Method, apparatus and system for predicting the severity of renal disease

Abstract

A method, apparatus, and system for predicting and processing the state of a human organ are disclosed. A method for predicting and processing the state of a human organ according to at least one of the various embodiments of the present disclosure may include: collecting image data for a specific organ of the human body; extracting radiomix data from the collected image data for the specific organ; machine learning the extracted radiomix data as a dataset; and distinguishing between a normal state and an abnormal state of the organ based on the extracted radiomix data.

Description

Technology Field

[0001] The present disclosure relates to a method, apparatus, and system for predicting and treating the condition of human organs. Background Technology

[0002] In the medical field, there are attempts to utilize various cutting-edge digital technologies to improve the accuracy of patient diagnoses and other related tasks.

[0003] Typically, in the medical field, diagnoses are made by assessing a patient's condition based on images acquired through ultrasound, CT, MRI, etc.

[0004] However, according to the traditional method, the images obtained in this way are interpreted directly by medical professionals, and in this case, there is a risk that the diagnosis may vary or be misdiagnosed depending on the competence of the medical professional.

[0005] Therefore, there is a demand for methods such as more objective interpretation. Prior art literature

[0006] (0001) Korean Registered Patent No. 10-2462099 (Registration Date: 2016.03.14.) The problem to be solved

[0007] The present disclosure provides a method, apparatus, and system for predicting and processing human organs based on artificial intelligence technology by more accurately determining whether the organs are normal and the state of abnormal organs. means of solving the problem

[0008] A method for predicting and processing the state of a human organ according to at least one of the various embodiments of the present disclosure may include: collecting image data for a specific organ of the human body; extracting radiomix data from the collected image data for the specific organ; machine learning the extracted radiomix data as a dataset; and distinguishing between a normal state and an abnormal state of the organ based on the extracted radiomix data.

[0009] In a method according to at least one of the various embodiments of the present disclosure, the step of extracting radiomix data from image data of a specific organ collected may include: a step of preprocessing the image data; a step of segmenting a region of interest (RoI) from the preprocessed image data; a step of extracting features from the segmented image data; a step of selecting the extracted features; and a step of classifying the selected features.

[0010] In a method according to at least one of the various embodiments of the present disclosure, the method may further include the step of acquiring image data of a user's organ; the step of extracting radiomix data from the acquired image data; and the step of determining whether the user's organ is in an abnormal state based on the extracted radiomix data.

[0011] In a method according to at least one of the various embodiments of the present disclosure, if the result of determining whether the user's organ is in an abnormal state is that it is in an abnormal state, the method may further include the step of obtaining clinical information.

[0012] In a method according to at least one of the various embodiments of the present disclosure, the method may further include the step of integrating the acquired clinical information and the extracted radiomix data for the user's organ.

[0013] In a method according to at least one of the various embodiments of the present disclosure, the method may further include the step of predicting an abnormal state level for the user's organ based on the integrated radiomix data and clinical information.

[0014] In a method according to at least one of the various embodiments of the present disclosure, the step of transmitting information regarding the abnormal state level of the predicted user's organ to a terminal may be further included.

[0015] In a method according to at least one of the various embodiments of the present disclosure, the organ may include a kidney.

[0016] In a method according to at least one of the various embodiments of the present disclosure, the image data may be computed tomography image data.

[0017] In a method according to at least one of the various embodiments of the present disclosure, the clinical information may include estimated glomerular filtration rate (eGFR) information.

[0018] An apparatus for predicting and processing the state of a human organ according to at least one of the various embodiments of the present disclosure may include: a memory; and a processor that collects image data for a specific organ of the human body, extracts radiomix data from the collected image data for the specific organ, performs machine learning on the extracted radiomix data as a dataset, and distinguishes between a normal state and an abnormal state of the organ based on the extracted radiomix data.

[0019] In an apparatus according to at least one of the various embodiments of the present disclosure, the processor can preprocess the image data, segment a region of interest (RoI) from the preprocessed image data, extract features from the segmented image data, select the extracted features, and classify the selected features to extract radiomix data from the collected image data for a specific organ.

[0020] In an apparatus according to at least one of the various embodiments of the present disclosure, the processor may acquire image data of a user's organ, extract radiomix data from the acquired image data, and determine whether the user's organ is in an abnormal state based on the extracted radiomix data.

[0021] In an apparatus according to at least one of the various embodiments of the present disclosure, the processor can obtain clinical information if, as a result of determining whether the user's organ is in an abnormal state, it is in an abnormal state.

[0022] In an apparatus according to at least one of the various embodiments of the present disclosure, the processor can integrate the acquired clinical information and the extracted radiomix data for the user's organ.

[0023] In an apparatus according to at least one of the various embodiments of the present disclosure, the processor can predict the abnormal state level of the user's organ based on the integrated radiomix data and clinical information.

