated in continuous space, and use an artificial intelligence algorithm to derive the condition and prognosis of the disease based on the brain blood vessel characteristics.

Therefore, by non-invasively and automatically quantifying the brain vascular structure, microstructural lesion analysis, which was difficult to perform with the conventional visual judgment method, can be performed more precisely. Additionally, errors that may occur when performing geometric analysis of fine structures in discrete space can be minimized.

The technical effects of the present invention are not limited to those mentioned above, and other technical effects not mentioned will be clearly understood by those skilled in the art from the description below.

1 is a flowchart explaining a lesion analysis method using cerebrovascular modeling according to an embodiment of the present invention.
FIG. 2 is a flowchart explaining in more detail the steps for generating the brain blood vessel model shown in FIG. 1.
FIG. 3 is a flowchart explaining in more detail the steps of extracting the center line of the brain blood vessel shown in FIG. 2.
FIG. 4 is a flowchart explaining the endpoint determination step shown in FIG. 3 in more detail.
FIG. 5 is a flowchart illustrating in more detail the steps for extracting cerebrovascular characteristics shown in FIG. 1.
Figure 6 is a diagram illustrating a brain blood vessel model created according to an embodiment of the present invention.
Figure 7 is a block diagram illustrating a lesion analysis device using cerebrovascular modeling according to an embodiment of the present invention.
Figure 8 is a conceptual diagram explaining a lesion analysis system using cerebrovascular modeling according to an embodiment of the present invention.
Figure 9 is a diagram showing a screen of a lesion analysis program using cerebrovascular modeling according to an embodiment of the present invention.

While the invention is susceptible to various modifications and variations, specific embodiments thereof are illustrated in the drawings and will be described in detail below. However, it is not intended to limit the invention to the particular form disclosed, but rather, the invention is intended to cover all modifications, equivalents and substitutions consistent with the spirit of the invention as defined by the claims.

It will be understood that when an element such as a layer, region or substrate is referred to as being “on” another element, it may be present directly on the other element or there may be intermediate elements in between. .

The invention may be represented by functional block configurations and various processing steps. These functional blocks may be implemented in various numbers of hardware or/and software configurations that execute specific functions. For example, the present invention provides integrated circuit configurations, such as memory, processing, logic, look-up tables, etc., that can execute various functions under the control of one or more microprocessors or other control devices. can be hired. Similar to the fact that the components of the invention can be implemented as software programming or software elements, the invention also includes various algorithms implemented as combinations of data structures, processes, routines or other programming constructs, including C, C++, , may be implemented in a programming or scripting language such as Java, assembler, etc.

Additionally, the present invention can employ conventional technologies for electronic environment settings, signal processing, and/or data processing. Terms such as “part,” “element,” “means,” and “configuration” may be used broadly, and the components of the present invention are not limited to mechanical and physical configurations. The term may include the meaning of a series of software routines in connection with a processor, etc.

Although the terms first, second, etc. may be used to describe various elements, components, regions, layers, stages and/or regions, these elements, components, regions, layers, stages, etc. It will be understood that fields and/or regions should not be limited by these terms.

1 is a flowchart explaining a lesion analysis method using cerebrovascular modeling according to an embodiment of the present invention.

Referring to Figure 1, the lesion analysis method using cerebrovascular modeling according to an embodiment of the present invention includes receiving a cerebrovascular image (S110), extracting a cerebrovascular region (S120), and creating a cerebrovascular model. It includes a step of generating (S130), a step of extracting cerebrovascular characteristics (S140), and a step of deriving the disease state and prognosis (S150).

In the step of receiving the brain blood vessel image (S110), the lesion analysis device receives the brain blood vessel image of the subject. The brain vascular image was obtained using brain magnetic resonance imaging (MRI), brain magnetic resonance angiography (MRA), or TOF-brain magnetic resonance angiography (Time-of-flight MRA). It could be a video. The lesion analysis device may acquire a cerebrovascular image from a cerebrovascular imaging device or acquire a cerebrovascular image through a medical data system. The medical data system includes an integrated medical information system, electronic medical record, laboratory information system, data warehouse, and clinical device information system. It may be a system necessary for medical services and research, such as the System), but is not limited to this.

