Apparatus and method for measuring muscle for each part and detecting vertebral segment in medical image

Abstract

The present disclosure relates to an apparatus and a method for measuring muscles for each part and detecting vertebral segments in a medical image. The apparatus comprises: a memory in which at least one process for performing muscle measurement for each part for an L3 slice in a CT image is stored; and a processor for performing an operation according to the process, wherein the processor generates each of a first muscle mask and a second muscle mask, marks an intersection region of the first muscle mask and the second muscle mask as an actual muscle, and divides the actual muscle into a plurality of regions to indicate the quality of the muscle for each region.

Description

Device and method for measuring muscle segments and detecting spinal segments in medical images

[0001] The present disclosure relates to a device and method for measuring muscle segments and detecting spinal segments in medical images.

[0002] Recent advances in artificial intelligence technology are being widely used to mark specific areas, such as diseases or muscles, within medical images.

[0003] Sarcopenia is a disease characterized by a decline in muscle mass and function associated with aging. This weakens the strength that supports bones and joints, increasing stress on the skeletal system, exacerbating joint pain, reducing stability, and increasing the risk of falls and fractures. It also increases the incidence of diabetes, high blood pressure, and cardiovascular disease, and can even lead to secondary conditions such as dysphagia in the muscular digestive system.

[0004] This type of sarcopenia is diagnosed comprehensively through body composition analysis, muscle mass measurement to check the thigh muscles, muscle strength assessment to evaluate grip strength or lower extremity muscle strength, and physical activity ability assessment such as walking or sitting and standing.

[0005] With the advancement of artificial intelligence, algorithms for measuring muscle mass based on images such as MRI, CT, and ultrasound images have been developed recently. There is a need for methods that can utilize these technologies to help in the early diagnosis and tracking of sarcopenia.

[0006] Furthermore, recent advancements in artificial intelligence are being used to identify specific structures within medical images. When measuring muscle mass, for example, doctors typically identify slices based on spinal segments and then interpret those slices.

[0007] One preprocessing technique that allows for better visualization of bones in CT is the Maximum Intensity Projection (MIP) technique. This is a 3D technique that displays only the highest HU value among the voxels that compose the image. For the purpose of obtaining a 3D technique image, when the projection angle is specified in the direction in which the reconstructed images (transversal axial images) are to be observed, this technique selects and displays only the voxels with the highest HU value among the voxels arranged parallel to the specified direction.

[0008] Techniques for identifying specific spinal segments using MIP have already been published in academic papers and other publications. However, this technique suffers from performance degradation in CT scans with a high number of brightly displayed artifacts (calcified vessels, iron cores, etc.) or in cases where the spinal image quality is poor for various reasons.

[0009] The purpose of the embodiments disclosed in the present disclosure is to provide a device and method capable of measuring muscles and detecting spinal segments by region in medical images.

[0010] Specifically, the embodiment disclosed in the present disclosure aims to provide a device and method capable of measuring muscle by region for an L3 slice in a CT image.

[0011] In addition, specifically, the embodiment disclosed in the present disclosure aims to provide a device and method capable of detecting spinal segments in a MIP image based on artificial intelligence.

[0012] According to one embodiment of the present disclosure for achieving the above-described technical problem, a device for measuring a muscle by region for an L3 slice in a CT image includes a memory storing at least one process for performing a muscle measurement by region for an L3 slice in a CT image and a processor performing an operation according to the process, wherein the processor generates a first muscle mask and a second muscle mask, respectively, and marks an intersection area of ​​the first muscle mask and the second muscle mask as an actual muscle, and divides the actual muscle into a plurality of areas to indicate the quality of the muscle by area.

[0013] Additionally, the first muscle mask may be extracted from the L3 slice image based on a full muscle segmentation model, and the second muscle mask may be extracted from the L3 slice image based on a detailed region muscle segmentation model.

[0014] Additionally, the plurality of regions may include anterior muscles, lateral muscles, posterior muscles, and psoas muscles.

[0015] Additionally, the entire muscle segmentation model and the detailed region muscle segmentation model may have different regions of interest.

