Method and device for extracting target region from chest CT data and detecting lesion by using artificial intelligence

Abstract

A method for extracting a target region from chest CT data and detecting a lesion is disclosed, the method comprising the steps of: obtaining chest CT data of a subject; segmenting an image obtained by extracting a heart region from the chest CT data by using an artificial intelligence model; overlaying, onto the chest CT data, the image obtained by extracting the heart region; analyzing the image obtained by extracting the heart region, so as to detect a calcified portion; and displaying the calcified portion on the image obtained by extracting the heart region.

Description

Method and device for extracting target areas and detecting lesions from chest CT data using artificial intelligence

[0001] The present invention relates to a method and device for extracting a target area from chest CT data and detecting a lesion, and more particularly, to a technology for detecting a calcified area in a heart area using artificial intelligence and displaying the result.

[0002] Cardiovascular disease is a leading cause of death. Therefore, early detection of calcification, a major risk factor for cardiovascular disease, is crucial before symptoms develop. A high calcification index typically indicates severe atherosclerosis, which in turn increases the risk of cardiovascular diseases such as angina and myocardial infarction, which are triggered by atherosclerosis.

[0003] To detect calcified areas in the cardiovascular system, doctors must find the heart area among the various organs that appear on the chest CT image and look for areas where calcium and other substances have accumulated. However, since the heart area is not clearly visible on low-dose CT images before contrast enhancement, the doctor must rely on his or her experience, which has the disadvantage of low accuracy and objectivity.

[0004] Additionally, even if the heart area is accurately distinguished, it is time-consuming because hundreds of CT images must be checked per single subject to find the calcified area.

[0005] Accordingly, there is a growing demand for technologies that utilize artificial intelligence to extract the heart region from low-dose CT images of all subjects undergoing chest CT scans and accurately detect calcified areas, enabling early detection of cardiovascular disease. Furthermore, there is a growing demand for technologies that can clearly mark calcified areas within the heart region for easier recognition by medical professionals.

[0006] The present invention aims to provide a method and device for extracting a heart region from chest CT data using artificial intelligence, analyzing the heart region, and detecting and displaying calcified areas.

[0007] As a technical means for achieving the above-described task, a first aspect of the present invention can provide a method for extracting a target region and detecting a lesion from chest CT data, including the steps of acquiring chest CT data of a subject, segmenting an image from which a heart region is extracted from the chest CT data using an artificial intelligence model, overlaying the image from which the heart region is extracted onto the chest CT data, analyzing the image from which the heart region is extracted to detect a calcified region, and marking the calcified region on the image from which the heart region is extracted.

[0008] In addition, the artificial intelligence model may apply an algorithm based on a CNN (Convolutional Neural Network), and the artificial intelligence model may learn a criterion for segmenting an image from which the heart region is extracted based on chest CT data indicating the heart region.

[0009] Additionally, the image from which the heart region is extracted can be used to calculate a calcification intensity score based on the range value of Hounsfield Unit (HU).

[0010] Additionally, the calcification intensity score is a score calculated for an object that groups adjacent pixels based on the range value in Hounsfield units, and the calcification intensity score can be calculated by multiplying the number of pixels included in the object, the weight, and the interval at which chest CT data were captured.

[0011] Additionally, the weight of the pixel with the highest Hounsfield unit range value among the pixels included in the object can be determined as the representative value of the object.

[0012] Additionally, the method can display the color of an object determined according to a representative value.

[0013] Additionally, the method can display a heart region with calcified areas as a three-dimensional model.

[0014] Additionally, the 3D model may rotate as it receives user input, and the calcified areas may be displayed in a different color than the heart area.

[0015] Additionally, the method can display the heart region with the calcified portion marked as two-dimensional data by overlaying it on chest CT data.

[0016] Additionally, the method can display information about the calcified portion together with the calcified portion.

[0017] Additionally, information about the calcified portion may include at least one of information about the calcification intensity score, the calcification intensity rank, the number of pixels, and the size of the calcified portion.

