Method for determining region of interest and electronic device performing same

Abstract

A method for determining a region of interest according to an embodiment comprises the steps of: obtaining blood vessel data including first points in a three-dimensional space; generating three-dimensional shapes in the three-dimensional space; determining candidate shapes among the three-dimensional shapes on the basis of the first points; determining second points among the first points on the basis of the candidate shapes; calculating a confidence score of each of the second points; and determining a region of interest on the basis of the confidence score.

Description

Method for determining a region of interest and electronic device for performing the same

[0001] The disclosure relates to a method for determining a region of interest and an electronic device for performing the same.

[0002] An aneurysm is a disease in which a portion of an artery expands like a balloon due to weakened artery walls; related conditions may include cerebral aneurysms, aortic aneurysms, renal artery aneurysms, and splenic artery aneurysms. Among the methods for treating such aneurysms, coil embolization is a representative treatment method that involves inserting a thin coil into the aneurysm to prevent blood from flowing into the cerebral aneurysm.

[0003] For the treatment of such coil embolization, the amount of coil is determined based on size information such as the volume and length of the aneurysm; therefore, precisely measuring size information along with location information is considered the most important factor in aneurysm treatment. Previously, medical staff determined the location of the aneurysm by viewing images of the patient's aneurysm and predicted its size using polygons such as octahedrons. However, there is a need for a method that enables medical staff to detect aneurysms more conveniently and efficiently, or to determine size information more precisely to be identical to the actual aneurysm.

[0004] One embodiment aims to provide a method and apparatus for rapidly and efficiently detecting a region of interest.

[0005] One embodiment aims to provide a method and apparatus for determining a region of interest using a point or a plane.

[0006] One embodiment aims to provide a method and apparatus for rapidly calculating quantitative parameters of a three-dimensional region of interest.

[0007] However, the problems that the present invention aims to solve are not limited to those mentioned above, and may include problems that are not mentioned but can be clearly understood by those skilled in the art from the description below.

[0008] A method for determining a region of interest according to one embodiment for solving such technical problems includes the steps of: acquiring blood vessel data including first points in a three-dimensional space; generating three-dimensional shapes in the three-dimensional space; determining candidate shapes among the three-dimensional shapes based on the first points; determining second points among the first points based on the candidate shapes; calculating a reliability score for each of the second points; and determining a region of interest based on the reliability score.

[0009] An electronic device according to one embodiment includes a processor and a memory connected to the processor, the memory is configured to store a program, the processor is configured to execute the program, and when the program is executed, the steps of the method according to the embodiments are implemented.

[0010] FIG. 1 is a schematic block diagram of a computing system according to one embodiment.

[0011] FIG. 2 is a block diagram of an electronic device according to one embodiment.

[0012] FIG. 3 is a flowchart illustrating a method for determining a region of interest according to one embodiment.

[0013] FIG. 4 is an example of a blood vessel mesh according to one embodiment.

[0014] FIG. 5 is an example of a point cloud according to one embodiment.

[0015] FIG. 6 is an example of a voxel grid according to one embodiment.

[0016] FIG. 7 is an example of a three-dimensional shape according to one embodiment.

[0017] FIG. 8 is an example of a filtered three-dimensional shape according to one embodiment.

[0018] FIG. 9 is an example of points extracted from a filtered three-dimensional shape according to one embodiment.

[0019] FIG. 10 is an example of points where sampling was performed according to one embodiment.

[0020] FIG. 11 is an example of normal vectors of vertices according to one embodiment.

[0021] FIG. 12 is an example of points to which normal vectors are assigned according to one embodiment.

[0022] FIG. 13 is a drawing for explaining a configuration for classifying points according to one embodiment.

[0023] FIG. 14 is a diagram illustrating the operation of an aneurysm localization model according to one embodiment.

[0024] FIG. 15 is a drawing for explaining a configuration for performing subdivision at a point according to one embodiment.

[0025] FIG. 16 is a drawing for explaining a configuration for displaying a region of interest according to one embodiment.

[0026] FIG. 17 is a drawing for explaining a configuration for displaying a region of interest according to one embodiment.

[0027] The various embodiments described in this specification are illustrative for the purpose of clearly explaining the technical concept of this disclosure and are not intended to limit it to specific embodiments. The technical concept of this disclosure includes various modifications, equivalents, alternatives, and embodiments optionally combined from all or part of each embodiment described in this specification. Furthermore, the scope of the technical concept of this disclosure is not limited to the various embodiments presented below or the specific descriptions thereof.

[0028] Terms used in this specification, including technical or scientific terms, may have the meaning generally understood by those skilled in the art to which this disclosure pertains, unless otherwise defined.

[0029] Expressions used herein such as “comprising,” “may compose,” “possessing,” “possessing,” “having,” and “possessing” imply the existence of the subject feature (e.g., function, operation, or component, etc.) and do not exclude the existence of other additional features. That is, such expressions should be understood as open-ended terms implying the possibility of including a second embodiment.

[0030] In this specification, singular expressions include plural expressions unless the context clearly specifies them as singular. Additionally, plural expressions include singular expressions unless the context clearly specifies them as plural. Throughout the specification, when a part is described as including a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0031] Additionally, the terms 'module' or 'part' as used in the specification refer to software or hardware components, and the 'module' or 'part' performs certain roles. However, the meaning of 'module' or 'part' is not limited to software or hardware. The 'module' or 'part' may be configured to reside in an addressable storage medium or configured to run on one or more processors. Thus, as an example, the 'module' or 'part' may include components such as software components, object-oriented software components, class components, and task components, and at least one of processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, or variables. The components and the functions provided within the 'module' or 'part' may be combined into a smaller number of components and 'modules' or 'parts', or further separated into additional components and 'modules' or 'parts'.

[0032] According to one embodiment of the present disclosure, a ‘module’ or ‘part’ may be implemented as a processor and memory. The term ‘processor’ should be broadly interpreted to include a general-purpose processor, a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a controller, a microcontroller, a state machine, etc. In some environments, the term ‘processor’ may refer to an application-specific integrated circuit (ASIC), a programmable logic device (PLD), a field programmable gate array (FPGA), etc. The term ‘processor’ may also refer to a combination of processing devices, such as, for example, a combination of a DSP and a microprocessor, a combination of multiple microprocessors, a combination of one or more microprocessors combined with a DSP core, or any other combination of such configurations. Additionally, the term ‘memory’ should be broadly interpreted to include any electronic component capable of storing electronic information. 'Memory' may refer to various types of processor-readable media, such as Random Access Memory (RAM), Read-Only Memory (ROM), Non-Volatile Random Access Memory (NVRAM), Programmable Read-Only Memory (PROM), Erasable-Programmable Read-Only Memory (EPROM), Electrically Erasable PROM (EEPROM), Flash Memory, Magnetic or Optical Data Storage Devices, Registers, etc. If a processor can read information from memory and / or write information to memory, the memory is said to be in an electronic communication state with the processor. Memory integrated into a processor is in an electronic communication state with the processor.