[0016] In addition, a method for measuring muscle by region for an L3 slice in a CT image performed by a processor of a device according to an embodiment of the present disclosure for achieving the above-described technical task may include the steps of: generating a first muscle mask and a second muscle mask, respectively; marking an intersection area of ​​the first muscle mask and the second muscle mask as an actual muscle; and dividing the actual muscle into a plurality of areas and indicating the quality of the muscle by area.

[0017] In addition, according to one embodiment of the present disclosure for achieving the above-described technical problem, a device for detecting a spinal node in an AI-based MIP image includes a memory storing at least one process for performing AI-based spinal node detection in a MIP image, and a processor performing an operation according to the process, wherein the processor divides the MIP image into a first region and a second region using an AI-based segmentation model, removes a portion other than a spine portion from each of the MIP image of the first region and the MIP image of the second region, and infers an inference value for each spine portion from the MIP image of the first region and the MIP image of the second region, in which only the spine portion remains, using an AI-based detection model, and determines a final slice through the inference value.

[0018] Additionally, the first region may be a sagittal plane region, and the second region may be a coronal plane region.

[0019] Additionally, the detection model may include a first detection model for the first region and a second detection model for the second region.

[0020] Additionally, the processor can find a specific spinal segment in the MIP image of the first region and the MIP image of the second region, in which only the spinal portion remains, and determine the final slice using the inference value of the first detection model and the inference value of the second detection model for the specific spinal segment.

[0021] In addition, a method for detecting a spinal segment in an AI-based MIP image, performed by a processor of a device according to an embodiment of the present disclosure for achieving the above-described technical problem, may include: dividing a MIP image into a first region and a second region using an AI-based segmentation model; removing a portion other than a spine portion from each of the MIP image of the first region and the MIP image of the second region; inferring an inference value for each spinal segment from the MIP image of the first region and the MIP image of the second region, in which only the spine portion remains, using an AI-based detection model; and determining a final slice using the inference value.

[0022] According to the aforementioned problem solving means of the present disclosure, the performance of a method of representing qualitative characteristics by region by performing muscle segmentation by region in an existing L3 slice of a CT image can be increased.

[0023] In addition, by improving the performance of the algorithm that can express the quality and quantity of muscles in CT images by detailed region, it can help develop a customized treatment plan through quantitative indicators for each individual part.

[0024] Additionally, it can be crucial to know which muscle area of ​​each patient is problematic, which can be particularly helpful in developing a personalized treatment plan tailored to each individual patient.

[0025] Furthermore, the aforementioned problem-solving method of the present disclosure can improve the performance of existing techniques that use MIP to locate specific segments of the spine in CT scans. Specifically, the performance of an artificial intelligence model that locates slices corresponding to specific spinal segments in abdomen CT scans can be improved.

[0026] In addition, according to the aforementioned problem solving means of the present disclosure, it is possible to provide an algorithm with excellent performance that can more accurately find a slice corresponding to a spinal segment even in low-quality CT images that may not be detected or may be measured incorrectly in the past in detecting spinal segments.

[0027] The effects of the present disclosure are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description below.

[0028] FIG. 1 is a block diagram of a device for measuring muscle segments and detecting spinal segments in medical images according to one embodiment of the present disclosure.

[0029] FIG. 2 is a flowchart illustrating a process for measuring muscles by region for an L3 slice in a CT image according to one embodiment of the present disclosure.

[0030] FIG. 3 is a diagram illustrating a process for generating muscle-specific images according to one embodiment of the present disclosure.

[0031] FIG. 4 is a flowchart illustrating a process for detecting spinal segments in a MIP image based on artificial intelligence according to one embodiment of the present disclosure.

[0032] FIG. 5 is a diagram illustrating a process of determining a final slice through inference values ​​for each spinal segment according to one embodiment of the present disclosure.

[0033] Throughout this disclosure, the same reference numerals denote the same components. This disclosure does not describe all elements of the embodiments, and any content that is common in the technical field to which this disclosure belongs or that overlaps between embodiments is omitted. The terms "part, module, element, block" used in the specification may be implemented in software or hardware, and depending on the embodiment, multiple "parts, modules, elements, blocks" may be implemented as a single component, or a single "part, module, element, block" may include multiple components. Throughout the specification, when a part is said to be "connected" to another part, this includes not only cases where it is directly connected, but also cases where it is indirectly connected, and an indirect connection includes a connection via a wireless communication network.