[0018] A second aspect of the present invention can provide a computer-readable storage medium having recorded thereon a program for executing the method of the first aspect on a computer.

[0019] The present invention has the advantage of more accurately extracting the heart region by eliminating human error through artificial intelligence. Furthermore, the invention enables monitoring of calcified areas in all individuals undergoing low-dose CT scans during health screenings, enabling early treatment of heart disease before symptoms appear. Furthermore, the use of low-dose CT reduces radiation exposure for the examinee, and offers the advantages of being economical and time-saving.

[0020] In addition, the algorithm of the present invention can eliminate the inconvenience of having to individually check hundreds of CT image data per subject to detect cardiac calcification.

[0021] In addition, the present invention has the advantage of providing information that can assist in a final diagnosis by enabling practitioners to more accurately confirm the information and location of the calcified portion by conveniently displaying the heart area where the calcified portion is indicated.

[0022] FIG. 1 is a diagram illustrating an example of extracting a heart region from chest CT data according to one embodiment.

[0023] Figure 2 is a flowchart illustrating a lesion detection method according to one embodiment.

[0024] FIGS. 3 and 4 are diagrams showing examples of learning an artificial intelligence model according to one embodiment.

[0025] FIG. 5 is a drawing showing an example of detecting a calcified portion according to one embodiment.

[0026] Figure 6 is a block diagram illustrating the configuration of a device according to one embodiment.

[0027] Figures 7 and 8 are drawings showing examples of calcified areas in the heart area.

[0028] The invention will now be described in detail by way of exemplary embodiments, with reference to the attached drawings. The following examples are intended only to illustrate the invention and do not limit or restrict the scope of the invention. Anything readily inferred by an expert in the technical field to which the invention pertains from the detailed description and examples is deemed to fall within the scope of the invention.

[0029] The terms “comprises” or “includes” used herein should not be construed to necessarily include all components or steps, but rather to mean that some of the components or steps may not be included, or that additional components or steps may be included.

[0030] The terms used in this specification are described as currently common terms, taking into account the functions mentioned herein. However, these terms may mean various other terms depending on the intentions of engineers working in the field, precedents, the emergence of new technologies, etc. Therefore, the terms used in this specification should not be interpreted solely based on their names, but rather based on the meaning of the terms and the overall content of this specification. Furthermore, singular expressions include plural meanings unless the context clearly indicates a singular meaning.

[0031] As used herein (and particularly in the claims), the terms "said" and "said" and similar referents may refer to both the singular and the plural. Furthermore, unless the order of steps in a method described herein is explicitly specified, the steps described may be performed in any appropriate order. The present invention is not limited to the order in which the steps are described.

[0032] The appearance of phrases such as “in one embodiment” in various places throughout this specification is not necessarily all referring to the same embodiment.

[0033] Some embodiments of this specification may be represented by functional block configurations and various processing steps. Some or all of these functional blocks may be implemented by various numbers of hardware and / or software components that perform specific functions.

[0034] Additionally, the connecting lines or connecting members between components depicted in the drawings are merely exemplary representations of functional connections and / or physical or circuit connections. In an actual device, connections between components may be represented by various functional connections, physical connections, or circuit connections that may be replaced or added.

[0035] The present embodiments relate to a method and device for extracting target areas and detecting lesions from chest CT data using artificial intelligence. Details well known to those skilled in the art to which the embodiments pertain will be omitted. The present invention will now be described in detail with reference to the accompanying drawings.

[0036] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings.

[0037] FIG. 1 is a diagram illustrating an example of extracting a heart region from chest CT data according to one embodiment.

[0038] The subject is undergoing a CT scan of the chest for a health checkup. The method according to the present invention relates to a technique for detecting calcified areas from a CT image of the subject's chest to measure the condition of the subject's heart and detect cardiovascular disease at an early stage. To this end, a computed tomography (CT) scan image of the subject's chest may first be captured, and / or the captured image data may be loaded.