[0033] Expressions such as "first," "second," or "first," "second" as used in this specification are used to distinguish one object from another when referring to a plurality of objects of the same kind, unless otherwise indicated by the context, and do not limit the order or importance of said objects.

[0034] Expressions used herein such as “A, B, and C,” “A, B, or C,” “A, B, and / or C,” or “at least one of A, B, and C,” “at least one of A, B, or C,” “at least one of A, B, and / or C,” “at least one selected from A, B, and C,” “at least one selected from A, B, or C,” “at least one selected from A, B, and / or C,” etc., may mean each of the listed items or all possible combinations of the listed items. For example, “at least one selected from A and B” may refer to (1) A, (2) at least one of A, (3) B, (4) at least one of B, (5) at least one of A and at least one of B, (6) at least one of A and B, (7) at least one of B and A, and (8) all of A and B.

[0035] As used herein, the expression “based on” is used to describe one or more factors affecting an act or action of a decision or judgment described in the phrase or sentence containing such expression, and such expression does not exclude additional factors affecting said act or action of a decision or judgment.

[0036] As used in this specification, the expression that a certain component (e.g., a first component) is "connected" or "connected" to another component (e.g., a second component) may mean that the said certain component is not only directly connected or connected to the said other component, but is also connected or connected through a new other component (e.g., a third component).

[0037] As used herein, the expression "configured to" may have meanings such as "set to," "capable of," "modified to," "made to," or "capable of." Such expression is not limited to the meaning of "specifically designed in hardware," and, for example, a processor configured to perform a specific operation may mean a generic-purpose processor capable of performing that specific operation by executing software.

[0038] Various embodiments of the present disclosure will be described below with reference to the accompanying drawings. In the accompanying drawings and the description thereof, identical or substantially equivalent components may be given the same reference numerals. Furthermore, in the description of the various embodiments below, the description of identical or corresponding components may be omitted, but this does not mean that such components are not included in the embodiments.

[0039]

[0040] FIG. 1 is a schematic block diagram of a computing system according to one embodiment, and FIG. 2 is a block diagram of an electronic device according to one embodiment.

[0041] Referring to FIG. 1, a computing system (10) according to one embodiment may acquire a blood vessel image of a first user, perform image processing on the acquired blood vessel image, and display the processed image. The computing system (10) may provide the processed image to a second user. For example, the first user may be a patient, and the second user may be a medical professional.

[0042] In one embodiment, the computing system (10) may generate a region of interest by photographing the blood vessel of the first user to obtain a blood vessel image and performing image processing on the blood vessel image. The computing system (10) may display the region of interest through points, planes, etc. The computing system (10) may photograph blood vessels that are subject to angiography, such as cerebral blood vessels, cardiovascular vessels, gastrointestinal blood vessels, etc., but the embodiments are not necessarily limited thereto.

[0043] A computing system (10) according to one embodiment includes a shooting device (100) and an electronic device (200). The shooting device (100) may be a device configured to capture a first user and acquire an image. For example, the shooting device (100) may be an X-ray imaging device (e.g., C-arm X-ray device), angiography device (e.g., angio device), optical coherence tomography (OCT) device, computed tomography (CT) device, magnetic resonance imaging (MRI), magnetic resonance angiography (MRA) device, etc., but the embodiment is not necessarily limited thereto and may be implemented as various devices configured to capture the blood vessels of the first user.

[0044] The imaging device (100) can photograph a first user at multiple shooting points and acquire multiple images. In one embodiment, the imaging device (100) can photograph the first user while rotating around the first user. In another embodiment, the imaging device (100) can rotate the first user and photograph the rotating first user. The imaging device (100) can transmit the acquired multiple images to an electronic device (200). Depending on the embodiment, the imaging device (100) may be implemented as a mono-plane device or a biplane device.

[0045] The electronic device (200) can generate a region of interest based on a plurality of images. For example, the region of interest may refer to an aneurysm region. The electronic device (200) can perform the method of determining the region of interest of FIG. 3. For example, the electronic device (200) can prepare first points corresponding to three-dimensional shapes and blood vessels in three-dimensional space. The electronic device (200) can determine candidate shapes among the three-dimensional shapes based on the first points. The electronic device (200) can determine second points among the first points based on the candidate shapes and calculate a reliability score for each of the second points. The electronic device (200) can determine the region of interest based on the reliability score.

[0046] In this way, the electronic device (200) can conveniently and quickly determine the area of ​​interest by processing vascular data such as the first point in three dimensions. Through this, the second user can reduce reading errors and achieve efficient processing speed, such as examining even the area of ​​a minute aneurysm.

[0047] The electronic device (200) may be a server, data center, artificial intelligence (AI) device, personal computer (PC), laptop computer, mobile phone, smartphone, tablet PC, wearable device, healthcare device, etc.

[0048] In some embodiments, where the electronic device (200) is implemented as a server, data center, etc., the computing system (10) may further include an electronic device for interacting with a second user. For example, such an electronic device may interact with the second user, such as displaying an image to the second user and receiving input from the second user. The electronic device of the second user may communicate with the shooting device (100) and / or the electronic device (200).

[0049] Referring to FIG. 2, an electronic device (200) according to one embodiment may include a processor (210) and a memory (220) connected to the processor (210). The memory (220) may be configured to store a program. The processor (210) may be configured to execute a program in the memory (220). When the program is executed, a method for determining a region of interest (or steps of the method) according to embodiments may be implemented.

[0050] The memory (220) may be configured to store images received from the imaging device (100). Images stored in the memory (220) may be loaded by the processor (210) and used to generate three-dimensional data (e.g., a blood vessel mesh). The processor (210) may perform calculations based on the images received from the imaging device (100), generate three-dimensional data, and determine a region of interest in the three-dimensional data.

[0051] In FIG. 2, for convenience of explanation, the electronic device (200) is illustrated as including a processor (210) and a memory (220), but the embodiment is not necessarily limited thereto, and the electronic device (200) may be implemented to include at least one additional component. For example, the electronic device (200) may further include components such as an input / output interface, a communication module, and a display. The input / output interface may be a component for an interface between an input / output device and the electronic device (200). The communication module may be a component for communication with a device outside the electronic device (200) (e.g., a camera (100), a server, etc.). The display may be a component for displaying the output of the processor (210) (e.g., an input image, 3D data, a region of interest, etc.).