[0034] Additionally, when a part is said to "include" a component, this does not mean that it excludes other components, but rather that it may include other components, unless otherwise specifically stated.

[0035] Throughout the specification, when we say that an element is "on" another element, this includes not only cases where the element is in contact with the other element, but also cases where another element exists between the two elements.

[0036] The terms first, second, etc. are used to distinguish one component from another, and the components are not limited by the aforementioned terms.

[0037] Singular expressions include plural expressions unless the context clearly indicates otherwise.

[0038] The identification codes for each step are used for convenience of explanation and do not describe the order of each step. Each step may be performed in a different order than specified unless the context clearly indicates a specific order.

[0039] The operating principle and embodiments of the present disclosure are described below with reference to the attached drawings.

[0040] Before proceeding, the meanings of terms used in this specification will be briefly explained. However, it should be noted that the explanation of terms is intended to aid understanding of this specification, and therefore, unless explicitly stated to limit the disclosure, they are not intended to limit the technical concepts of this disclosure.

[0041] In this specification, the term "device" encompasses a variety of devices capable of performing computational processing and providing results to a user. For example, a device may include a computer, a server, or a portable terminal, or may be any one of these.

[0042] Here, the computer may include, for example, a notebook, desktop, laptop, tablet PC, slate PC, etc. equipped with a web browser.

[0043] The above server device is a server that processes information by communicating with an external device, and may include an application server, a computing server, a database server, a file server, a game server, a mail server, a proxy server, and a web server.

[0044] The above portable terminal may include, for example, a wireless communication device that ensures portability and mobility, and may include all kinds of handheld-based wireless communication devices such as a PCS (Personal Communication System), GSM (Global System for Mobile communications), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), WiBro (Wireless Broadband Internet) terminal, a smart phone, and a wearable device such as a watch, a ring, a bracelet, an anklet, a necklace, glasses, contact lenses, or a head-mounted device (HMD).

[0045] The artificial intelligence-related functions according to the present disclosure are operated through a processor and memory. The processor may be composed of one or more processors. In this case, one or more processors may be a general-purpose processor such as a CPU, an AP, a DSP (Digital Signal Processor), a graphics-only processor such as a GPU or a VPU (Vision Processing Unit), or an artificial intelligence-only processor such as an NPU. One or more processors control the processing of input data according to predefined operation rules or artificial intelligence models stored in memory. Alternatively, if one or more processors are artificial intelligence-only processors, the artificial intelligence-only processors may be designed with a hardware structure specialized for processing a specific artificial intelligence model.

[0046] The predefined operation rules or artificial intelligence models are characterized by being created through learning. Here, being created through learning means that the basic artificial intelligence model is learned by a learning algorithm using a plurality of learning data, thereby creating a predefined operation rules or artificial intelligence model set to perform a desired characteristic (or purpose). This learning may be performed on the device itself on which the artificial intelligence according to the present disclosure is performed, or may be performed through a separate server and / or system. Examples of the learning algorithm include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.

[0047] An artificial intelligence model may be composed of multiple neural network layers. Each of the multiple neural network layers has multiple weight values, and performs neural network operations through operations between the operation results of the previous layer and the multiple weights. The multiple weights of the multiple neural network layers may be optimized based on the learning results of the artificial intelligence model. For example, the multiple weights may be updated so that the loss value or cost value obtained from the artificial intelligence model is reduced or minimized during the learning process. The artificial neural network may include a deep neural network (DNN), and examples thereof include, but are not limited to, a convolutional neural network (CNN), a deep neural network (DNN), a recurrent neural network (RNN), a restricted boltzmann machine (RBM), a deep belief network (DBN), a bidirectional recurrent deep neural network (BRDNN), or deep Q-networks.

[0048] The processor can create a neural network, train (or learn) a neural network, perform computations based on received input data, and generate information signals based on the results of the computations, or retrain the neural network.

