WO2022182182A1 - 딥러닝 알고리즘을 기반으로 하는 심전도 생성 시스템 및 그 방법 - Google Patents
딥러닝 알고리즘을 기반으로 하는 심전도 생성 시스템 및 그 방법 Download PDFInfo
- Publication number
- WO2022182182A1 WO2022182182A1 PCT/KR2022/002747 KR2022002747W WO2022182182A1 WO 2022182182 A1 WO2022182182 A1 WO 2022182182A1 KR 2022002747 W KR2022002747 W KR 2022002747W WO 2022182182 A1 WO2022182182 A1 WO 2022182182A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- electrocardiogram
- ecg
- learning
- virtual
- electrocardiograms
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000013135 deep learning Methods 0.000 title claims abstract description 17
- 230000001360 synchronised effect Effects 0.000 claims abstract description 18
- 238000013075 data extraction Methods 0.000 claims abstract description 10
- 238000005259 measurement Methods 0.000 claims abstract description 6
- 208000019622 heart disease Diseases 0.000 claims description 10
- 230000036541 health Effects 0.000 claims description 6
- 230000005856 abnormality Effects 0.000 claims description 3
- 238000004070 electrodeposition Methods 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 2
- 239000000203 mixture Substances 0.000 claims 2
- 238000012549 training Methods 0.000 abstract description 8
- 238000002565 electrocardiography Methods 0.000 description 63
- 210000000038 chest Anatomy 0.000 description 14
- 210000003414 extremity Anatomy 0.000 description 10
- 238000010586 diagram Methods 0.000 description 9
- 230000006698 induction Effects 0.000 description 9
- 239000000284 extract Substances 0.000 description 6
- 238000012544 monitoring process Methods 0.000 description 3
- 230000007423 decrease Effects 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 210000004072 lung Anatomy 0.000 description 2
- 238000012806 monitoring device Methods 0.000 description 2
- 210000002784 stomach Anatomy 0.000 description 2
- 208000006545 Chronic Obstructive Pulmonary Disease Diseases 0.000 description 1
- 208000019693 Lung disease Diseases 0.000 description 1
- 210000003423 ankle Anatomy 0.000 description 1
- 210000000481 breast Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000001684 chronic effect Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000007429 general method Methods 0.000 description 1
- 210000004247 hand Anatomy 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000004165 myocardium Anatomy 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 210000000707 wrist Anatomy 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/327—Generation of artificial ECG signals based on measured signals, e.g. to compensate for missing leads
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/332—Portable devices specially adapted therefor
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7271—Specific aspects of physiological measurement analysis
- A61B5/7278—Artificial waveform generation or derivation, e.g. synthesising signals from measured signals
Definitions
- the present invention relates to an electrocardiogram generating system and method based on a deep learning algorithm, and more particularly, to an electrocardiogram generating system and method for generating a plurality of electrocardiograms from one or more induced electrocardiograms using a deep learning algorithm will be.
- the 12-lead electrocardiogram records the potential of the heart in 12 electrical directions centered on the heart. Through this, the state of the heart can be judged from various directions, so that the heart disease confined to one area can be accurately read.
- Measuring the potential of the heart in various directions is meaningful in that the characteristics of the heart can be grasped in each direction. have.
- a single-lead ECG device using two electrodes includes a watch-type ECG device (Apple Watch or Galaxy Watch).
- a watch-type ECG device Apple Watch or Galaxy Watch.
- the back of the watch and the left wrist are in contact, and the right finger and the crown of the watch are in contact, so that the left and right arm electrodes come into contact, and the I-guided ECG is measured using the potential difference between the two electrodes.
- the watch-type electrocardiogram device is worn on the left arm and measured by touching the crown with the right hand to measure I-leading, placing the watch on the stomach and measuring the II-leading by touching the crown with the right hand with the left hand while placing the watch on the stomach. Measure III induction by touching the crown. Then, with the left hand in contact with the watch crown, the V1-6 induced electrocardiogram is measured by touching the back of the watch to the position of the V1-6 electrode.
- the above method has poor usability in that the user must precisely contact the electrocardiogram at the V1-6 position, and in the case of the V1-6-guided electrocardiogram, the standard chest that needs to show the potential difference between the imaginary center point and the V electrode even when touching the corresponding part Unlike guided electrocardiogram, there was a problem in that standard chest guidance could not be implemented in that it showed the potential difference between the right arm electrode and the V electrode.