[0052]

[0053] FIG. 3 is a flowchart illustrating a method for determining a region of interest according to one embodiment.

[0054] Referring to FIGS. 1 and FIGS. 3, an electronic device (200) according to one embodiment can acquire blood vessel data including first points in a three-dimensional space (S310). The electronic device (200) can acquire a vessel mesh and acquire blood vessel data from the vessel mesh. The vessel mesh is data representing blood vessels in three dimensions and can be generated based on a plurality of images captured by a imaging device (100). The plurality of images include blood vessels and may include, for example, X-ray images, ultrasound (or sonography) images, CT (Computed Tomography) images, PET (Positron Emission Tomography) images, MRI (Magnetic Resonance Imaging) images, fMRI (functional Magnetic Resonance Imaging) images, digital pathology WSI (digital pathology Whole Slide Image), DBT (Digital Breast Tomosynthesis) images, etc.

[0055] According to an embodiment, the electronic device (200) may generate a blood vessel mesh generated by the imaging device (100) or generate a blood vessel mesh by receiving a plurality of images from the imaging device (100). For example, the electronic device (200) may generate a blood vessel mesh using a marching cube algorithm, but the embodiment is not necessarily limited thereto. The electronic device (200) may obtain vertex information included in the blood vessel mesh as blood vessel data. For example, the electronic device (200) may remove edge information from the blood vessel mesh and extract only vertex information. The electronic device (200) may perform voxelization based on the vertex information.

[0056] The electronic device (200) can generate three-dimensional shapes in three-dimensional space (S320). The electronic device (200) can generate a voxel grid that divides three-dimensional space. The voxel grid has a predetermined size based on each axis of three-dimensional space and may include a plurality of grid cells.

[0057] The electronic device (200) can generate three-dimensional shapes based on a voxel grid. The three-dimensional shapes may represent anchor boxes. The electronic device (200) can generate anchor boxes by moving along each axis by a predetermined interval from the origin. For example, the electronic device (200) can generate anchor boxes based on a plurality of predetermined ratios for width, height, and depth.

[0058] In one embodiment, the electronic device (200) can obtain the size of a voxel grid. The electronic device (200) can obtain a scale value for scaling anchor boxes. The electronic device (200) can determine a predetermined interval based on the size of the voxel grid and the scale value. The electronic device (200) can determine the volume of anchor boxes based on the scale value.

[0059] The electronic device (200) can determine candidate shapes among three-dimensional shapes based on first points (S330). That is, the electronic device (200) can filter the three-dimensional shapes. The electronic device (200) can determine the number of first points included in each of the three-dimensional shapes. The electronic device (200) can determine candidate shapes among the three-dimensional shapes based on the determined number of first points.

[0060] The electronic device (200) can determine first shapes among three-dimensional shapes in which the number of determined first points is greater than a first reference value. The electronic device (200) can determine candidate shapes among the first shapes based on the degree of overlap.

[0061] For example, the electronic device (200) can determine second shapes among the first shapes whose degree of overlap is greater than a second reference value. The electronic device (200) can determine one of the second shapes as a candidate shape. In this way, the electronic device (200) can effectively detect a region of interest by selecting an anchor box that meets specific conditions through two filtering steps. According to an embodiment, the electronic device (200) may use at least one of the two filtering methods.

[0062] The electronic device (200) can determine second points among first points based on candidate shapes (S340). The electronic device (200) can extract points corresponding to the candidate shapes. The electronic device (200) can determine grid cells included in the candidate shapes. The electronic device (200) can determine points among the first points that are included in the grid cells as second points.

[0063] In the embodiments, the electronic device (200) can obtain third points by performing sampling on the second points. For example, sampling may include upsampling, downsampling, uniform sampling, etc., and may make the distribution of the second points uniform. As an example, the electronic device (200) can obtain uniformly distributed third points by performing uniform sampling on the second points. As another example, the electronic device (200) can obtain third points in which points are newly created by performing upsampling on the second points.

[0064] The electronic device (200) can calculate a confidence score for each of the second points (S350). The electronic device (200) can acquire a blood vessel mesh corresponding to the blood vessel data and generate normal vectors for the vertices of the blood vessel mesh. The electronic device (200) can assign the normal vectors to the second points. The electronic device (200) can calculate a confidence score based on the first coordinates of the second points and the normal vectors.

[0065] In one embodiment, the electronic device (200) can determine the vertex closest to the third point among the second points. The electronic device (200) can assign the normal vector of the vertex to the third point.

[0066] The electronic device (200) can perform centroid normalization on the second points. For example, the electronic device (200) can calculate the average coordinate of the first coordinates of the second points. The electronic device (200) can obtain the second coordinates from the first coordinates based on the average coordinate. The electronic device (200) can calculate a confidence score based on the second coordinates and normal vectors. For example, the electronic device (200) can use a three-dimensional point cloud classification model.

[0067] In the embodiments, the electronic device (200) may calculate the reliability score of the points where sampling was performed.

[0068] The electronic device (200) can determine a region of interest based on a reliability score (S360). For example, the electronic device (200) can determine third points among the second points where the reliability score is greater than a third reference value. The electronic device (200) can perform point cloud segmentation on the third points. The electronic device (200) can determine a region of interest based on the segmentation results.

[0069] In one embodiment, the electronic device (200) can classify the third points by class. The electronic device (200) can determine the fourth points with the highest reliability scores in each class. The electronic device (200) can determine a region of interest based on the fourth points.

[0070] In one embodiment, the electronic device (200) can determine fifth points among the fourth points where the reliability score is greater than the fourth reference value. The electronic device (200) can determine a region of interest based on the fifth points.

[0071] The electronic device (200) can obtain a method for displaying the region of interest. For example, the user can input whether to display the region of interest using points or using planes to display the region of interest to the electronic device (200).

[0072] In one embodiment, when the display method is a point, the electronic device (200) may cluster the fifth points into a plurality of groups, determine the sixth point with the highest reliability score in each of the plurality of groups, and display the sixth point as a region of interest.

[0073] In one embodiment, the electronic device (200) can display a region of interest by generating a separation surface based on fifth points and separating the blood vessel mesh based on the separation surface, when the display method is planar.