[0049] Neural networks include CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), perceptron, multilayer perceptron, FF (Feed Forward), RBF (Radial Basis Network), DFF (Deep Feed Forward), LSTM (Long Short Term Memory), GRU (Gated Recurrent Unit), AE (Auto Encoder), VAE (Variational Auto) Encoder), DAE (Denoising Auto Encoder), SAE (Sparse Auto Encoder), MC (Markov Chain), HN (Hopfield Network), BM (Boltzmann Machine), RBM (Restricted Boltzmann Machine), DBN (Depp Belief Network), DCN (Deep Convolutional Network), DN (Deconvolutional Network), DCIGN (Deep Convolutional Inverse Graphics Network), Generative Adversarial Network (GAN), Liquid State Machine (LSM), Extreme Learning Machine (ELM), It will be understood by those skilled in the art that any neural network may be included, including but not limited to ESN (Echo State Network), DRN (Deep Residual Network), DNC (Differentiable Neural Computer), NTM (Neural Turning Machine), CN (Capsule Network), KN (Kohonen Network), and AN (Attention Network).

[0050] According to an exemplary embodiment of the present disclosure, the processor may be configured to perform data intelligence, such as CNN (Convolution Neural Network), R-CNN (Region with Convolution Neural Network), RPN (Region Proposal Network), RNN (Recurrent Neural Network), S-DNN (Stacking-based deep Neural Network), S-SDNN (State-Space Dynamic Neural Network), Deconvolution Network, DBN (Deep Belief Network), RBM (Restrcted Boltzman Machine), Fully Convolutional Network, LSTM (Long Short-Term Memory) Network, Classification Network, Generative Modeling, eXplainable AI, Continual AI, Representation Learning, AI for Material Design, BERT for natural language processing, SP-BERT, MRC / QA, Text Analysis, Dialog System, LLM (Large Language Model), Generative AI, GPT (Generative Pre-trained Transformer), Visual Analytics for vision processing, Visual Understanding, Video Synthesis, ResNet. Various artificial intelligence structures and algorithms can be used, including but not limited to Anomaly Detection, Prediction, Time-Series Forecasting, Optimization, Recommendation, and Data Creation.

[0051] The present disclosure relates to a method for segmenting muscles in an abdomen CT image and evaluating the quality of muscles, thereby representing qualitative characteristics for each muscle region.

[0052] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the attached drawings.

[0053] FIG. 1 is a block diagram of a device for measuring muscle segments and detecting spinal segments in medical images according to one embodiment of the present disclosure.

[0054] FIG. 2 is a flowchart illustrating a process for measuring muscles by region for an L3 slice in a CT image according to one embodiment of the present disclosure.

[0055] FIG. 3 is a diagram illustrating a process for generating muscle-specific images according to one embodiment of the present disclosure.

[0056] FIG. 4 is a flowchart illustrating a process for detecting spinal segments in a MIP image based on artificial intelligence according to one embodiment of the present disclosure.

[0057] FIG. 5 is a diagram illustrating a process of determining a final slice through inference values ​​for each spinal segment according to one embodiment of the present disclosure.

[0058] The device according to the present disclosure can measure muscle segments for an L3 slice in a CT image and detect spinal segments in a MIP image based on artificial intelligence.

[0059] A device (10) according to one embodiment of the present disclosure may include a communication unit (11), a memory (12), and a processor (13). However, in some embodiments, the device (10) may include fewer or more components than the components illustrated in FIG. 1.

[0060] The communication unit (11) may include one or more modules that enable wireless or wired communication between the device (10) and an external device (not shown), or between the device (10) and a communication network. For example, it may include at least one of a wired communication module, a wireless communication module, a short-range communication module, and a location information module.

[0061] The communication network can use various types of communication networks, for example, wireless communication methods such as WLAN (Wireless LAN), Wi-Fi, Wibro, WiMAX, and HSDPA (High Speed ​​Downlink Packet Access), or wired communication methods such as Ethernet, xDSL (ADSL, VDSL), HFC (Hybrid Fiber Coax), FTTC (Fiber to The Curb), and FTTH (Fiber to The Home) can be used.

[0062] Meanwhile, the communication network is not limited to the communication methods presented above, and may include all other widely known or future-developed communication methods in addition to the above-described communication methods.