[0024] In an apparatus according to at least one of the various embodiments of the present disclosure, the processor may transmit information regarding the abnormal state level of the predicted user's organ to a terminal.

[0025] In an apparatus according to at least one of the various embodiments of the present disclosure, the organ may include a kidney.

[0026] In an apparatus according to at least one of the various embodiments of the present disclosure, the image data may be computed tomography image data.

[0027] In an apparatus according to at least one of the various embodiments of the present disclosure, the clinical information may include estimated glomerular filtration rate (eGFR) information.

[0028] A system for predicting and processing the state of a human organ according to at least one of the various embodiments of the present disclosure may include: an image acquisition device; a terminal; and a computing device that collects image data for a specific organ of the human body, extracts radiomix data from the collected image data for the specific organ, performs machine learning on the extracted radiomix data as a dataset, and distinguishes between a normal state and an abnormal state of the organ based on the extracted radiomix data.

[0029] In a system according to at least one of the various embodiments of the present disclosure, the computing device can preprocess the image data, segment a region of interest (RoI) from the preprocessed image data, extract features from the segmented image data, select the extracted features, and classify the selected features to extract radiomix data from the collected image data for a specific organ.

[0030] In a system according to at least one of the various embodiments of the present disclosure, the computing device acquires image data of a user's organ and extracts radiomix data from the acquired image data, and if the user's organ is in an abnormal state based on the extracted radiomix data, acquires clinical information, integrates the acquired clinical information and the extracted radiomix data of the user's organ, predicts an abnormal state level of the user's organ based on the integrated radiomix data and clinical information, and transmits information regarding the predicted abnormal state level of the user's organ to the terminal. Effects of the invention

[0031] According to at least one of the various embodiments of the present disclosure, there is an effect of predicting and processing human organs more accurately by determining whether the organ is normal and the condition of an abnormal organ based on artificial intelligence technology. Brief explanation of the drawing

[0032] FIG. 1 is a schematic diagram of a system for predicting and processing the state of human organs according to one embodiment of the present disclosure. Figure 2 is a configuration block diagram of the computing device of Figure 1. FIGS. 3 to 5 are flowcharts illustrated to explain a method for predicting and processing the condition of a human organ according to an embodiment of the present disclosure. FIGS. 6 to 10 are drawings illustrating a radio mix analysis method according to an embodiment of the present disclosure. FIG. 11 is a drawing illustrating a process for predicting and treating the condition of human organs according to an embodiment of the present disclosure. FIGS. 12 and 13 are drawings illustrated to explain baseline features related to human organ state prediction and processing according to an embodiment of the present disclosure. FIGS. 14 and 15 are drawings illustrated to explain baseline features related to human organ state prediction and processing according to an embodiment of the present disclosure. FIGS. 16 and 17 are drawings illustrating the correlation between eGFR and radiomix features related to human organ state prediction and processing according to an embodiment of the present disclosure. Specific details for implementing the invention

[0033] Throughout this disclosure, the same reference numerals denote the same components. This disclosure does not describe all elements of the embodiments, and general content in the art to which this disclosure pertains or content that overlaps between embodiments is omitted. The terms 'part, module, component, block' as used in the specification may be implemented in software or hardware, and depending on the embodiments, a plurality of 'parts, modules, components, blocks' may be implemented as a single component, or a single 'part, module, component, block' may include a plurality of components.

[0034] Throughout the specification, when a part is described as being "connected" to another part, this includes not only cases where they are directly connected but also cases where they are indirectly connected, and indirect connections include connections made via a wireless communication network.

[0035] Furthermore, when it is stated that a part "includes" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0036] Throughout the specification, when it is stated that a component is located "on" another component, this includes not only cases where a component is in contact with another component, but also cases where another component exists between the two components.

[0037] Terms such as "first," "second," etc., are used to distinguish one component from another, and the components are not limited by the aforementioned terms.

[0038] Singular expressions include plural expressions unless there is an obvious exception in the context.

[0039] In each step, identification codes are used for convenience of explanation and do not describe the order of the steps; the steps may be performed differently from the specified order unless a specific order is clearly indicated in the context.

[0040] The operating principles and embodiments of the present disclosure will be described below with reference to the attached drawings.

[0041] In this specification, the term "device according to the present disclosure" includes all various devices capable of performing computational processing and providing results to a user. For example, the device according to the present disclosure may include all of a computer, a server device, and a portable terminal, or may be in the form of 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, a web server, etc.

[0044] The above portable terminal may include, for example, all types of handheld-based wireless communication devices such as 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) terminals, smartphones, etc., as well as wearable devices such as watches, rings, bracelets, anklets, necklaces, glasses, contact lenses, or head-mounted devices (HMD).