[0074] In this way, the electronic device (200) can automatically detect and display the location of an area of ​​interest, such as a cerebral aneurysm, and derive an automatic plane that distinguishes between normal blood vessels and cerebral aneurysms, thereby enabling the rapid calculation of quantitative information such as the volume of the cerebral aneurysm. Accordingly, the electronic device (200) significantly reduces the burden of interpretation on medical personnel and makes it easier to detect even minute cerebral aneurysms. In addition, the electronic device (200) enables objective data-based follow-up observation, thereby increasing accuracy in the diagnosis and treatment process and providing the effect of supporting rapid diagnosis even in emergency situations.

[0075]

[0076] FIGS. 4 to 17 are drawings for explaining a method for determining a region of interest according to one embodiment.

[0077] Referring to FIGS. 4 through 17, an electronic device according to one embodiment (e.g., 200 in FIG. 1) can acquire a three-dimensional blood vessel mesh. The mesh may refer to data composed of vertices, edges, and polygons (or faces). The electronic device may receive a plurality of blood vessel images from a imaging device (e.g., 100 in FIG. 1) and generate a blood vessel mesh based on the plurality of blood vessel images. The electronic device may use a marching cube algorithm, but the embodiments are not necessarily limited thereto. According to an embodiment, the imaging device may generate a blood vessel mesh from a plurality of blood vessel images and transmit it to the electronic device. An example of a blood vessel mesh acquired by the electronic device according to one embodiment may be as shown in FIG. 4.

[0078] An electronic device can generate a point cloud from a blood vessel mesh. In one embodiment, the electronic device can generate a point cloud by extracting vertex information from a blood vessel mesh. For example, the electronic device can obtain vertex information by removing edge information and polygon information from a blood vessel mesh. Vertex information may include coordinates of (x, y, z). That is, the point cloud may be a set of coordinates. An example of a point cloud generated from a blood vessel mesh by an electronic device according to one embodiment may be as shown in FIG. 5. In FIG. 5, the point cloud may include a plurality of points such as point (P1).

[0079] In one embodiment, the electronic device may perform discretization (or adaptive voxelization) on the point cloud. The electronic device may originate the point cloud so that its minimum value becomes (0,0,0). The electronic device may calculate a voxel index based on the originated coordinates and the unit voxel size. The voxel index is a value indicating which cell a point is located in at the voxel level; for example, the electronic device may calculate the voxel index by dividing the coordinates by the unit voxel size. The electronic device may calculate the total voxel size by calculating the maximum and minimum values ​​for each axis of the point cloud and calculating the rounding up of the difference between the maximum and minimum values ​​on one axis. For example, the electronic device may calculate the x-axis size based on the difference between the maximum and minimum values ​​on the x-axis and calculate the total axis size in the same manner to calculate the total voxel size. Accordingly, an adaptive voxel grid that takes into account the original mesh size may be generated.

[0080] In one embodiment, the electronic device can generate a voxel grid and map a point cloud to the voxel grid. The electronic device can generate a voxel grid having a predetermined size. The voxel grid is a three-dimensional space divided into equal intervals and may include a plurality of grid cells. The voxel grid may have a size of a*b*c. That is, the voxel grid may have a size of a in the x-axis direction, a size of b in the y-axis direction, and a size of c in the z-axis direction, and may include a number of grid cells of a*b*c.

[0081] An electronic device can map a point cloud to a voxel grid. In one embodiment, the electronic device may perform normalization on the point cloud prior to mapping. For example, the electronic device may perform normalization based on the range of values ​​that the points of the point cloud have. The range may include a minimum value and a maximum value. In one embodiment, the minimum value may be 0 and the maximum value may be 1, but the embodiment is not necessarily limited thereto.

[0082] The electronic device can set the value of a grid cell among a plurality of grid cells of a voxel grid that has a mapped point to '1'. The electronic device can set the value of a grid cell among a plurality of grid cells of a voxel grid that does not have a mapped point to '0'. The electronic device can store mapping information that matches a grid cell with a point included in that grid cell. An example of the result of an electronic device according to one embodiment mapping three-dimensional point cloud data to a voxel grid may be as shown in FIG. 6.

[0083] An electronic device can generate three-dimensional shapes on a voxel grid. For example, the electronic device can generate a three-dimensional anchor box. The anchor box may have parameters of width, height, and depth (or width, length, and height). Width, height, and depth may be mutually orthogonal parameters. The electronic device can generate an anchor box based on parameters having a predetermined ratio. For example, the electronic device may determine the ratio of the parameters to be 1:1:1 or greater. The electronic device may randomly select one of several size ratios and generate an anchor box based on the selected size ratio. That is, the electronic device may randomly determine the width, height, and depth. Accordingly, the anchor boxes generated by the electronic device may be implemented to have different sizes.

[0084] In one embodiment, the electronic device can generate anchor boxes by moving along each axis at predetermined intervals starting from the coordinates of the origin. The electronic device can generate anchor boxes using each point moved from the origin as a center point. The coordinates of the origin may be (0, 0, 0), but depending on the embodiment, the electronic device may designate the coordinates of another point as the origin.

[0085] The electronic device can set the center point of the anchor box to (i * ST, j * ST, k * ST). Here, ST is a stride value indicating the interval at which the center point moves, and i, j, and k can each represent a voxel grid index.

[0086] When the actual size of each axis of the voxel grid is denoted as G_x, G_y, and G_z, the number of anchor centers for each axis is defined as N_x = G_x / ST, N_y = G_y / ST, and N_z = G_z / ST, and the total number of anchor boxes is N_x * N_y * N_z, where ST can be a stride value.

[0087] The electronic device can set the volume of each anchor box to be equal. The electronic device can generate anchor boxes based on a stride value and a scale value. The stride value is the interval of movement to generate the anchor box, and the scale value may be a parameter for scaling the anchor box. The electronic device can determine the stride value based on the size of the voxel grid. For example, the electronic device can determine the stride value based on Equation 1.

[0088]

[0089] Here, ST is the stride value, SC is the scale value, and GSZ may be the size of the voxel grid. For example, if the voxel grid has a size of 128*128*128 and the scale value is 16, the electronic device may determine the stride value as 8. That is, the electronic device can generate anchor boxes by moving the center point by a distance of 8 on each axis. According to an embodiment, the electronic device may receive the scale value from the user or automatically determine the scale value.

[0090] The electronic device can determine the volume of the anchor box based on the scale value. For example, the electronic device can determine the volume of the anchor box based on Equation 2.