[0039] As illustrated in Fig. 1, the subject's chest CT data input into the AI ​​model includes body tissues other than the heart region. Therefore, if an attempt is made to identify areas of calcium accumulation and calcification, or areas with high density, in the entire chest CT data, calcified tissues other than the heart region may be identified, or even ribs, etc. may be detected. Therefore, it is necessary for the AI ​​model to automatically identify the region corresponding to the heart (ROI: Region Of Interest). Accordingly, the AI ​​model can be used to segment an image (mask image) from which only the heart region is extracted from the chest CT data, and the image from which the heart region is extracted can be overlaid on the original chest CT data to display the heart region so that it is clearly distinguished. This allows for more accurate distinction of the heart region even in low-dose CT images, which has the advantage of being used as information to assist medical staff. In one embodiment, as illustrated in Fig. 1, when the "get heart" button is clicked in the program, only the heart region in the chest CT data can be displayed in color, separated from the black and white CT data. For example, the entire area outside the heart could be black and white, with only the heart region highlighted in red. This would indicate the area where the AI ​​model extracted the heart region.

[0040] This facilitates the analysis of calcified areas in the blood vessels of the heart, as described later in this specification. For example, the deposition of calcium and waste products (calcium clumps) in the blood vessels supplying blood to the heart muscle can lead to arteriosclerosis, and the subsequent development of heart diseases such as angina and myocardial infarction. Through analysis of chest CT data, this can be used for early prevention or more accurately detecting even low-level calcification.

[0041] Figure 2 is a flowchart for explaining a medical image processing method according to one embodiment.

[0042] In step S210, chest CT data of the subject may be acquired. For example, the chest CT data may be an image captured by the device (1000), or an image received from an external device or server and loaded into the device (1000). An example of CT data of the subject's chest loaded into the device (1000) is illustrated in FIG. 1. The chest CT data may include a coronal view, a sagittal view, an axial view, and / or a three-dimensional model of the chest.

[0043] Example CT image data of a human subject's chest may include other organs, such as the lungs, in addition to the heart. Low-dose chest CT images intended for health screenings often have low resolution and lack contrast agents, making the boundaries of each organ unclear. Therefore, it is crucial to more clearly distinguish the heart region. Methods for generating images that extract the heart region will be described later.

[0044] At step S220, an image in which a heart region is extracted from chest CT data can be generated using an artificial intelligence model. In one embodiment, the artificial intelligence model may apply an algorithm based on a convolutional neural network (CNN). An image in which a heart region is extracted from chest CT data using the artificial intelligence model is illustrated in FIG. 3. The artificial intelligence model can learn criteria for generating an image in which a heart region is extracted based on chest CT data that display the heart region, which will be described later with reference to FIG. 4.

[0045] At step S230, the image from which the heart region is extracted can be overlaid on chest CT data. An example of extracting only the heart region from chest CT data using an AI model and then overlaying it on chest CT data is illustrated in Figure 3 (c). This allows only the heart region to be clearly distinguished from the existing CT data, making it convenient for doctors to diagnose heart-related diseases.

[0046] In step S240, the image from which the heart region is extracted may be analyzed to detect calcified areas. In one embodiment, a calcification intensity score may be calculated based on the range value of Hounsfield Units (HU) from the image from which the heart region is extracted. The calcification intensity score may be a score calculated for an object that groups adjacent pixels based on the range value of Hounsfield Units in the image from which the heart region is extracted. The calcification intensity score may be calculated by multiplying the number of pixels included in the object, a weight, and the interval at which chest CT data were captured. For example, the object may indicate a calcified portion within the heart region. If there are three calcified portions within the heart region of a subject, there may also be three objects. Furthermore, in one embodiment, the weight of the pixel with the highest Hounsfield Unit range value among the pixels included in the object may be determined as the representative value of the object. Furthermore, the color of the object may be determined based on the representative value of the object. This will be described later with reference to FIG. 5.

[0047] At step S250, the calcified area can be marked on the image from which the heart region is extracted.