- two or more induced electrocardiograms can be measured while holding electrodes in both hands, touching the electrodes on the back of the device to the legs or ankles, and touching the electrodes to each of the left arm, right arm, and left leg.
- the above method is a method that cannot be monitored for a long time (1 hour, 24 hours, 7 days, etc.), and has a disadvantage that the device itself must have three electrodes.
- an electrocardiogram generating system and method for generating a plurality of electrocardiograms from one or more induced electrocardiograms using a deep learning algorithm are provided.
- an electrocardiogram generating system based on a deep learning algorithm includes a 12-guided electrocardiogram measured by a plurality of patients and a data input unit for receiving patient information corresponding to the 12-guided electrocardiogram; A data extraction unit for classifying and storing the input 12-guide ECG according to the patient information, extracting learning data from the stored 12-guide ECG, and inputting the extracted learning data into one or more learning models to determine the characteristics of the ECG
- a learning unit for learning an electrocardiogram generator that receives one or more reference electrocardiograms from the measurement target, and inputs the input reference electrocardiogram into one or more learning models that have completed learning to generate a virtual electrocardiogram, and the reference electrocardiogram and generation and a controller for mutually synchronizing the synchronized virtual ECG and outputting waveforms for the synchronized reference ECG and the virtual ECG.
- the patient information may include at least one of gender, age, whether or not a heart disease is present, and a potential vector of a measured electrocardiogram.
- the learning unit inputs n-guided electrocardiograms, which are part of the extremity 6-guided electrocardiogram and the chest 6-guided electrocardiogram, into the first learning model, and causes the first learning model to determine the electric potential vector for the input electrocardiogram, and conduct 12-n electrocardiograms. can be trained to create
- the learning unit builds 12 second learning models for the extremity 6-lead ECG and the chest 6-guide ECG, and inputs 12-n generated ECGs from the constructed first learning model to the second learning model. , it is possible to learn to convert the input induced electrocardiogram into a style having a corresponding potential vector and output it as n virtual electrocardiograms.
- the first learning model and the second learning model may be constructed by mixing an adversarial generation network and an autoencoder method.
- the electrocardiogram generating unit inputs the input reference electrocardiogram into a first learning model, extracts a potential vector for the reference electrocardiogram, generates a plurality of remaining virtual electrocardiograms, and generates the generated second learning model with the extracted potential vector. By inputting the obtained virtual ECG, a virtual ECG having the same induction as the reference ECG may be generated again.
- the controller matches and synchronizes the reference ECG with the plurality of virtual ECGs, and outputs the synchronized plurality of ECGs through a monitor.
- the electrocardiogram generating system based on a deep learning algorithm generates 12-guided electrocardiograms measured by a plurality of patients and patient information corresponding to the 12-guided electrocardiograms.
- heart disease can be accurately read like a 12-guide electrocardiogram.
- the present invention since two guided electrocardiograms are used, it can be measured at home or in daily life, and it can be measured even in a moving state, so real-time monitoring is possible. , so that the medical staff can read the synchronized guided electrocardiogram information tailored to the corresponding beat, and based on this, more accurate diagnosis is possible.
- FIG. 1 is a configuration diagram for explaining an electrocardiogram generating system according to an embodiment of the present invention.
- FIG. 2 is a flowchart illustrating a method for generating an electrocardiogram using an electrocardiogram generating system according to an embodiment of the present invention.
- FIG. 3 is an exemplary diagram for explaining a method of measuring a general 12-lead ECG.
- FIG. 4 is an exemplary diagram for explaining step S240 shown in FIG. 2 .
- FIG. 5 is an exemplary diagram for explaining step S260 shown in FIG. 2 .
- the electrocardiogram generating system 100 includes a data input unit 110 , a data extracting unit 120 , a learning unit 130 , an electrocardiogram generating unit 140 , and a controller 150 .
- the data input unit 110 receives 12-guided electrocardiograms and patient information corresponding to 12-guided electrocardiograms measured through a plurality of patients.
- the 12-lead ECG includes a six-lead ECG of a limb and a six-lead ECG of the chest
- the patient information includes at least one of gender, age, whether or not a heart disease is present, and a potential vector of the measured ECG.