[0091]

[0092] Here, VOL is the volume of the anchor box, and SC may be a scale value. For example, if the scale is 8, the electronic device may determine the volume of each anchor box as 4,096. The anchor boxes generated by the electronic device may be candidates for detecting a region of interest (e.g., a cerebral aneurysm region) in three-dimensional space. That is, the electronic device may determine the region of interest based on the anchor boxes. An example of an anchor box generated by the electronic device according to one embodiment may be as shown in FIG. 7. In FIG. 7, the anchor box may include a plurality of boxes such as box (BX1).

[0093] The electronic device can determine a candidate box among the anchor boxes. The electronic device can determine a first candidate box based on the anchor boxes and the voxel grid. The first candidate box may refer to a box selected through first-order filtering among the anchor boxes. For example, the electronic device can count the values ​​of the grid cells contained within the anchor boxes. Since the value of a grid cell containing a point in the point cloud is '1' and the value of a grid cell without a point is '0', the electronic device can filter anchor boxes that do not contain a point or contain a relatively small number of points by adding the values ​​of the grid cells contained within the anchor box. That is, the electronic device can count the number of grid cells with a value of '1' within the anchor box.

[0094] The electronic device can determine anchor boxes whose sum of the values ​​of the grid cells is greater than (or greater than or equal to) a first reference value as first candidate boxes. According to an embodiment, the first reference value may be determined by a user or automatically determined by the electronic device.

[0095] The electronic device can obtain second candidate boxes by performing second filtering on first candidate boxes. The electronic device can perform second filtering based on the degree of overlap of the first candidate boxes (e.g., Intersection over Union (IoU)). For example, among two first candidate boxes whose degree of overlap is greater than (or greater than or equal to) a second reference value, the electronic device may retain only one first candidate box and remove the other first candidate box. The electronic device may randomly select the first candidate box to retain or remove.

[0096] According to an embodiment, the electronic device may pre-set a priority for maintaining or removing, and may maintain or remove a first candidate box according to the priority.

[0097] In one embodiment, if there are multiple other fourth candidate boxes among the first candidate boxes that have a degree of overlap with the third candidate box greater than the second reference value, the electronic device may determine a second candidate box based on the sum of the grid cell values. For example, the electronic device may determine the box with the largest sum of grid cell values ​​among the third candidate box and the fourth candidate boxes as the second candidate box and remove the remaining boxes. However, the embodiment is not necessarily limited thereto, and the electronic device may perform filtering by removing at least one box with a relatively low sum of grid cell values.

[0098] In another embodiment, if there are multiple other fourth candidate boxes among the first candidate boxes whose degree of overlap with the third candidate box is greater than the second reference value, the electronic device may retain the third candidate box and remove the fourth candidate boxes. For example, there may be candidate box B and anchor box C whose IoU with candidate box A is greater than the second reference value. The IoU between candidate box B and candidate box C may be less than the second reference value. The electronic device may retain candidate box A and remove candidate box B and candidate box C. The electronic device may perform secondary filtering on all first candidate boxes and determine the remaining boxes as second candidate boxes.

[0099] The electronic device can prevent excessive overlap of anchor boxes covering similar adjacent spaces through secondary filtering. The electronic device can maintain only anchor boxes where points are sufficiently distributed and not excessively overlapped through primary and secondary filtering. An example of a candidate box determined from anchor boxes by the electronic device according to one embodiment may be as shown in FIG. 8. The candidate box may be filtered from the anchor box. In FIG. 8, the candidate box may include a plurality of boxes such as box (FB1).

[0100] Although the above description describes an electronic device performing both first-order filtering and second-order filtering, the embodiments are not necessarily limited thereto. That is, depending on the embodiment, the electronic device may be implemented to obtain candidate boxes by performing either first-order filtering or second-order filtering.

[0101] The electronic device can extract grid cells included in the secondary candidate boxes. The electronic device can extract cells with a value of '1' from among the extracted grid cells. The electronic device can obtain original point information using mapping information. Based on the mapping information, the electronic device can determine the points corresponding to the cells with a value of '1'. That is, the electronic device can obtain the coordinates of the points. By utilizing the mapping information, the electronic device can extract the regions in space where points are distributed.

[0102] An example of points extracted from a candidate box by an electronic device according to one embodiment may be as shown in FIG. 9.

[0103] The electronic device can perform sampling based on determined points. For example, the points of the extracted grid cells may be distributed at different densities. Accordingly, the electronic device can perform upsampling to duplicate points, downsampling to reduce the number of points, or uniform sampling to make the placement of points uniform. In other words, if the number of determined points is too large, the electronic device can appropriately reduce it, and conversely, if it is insufficient, it can generate new points to meet a fixed number. The electronic device can determine that there are too many points if the number of determined points is greater than (or equal to) a threshold, and determine that there are too few points if the number of points is less than (or equal to) the threshold.

[0104] In one embodiment, when the number of determined points is large, the electronic device can utilize the Farthest Point Sampling (FPS) technique to uniformly extract a specific number of points. For example, the electronic device can select a reference point among the determined points and select the point farthest from the reference point. The electronic device can repeatedly select the point farthest from the selected points. The selected points may represent the set of all points previously selected by the electronic device. Accordingly, the electronic device can obtain a set of points that are evenly distributed across the entire space. Through sampling, the electronic device can obtain an even sample even when the point distribution is skewed in a specific area.

[0105] According to the embodiments, the electronic device may use random sampling, grid-based downsampling, etc. The electronic device can reduce the computational load by simplifying overly dense points through various sampling methods.

[0106] In one embodiment, if the number of determined points is insufficient, the electronic device can generate additional points to meet a certain number. For example, the electronic device can generate a new point by duplicating at least one of the determined points. As another example, the electronic device can generate a new point using a deep learning-based PU-Net (Point Cloud Upsampling Network). For example, if the number of determined points is less than a specific number, the electronic device can meet the number by repeatedly duplicating some points to make up the missing number, or by generating coordinate values ​​through an upsampling network. The upsampling network can predict new points that reflect the distribution characteristics of existing points using a trained model.

[0107] An example of the result of an electronic device performing sampling on points according to one embodiment may be as illustrated in FIG. 10. In this way, since a uniform range of points remains for each anchor box through sampling, the size of the data input in the subsequent step is maintained evenly, so the electronic device can effectively and precisely detect the region of interest.

[0108] The electronic device can generate normal vectors from a blood vessel mesh. The electronic device can detect open surfaces in the blood vessel mesh and generate a blood vessel mesh of a continuous three-dimensional structure by closing the open surfaces. That is, the electronic device can obtain a blood vessel mesh of a closed structure (hereinafter referred to as a 'closed mesh').