[0048] In one embodiment, a heart region with calcified areas can be overlaid on chest CT data and displayed as two-dimensional data. At this time, information about the calcified area can also be displayed near the area suspected of being calcified. The information about the calcified area may include at least one of the following: a calcification intensity score, a calcification intensity ranking (weight), the number of pixels, and the size of the calcified area. This will be described later with reference to FIG. 7.

[0049] In another embodiment, a heart region with calcified portions may be displayed as a 3D model. For example, only the heart region with calcified portions may be displayed as a separate 3D model. In this case, the 3D model may rotate 360 ​​degrees upon receiving user input, and the calcified portions may be displayed in a different color from the heart region. This will be described later with reference to FIG. 8.

[0050] By identifying the areas of the heart where calcification is present, the doctor can diagnose the condition of the heart and decide whether further treatment is necessary.

[0051] FIG. 3 and FIG. 4 are diagrams showing examples of learning an artificial intelligence model according to one embodiment.

[0052] Referring to Figure 4, chest CT data can be input into an AI model to detect calcified areas in the cardiac region from chest CT data. For example, the chest CT data may be a two-dimensional dicom file, such as low-dose chest CT data used for general health checkups. As illustrated in Figure 3 (a), the input low-dose chest CT data may be images of CT images taken of the chest of a single subject.

[0053] To enable the acquired data to be used for learning the criteria for generating images from which the heart region is extracted, the acquired data can be preprocessed and input into an AI model. In other words, the acquired data can be processed into a preset format to enable learning for heart region extraction.

[0054] In one embodiment, the subject's chest CT data can be transformed through image processing, and preprocessing can be performed to enhance the data. For example, image processing may include transforming the size or arrangement of the image, or normalizing it. Furthermore, data enhancement methods such as rotating the image or zooming in and out can be used.

[0055] The artificial intelligence model may be, for example, a model based on a neural network. For example, the artificial intelligence model may be a model based on a convolutional neural network (CNN) and may apply the U-net algorithm (segmentation), but is not limited thereto. In one embodiment, the identification of the heart region in the input CT image may be performed based on learning according to CNN technology. That is, the artificial intelligence model may be trained to separate and extract the heart region from a low-dose chest CT image. For example, the artificial intelligence model may be trained using training data as input values. At this time, the training data, the chest CT data, may be a grayscale image (DICOM) file (see (a) of FIG. 3), and the heart mask data may be a binary image (DICOM) file with the heart region masked. That is, a region determined to be the heart region may be indicated as 1 (e.g., white), and other regions may be indicated as 0 (e.g., black). A cardiac mask image can be generated for each slice containing the cardiac region from all images of chest CT image data. As illustrated in Fig. 4, the AI ​​model can be trained to predict a cardiac mask image, such as that shown in Fig. 3 (b), by passing through layers.

[0056] In other words, an AI model can learn the criteria for generating images with the heart region extracted from low-dose CT images of the chest based on chest CT images with the heart region marked. For example, a doctor could label the heart region in chest CT data from multiple subjects, and the AI ​​model could be trained using this labeled data as training data. This would enable the model to extract the heart region even when a chest CT image from a subject is input that does not have the heart region marked.

[0057] FIG. 5 is a diagram illustrating an example of detecting a calcified portion according to one embodiment. In one embodiment, the detection of the calcified portion may be performed based on the Hounsfield Unit (HU) value within the image from which the heart region is extracted after extracting the heart region. Since the Hounsfield Unit value is proportional to the density of human tissue in a CT scan, a large Hounsfield Unit value may indicate the presence of calcified plaque.

[0058] As shown in (a) of Fig. 5, first, all pixels are extracted from one slice of chest CT data and an HU value is determined for each pixel.