[0045] Functions related to artificial intelligence 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, the one or more processors may be general-purpose processors such as CPUs, APs, and DSPs (Digital Signal Processors), graphics-dedicated processors such as GPUs and VPUs (Vision Processing Units), or artificial intelligence-dedicated processors such as NPUs. The one or more processors control the processing of input data according to predefined operation rules or artificial intelligence models stored in memory. Alternatively, if the one or more processors are artificial intelligence-dedicated processors, the artificial intelligence-dedicated processors may be designed with a hardware structure specialized for processing a specific artificial intelligence model.

[0046] The predefined rules of operation or artificial intelligence models are characterized by being created through learning. Here, being created through learning means that a predefined rules of operation or artificial intelligence models configured to perform a desired characteristic (or objective) are created by a basic artificial intelligence model being trained using a number of training data by a learning algorithm. Such learning may be performed on the device itself where the artificial intelligence according to the present disclosure is executed, or it may be performed through a separate server and / or system. Examples of learning algorithms include supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but are not limited to the examples described above.

[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 results of previous layers and the multiple weights. The multiple weights possessed by the multiple neural network layers can 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 during the learning process is reduced or minimized. Artificial neural networks may include deep neural networks (DNNs), such as Convolutional Neural Networks (CNNs), Deep Neural Networks (DNNs), Recurrent Neural Networks (RNNs), Restricted Boltzmann Machines (RBMs), Deep Belief Networks (DBNs), Bidirectional Recurrent Deep Neural Networks (BRDNNs), or Deep Q-Networks, but are not limited to the examples mentioned above.

[0048] According to an exemplary embodiment of the present disclosure, a processor can implement artificial intelligence. Artificial intelligence refers to a machine learning method based on an artificial neural network that enables a machine to learn by mimicking human biological neurons. Methodologies of artificial intelligence can be classified according to the learning method into supervised learning, where input and output data are provided together as training data and the solution (output data) to the problem (input data) is predetermined; unsupervised learning, where only input data is provided without output data and the solution (output data) to the problem (input data) is not predetermined; and reinforcement learning, where a reward is given from an external environment whenever an action is taken from the current state, and learning proceeds in a direction that maximizes such reward. In addition, artificial intelligence methodologies can be classified according to the architecture, which is the structure of the learning model. The architectures of widely used deep learning technologies can be classified into Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Transformers, and Generative Adversarial Networks (GAN).

[0049] The device and system may include an artificial intelligence model. The artificial intelligence model may be a single model or may be implemented as multiple models. The artificial intelligence model may be composed of a neural network (or artificial neural network) and may include statistical learning algorithms in machine learning and cognitive science that mimic biological neurons. A neural network may refer to a model that possesses problem-solving capabilities by having artificial neurons (nodes) that form a network through synaptic connections and change the strength of synaptic connections through learning. The neurons of a neural network may include combinations of weights or biases. A neural network may include one or more layers composed of one or more neurons or nodes. For example, the device may include an input layer, a hidden layer, and an output layer. The neural network constituting the device can infer a result (output) to be predicted from an arbitrary input by changing the weights of the neurons through learning.

[0050] The processor can create a neural network, train or learn a neural network, perform computations based on received input data, generate an information signal based on the results of the computation, or retrain the neural network. The neural network models may include, but are not limited to, various types of models such as Convolutional Neural Networks (CNN), Region with Convolutional Neural Networks (R-CNN), Region Proposal Networks (RPN), Recurrent Neural Networks (RNN), Stacking-based Deep Neural Networks (S-DNN), State-Space Dynamic Neural Networks (S-SDNN), Deconvolution Networks, Deep Belief Networks (DBN), Restructured Boltzmann Machines (RBM), Fully Convolutional Networks, Long Short-Term Memory Networks (LSTM), and Classification Networks, such as GoogleNet, AlexNet, and VGG Network. The processor may include one or more processors to perform computations according to the neural network models. For example, a neural network is a deep neural It may include a network (Deep Neural Network).

[0051] 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), GAN (Generative Adversarial Network), LSM (Liquid State Machine), ELM (Extreme Learning Machine), ESN (Echo It will be understood by a person skilled in the art that any neural network may be included, but is not limited to, State Network, Deep Residual Network, Differential Neural Computer, Neural Turning Machine, Capsule Network, Kohonen Network, and Attention Network.