[0109] Electronic devices can generate normal vectors based on a closed mesh. For example, electronic devices can generate normal vectors using a normal calculation function on a closed mesh. Electronic devices can load the normal calculation function from memory or an external library. The normal calculation function can calculate a normal vector corresponding to each vertex of the closed mesh. For example, the normal calculation function can calculate a normal vector based on the geometry of the face to which the vertex belongs. The normal calculation function can extract a normal vector based on surface characteristics defined as the outward or inward direction.

[0110] The electronic device can assign the extracted normal vectors to the vertices of a closed mesh. Accordingly, each vertex can have a unique normal vector. The electronic device can store the vertices and their normal vectors in memory. An example of normal vectors generated by the electronic device based on the mesh according to one embodiment may be as shown in FIG. 11. In FIG. 11, the normal vectors may include a plurality of normal vectors such as normal vector (NV1).

[0111] An electronic device can assign a normal vector to a sampled point. In one embodiment, if the sampled point and the vertex location match, the electronic device can assign the normal vector of the vertex to the sampled point. In one embodiment, if the sampled point and the vertex location do not match, the electronic device can determine the vertex closest to the sampled point. For example, a sampled point may be generated in a new space due to upsampling, etc. The electronic device can assign the normal vector of the determined vertex to the sampled point. In one embodiment, the electronic device can determine the nearest vertex within a predetermined distance range and assign the normal vector of the determined vertex to the sampled point. That is, by assigning a normal vector in addition to the coordinate information possessed by the sampled point, the sampled point can additionally include information regarding the orientation of the surface. An example of the result of the electronic device assigning normal vectors to sampled points according to one embodiment may be as shown in FIG. 12. Referring to Fig. 12, multiple points to which normal vectors are assigned can be identified, such as the point (PP1) to which a normal vector (VT1) is assigned.

[0112] The electronic device can determine a region of interest by classifying points and performing segmentation based on the classification results. In other words, the electronic device can determine whether a region of interest exists in the blood vessel mesh through classification, and determine whether a point corresponds to the region of interest through segmentation.

[0113] An electronic device can perform center normalization on sampled points. For example, the electronic device can calculate the average of the coordinates of all sampled points. The electronic device can subtract the average coordinate from each of the coordinates of the sampled points. That is, the electronic device can shift the sampled points so that the average coordinate becomes (0, 0, 0). By performing center normalization on the points, the electronic device can reduce the deviation of the points' spatial positions.

[0114] The electronic device can input normalized points and normal vectors assigned to those normalized points into a classification model. Based on the normalized points and normal vectors, the classification model can calculate the confidence score of candidate boxes containing the corresponding normalized points. Based on the confidence score, the electronic device can determine whether a region of interest exists. The confidence score represents the probability for each class to which the point belongs, and for example, it can represent the probability that the point corresponds to the region of interest. The classification model is a 3D point cloud classification model and can calculate the confidence score using functions such as softmax, sigmoid, or scoring. The classification model may be trained to output a confidence score corresponding to the points and normal vectors when receiving them as input. Finally, the electronic device can calculate the confidence score for each of the filtered candidate boxes.

[0115] The electronic device can extract a box among the candidate boxes whose reliability score is greater than (or greater than or equal to) a third threshold value. The electronic device can extract first points contained in the extracted box. According to an embodiment, the electronic device can remove a box among the candidate boxes whose reliability score is less than or equal to (or less than) the third threshold value. An example of the first points of the boxes extracted from the candidate boxes based on the reliability score by the electronic device according to one embodiment may be as shown in FIG. 13. Referring to FIG. 13, boxes with a high probability of corresponding to the region of interest are marked to be distinguished using coloring. The box extracted by the electronic device includes first points such as point (CF1). The colors of the first points may correspond to the colors of the anchor boxes in FIG. 7.

[0116] The electronic device can determine clustering points from first points through localization and clustering. The electronic device can input the first points of the extracted boxes into an aneurysm localization model. For convenience of explanation, an aneurysm has been described as an example of a region of interest, but the embodiments are not necessarily limited thereto, and various parts of blood vessels can be set as regions of interest to implement the embodiments. The aneurysm localization model may be a machine learning model trained to predict a single point at a location where an aneurysm is likely to exist from each box that exceeds a classification threshold. The electronic device or the aneurysm localization model may select a box that exceeds the threshold among the anchor boxes.

[0117] Points generated by the aneurysm localization model are not limited to mesh vertices and can be located inside or outside the mesh. Points generated by the aneurysm localization model can simultaneously have (x, y, z) coordinates and a probability value of being located in an aneurysm. In this case, each probability value can serve as a basis for determining whether it is an aneurysm point based on a threshold.

[0118] The electronic device can display points determined to be located in an aneurysm using a first color through an aneurysm localization model. The electronic device can display points determined not to be located in an aneurysm using a second color different from the first color through an aneurysm localization model. For example, referring together with FIG. 14, points of the first color may be located as first points (POP), and points of the second color may be located as second points (NGP). According to an embodiment, the electronic device may distinguish between the first points (POP) and the second points (NGP) by displaying points corresponding to the aneurysm and points not corresponding to the aneurysm differently in a different way.

[0119] The electronic device can generate a third point (CP) by performing clustering on a first point (POP). For example, since the first point (POP) does not exist on a mesh vertex, the electronic device can map the first point (POP) to the nearest mesh vertex. That is, the electronic device can generate a third point (CP) by determining the mesh vertex closest to the first point (POP) in the vascular data or mesh and performing clustering on the mesh vertex using a clustering algorithm. The third point (CP) may represent a clustered point. In one embodiment, the algorithm may be a density-based spatial clustering nonparametric algorithm (DBSCAN), but the embodiment is not necessarily limited thereto.

[0120] A third point (CP) can be assumed to be a point located on the aneurysm, and the electronic device can generate an anchor box (ACB) centered on the third point (CP). The electronic device can determine a plurality of points included in the anchor box (ACB) and determine a region of interest based on the plurality of points.

[0121] For example, the electronic device can generate an anchor box (ACB) in the manner described with reference to FIG. 7, and can generate coordinates and normal values ​​of points inside the anchor box (ACB). The coordinates and normal values ​​can be inputs for subdivision. According to an embodiment, there may be multiple third points (CP), and accordingly, there may also be multiple anchor boxes (ACB) generated.

[0122] Referring to FIG. 15, points inside the anchor box (ACB) generated based on the third point (CP) of FIG. 14 can be input into a point cloud segmentation model. Point cloud segmentation may be an operation that classifies the segment to which a point belongs. Among the outputs of the point cloud segmentation model, points above a threshold may be the same as the fourth point (TSP) indicated in a dark color.