[0059] And as shown in (b) of Fig. 5, adjacent pixels in the four directions of adjacent up, down, left, and right from the extracted pixels are grouped and treated as a single object. For example, pixels with a Hounsfield unit value ranging from 0 to 129 are grouped by setting the weight to 0, and adjacent pixels with a Hounsfield unit value of 130 or more are grouped as an object. Since a Hounsfield unit value of less than 130 is in the normal range, if pixels with a Hounsfield unit value of more than that are grouped together, they can be viewed as calcified areas and are designated as objects. For example, if there are three or more adjacent pixels with a Hounsfield unit value of 130 or more, they can be regarded as an object. Depending on the weight, if the number of pixels included in the object is less than a certain number, the object is deleted, which will be explained later.

[0060] Also, as shown in (c) of FIG. 5, the Hounsfield unit value of the object is displayed by changing the weight according to the range. For example, when the Hounsfield unit value is in the range of 0 to 129, the weight factor is 0, when the Hounsfield unit is in the range of 130 to 199, the rank is Weak and the weight is 1, when the Hounsfield unit is in the range of 200 to 299, the rank is Normal and the weight is 2, when the Hounsfield unit is in the range of 300 to 399, the rank is Storng and the weight is 3, and when the Hounsfield unit is 400 or more, the rank is Hardened and the weight is 4. At this time, the weight is a number applied to distinguish the importance based on the range value of the Hounsfield. At this time, the color of the corresponding pixel can be set differently for each weight. For example, you can set the color to black when the weight is 0, light green when the weight is 1, yellow when the weight is 2, orange when the weight is 3, and red when the weight is 4.

[0061] As illustrated in (d) of Fig. 5, the weight of the pixel having the highest Hounsfield unit range value among the pixels included in the object can be determined as the representative value of the object. At this time, if the number of pixels included in the object is less than or equal to a predetermined number according to the weight, the object is deleted and not determined as a calcified part. For example, if the weight is 1, the number of pixels is 6 or less, if the weight is 2, the number of pixels is 5 or less, if the weight is 3, the number of pixels is 4 or less, and if the weight is 3, the number of pixels is 3 or less, the object is deleted.

[0062] This process can be repeated for each slice of every CT image in which an image from which the heart region has been extracted exists.

[0063] Additionally, a calcification intensity score can be calculated for each object. In one embodiment, the calcification intensity score can be calculated by multiplying the number of pixels contained in the object, a weight, and the interval at which chest CT data were captured for a single subject. The calculation of the calcification intensity score is as shown in Equation 1 below.

[0064]

[0065] At this time, the interval at which chest data are captured refers to the thickness of the CT image slice.

[0066]

[0067] At this time, the interval at which chest data are captured refers to the thickness of the CT image slice. For example, the slice thickness can be based on 3 mm, but is not limited thereto. At this time, to correct slices of different thicknesses to a consistent value, the slice thickness is divided by 3, the standard slice thickness of this specification. However, this is not limited thereto, and the constant for correction can be any integer n other than 3.

[0068] In one embodiment, by quantifying the intensity of calcification, information can also be obtained regarding which objects have the most calcification. In other words, this technology can calculate a calcification intensity score, which can then be used to provide diagnostic assistance regarding the likely location and risk of cardiovascular disease.

[0069] FIG. 6 is a block diagram illustrating a configuration of a device according to one embodiment. As illustrated in FIG. 6, a device (1000) according to one embodiment of the present invention may include a control unit (1300) and a memory (1700). However, not all illustrated components are essential. The device (1000) may be implemented with more components than the illustrated components, or may be implemented with fewer components. The above components will be described in turn below.

[0070] The control unit (1300) typically controls the overall operation of the device (1000). The control unit (1300) may include at least one processor. Depending on its function and role, the control unit (1300) may include multiple processors or a single processor in an integrated form. The processor may mainly refer to a central processing unit (CPU), an application processor (AP), a graphics processing unit (GPU), etc. In addition, the CPU, AP, or GPU may include one or more cores therein, and the CPU, AP, or GPU may operate using an operating voltage and a clock signal. However, while the CPU or AP may be composed of a few cores optimized for serial processing, the GPU may be composed of thousands of smaller and more efficient cores designed for parallel processing.