[0052] According to an exemplary embodiment of the present disclosure, the processor comprises a Convolutional Neural Network (CNN) such as GoogleNet, AlexNet, VGG Network, Region with Convolutional Neural Network (R-CNN), Region Proposal Network (RPN), Recurrent Neural Network (RNN), Stacking-based Deep Neural Network (S-DNN), State-Space Dynamic Neural Network (S-SDNN), Deconvolution Network, Deep Belief Network (DBN), Restructured Boltzmann Machine (RBM), Fully Convolutional Network, Long Short-Term Memory (LSTM) Network, Classification Network, Generative Modeling, eXplainable AI, Continual AI, Representation Learning, AI for Material Design, BERT, SP-BERT, MRC / QA, Text Analysis, Dialog System, GPT-3, GPT-4 for Natural Language Processing, Visual Analytics, Visual Understanding, Video Synthesis for Vision Processing, Anomaly Detection, Prediction, Time-Series Forecasting, Optimization for ResNet Data Intelligence, Various artificial intelligence structures and algorithms, such as recommendation and data creation, may be used, but are not limited thereto. Hereinafter, embodiments of the present disclosure will be described in detail with reference to the attached drawings.

[0053] The present disclosure discloses a method, apparatus, and system for predicting and processing the condition of human organs.

[0054] In this case, the kidney is used as an example of a human organ to aid in understanding the present disclosure and for convenience of explanation, but is not limited thereto.

[0055] FIG. 1 is a schematic diagram of a human organ condition prediction and processing system (1) according to one embodiment of the present disclosure.

[0056] Figure 2 is a configuration block diagram of the computing device (200) of Figure 1.

[0057] A human organ state prediction and processing system (1) according to at least one of the various embodiments of the present invention may be configured to include a terminal (100), a computing device (200), and an image acquisition device (300).

[0058] The terminal (100) may be a terminal that requests the prediction or / and processing of the condition of a human organ.

[0059] The terminal (100) may be a terminal owned by medical personnel, medical institutions, etc. Such a terminal may be a terminal equipped with software or hardware, such as a specific program.

[0060] The terminal (100) may be a terminal owned by a user such as a patient.

[0061] The terminal (100) can be any one of a fixed terminal such as a TV, PC, monitor, or digital signage, or a mobile terminal such as a smartphone, tablet PC, or laptop.

[0062] The image acquisition device (300) may include all devices for acquiring images of a target, such as ultrasonography, CT, magnetic resonance imaging.

[0063] The computing device (200) can predict the condition of a specific organ of a target user (e.g., a patient) from image data of the organ acquired automatically by the image acquisition device (300) or at the request of the terminal (100), and perform a processing operation based on the predicted result.

[0064] These computing devices (200) may be computing devices equipped in a medical institution, located remotely, or in the form of a server in the form of a cloud.

[0065] Referring to FIG. 2, the computing device (200) is described in more detail as follows.

[0066] The computing device (200) may be configured to include memory (260) and a processor.

[0067] As illustrated in FIG. 2, the processor may be configured to include a communication module (210), a data acquisition unit (220), a data preprocessing unit (230), a data processing unit (240), and a control unit (250).

[0068] The communication module (210) provides a wired / wireless communication interface environment, which can support the computing device (200) in exchanging data with the terminal (100) and / or the image acquisition device (300).

[0069] The communication module (210) provides a wired / wireless communication interface environment to support the computing device (200) in exchanging data with an external database. At this time, the external database may include, for example, a database related to the medical field, such as a medical institution or a medical society. However, it is not limited thereto.

[0070] In particular, wireless communication interfaces may include LTE, LTE-A, 5G, and Wi-Fi communication methods.

[0071] The data acquisition unit (220) can acquire image data containing the user's target organ from the image acquisition device (300) through the communication module (210).

[0072] The data acquisition unit (220) may receive a signal from the terminal (100) through the communication module (210).

[0073] The above signal may include image data containing a target organ transmitted directly by the user.

[0074] The above signal may include data regarding requests for prediction and processing of the organ's state, suspension or cessation, termination, etc.

[0075] The data preprocessing unit (230) can preprocess image data containing a target organ acquired from the data acquisition unit (220) so that it can be processed by the data processing unit (240).

[0076] The data processing unit (240) can process image data containing the target organ preprocessed in the data preprocessing unit (230).

[0077] The memory (260) can store various information in advance or update the stored information. The memory (260) can temporarily store data acquired from a computing device (200) or data collected / acquired from a terminal (100) or an image acquisition device (300).

[0078] The memory (260) may store information regarding software, such as applications or programs, or related artificial intelligence processing engines, which are necessary for data preprocessing or data processing, as needed. The artificial intelligence processing engine may include, for example, an artificial intelligence learning engine that learns using a training dataset for learning, and this may be implemented within the control unit (250).

[0079] The control unit (250) can control the operation of each component constituting the processor and various devices within the computing device (200).

[0080] In FIG. 2, the memory (260) may include at least one type of storage medium among flash memory type, hard disk type, SSD type (Solid State Disk type), SSD type (Silicon Disk Drive type), multimedia card micro type, card type memory (e.g., SD or XD memory, etc.), RAM (random access memory; RAM), SRAM (static random access memory), ROM (read-only memory; ROM), EEPROM (electrically erasable programmable read-only memory), PROM (programmable read-only memory), magnetic memory, magnetic disk, and optical disk.