[0123] The electronic device can determine a representative point among the fourth points (TSPs) predicted as aneurysms. For example, the electronic device can select at least one point (FPS1 and FPS2) among the fourth points (TSPs) as a representative point using the Farthest Point Sampling (FPS) technique. This process applies the results of all anchor box segmentation and may mean that multiple points are selected from the entire mesh rather than from a single anchor box.

[0124] At least one selected point (FPS1 and FPS2) may occupy the fourth point (TSP) while drawing a sphere with a predetermined radius. The electronic device may verify the validity of the sphere. For example, the electronic device may determine that the sphere is valid if the number of points belonging to the sphere satisfies a certain number or more. The electronic device may select the point with the highest reliability score from the valid sphere as one of the final aneurysm points. The fourth point (TSP) may belong to only one sphere and may not belong to multiple spheres.

[0125] The electronic device can determine a region of interest based on at least one point (FPS1 and FPS2). The electronic device can display the region of interest using a point or a plane. That is, the electronic device can determine a point or a plane. For example, the electronic device can receive input from the user regarding how the region of interest should be displayed. That is, the user can select whether to display the region of interest as a point or as a plane.

[0126] For example, if the electronic device receives input from a user stating, "[I want to identify the region of interest as points]," it can display a list of representative points using the results of point clustering. If the electronic device receives input stating, "[I want to identify the region of interest as a plane]," it can generate a plane. Additionally, the electronic device can calculate the parameters of the region of interest separated by the plane. If the user inputs both point and plane methods, the electronic device can first select representative points through clustering and then input the entire set of identical points into a plane detection algorithm to determine the plane. In this way, the electronic device can intuitively display the location of the region of interest using a single representative point, and / or provide information in a comprehensive form through planes and parameters.

[0127] In one embodiment, the electronic device may determine a point to display a region of interest based on at least one point (FPS1 and FPS2). For example, the electronic device may apply a clustering algorithm to at least one point (FPS1 and FPS2) to divide the at least one point (FPS1 and FPS2) into multiple groups. Each group may contain at least one point (FPS1 and FPS2). The electronic device may determine the point (FPS1 or FPS2) having the highest confidence score in each group as the representative point. If cerebral aneurysms are distributed across multiple regions of the blood vessel, one representative point is selected for each group, allowing the user to identify cerebral aneurysms at multiple locations through the electronic device. An example of points to display a region of interest determined by the electronic device according to one embodiment may be as shown in FIG. 16. In FIG. 16, the representative point (RP1) having the highest confidence score among at least one point (FPS1 and FPS2) of FIG. 15 can be identified. There may be multiple representative points (RP1).

[0128] In one embodiment, the electronic device may determine a plane to display a region of interest based on at least one point (FPS1 and FPS2) or a representative point (RP1). The plane may be a plane that separates a normal blood vessel region from an aneurysm region in a blood vessel. For example, the electronic device may determine a plane based on a representative point (RP1) of one of a plurality of groups. For example, the electronic device may extract the boundary of a representative point (RP1) belonging to one group and determine a plane based on the extracted boundary.

[0129] An example of a plane to display a region of interest, determined by an electronic device according to one embodiment, may be as shown in FIG. 17. In FIG. 17, the boundary of a representative point (RP1) indicated by c, x, and z can be detected, and a region of interest (e.g., an aneurysm region) can be separated from a blood vessel based on the boundary. The electronic device can define a plane by three points c, x, and z. The electronic device can generate a contour based on the points where the defined plane touches the blood vessel mesh. The electronic device can use the contour to separate the blood vessel into an aneurysm region and a normal blood vessel region. The electronic device can determine an aneurysm point in the separated aneurysm region. For example, the electronic device determines an aneurysm point based on the representative point (RP1), which may be indicated as an point.

[0130] According to an embodiment, the electronic device may determine one or more planes based on a representative point (RP1) in the case of a kissing vessel.

[0131] The electronic device can separate and display the normal blood vessel region and the aneurysm region based on a determined plane. The electronic device can calculate the parameters of the separated aneurysm region. For example, the parameters of the aneurysm region may include volume, length, circumference, diameter, height, width, angle, etc. Accordingly, the user can check how the aneurysm and the normal blood vessel are visually distinguished and can also check the parameters of the aneurysm.

[0132] In this way, the electronic device automatically detects and displays the location of regions of interest, such as cerebral aneurysms, and can rapidly calculate quantitative information, such as aneurysm volume, by deriving an automatic plane that distinguishes normal blood vessels from aneurysms. Accordingly, the electronic device significantly reduces the reading burden on medical staff and enables easier detection of even minute cerebral aneurysms. Furthermore, by enabling objective data-based follow-up monitoring, the device can enhance accuracy in the diagnostic and treatment processes and provide the advantage of supporting rapid diagnosis even in emergency situations.

[0133]

[0134] It is obvious that each step or operation of the method according to the embodiments of the present disclosure may be performed by a computer comprising one or more processors in accordance with the execution of a computer program stored in a computer-readable recording medium.

[0135] The computer-executable instructions stored on the aforementioned recording medium can be implemented through a computer program programmed to perform each corresponding step, and such a computer program can be stored on a computer-readable recording medium and executed by a processor. The computer-readable recording medium may be a non-transitory readable medium. In this case, a non-transitory readable medium refers to a medium that stores data semi-permanently and is readable by a device, rather than a medium that stores data for a short moment, such as a register, cache, or memory. Specifically, programs for performing the various methods described above may be provided by being stored on a non-transitory readable medium, such as semiconductor memory devices including erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), and flash memory devices; magnetic disks such as internal hard disks and removable disks; optical-magnetic disks; and non-volatile memory including CD-ROMs and DVD-ROMs.

[0136] Methods according to the various examples disclosed in this document may be provided by being included in a computer program product. The computer program product may be distributed in the form of a device-readable storage medium (e.g., compact disc read-only memory (CD-ROM)) or online through an application store (e.g., Play Store™). In the case of online distribution, at least a portion of the computer program product may be temporarily stored or temporarily created on a storage medium, such as the memory of a manufacturer's server, an application store's server, or a relay server.

[0137] As explained above, a person skilled in the art to which this disclosure pertains will understand that this disclosure may be implemented in other specific forms without altering its technical concept or essential features. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. The scope of this disclosure is defined by the claims set forth below rather than by the detailed description, and all modifications or variations derived from the meaning and scope of the claims and equivalent concepts should be interpreted as being included within the scope of this disclosure.