In the step of extracting the brain blood vessel area (S120), the lesion analysis device extracts the brain blood vessel area from the received brain blood vessel image. Here, brain blood vessel area extraction may be a process of removing background and noise from the brain blood vessel image and extracting pixels corresponding to the brain blood vessels.

In one embodiment, the lesion analysis device selects pixels above a preset threshold based on the brightness of pixels of the cerebrovascular image, selects pixels that satisfy a specific condition among pixels adjacent to the selected pixels, and selects pixels that satisfy a specific condition to determine the cerebrovascular image. Pixels corresponding to can be extracted. Additionally, the lesion analysis device may extract the cerebrovascular region using a region-based image segmentation method, a border-based image segmentation method, or a combination thereof. Here, the region-based image segmentation method may include a watershed algorithm.

In the step of generating a cerebrovascular model (S130), the lesion analysis device generates a cerebrovascular model in continuous space based on the extracted cerebrovascular region. Here, the cerebrovascular model in continuous space refers to a cerebrovascular model created based on a continuous relationship between coordinate points, rather than a cerebrovascular model created based on conventional isolated coordinate points. As conventional cerebrovascular models were created based on isolated coordinate points, it was difficult to calculate geometric cerebrovascular characteristics due to breaks, distortions, or incorrect connections in the fine structure. On the other hand, since the present invention uses a brain blood vessel model created in continuous space, the geometric characteristics of the brain blood vessel, such as the cross-section of the fine structure, direction of movement, and distortion, can be calculated more accurately. The step of creating a brain blood vessel model (S130) is described in more detail in the description of FIG. 2.

In the step of extracting cerebrovascular characteristics (S140), the lesion analysis device calculates a plurality of cerebrovascular characteristics using the generated cerebrovascular model. At this time, the plurality of cerebrovascular characteristics include cerebrovascular cross-sectional area, maximum inscribed hole radius, maximum radius, minimum radius, maximum-minimum radius ratio, surface perimeter, twist, curvature, circularity, branch angle, and It may contain at least one of the calculated values. At this time, the lesion analysis device may group the brain blood vessels of the brain blood vessel model by node and derive characteristics of the grouped brain blood vessels.

In the step of deriving the disease state and prognosis (S150), the lesion analysis device may use an artificial intelligence algorithm to derive the disease state and prognosis based on cerebrovascular characteristics.

Here, the artificial intelligence algorithm may be repeatedly learned from a plurality of learning data sets including brain blood vessel characteristics. Here, the plurality of learning data sets may include cerebrovascular images or cerebrovascular characteristics calculated from cerebrovascular images, and may include correct answer data on the state and prognosis of the disease corresponding to each cerebrovascular image.

The artificial intelligence algorithm learns multiple learning data sets, extracts disease-linked attributes, derives the disease status and prognosis of each learning data set based on the extracted disease-linked attributes, and evaluates the quality of the derived results. It may be evaluated and compared with the correct answer data included in the learning data set, and learned repeatedly so that the comparison value is greater than or equal to a preset reference value.

In addition, the artificial intelligence algorithm receives new cerebrovascular images or cerebrovascular characteristics calculated from them, extracts disease-related properties based on this, derives the status and prognosis of each disease based on the extracted disease-related properties, and , you can learn using the results.

In the step of deriving the disease state and prognosis (S150), the artificial intelligence algorithm extracts disease-linked attributes through feature extraction from cerebrovascular characteristics, and the artificial intelligence algorithm extracts the extracted disease-linked attributes. Based on this, it may include the step of deriving the condition and prognosis of the test subject's disease.

The artificial intelligence algorithm can use the Cox-Lasso model as a survival risk model. In another embodiment, when the result value that the artificial intelligence algorithm seeks to derive is a binary value, the logistic lasso model can be used as a survival risk model.

The artificial intelligence algorithm can use known machine learning classification models to classify disease markers and disease states. For example, the artificial intelligence algorithm may use a support vector machine, a random forest algorithm, an artificial neural network, or a combination thereof.

FIG. 2 is a flowchart explaining in more detail the steps for generating the brain blood vessel model shown in FIG. 1.