[0081] In FIG. 2, only one memory (260) is shown for convenience, but it is not limited to this and may be multiple.

[0082] In FIG. 2, the memory (260) may be built into a computing device (200) or terminal (100) for convenience, but is not limited thereto. For example, in FIG. 2, the memory (260) may be implemented in the form of a DB server located externally or remotely.

[0083] In FIG. 2, the memory and the processor may each be implemented as separate chips. Alternatively, the memory and the processor may be implemented as a single chip.

[0084] At least one component may be added or removed in response to the performance of the components shown in FIG. 2.

[0085] In addition, it will be readily understood by those skilled in the art that the inter-arrangement relationships of the components may be changed in response to the performance or structure of the system.

[0086] Meanwhile, each component illustrated in FIG. 2 may refer to software and / or hardware components such as Field Programmable Gate Arrays (FPGAs) and Application Specific Integrated Circuits (ASICs).

[0087] In FIGS. 3 to 5, the state prediction and processing operations regarding the target organ of the computing device (200), particularly the processor, are described in detail below.

[0088] As mentioned above, the target organ below will be explained using the kidney as an example for the convenience of explanation.

[0089] In addition, image data is explained using computed tomography (CT) image data as an example.

[0090] And clinical information may include estimated glomerular filtration rate (eGFR) information.

[0091] However, it is not limited to the above contents.

[0092] FIGS. 3 to 5 are flowcharts illustrated to explain a method for predicting and processing the condition of a human organ according to an embodiment of the present disclosure.

[0093] FIGS. 3 to 5 are described for convenience in terms of a computing device (200), i.e., a processor, but are not limited thereto.

[0094] That is, the method for predicting and processing the condition of a human organ according to at least one of the various embodiments of the present invention is performed as follows.

[0095] First, Fig. 3 may be a preparatory step for predicting the condition of human organs.

[0096] In step S110, the processor can collect image data of specific organs of the human body.

[0097] The above collection can be carried out not only through an internal database but also through at least one of the aforementioned external databases, etc.

[0098] In step S120, the processor can extract radiomix data from image data of a specific organ collected.

[0099] In step S130, the processor can machine learn using the extracted radiomix data as a dataset.

[0100] In step S140, the processor can distinguish between the normal and abnormal states of the organ based on the extracted radiomix data.

[0101] Next, FIG. 4 may be a detailed operation of step S120 of FIG. 3, that is, a detailed process of extracting radiomix data from image data of a specific organ collected.

[0102] In step S210, the processor can preprocess image data acquired from the image acquisition device (300).

[0103] In step S220, the processor can segment the region of interest (RoI) from the preprocessed image data.

[0104] In step S230, the processor can extract features from the segmented image data.

[0105] In step S240, the processor can select extracted features (feature selection).

[0106] In step S250, the processor can classify the selected features.

[0107] Finally, Fig. 5 can show the organ state prediction and processing process for the target organ of an actual user, i.e., a patient, in a computing device (200) after Figs. 3 and 4 described above.

[0108] In step S310, the processor can acquire image data of the user's organ.

[0109] In step S320, the processor can extract radiomix data from the acquired image data.

[0110] In step S330, the processor can determine whether the user's organ is in an abnormal state based on the extracted radiomix data.

[0111] In step S340, if the processor determines whether the user's organ is in an abnormal state and it is, it acquires clinical information, and can integrate the acquired clinical information with the radiomix data extracted for the user's organ.

[0112] In step S350, the processor can predict the abnormal state level of the user's organ based on integrated radiomix data and clinical information.

[0113] In step S360, the processor can transmit information regarding the abnormal state level of the predicted user's organ to the terminal.

[0114] If the process of FIG. 5 is performed on at least one user (i.e., a patient) or on at least one target organ, the data of FIG. 3 or FIG. 4 may be updated based on the information or data. In the update process, feedback data from the holder of the terminal (100), for example, a medical institution, may be referenced.

[0115] In the following sections, we can specifically describe the analysis of radiomix data for a specific organ, namely diabetic nephropathy, using a machine learning model derived from computed tomography, i.e., CT.

[0116] Diabetic nephropathy can be caused by hyperglycemia, diabetic nephropathy, metabolic factors, etc. It can also be caused by excessive fatty acids, carbonyls, and oxidative stress, etc., and hemodynamic factors such as hypertension, impaired self-regulation, hyperperfusion / hyperperfusion, RAAS activation, etc.

[0117] In connection with the present disclosure, renal function assessment may be based on serum creatinine, gross renal failure rate (GFR), and urinary protein ratio (ACR). However, these data may not be perfect for distinguishing the difference between AKI and CKD and distinguishing the type of renal damage.