[0138] The features and advantages described herein are not all included, and in particular, many additional features and advantages will become apparent to those skilled in the art by considering the drawings, the specification, and the claims. Furthermore, it should be noted that the language used in this specification has been chosen primarily for readability and instructional purposes and may not be chosen to describe or limit the subject matter of this disclosure.

[0139] The foregoing description of the embodiments of the present disclosure is provided for illustrative purposes only. It is not intended to limit the present disclosure to the exact form disclosed or to make it incomplete. Those skilled in the art will understand that many modifications and variations are possible in light of the foregoing disclosure.

[0140] Therefore, the scope of the present disclosure is not limited by the detailed description but by any of the claims of the application based thereon. Accordingly, the disclosure of embodiments of the present disclosure is illustrative and does not limit the scope of the present disclosure as set forth in the following claims.

Claims

1. A step of acquiring blood vessel data including first points in a three-dimensional space; A step of generating three-dimensional shapes in the above three-dimensional space; A step of determining candidate shapes among the three-dimensional shapes based on the first points above; A step of determining second points among the first points based on the above candidate shapes; A step of calculating a reliability score for each of the above second points; and Step of determining the region of interest based on the above reliability score A method for determining a region of interest including 2. In Paragraph 1, The step of generating three-dimensional shapes in the above three-dimensional space is, A step of generating a voxel grid that divides the above three-dimensional space; and Step of generating the three-dimensional shapes based on the above voxel grid A method for determining a region of interest, including 3. In Paragraph 2, The above three-dimensional shapes are anchor boxes, and The step of generating the three-dimensional shapes based on the above voxel grid is, A step of generating anchor boxes while moving along each axis by a predetermined interval from the origin. A method for determining a region of interest, including 4. In Paragraph 3, The step of generating anchor boxes while moving along each axis by a predetermined interval from the above origin is, A step of generating the anchor boxes based on a plurality of predetermined ratios for width, height, and depth. A method for determining a region of interest, including 5. In Paragraph 3, The step of generating anchor boxes while moving along each axis by a predetermined interval from the above origin is, A step of obtaining the size of the above voxel grid; A step of obtaining a scale value for scaling the above anchor boxes; and A step of determining the predetermined interval based on the size of the voxel grid and the scale value. A method for determining a region of interest, including 6. In Paragraph 5, The step of generating anchor boxes while moving along each axis by a predetermined interval from the above origin is, Step of determining the volume of the anchor boxes based on the above scale value A method for determining a region of interest that further includes 7. In Paragraph 1, The step of determining candidate shapes among the three-dimensional shapes based on the first points above is: A step of determining the number of first points included in each of the above three-dimensional shapes; and A step of determining the candidate shapes among the three-dimensional shapes based on the number of first points determined above. A method for determining a region of interest, including 8. In Paragraph 7, The step of determining the candidate shapes among the three-dimensional shapes based on the number of first points determined above is, A step of determining first shapes among the above three-dimensional shapes in which the number of the determined first points is greater than a first reference value; and Step of determining the candidate shapes among the first shapes based on the degree of overlap A method for determining a region of interest that further includes 9. In Paragraph 8, The step of determining the candidate shapes among the first shapes based on the degree of overlap of the first shapes is, A step of determining second shapes among the first shapes in which the degree of overlap is greater than a second reference value; and Step of determining one of the second shapes as a candidate shape A method for determining a region of interest, including 10. In Paragraph 1, The step of determining the second points among the first points based on the above candidate shapes is, A step of determining grid cells included in the above candidate shapes; and A step of determining the points included in the grid cells among the first points as the second points. A method for determining a region of interest, including 11. In Paragraph 1, A step of obtaining third points by performing sampling on the second points above. Includes more, The step of calculating the reliability score for each of the above second points is, Step of calculating the reliability score for each of the above third points A method for determining a region of interest, including 12. In Paragraph 11, The step of obtaining third points by performing sampling on the second points above is, A step of obtaining the third points uniformly distributed by performing uniform sampling on the second points; or A step of obtaining the third points in which new points are generated by performing upsampling on the second points. A method for determining a region of interest, including 13. In Paragraph 1, The step of calculating the reliability score for each of the above second points is, A step of obtaining a blood vessel mesh corresponding to the above blood vessel data; A step of generating normal vectors of the vertices of the above-mentioned blood vessel mesh; A step of assigning the above normal vectors to the above second points; and A step of calculating the reliability score based on the first coordinates of the second points and the normal vectors. A method for determining a region of interest, including 14. In Paragraph 13, The step of assigning the above normal vectors to the above second points is, A step of determining the vertex closest to the third point among the second points; and The step of assigning the normal vector of the above vertex to the above third point A method for determining a region of interest, including 15. In Paragraph 13, The step of calculating the reliability score for each of the above second points is, A step of calculating the average coordinate of the first coordinates of the second points; A step of obtaining second coordinates from the first coordinates based on the above average coordinates; and Step of calculating the reliability score based on the second coordinates and the normal vectors A method for determining a region of interest, including 16. In Paragraph 1, The step of determining the region of interest based on the above reliability score is, A step of determining third points among the second points, wherein the reliability score is greater than the first reference value; A step of performing point cloud segmentation on the above third points; and Step of determining the above region of interest based on the subdivision results A method for determining a region of interest, including 17. In Paragraph 16, The step of performing point cloud segmentation on the above third points is, A step of determining a fourth point from the third point through localization and clustering; A step of creating an anchor box centered on the fourth point; Step of determining the fifth point included in the above anchor box Includes, The step of determining the above region of interest based on the subdivision results is, Step of determining the region of interest based on the above 5 points A method for determining a region of interest, including 18. In Paragraph 17, The step of determining the fourth point from the third point through the localization and clustering above is, A step of determining candidate points of interest from the third points through a machine learning model; A step of determining the mesh vertex closest to the candidate point of interest among the above blood vessel data; and A step of generating the fourth point by clustering mesh vertices determined through a clustering algorithm A method for determining a region of interest, including 19. In Paragraph 17, The step of determining the region of interest based on the above-mentioned fifth points is, A step of obtaining a display method for the above-mentioned region of interest; If the above display method is a point, the step of clustering the fifth points into a plurality of groups, determining the sixth point with the highest reliability score in each of the plurality of groups, and displaying the sixth point as the region of interest; and If the above display method is planar, the step of generating a separation surface based on the fifth points and displaying the region of interest by separating the blood vessel mesh based on the separation surface A method for determining a region of interest, including 20. A processor and a memory connected to the processor, and The above memory is configured to store a program, and The above processor is configured to execute the above program, and When the above program is executed, the steps of the method of any one of claims 1 to 19 are implemented, Electronic device.

Images