Referring to FIG. 2, the step of generating a cerebrovascular model (S130) includes dividing the surface of the cerebrovascular region into equivalent curved surfaces (S210) and dividing the surface into a plurality of cells based on the vertices of the equivalent curved surface (S130). S220), and a step of extracting the center line of the brain blood vessel (S230).

The step of dividing the surface of the brain blood vessel region into isosurfaces (S210) is a step of dividing the surface of the extracted brain blood vessel region into isosurfaces. The lesion analysis device converts the surface of the cerebrovascular region in the form of a volume data set into isosurfaces of a triangular mesh structure using a marching cubes algorithm. Here, the isosurfaces can be a set of levels of a continuous function that share a vertex.

The lesion analysis device can delete vertices that are too close or overlapping and rearrange the vertices so that each isosurface is close to an equilateral triangle. The vertices of the equivalent curves and the connection relationship information between the vertices can construct a surface model of the cerebrovascular region.

In the step of dividing into a plurality of cells based on the vertices of the equivalent curve (S220), the lesion analysis device evenly divides the three-dimensional continuous space into a plurality of cells based on each vertex of the equivalent curve. Here, dividing equally into a plurality of cells means setting a set of spatial coordinates in which the closest vertex among the vertices of the isoform surface is the same as one cell. In one embodiment, the plurality of cells may be 3D Voronoi cells.

In the step of extracting the center line of the brain blood vessel (S230), the lesion analysis device may extract the center line of the brain blood vessel from the boundary surface of each cell. Hereinafter, the step of extracting the center line of the brain blood vessel will be described in more detail in the description of FIG. 3.

FIG. 3 is a flowchart explaining in more detail the steps of extracting the center line of the brain blood vessel shown in FIG. 2.

Referring to FIG. 3, the step of extracting the center line of a brain blood vessel (S230) includes a step of determining a starting point (S310), a step of determining an end point (S320), and a step of extracting the center line by tracking the boundary surfaces of cells (S330). can do.

The starting point determination step (S310) may be a step of determining the starting point of the center line from the lower slice of the brain blood vessel region. The lesion analysis device may determine the starting point of the center line from the center of gravity of the clusters appearing in the slice containing the lowermost part of the cerebrovascular region. In one embodiment, the lesion analysis device may select the lowest slice or a plurality of consecutive slices including the lowest slice to generate a cross-sectional profile of the lower part of the blood vessel. A group index can be assigned to all areas of the lower blood vessel cross-sectional profile generated using a segmentation algorithm or a search algorithm. In one embodiment, the segmentation algorithm may be a Region Growing algorithm. In one embodiment, the search algorithm may be a Breadth First Search algorithm. The center of gravity of each created group can be calculated, and the calculated centers of gravity can be determined as the starting point.

In the endpoint determination step (S320), the lesion analysis device generates a skeletal structure by skeletonizing the brain blood vessel surface model composed of a set of equivalent curves, converts the generated skeletal structure into a data structure, and specifies the leaf nodes to determine the endpoint. decide. Each step of determining the end point will be described in detail in FIG. 4.

In the step of extracting the center line by tracking the boundaries of the cells (S330), the lesion analysis device can extract the center line by tracking the edges of a plurality of cells created based on the vertices of the equivalent curves from the start point to the end point. there is.

Additionally, the lesion analysis device tracks the boundary surface and can obtain the tangent vector, normal vector, and binormal vector of the center line.

FIG. 4 is a flowchart explaining the endpoint determination step shown in FIG. 3 in more detail.

Referring to FIG. 4, the endpoint determination step (S320) includes the skeletal structure creation step (S410), the skeletal structure refining step (S420), the creation of a linked list based on the skeletal structure (S430), and the leaf node It may include a step (S440) of determining end points by specifying them.

The skeletal structure generation step (S410) may be a step of creating a skeletal structure by skeletonizing the brain blood vessel region or a surface model of the brain blood vessel region.

In the step of refining the skeletal structure (S420), the lesion analysis device may refine the skeletal structure by pruning branches of a certain length or less among the blood vessel branches forming the skeletal structure.

In the step of generating a linked list based on the skeletal structure (S430), the lesion analysis device may generate a linked list with a tree structure based on the skeletal structure.

In the step of determining endpoints by specifying leaf nodes (S440), the lesion analysis device may specify leaf nodes that do not have child nodes and determine endpoints corresponding to the leaf nodes.