[0118] In this regard, regarding the role of imaging for kidney evaluation, ultrasound is used to determine size, echogenicity of kidney tissue, presence of aqueduct formation or focal lesions, vascularity, or resistance index (RI); CT is used to determine the cause of aqueduct formation (calculi, inflammation, secondary ureteral carcinoma, etc.), kidney tumors, infections, renal incontinence, and vascular diseases (infarction, renal vein thrombosis); and MRI can be used to characterize ambiguous cystic or solid kidney tumors. However, there may be limitations in evaluating kidney function or the severity of kidney disease.

[0119] Accordingly, the present disclosure particularly discloses a method for extracting and analyzing radiomix data.

[0120] In this case, for the analysis operation, image (preprocessing), segmentation (2D or 3D ROI), feature extraction, shape features (circularity, degree of roundness, etc.), grayscale histogram, mean of pixel values, median, kurtosis, asymmetry, entropy, etc. may be referenced. In addition, high-dimensional features based on texture features (GLCM, GMRLM) and image filters (texture features) may be referenced.

[0121] Regarding feature selection, machine learning methods such as feature dimensionality reduction and regression can be used, and regarding classification, machine learning methods such as regression (SVM, DT, RF, KNN, etc.) can be used.

[0122] Meanwhile, in the present disclosure, the condition of an organ can be determined and predicted by referring to additional information from clinical images without acquiring quantitative information and images that cannot be determined by visual inspection.

[0123] In the present disclosure, radiomics studies of the kidney can be referenced, in particular, for the classification of tumor tissues, prognosis prediction (CT, MRI) of kidney tumors, for example, classification of the severity of Diabetic Kidney Disease (DKD) using ultrasound, classification of GN, prediction of renal fibrosis, and prediction of the burden of kidney stones using CT, prediction of radiation-induced kidney damage, etc.

[0124] Accordingly, this disclosure describes the prediction and processing of DKD levels through radiomix data analysis related to CT-based DKD.

[0125] FIGS. 6 to 10 are drawings illustrating a radio mix analysis method according to an embodiment of the present disclosure.

[0126] FIGS. 6 to 10 are drawings illustrating a method for analyzing radiomix data extracted from an image containing a patient's target organ acquired from an image acquisition device (300) in a processor.

[0127] Figure 6 shows the segmentation operation after image preprocessing.

[0128] In FIG. 6 (a), the image (610) may be a preprocessed image.

[0129] The processor can segment only the region of interest (RoI), i.e., the target organ region, from the preprocessed image. The image (630) obtained through segmentation is shown in FIG. 6 (a) and (b).

[0130] The segmentation illustrated in FIG. 6 can be performed automatically, and deep learning techniques may be used for this purpose. However, it is not limited thereto.

[0131] The segmented image in Fig. 6 may be a 2D (Fig. 6(a)) or 3D (Fig. 6(b)) image.

[0132] For the segmentation of Fig. 6, specific software can be used for deep learning-based automatic segmentation of both kidneys.

[0133] Figure 7 illustrates the operation of extracting features for predicting the state of a target organ from a segmented image in a processor.

[0134] For feature extraction, the processor may extract shape features using a histogram of grayscale pixel values ​​as shown in FIG. 7 (a), or extract textural features as shown in FIG. 7 (b).

[0135] In relation to the extraction of shape features of electrons, circularity, roundness, etc., may be used. Meanwhile, in histograms, mean, median, kurtosis, skewness, entropy, etc., may be referenced.

[0136] Regarding the extraction of texture features in the latter case, higher-order features using image filters can be referenced.

[0137] In relation to the feature extraction operation of FIG. 7 described above, radiomix features can be extracted from a segmented 3D ROI using specific software.

[0138] Although not explicitly stated, in the feature selection process, the processor can utilize machine learning methods such as feature dimension reduction and regression.

[0139] In relation to the aforementioned feature selection operations, Constant Feature Removal (CST), Linearly Associated Feature Removal (Corr), Relief in Estimating Features based on the importance of feature estimation (RELevance in Estimating Features), Lesser Absolute Value Shrinkage and Selection Operator (LASSO), Principal Component Analysis (PCA), and Forward Feature Selection (FFS) may be referenced.

[0140] Figure 8 may be a graph regarding machine learning methods, such as regression, used in a processor in relation to classification. In this case, a binary classifier may be used as a machine learning method. Examples of such binary classifiers include Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), K-Nearest Neighbor (KNN), ADAB, SGD, LDA, QDA, Naive Bayesian, etc.

[0141] Figure 9 is a diagram illustrating the use of ultrasound in the radiomix data analysis process of abnormal kidney disease (DKD).