[0071] For example, the control unit (1300) can control the user input unit (not shown), the output unit (not shown), the sensing unit (not shown), the communication unit (not shown), the A / V input unit (not shown), etc. in general by executing programs stored in the memory (1700). In addition, the control unit (1300) can cause the device (1000) to extract the heart region from chest CT data and detect calcified parts.

[0072] The artificial intelligence model may be manufactured in the form of a hardware chip and loaded onto the device (1000). For example, the artificial intelligence model may be manufactured in the form of a dedicated hardware chip for artificial intelligence (AI), or may be manufactured as part of an existing general-purpose processor (e.g., CPU or application processor) or a graphics-only processor (e.g., GPU) and loaded onto various devices. In this case, the artificial intelligence model may be loaded onto the device (1000), a server, or a separate device. In addition, the device (1000) and a server (not shown) may provide model information constructed by the server (not shown) to the device (1000) via wired or wireless communication. Alternatively, the artificial intelligence model may be loaded onto the server, and the device (1000) may provide chest CT data to the server, so that the server may transmit an image from which a heart region is extracted to the device (1000). In addition, the process of detecting a calcified portion from the image from which a heart region is extracted may be performed on the device (1000).

[0073] The memory (1700) stores data supporting various functions of the control unit (1300). The memory (1700) can store a plurality of application programs (or applications) run by the control unit (1300), data for the operation of the control unit (1300), and commands. At least some of these application programs can be downloaded from an external server via wireless communication. In addition, the application programs can be stored in the memory (1700), installed in the control unit (1300), and driven by the processor to perform the operation (or function) of the control unit (1300).

[0074] The memory (1700) may include at least one type of storage medium among a flash memory type, a hard disk type, an SSD (Solid State Disk type), an SDD (Silicon Disk Drive type), a multimedia card micro type, a card type memory (e.g., SD or XD memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk. In addition, the memory (1700) may also include a web storage that performs a storage function on the Internet.

[0075] Meanwhile, the configuration of the device (1000) illustrated in FIG. 6 is an example, and each component of the device (1000) may be integrated, added, or omitted depending on the specifications of the device (1000) being implemented. That is, two or more components may be combined into one component, or one component may be subdivided into two or more components, as needed. In addition, the functions performed by each component (or module) are for describing embodiments, and the specific operations or devices thereof do not limit the scope of the present invention.

[0076] Figures 7 and 8 are diagrams illustrating examples of displaying calcified areas in the heart region. Figure 7 displays a heart region with calcified areas overlaid on chest CT data as two-dimensional data, and Figure 8 illustrates an example of displaying a heart region with calcified areas separately as a three-dimensional model.

[0077] First, after the heart area is separated and extracted through the “predict Heart Area” function, as shown in Fig. 7, when the user of the device clicks the “Inspect” button to analyze the heart area, the calcified part (object) can be displayed in a color (e.g., red) corresponding to the weight of the pixel with the highest Hounsfield unit range value among the pixels included in the object.

[0078] In one embodiment, when a user hovers the mouse over an object, information about the calcified portion may be displayed. The information about the calcified portion may include at least one of a calcification intensity score of the calcified portion, a weight (calcification intensity rank), the number of pixels included in the calcified portion, and the size of the calcified portion. For example, information about the calcified portion may be displayed such that the calcification intensity rank is Hardened, the weight is 4, the size (W / H) is 9.63mm / 9.63mm, the area is 92.64mm2, the number of pixels is 49, and the calcification intensity score is 38.60. In addition, in “Inspect Result,” statistical values ​​by calcification intensity rank of all detected objects can be checked.

[0079] Finally, among the analyzed CT data, the slice containing the object with the highest calcification intensity score can be output as a representative image along with the analyzed statistical data as a pdf and dicom image.

[0080] Referring to Fig. 8, two-dimensional images of a heart region showing calcified areas can be converted into a three-dimensional model and displayed. Unlike Fig. 7, only the heart region can be displayed as a three-dimensional model. However, the method for converting two-dimensional images into a three-dimensional model will not be described in detail herein.