The lesion analysis device according to an embodiment of the present invention can automatically extract the start and end points of the center line, thereby reducing errors due to the user's arbitrary selection.

FIG. 5 is a flowchart illustrating in more detail the steps for extracting cerebrovascular characteristics shown in FIG. 1.

Referring to FIG. 5, the step of extracting cerebrovascular characteristics (S140) includes grouping based on branch points (S510), and calculating cerebrovascular characteristics for each group (S520).

The grouping step (S510) based on the branch point may be a step in which the lesion analysis device groups brain blood vessels into a plurality of groups based on the branch point of the center line of the brain blood vessel model. Here, the branch point may be a point in the skeletal structure of brain blood vessels at which one blood vessel branches into one or more blood vessels.

The center lines extracted in step S230 are not grouped according to each branch point. Therefore, in order to calculate the geometric characteristics of each location and the characteristics of each node, blood vessels can be grouped or divided based on branch points.

Grouped sections of brain blood vessels can be implemented as a linked list, which is a data structure that is connected to each other.

In the step of calculating cerebrovascular characteristics for each group (S520), the lesion analysis device may calculate the plurality of cerebrovascular characteristics for each of the groups. The lesion analysis device may calculate a plurality of cerebrovascular characteristics at all locations and in each group based on the vascular groups based on the cerebrovascular model. Here, the brain vessel model may include the above-described brain vessel surface model, center line model, and grouping connection list.

The plurality of cerebrovascular characteristics include cerebrovascular cross-sectional area, maximum inscribed hole radius, maximum radius, minimum radius, maximum-minimum radius ratio, surface circumference, tortuosity, curvature, circularity, branch angle, and calculated values calculated based on these. It may include at least one of: The cerebrovascular characteristics may be calculated based on the tangent vector, normal vector, and dual method vector of the center line described above. Therefore, compared to the conventional lesion analysis method that generates a cerebrovascular model based on a volumetric data set in a discrete space and calculates geometric characteristics based on this, the lesion analysis method according to an embodiment of the present invention is based on the cerebral blood vessel terminals. It is possible to calculate more accurate brain blood vessel characteristics down to the fine structure of the brain.

Figure 6 is a diagram illustrating a brain blood vessel model created according to an embodiment of the present invention.

Referring to FIG. 6, the brain blood vessel model includes a brain blood vessel surface model 10 and a center line 20.

The cerebrovascular surface model 10 converts the cerebrovascular region consisting of a volume data set into a set of equivalent surfaces using the marching cube algorithm, and uses the vertices of the equivalent curves and the connection information of the vertices to create a continuous model. It is a model in space.

The center line 20 is created by setting a set of spatial coordinates in which the closest vertex among the vertices of the equivalent curves is the same as one cell and extracting it by tracking along the boundary of the cell connecting the start point (S) and the end point (E). It is a model in continuous space.

A branch point P is located at a point where one center line 20 branches off into a plurality of center lines 20. Based on branch points, brain blood vessels can be grouped. As an example, the cerebral blood vessels may be divided into sections based on the branch point P, and the positional relationship between each cerebral blood vessel section may be stored as a linked list.

The lesion analysis method according to various embodiments of the present invention can be written as a program that can be executed on a computer, and can be implemented in a general-purpose digital computer that operates the program using a computer-readable recording medium. The computer-readable recording media includes storage media such as magnetic storage media (eg, ROM, floppy disk, hard disk, etc.) and optical read media (eg, CD-ROM, DVD, etc.).

Figure 7 is a block diagram illustrating a lesion analysis device using cerebrovascular modeling according to an embodiment of the present invention.

Referring to FIG. 7 , the lesion analysis device 100 includes a memory 110 and a processor 120.

Memory 110 may store brain blood vessel images, software for brain blood vessel modeling and lesion analysis, and data. The memory 110 may further store various data for the overall operation of the lesion analysis device 100, such as a program for processing or controlling the processor 120. As an example, the OS, data, and commands for driving the lesion analysis device 100 may be stored. At least some of these data and applications may be downloaded from an external server through the communication unit.

The memory 110 may be implemented as internal memory such as ROM or RAM included in the processor 120, or may be implemented as a device separate from the processor 120.