[0142] On the other hand, Figure 10 is a diagram illustrating the use of CT in the radiomix data analysis process of abnormal kidney disease (DKD).

[0143] As illustrated in FIGS. 9 and 10, ultrasound and CT can be referenced in relation to the radiomix analysis of the kidney, i.e., DKD.

[0144] However, ultrasound may have limitations in radiomix data analysis due to 2D images, variability between operators during image acquisition, and the difficulty in acquiring homogeneous and reproducible images.

[0145] On the other hand, CT can assess the risk of diabetes in abdominal CT, and a combination of five features of CT images composed of radiomix features of the pancreas can be effective for early diabetes screening and future diabetes prediction (e.g., AUC > 0.8).

[0146] FIG. 11 is a drawing illustrating a process for predicting and treating the condition of human organs according to an embodiment of the present disclosure.

[0147] Figure 11 is illustrated to explain the determination of the abnormal kidney disease (DKD) condition and the classification of more specific grades of the determination condition.

[0148] 1110 can represent a basic premise, that is, data. For example, data from 90 healthy controls and 343 DKD patients may be used as data. However, this is for the convenience of explanation and is not limited thereto.

[0149] 1120 may represent a DKD level, i.e., a grade. Grades can be defined as normal, DKD grade I, DKD grade II, etc. This can be changed arbitrarily and depends on the settings.

[0150] Subsequently, the sample separated in 1120 can be enhanced. At this time, SMOTE may be used for the enhancement, but is not limited thereto.

[0151] In 1130, the set was constructed through the above sample augmentation.

[0152] 1140 provides examples of training and test sets for a configured set.

[0153] FIGS. 12 and 13 are drawings illustrated to explain baseline features related to human organ state prediction and processing according to an embodiment of the present disclosure.

[0154] FIGS. 14 and 15 are drawings illustrated to explain baseline features related to human organ state prediction and processing according to an embodiment of the present disclosure.

[0155] FIGS. 16 and 17 are drawings illustrating the correlation between eGFR and radiomix features related to human organ state prediction and processing according to an embodiment of the present disclosure.

[0156] Figures 12 and 13 show statistical analysis data.

[0157] Performance evaluation of CT radiomix features can be performed using the distinction between HC and DKD, and the distinction between DKD groups based on clinical information (e.g., eGFR).

[0158] In this case, the area under the receiver operation characteristic curve (AUC), sensitivity, specificity, and accuracy may be referenced.

[0159] Meanwhile, it may be necessary to define the correlation between radiomix characteristics and clinical information, specifically eGFR. This may be for the integrated processing of the aforementioned radiomix data and clinical information. Pearson's correlation coefficient may be used for the above correlation.

[0160] As a result, baseline characteristics are illustrated in FIGS. 12 to 13, the difference between HC and DKD grades is illustrated in FIGS. 14 (a) to (b), and the difference between DKD grades according to clinical information is illustrated in FIGS. 15 (a) to (c).

[0161] In Figures 14 and 15, a machine learning model was used, and LASSO x Random Forest was used as the machine learning model, but is not limited thereto.

[0162] In Fig. 16, numerical values ​​regarding the correlation between clinical information and radiomix characteristics for both kidneys are shown, and in Fig. 17, a correlation graph based on the numerical values ​​of Fig. 16 is shown.

[0163] In conclusion, the present disclosure aims to distinguish between normal kidneys and DKD, i.e., abnormal kidneys, using radiomix features extracted from non-contrast CT, and to evaluate and predict the severity of abnormal kidneys (DKD).

[0164] In other words, the present disclosure constructs a CT radiomix-based machine learning model, distinguishes between normal kidneys and DKD, and evaluates and predicts the performance of distinguishing DKD groups based on clinical information (e.g., eGFR).

[0165] Through this, the present disclosure can distinguish between abnormal kidney disease (DKD) and normal kidneys based on radiomix analysis of both kidneys via non-contrast abdominal CT using a machine learning model. Additionally, the status level, i.e., severity, of the distinguished abnormal kidney disease (DKD) can be evaluated and predicted.

[0166] Meanwhile, the disclosed embodiments may be implemented in the form of a recording medium that stores instructions executable by a computer. The instructions may be stored in the form of program code and, when executed by a processor, may generate a program module to perform the operation of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

[0167] Computer-readable recording media include all types of recording media that store instructions that can be decoded by a computer. Examples include ROM (Read Only Memory), RAM (Random Access Memory), magnetic tape, magnetic disk, flash memory, optical data storage devices, etc.