[0081] Users of the device (e.g., medical professionals including doctors) can more easily identify calcified areas (810 to 860) through this 3D heart model, and the 3D model can be displayed while rotating 360 degrees in all directions according to user input. At this time, the calcified areas (810 to 860) can be displayed in a different color from the heart region depending on the intensity of the calcification. For example, areas with high calcification intensity can be displayed in red.

[0082] By displaying the calcified area as a 3D model that is similar to an actual heart, it has the advantage of being able to see even the parts that are not visible depending on the actual location or direction of the lesion that is difficult to confirm in 2D, and this can be easily utilized as auxiliary information when performing procedures / surgeries.

[0083] Meanwhile, the embodiments of the present invention described above can be written as a program that can be executed on a computer or cloud, and can be implemented in a general-purpose digital computer that operates the program using a computer-readable recording medium.

[0084] The computer-readable recording medium includes storage media such as magnetic storage media (e.g., ROM, floppy disk, hard disk, etc.) and optical readable media (e.g., CD-ROM, DVD, etc.).

[0085] Although the embodiments of the present invention have been described above with reference to the accompanying drawings, those skilled in the art will appreciate that the present invention can be implemented in other specific forms without altering the technical concept or essential features thereof. Therefore, the embodiments described above should be understood to be illustrative in all respects and not restrictive.

[0086] The present invention relates to a method and device for extracting a target area from chest CT data and detecting a lesion, and more particularly, to a technology for detecting a calcified area in a heart area using artificial intelligence and displaying the result.

Claims

1. A method for extracting a target area and detecting a lesion from chest CT data, A step of acquiring chest CT data of a subject; A step of segmenting an image in which a heart region is extracted from the chest CT data using an artificial intelligence model; A step of overlaying the image from which the heart region is extracted onto the chest CT data; A step of analyzing the image from which the above heart region is extracted to detect the calcified portion; and A step of displaying the calcified portion on the image from which the heart region is extracted; A method comprising:

2. In paragraph 1, The above artificial intelligence model applies an algorithm based on CNN (Convolutional Neural Network). A method wherein the artificial intelligence model learns criteria for segmenting an image from which the heart region is extracted based on chest CT data indicating the heart region.

3. In paragraph 1, The step of detecting the above calcified part is: A method for calculating a calcification intensity score based on a range value of Hounsfield Unit (HU) from an image from which the above heart region is extracted.

4. In paragraph 3, The above calcification intensity score is a score calculated for an object that groups adjacent pixels based on the range value of the Hounsfield unit, A method wherein the above calcification intensity score is calculated by multiplying the number of pixels included in the object, the weight, and the interval at which chest CT data were captured.

5. In paragraph 4, A method for determining the weight of a pixel having the highest Hounsfield unit range value among the pixels included in the object as a representative value of the object.

6. In paragraph 5, The step of indicating the above calcified part is: A method comprising the step of displaying the color of an object determined according to the above representative value.

7. In paragraph 1, A method further comprising the step of displaying the heart region in which the calcified portion is indicated as a three-dimensional model.

8. In paragraph 7, A method wherein the three-dimensional model rotates upon receiving user input, and the calcified portion is displayed in a different color from the heart region.

9. In paragraph 1, The step of indicating the above calcified part is: A method comprising the step of overlaying the heart region in which the calcified portion is indicated on the chest CT data and displaying it as two-dimensional data.

10. In paragraph 9, The step of indicating the above calcified part is: A method comprising the step of displaying information about the calcified portion together with the calcified portion.

11. In paragraph 10, A method wherein the information about the calcified portion includes at least one of information about the calcification intensity score, the calcification intensity rank, the number of pixels, and the size of the calcified portion.

12. Among the 1st to 11th clauses A computer-readable storage medium having recorded thereon a program for executing the method of any one of claims 1 to 11 on a computer.

Images