Although not shown in the drawing, the lesion analysis device 100 may further include a communication unit. The communication unit may be a wired communication device or a wireless communication device for transmitting cerebrovascular images, software for cerebrovascular modeling and lesion analysis, and data to the lesion analysis device 100. When the communication unit uses a Wi-Fi chip or a Bluetooth chip, various connection information such as SSID and session key are first transmitted and received, and various information can be transmitted and received after establishing a communication connection using this. A wireless communication chip refers to a chip that performs communication according to various communication standards such as IEEE, Zigbee, 3G (3rd Generation), 3GPP (3rd Generation Partnership Project), and LTE (Long Term Evolution). An NFC chip refers to a chip that operates in the NFC (Near Field Communication) method using the 13.56MHz band among various RF-ID frequency bands such as 135kHz, 13.56MHz, 433MHz, 860~960MHz, 2.45GHz, etc.

The processor 120 is configured to overall control the lesion analysis device 100. Specifically, the processor 120 controls the overall operation of the lesion analysis device 100 using various programs stored in the memory 110 of the lesion analysis device 100. For example, the processor 120 may include a CPU, RAM, ROM, and a system bus. According to an embodiment of the present invention, the processor 120 may be implemented as a digital signal processor (DSP), a microprocessor, or a time controller (TCON) that processes digital signals. However, It is not limited to, but is not limited to, a central processing unit (CPU), a micro controller unit (MCU), a micro processing unit (MPU), a controller, an application processor (AP), or a communication processor ( It may include one or more of a communication processor (CP), an ARM processor, or be defined by the corresponding term. In addition, the processor 120 is a system on chip (SoC) with a built-in processing algorithm, and a large scale integration (LSI). ), or it may be implemented in the form of an FPGA (Field Programmable Gate Array).

The processor 120 extracts a cerebrovascular region from a cerebrovascular image, generates a cerebrovascular model in continuous space based on the extracted cerebrovascular region, calculates a plurality of cerebrovascular characteristics from the cerebrovascular model, and It may be configured to derive the condition and prognosis of the disease based on the characteristics.

The lesion analysis device 100 may receive a cerebrovascular image from the external device 200, analyze the image, derive the disease state and prognosis, and transmit the results to the external device 200. Here, the external device 200 may be, but is not limited to, a cerebrovascular image acquisition device, a personal computer, a personal terminal, and a medical data storage at a hospital or research institute.

Figure 8 is a conceptual diagram explaining a lesion analysis system using cerebrovascular modeling according to an embodiment of the present invention.

Referring to FIG. 8, the lesion analysis device 300 may be connected to a network and provide a lesion analysis service to a plurality of external devices 400.

The lesion analysis device 300 may be operated in the form of a server, and may be a server provided by a hospital, company, or research institute, but is not limited thereto.

Here, the external devices 400 may be a cerebrovascular image acquisition device, a personal computer, a personal terminal, and a medical data storage. In one embodiment, the external devices 400 may be servers from different medical service providers, and each server may be connected to a personal computer, personal terminal, etc. to provide services.

The lesion analysis device 300 can receive the patient's cerebrovascular image from the external devices 400, analyze it, and provide the external devices 400 with results derived from the patient's disease state and prognosis. there is. Therefore, the lesion analysis device 300 according to this embodiment can provide lesion analysis services to private hospitals, companies, or individual users who do not have high-speed computing devices for artificial intelligence algorithms and image analysis.

Figure 9 is a diagram showing a screen of a lesion analysis program using cerebrovascular modeling according to an embodiment of the present invention.

Referring to FIG. 9, the lesion analysis method using cerebrovascular modeling according to an embodiment of the present invention may be an application running on a network as shown.

Users can upload cerebrovascular images and receive a cerebrovascular model or disease status and prognosis results derived based on the cerebrovascular model.

Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the attached drawings. Hereinafter, the same reference numerals will be used for the same components in the drawings, and duplicate descriptions for the same components will be omitted.