- the data extraction unit 120 classifies the 12-guided electrocardiogram using the input patient information and stores it in the database. Then, the data extraction unit 120 generates learning data by randomly extracting a plurality of 12-guided electrocardiograms stored in the database.
- the learning unit 130 learns the learning model built using the learning data.
- the learning unit 130 extracts the potential vector information for the input induced electrocardiogram to generate 12-n induced virtual electrocardiograms from the n induced reference electrocardiograms, the first learning model, and 12-n induced electrocardiograms.
- a second learning model is constructed that generates n-derived virtual electrocardiograms from the virtual electrocardiograms.
- the learning unit 130 learns the first learning model or the second learning model by inputting the learning data generated in the constructed first learning model or the second learning model.
- the electrocardiogram generator 140 acquires one or more reference electrocardiograms obtained from the measurement target. Then, the electrocardiogram generator 140 inputs the obtained reference electrocardiogram to the first learning model on which the learning is completed, and generates a first virtual electrocardiogram based on a potential vector with respect to the reference electrocardiogram. The electrocardiogram generator 140 inputs the first virtual electrocardiogram generated to the learned second learning model, and causes the second learning model to output the second virtual electrocardiogram.
- controller 150 synchronizes the reference ECG and the virtual ECG, and outputs a plurality of ECGs synchronized from the reference ECG and the virtual ECG through the monitor device.
- FIG. 1 is a configuration diagram for explaining an electrocardiogram generating system according to an embodiment of the present invention.
- the electrocardiogram generating system 100 includes a data input unit 110 , a data extractor 120 , a learning unit 130 , an electrocardiogram generating unit 140 , and a control unit ( 150).
- the data input unit 110 receives 12-guided electrocardiograms and patient information corresponding to 12-guided electrocardiograms measured through a plurality of patients.
- the 12-lead ECG includes a six-lead ECG of a limb and a six-lead ECG of the chest
- the patient information includes at least one of gender, age, whether or not a heart disease is present, and a potential vector of the measured ECG.
- the data extraction unit 120 classifies the 12-guided electrocardiogram using the input patient information and stores it in the database. Then, the data extraction unit 120 generates learning data by randomly extracting a plurality of 12-guided electrocardiograms stored in the database.
- the learning unit 130 learns the learning model built using the learning data.
- the learning unit 130 extracts the potential vector information for the input induced electrocardiogram to generate 12-n induced virtual electrocardiograms from the n induced reference electrocardiograms, the first learning model, and 12-n induced electrocardiograms.
- a second learning model is constructed that generates n-derived virtual electrocardiograms from the virtual electrocardiograms.
- the learning unit 130 learns the first learning model or the second learning model by inputting the learning data generated in the constructed first learning model or the second learning model.
- the electrocardiogram generator 140 acquires one or more reference electrocardiograms obtained from the measurement target. Then, the electrocardiogram generator 140 inputs the obtained reference electrocardiogram to the first learning model on which the learning is completed, and generates a first virtual electrocardiogram based on a potential vector with respect to the reference electrocardiogram. The electrocardiogram generator 140 inputs the first virtual electrocardiogram generated to the learned second learning model, and causes the second learning model to output the second virtual electrocardiogram.
- controller 150 synchronizes the reference ECG and the virtual ECG, and outputs a plurality of ECGs synchronized from the reference ECG and the virtual ECG through the monitor device.
- FIG. 2 is a flowchart illustrating a method for generating an electrocardiogram using an electrocardiogram generating system according to an embodiment of the present invention.
- the method for generating an electrocardiogram using the electrocardiogram generating system is divided into a step of learning a learning model and a step of generating an electrocardiogram using the learning model on which the learning is completed.
- the electrocardiogram generating system 100 receives the guided electrocardiogram measured through a plurality of patients and information about the patient (S210).
- the 12-guide electrocardiogram represents the recording of the potential of the heart in 12 electrical directions centered on the heart.
- the electrocardiogram measuring device measures the potential with dual arms and left leg electrodes, and the right leg electrode serves as a ground electrode.
- FIG. 3 is an exemplary diagram for explaining a method of measuring a general 12-lead ECG.