[0063] The wired communication module may include various wired communication modules such as a Local Area Network (LAN) module, a Wide Area Network (WAN) module, or a Value Added Network (VAN) module, as well as various cable communication modules such as a Universal Serial Bus (USB), a High Definition Multimedia Interface (HDMI), a Digital Visual Interface (DVI), RS-232 (recommended standard 232), power line communication, or plain old telephone service (POTS).

[0064] The wireless communication module may include a wireless communication module that supports various wireless communication methods such as GSM (global System for Mobile Communication), CDMA (Code Division Multiple Access), WCDMA (Wideband Code Division Multiple Access), UMTS (universal mobile telecommunications system), TDMA (Time Division Multiple Access), LTE (Long Term Evolution), 4G, 5G, and 6G, in addition to a WiFi module and a Wireless Broadband module.

[0065] The short-range communication module is for short-range communication, and can support short-range communication using at least one of Bluetooth™, RFID (Radio Frequency Identification), Infrared Data Association (IrDA), UWB (Ultra Wideband), ZigBee, NFC (Near Field Communication), Wi-Fi (Wireless-Fidelity), Wi-Fi Direct, and Wireless USB (Wireless Universal Serial Bus) technologies.

[0066] The memory (12) may store at least one process for performing region-specific muscle measurement for an L3 slice in a CT image according to one embodiment of the present disclosure, and the processor (13) may perform region-specific muscle measurement for an L3 slice in a CT image according to the process.

[0067] Additionally, the memory (12) may store at least one process for performing detection of a spinal node in a MIP image based on artificial intelligence according to one embodiment of the present disclosure, and the processor (13) may perform detection of a spinal node in a MIP image based on artificial intelligence according to the process.

[0068] The memory (12) can store data supporting various functions of the device (10), a program for the operation of the processor (13), can store input / output data (e.g., music files, still images, moving images, etc.), and can store a plurality of application programs (or applications) run on the device (10), data for the operation of the device (10), and commands. At least some of these application programs can be downloaded from an external server via wireless communication.

[0069] The memory (12) 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 (12) is separate from the muscle measurement device (10), but may be a database connected by wire or wirelessly.

[0070] The processor (13) can perform the aforementioned operations using a memory that stores data for an algorithm for controlling the operations of components within the device (10) or a program that reproduces the algorithm, and the data stored in the memory. In this case, the memory (12) and the processor (13) may each be implemented as separate chips. Alternatively, the memory (12) and the processor (13) may be implemented as a single chip.

[0071] In addition, the processor (13) can control any one or a combination of the components discussed above to implement various embodiments according to the present disclosure described in FIGS. 2 to 5 below on the device (10).

[0072] Hereinafter, with reference to FIGS. 2 and 3, a detailed description will be given of a muscle measurement process for each region of the L3 slice in a CT image according to one embodiment of the present disclosure.

[0073] Referring to FIG. 2, the processor (13) of the device (10) can generate a first muscle mask and a second muscle mask, respectively (S210).

[0074] The processor (13) can mark the intersection area of ​​the first muscle mask and the second muscle mask as an actual muscle (S220).

[0075] The processor (13) can divide the actual muscle into multiple areas and indicate the quality of the muscle for each area (S230).

[0076] Here, the first muscle mask may be an image extracted from the L3 slice image based on a first model that segments the entire muscle based on artificial intelligence, and the second muscle mask may be an image extracted from the L3 slice image based on a second model that segments the detailed muscle area based on artificial intelligence.

[0077] Here, the multiple areas may include the anterior muscles, lateral muscles, posterior muscles, and psoas muscles.

[0078] That is, the processor (13) can input the L3 slice image into the first model to extract and generate the first muscle mask, and input the L3 slice image into the second model to extract and generate the second muscle mask.

[0079] The processor (13) can indicate that the intersection area of ​​the first and second muscle masks is an actual muscle, and divide the indicated actual muscle into a plurality of areas (areas included in the detailed area division) to indicate the quality of the muscle for each area.

[0080] Here, the first model and the second model are trained independently, and the range of the region of interest may be different.

[0081] Referring to FIG. 3, the processor (13) can extract a first muscle mask (32) as a detailed result for a muscle area from the entire image of the L3 slice image (31) through the first model.

[0082] The processor (13) can extract a second muscle mask (33) that divides the areas corresponding to the front muscles, side muscles, back muscles, and psoas muscles in the muscle area of ​​the L3 slice image (31) through the first model.