[0168] As described above, the disclosed embodiments have been explained with reference to the attached drawings. Those skilled in the art will understand that the present disclosure may be practiced in forms different from the disclosed embodiments without changing the technical spirit or essential features of the present disclosure. The disclosed embodiments are illustrative and should not be interpreted restrictively. Explanation of the symbols

[0169] 100 : Terminal 200 : Computing device 300: Image acquisition device

Claims

Claim 1 A method for predicting the severity of kidney disease, comprising: a step of extracting radiomix data from image data for a kidney, performed by a processor of a device; a step of obtaining estimated glomerular filtration rate (eGFR) information for the kidney; a step of defining a correlation between the radiomix data and the estimated glomerular filtration rate information using Pearson's correlation; a step of integrating the radiomix data and the estimated glomerular filtration rate information based on the defined correlation; a step of predicting an abnormal state level of the kidney based on the integrated radiomix data and the estimated glomerular filtration rate information; and a step of obtaining feedback data for the abnormal state level of the kidney. Claim 2 A method for predicting the severity of kidney disease according to claim 1, further comprising: a step of collecting image data for the kidney; a step of machine learning the extracted radiomix data as a dataset; and a step of distinguishing between a normal state and an abnormal state of the kidney based on the extracted radiomix data, wherein the step of obtaining estimated glomerular filtration rate (eGFR) information comprises: a step of obtaining the estimated glomerular filtration rate (eGFR) information when the distinction result is an abnormal state. Claim 3 A method for predicting the severity of kidney disease according to claim 1, wherein the step of extracting radiomix data comprises: a step of preprocessing image data; a step of segmenting a region of interest (RoI) from the preprocessed image data; a step of extracting features from the segmented image data; a step of selecting the extracted features; and a step of classifying the selected features. Claim 4 A method for predicting kidney disease severity according to claim 1, further comprising the step of transmitting information regarding the abnormal state level of the predicted user's kidney to a terminal. Claim 5 A method for predicting the severity of kidney disease according to claim 1, wherein the image data is computed tomography image data. Claim 6 A kidney disease severity prediction device comprising: a memory; and a processor for predicting kidney disease severity from image data of a kidney, wherein the processor extracts radiomix data from image data of the kidney, obtains estimated glomerular filtration rate (eGFR) information of the kidney, defines the correlation between the radiomix data and the estimated glomerular filtration rate information using Pearson's correlation coefficient, integrates the radiomix data and the estimated glomerular filtration rate information based on the defined correlation, predicts the abnormal state level of the kidney based on the integrated radiomix data and the estimated glomerular filtration rate information, and obtains feedback data for the abnormal state level of the kidney. Claim 7 A kidney disease severity prediction device according to claim 6, wherein the processor collects image data for the kidney, machine learns the extracted radiomix data as a dataset, distinguishes between a normal state and an abnormal state of the kidney based on the extracted radiomix data, and obtains the estimated glomerular filtration rate (eGFR) information when the result of the distinction is an abnormal state. Claim 8 A kidney disease severity prediction device according to claim 6, wherein the processor preprocesses the image data, segments the region of interest (RoI) from the preprocessed image data, extracts features from the segmented image data, selects the extracted features, classifies the selected features, and extracts radiomix data from the collected image data for kidneys. Claim 9 A kidney disease severity prediction device according to claim 6, wherein the processor transmits information regarding the abnormal state level of the predicted user's kidney to a terminal. Claim 10 A kidney disease severity prediction device according to claim 6, wherein the image data is computed tomography image data. Claim 11 A kidney disease severity prediction system comprising: an image acquisition device; a terminal; and a computing device for predicting kidney disease severity from image data of a kidney; wherein the computing device extracts radiomix data from image data of a kidney, acquires estimated glomerular filtration rate (eGFR) information of the kidney, defines the correlation between the radiomix data and the estimated glomerular filtration rate information using Pearson's correlation coefficient, integrates the radiomix data and the estimated glomerular filtration rate information based on the defined correlation, predicts the abnormal state level of the kidney based on the integrated radiomix data and the estimated glomerular filtration rate information, and acquires feedback data regarding the abnormal state level of the kidney from the terminal. Claim 12 A kidney disease severity prediction system according to claim 11, wherein the computing device collects image data for the kidney, machine learns the extracted radiomix data as a dataset, distinguishes between a normal state and an abnormal state of the kidney based on the extracted radiomix data, and obtains the estimated glomerular filtration rate (eGFR) information when the result of the distinction is an abnormal state. Claim 13 A kidney disease severity prediction system according to claim 11, wherein the computing device preprocesses the image data, segments a region of interest (RoI) from the preprocessed image data, extracts features from the segmented image data, selects the extracted features, classifies the selected features, and extracts radiomix data from the collected image data for kidneys. Claim 14 A kidney disease severity prediction system according to claim 11, wherein the computing device transmits information regarding the abnormal state level of the predicted user's kidney to the terminal. Claim 15 A kidney disease severity prediction system according to claim 11, wherein the image data is computed tomography image data.

Images