10: Brain blood vessel surface model
20: center line
100, 300: Lesion analysis device
110: memory
120: processor
200, 400: External device

Claims (12)

In a lesion analysis method performed in a lesion analysis device using cerebrovascular modeling,
Receiving a cerebrovascular image by the lesion analysis device;
extracting, by the lesion analysis device, a brain blood vessel region from the brain blood vessel image;
generating a cerebrovascular model in continuous space based on the cerebrovascular region extracted by the lesion analysis device;
calculating, by the lesion analysis device, a plurality of cerebrovascular characteristics from the cerebrovascular model; and
The lesion analysis device inputs the plurality of cerebrovascular characteristics into an artificial intelligence algorithm, and the artificial intelligence algorithm extracts disease-linked attributes to derive a corresponding disease,
The artificial intelligence algorithm extracts the disease-linked properties by learning a plurality of learning data sets containing a plurality of learning cerebrovascular characteristics, and detects diseases corresponding to the learning cerebrovascular characteristics based on the extracted disease-linked properties. A lesion analysis method using cerebrovascular modeling that is derived, evaluated and compared with the correct answer data, and learned repeatedly so that the comparison value is higher than a preset standard value. According to paragraph 1,
The step of generating the cerebrovascular model by the lesion analysis device,
generating a brain blood vessel surface model by dividing the surface of the brain blood vessel region into isosurfaces;
Evenly dividing a three-dimensional continuous space into a plurality of cells based on each vertex of the equivalent curved surface; and
A lesion analysis method using brain blood vessel modeling, comprising extracting the center line of the brain blood vessel from the interface of each cell. According to paragraph 2,
The step of dividing equally into a plurality of cells is,
A lesion analysis method using cerebrovascular modeling, comprising setting as one cell a set of spatial coordinates in which the closest vertex among the vertices of the isosurface is the same. According to paragraph 2,
The step of extracting the central line of the brain blood vessel,
determining a starting point of the center line from a lower slice of the cerebrovascular region;
generating a skeletal structure from the cerebrovascular region or the cerebrovascular surface model, and refining the skeletal structure to determine endpoints of the center line; and
A lesion analysis method using cerebrovascular modeling, comprising: extracting a center line by tracking the boundary surface of each cell connecting the start point and the end point. According to paragraph 4,
The step of determining the end points of the center line is,
generating a skeletal structure by skeletonizing the brain blood vessel region or the brain blood vessel surface model;
pruning and purifying branches of a certain length or less from the skeletal structure;
Creating a linked list of a tree structure based on the refined skeletal structure; and
A lesion analysis method using cerebrovascular modeling, comprising determining endpoints by specifying leaf nodes from the linked list. According to paragraph 1,
The step of calculating the plurality of cerebrovascular characteristics by the lesion analysis device,
Grouping cerebral blood vessels into a plurality of groups based on a branch point of the center line of the cerebral blood vessel model; and
A lesion analysis method using cerebrovascular modeling, comprising: calculating the plurality of cerebrovascular characteristics for each of the groups. According to clause 6,
The characteristics of the plurality of cerebral blood vessels are,
Containing at least one of the cerebrovascular cross-sectional area, maximum inner port radius, maximum radius, minimum radius, maximum-minimum radius ratio, surface perimeter, tortuosity, curvature, circularity, divergence angle, and calculated values calculated based on these, Lesion analysis method using cerebrovascular modeling. delete delete delete Memory for storing cerebrovascular images, software for cerebrovascular modeling and lesion analysis, and data; and
Extract the cerebrovascular region from the cerebrovascular image,
Based on the extracted cerebrovascular region, a cerebrovascular model is created in continuous space,
At least configured to calculate a plurality of cerebrovascular characteristics from the cerebrovascular model, input the calculated plurality of cerebrovascular characteristics into an artificial intelligence algorithm, and the artificial intelligence algorithm extracts disease-linked properties to derive a corresponding disease. Contains one processor,
The artificial intelligence algorithm extracts the disease-linked properties by learning a plurality of learning data sets containing a plurality of learning cerebrovascular characteristics, and detects diseases corresponding to the learning cerebrovascular characteristics based on the extracted disease-linked properties. A lesion analysis device using cerebrovascular modeling that is derived, evaluated and compared with the correct answer data, and repeatedly learned so that the comparison value is greater than a preset reference value. A computer program stored in a medium for executing the method of any one of claims 1 to 7 using a computing device.

Images