- the electrocardiogram measuring device generates an I-guided electrocardiogram by subtracting the potential of the right arm from the potential of the left arm, subtracting the potential of the right arm from the potential of the left leg to generate a II-guided electrocardiogram, and the potential of the left arm from the potential of the left leg Subtracted to produce a III-guided electrocardiogram.
- the electrocardiogram device calculates the average of the potentials of the electrodes of both arms and the left leg to obtain the potential at the virtual center point. And the ECG measurement device generates an aVL-induced ECG by subtracting the potential of the imaginary central point from the potential of the left arm, and generates an aVR-induced ECG by subtracting the potential of the imaginary central point from the potential of the right arm. In addition, the electrocardiogram device generates an aVF-induced electrocardiogram by subtracting the potential of the imaginary central point from the potential of the left leg electrode. As described above, the electrocardiogram measuring device generates a total of 6 guided (6 lines) electrocardiograms using three (4) extremity electrodes.
- the electrocardiogram measuring device generates a chest 6-guided electrocardiogram using the potential difference between the virtual center point determined in extremity guidance and the potential difference between the 6 electrodes attached to the chest. That is, 6 electrodes (V1, V2, V3, V4, V5, V6) are attached to the chest from the front of the predetermined position to the left chest.
- the electrocardiogram measuring device generates a V1-induced electrocardiogram by subtracting the potential measured at the V1 electrode and the potential at the imaginary central point obtained through the average of the extremity electrodes above.
- the electrocardiogram measuring apparatus generates a 12-guided electrocardiogram using electrodes attached to a plurality of patients. Then, the generated 12-guided electrocardiogram is transmitted to the electrocardiogram generating system 100 .
- the electrocardiogram generating system 100 additionally receives the generated 12-guided electrocardiogram and patient information corresponding thereto.
- the patient information includes at least one of gender, age, whether or not a heart disease is present, and a potential vector of the measured electrocardiogram.
- step S210 the electrocardiogram generating system 100 extracts learning data using the collected 12-guided electrocardiogram and patient information (S220).
- the data extraction unit 120 classifies the collected 12-guided electrocardiogram according to patient information and stores it in the database. Then, the data extraction unit 120 generates training data by randomly extracting from the stored plurality of 12-guided electrocardiograms.
- the learning unit 130 trains the first learning model and the second learning model by using the generated learning data (S230).
- the learning unit 130 learns the characteristics of the electrocardiogram by inputting learning data composed of 12-guided electrocardiograms into the first learning model and the second learning model.
- the direction of electric flow is different according to the potential vector, and the electrocardiogram style is affected according to the age and gender of the patient.
- the amplitude of the electrocardiogram tends to decrease.
- the electrocardiogram electrode position is lowered due to the breast or the distance between the heart and the electrode increases, resulting in a change in the electrocardiogram shape.
- the learning unit 130 inputs the induced electrocardiogram stored separately according to the patient information, that is, the patient's age, sex, health level, and potential vector, to the first learning model and the second learning model. Then, the first learning model and the second learning model learn the characteristics of the input induced ECG and extract potential vector information of the input induced ECG. However, it is also possible to learn by inputting an electrocardiogram without distinguishing it according to patient information, and through this method, a learning model that is not limited to one characteristic can be learned.
- the learning unit 130 builds 12 second learning models corresponding to the 12-guided electrocardiogram. And the learning unit 130 learns each second learning model according to the characteristics of the electrocardiogram.
- the first learning model learns the association between the input electrocardiogram and the 12-guided electrocardiogram based on the deep learning algorithm to extract at least one of the characteristics of the input electrocardiogram, that is, age, gender, and potential vector, and A virtual ECG of 12-n inductions is generated from the reference ECG of n inductions.
- the second learning model generates n-guided virtual ECGs from 12-n-guided virtual ECGs. For example, assuming that the V1-induced ECG is generated, the learning unit 130 uses the virtual I, II, III, aVL, aVR, aVF, V2, V3, V4, V5 from the V1-induced ECG in the first learning model.
- the style is learned to generate an electrocardiogram of V6 induction
- the second learning model is an electrocardiogram of virtual V1 induction from the hypothetical I, II, III, aVL, aVR, aVF, V2, V3, V4, V5, V6 induction ECG. Teach the style to generate an electrocardiogram. Then, the second learning model converts the input electrocardiogram into a V1-guided style to generate a virtual electrocardiogram.