[0083] At this time, the processor (13) can extract an image (33) representing the intersection area of ​​the first model and the second model by ensembling (combining) the first model and the second model. In other words, the measurement precision of the actual muscle for each part is increased for the intersection area.

[0084] The ensemble techniques used at this time may include methodologies such as voting, bagging, and boosting.

[0085] The processor (13) can extract an image (34) visualizing the quality and quantity of muscles by part based on the ensemble results of the first model and the second model.

[0086] That is, the processor (13) projects images (32, 33) of the results inferred through the first model and the second model, respectively, onto the original image (31) to verify whether the images are actually muscle parts, and can visualize and provide the quality and quantity of muscles for each region by distinguishing them based on the regions segmented in the inference process through the second model.

[0087] Although FIG. 2 describes the steps as being executed sequentially, this is merely an example of the technical idea of ​​the present embodiment, and a person having ordinary skill in the technical field to which the present embodiment belongs can modify and apply various modifications and variations by changing the order described in FIG. 2 or executing them in parallel without departing from the essential characteristics of the present embodiment, and therefore FIG. 2 is not limited to a chronological order.

[0088] Meanwhile, in the above description, the steps described in FIG. 2 may be further divided into additional steps or combined into fewer steps, depending on the implementation example of the present disclosure. Furthermore, some steps may be omitted as needed, and the order of the steps may be changed.

[0089] Meanwhile, the disclosed embodiments may be implemented in the form of a recording medium storing computer-executable instructions. The instructions may be stored in the form of program code, and when executed by a processor, may generate program modules to perform the operations of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

[0090] Hereinafter, with reference to FIGS. 4 and 5, a process for detecting spinal segments in a MIP image based on artificial intelligence according to one embodiment of the present disclosure will be described in more detail.

[0091] Referring to FIG. 4, the processor (13) can divide the MIP image into a first region and a second region using an artificial intelligence-based segmentation model (S410).

[0092] The processor (13) can remove the remaining portion except for the spine portion from each of the MIP image of the first region and the MIP image of the second region (S420).

[0093] The processor (13) can infer inference values ​​for each spinal segment from the MIP image of the first region with only the spine remaining and the MIP image of the second region using an artificial intelligence-based detection model (S430).

[0094] The processor (13) can determine the final slice through the inference value (S440).

[0095] Here, the first region may be a sagittal plane region, and the second region may be a coronal plane region.

[0096] Here, the detection model may include a first detection model for a first region corresponding to the sagittal plane region and a second detection model for a second region corresponding to the coronal plane region.

[0097] The processor (13) can find a specific spinal segment in a MIP image of a first region in which only the spine portion remains and a MIP image of a second region in which only the spine portion remains, and can determine a final slice using an inference value of the first detection model for the specific spinal segment and an inference value of the second detection model for the specific spinal segment.

[0098] Referring to FIG. 5, the processor (13) can divide the MIP image (51) into sagittal and coronal regions, and generate a MIP image (52) of the sagittal region and a MIP image (53) of the coronal region.

[0099] The processor (13) can input a MIP image (52) of a sagittal plane region into the segmentation model to remove artifacts that interfere with the detection model and generate a first MIP image (54) of a sagittal plane region from which only the spinal bone region is extracted.

[0100] The processor (13) can input a MIP image (53) of a coronal region into the segmentation model to remove artifacts that interfere with the detection model and generate a second MIP image (55) of a coronal region from which only the spinal bone region is extracted.

[0101] The processor (13) can input the first MIP image (54) and the second MIP image (55) into the first detection model and the second detection model, respectively, to return candidate locations of expected slices for the images of the sagittal plane region and the images of the coronal plane region, respectively.

[0102] At this time, the final position of the slice can be determined by calculating a weighted average according to the confidence score of the first detection model and the second detection model, thereby generating a final slice image (36).

[0103] Although FIG. 4 describes the steps as being executed sequentially, this is merely an example of the technical idea of ​​the present embodiment, and a person having ordinary knowledge in the technical field to which the present embodiment belongs can modify and apply various modifications and variations by changing the order described in FIG. 4 or executing them in parallel without departing from the essential characteristics of the present embodiment, and therefore FIG. 4 is not limited to a chronological order.

[0104] Meanwhile, in the above description, the steps described in FIG. 4 may be further divided into additional steps or combined into fewer steps, depending on the implementation example of the present disclosure. Furthermore, some steps may be omitted as needed, and the order of the steps may be changed.

[0105] Meanwhile, the disclosed embodiments may be implemented in the form of a recording medium storing computer-executable instructions. The instructions may be stored in the form of program code, and when executed by a processor, may generate program modules to perform the operations of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

[0106] Computer-readable storage media include all types of storage media that store instructions that can be deciphered by a computer. Examples include read-only memory (ROM), random access memory (RAM), magnetic tape, magnetic disks, flash memory, and optical data storage devices.

[0107] The disclosed embodiments have been described with reference to the attached drawings as described above. Those skilled in the art will understand that the present disclosure can be implemented in forms other than the disclosed embodiments without altering the technical spirit or essential features of the present disclosure. The disclosed embodiments are illustrative and should not be construed as limiting.

Claims

1. A memory storing at least one process for performing regional muscle measurement for the L3 slice in a CT image; and Includes a processor that performs operations according to the above process, The above processor, Create the first muscle mask and the second muscle mask respectively, The intersection area of ​​the first muscle mask and the second muscle mask is marked as a real muscle. The above actual muscles are divided into multiple areas and the quality of the muscles is indicated by area. Segmental muscle measurement device for L3 slice in CT image.

2. In paragraph 1, The above first muscle mask is extracted from the L3 slice image based on the first model for segmentation of the entire muscle, The second muscle mask is extracted from the L3 slice image based on a second model for segmenting muscles in the detailed region. Segmental muscle measurement device for L3 slice in CT image.

3. In paragraph 1, The above multiple areas are, Including the anterior muscles, lateral muscles, posterior muscles and psoas muscles, Segmental muscle measurement device for L3 slice in CT image.

4. In paragraph 2, The first model and the second model have different ranges of interest. Segmental muscle measurement device for L3 slice in CT image.

5. A method for detecting spinal segments in an artificial intelligence-based MIP image performed by a processor of a device, A step of generating a first muscle mask and a second muscle mask, respectively; A step of indicating that the intersection area of ​​the first muscle mask and the second muscle mask is a real muscle; and A step of dividing the actual muscle into multiple regions and indicating the quality of the muscle for each region; including; Segmental muscle measurement method for L3 slice in CT images.

6. A memory storing at least one processor for performing artificial intelligence-based spinal node detection in a MIP image; and Includes a processor that performs operations according to the above process, The above processor, Using an artificial intelligence-based segmentation model, the MIP image is divided into the first and second regions, Remove the remaining portion except the spine portion from each of the MIP image of the first region and the MIP image of the second region, Using an artificial intelligence-based detection model, inference values ​​for each spinal segment are inferred from the MIP image of the first region and the MIP image of the second region, where only the spinal portion remains. The final slice is determined through the above inference value, An AI-based spinal node detection device in MIP images.

7. In paragraph 6, The above first region is the sagittal plane region, The above second region is the coronal plane region, An AI-based spinal node detection device in MIP images.

8. In paragraph 6, The above detection model is, comprising a first detection model for the first region and a second detection model for the second region, An AI-based spinal node detection device in MIP images.

9. In paragraph 8, The above processor, Find a specific spinal segment in the MIP image of the first region and the MIP image of the second region, where only the spinal portion remains, The final slice is determined using the inference value of the first detection model and the inference value of the second detection model for the specific spinal segment. An AI-based spinal node detection device in MIP images.

10. A method for detecting spinal segments in an artificial intelligence-based MIP image performed by a processor of a device, A step of dividing a MIP image into a first region and a second region using an artificial intelligence-based segmentation model; A step of removing the remaining portion except for the spine portion from each of the MIP image of the first region and the MIP image of the second region; A step of inferring an inference value for each spinal segment from the MIP image of the first region and the MIP image of the second region, wherein only the spinal portion remains, using an artificial intelligence-based detection model; and A step of determining the final slice through the above inference value; including; An AI-based spinal node detection method in MIP images.

Images