- the type of electrocardiogram is input and learned so as to determine which type of electrocardiogram the input electrocardiogram is.
- a 12-lead ECG can be generated even if an arbitrary ECG whose guidance is unknown is input.
- the first learning model and the second learning model are based on a deep learning algorithm composed of an autoencoder or an adversarial generating network, and the deep learning algorithm may be implemented using one selected from an autoencoder or an adversarial generating network, It can also be implemented by mixing autoencoders and adversarial generative networks.
- the electrocardiogram generating system 100 When the learning of the learning model is completed through steps S210 and S230, the electrocardiogram generating system 100 according to an embodiment of the present invention generates a heart diagram using the learned learning model.
- the electrocardiogram generating system 100 receives a reference electrocardiogram measured through an electrode attached to a body of a subject to be measured ( S240 ).
- the reference electrocardiogram does not include information on the potential vector.
- the electrocardiogram generator 140 generates a plurality of virtual electrocardiograms by inputting the input reference electrocardiograms to the first learning model and the second learning model ( S250 ).
- the electrocardiogram generator 140 inputs the reference electrocardiogram to the first learning model and extracts characteristics of the reference electrocardiogram.
- the characteristic includes at least one of age, gender, and a potential vector of the subject to be measured.
- the electrocardiogram generator 140 inputs the generated first virtual electrocardiogram to the second learning model. Then, the second learning model generates a second virtual ECG based on the input first virtual ECG.
- FIG. 4 is an exemplary diagram for explaining step S240 shown in FIG. 2 .
- the first electrocardiogram represents a reference electrocardiogram.
- the first learning model sets the characteristic for the reference ECG as L1.
- the electrocardiogram generating unit 140 generates virtual electrocardiograms of the remaining eleven inductions based on the L1 electrocardiogram of the first learning model.
- a virtual L1-induced ECG is generated.
- Each learning model is configured together with a generator that generates an ECG, and a discriminator to determine whether the generated ECG is correctly generated and to increase the accuracy of the generated ECG through feedback.
- step S250 the controller 150 outputs the reference ECG and 11 virtual ECGs through the monitoring device (S260).
- FIG. 5 is an exemplary diagram for explaining step S260 shown in FIG. 2 .
- the controller 150 outputs the reference ECG and 11 virtual ECGs. And, as shown in FIG. 5 , the controller 150 synchronizes the output reference electrocardiogram with 11 virtual electrocardiograms to determine whether the subject's health is abnormal. At this time, since the virtual electrocardiogram having the same induction as the reference electrocardiogram is also generated in the second learning model, it can be identified as 12 false electrocardiograms.
- the controller 150 outputs 12 electrocardiograms synchronized through the reference electrocardiogram and 11 virtual electrocardiograms through the monitoring device.
- heart disease can be accurately read like a 12-guide electrocardiogram.
- the electrocardiogram generating system uses two guided electrocardiograms, it can be measured at home or in daily life, and it can be measured even in a moving state, so real-time monitoring is possible.
- medical staff can read synchronized guided ECG information tailored to the corresponding beat, and based on this, more accurate diagnosis is possible.
- the present invention can generate a plurality of synchronized electrocardiograms from two electrocardiograms measured at different points using a deep learning algorithm, so it can accurately read heart diseases like a 12-guide electrocardiogram, and several kinds of deep learning algorithms It is industrially applicable to electrocardiogram generating systems based on
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Pathology (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Heart & Thoracic Surgery (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Cardiology (AREA)
- Artificial Intelligence (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
- Electrotherapy Devices (AREA)
Abstract
Description
Claims (14)
- 딥러닝 알고리즘을 기반으로 하는 심전도 생성 시스템에 있어서,복수의 환자에 의해 측정된 12 유도 심전도를 입력받는 데이터 입력부,상기 입력된 12 유도 심전도로부터 학습 데이터를 추출하는 데이터 추출부,상기 추출된 학습 데이터를 복수의 학습 모델에 입력하여 심전도에 대한 특성을 학습시키는 학습부,상기 측정대상자로부터 1개 이상의 기준 심전도를 입력받고, 상기 입력된 기준 심전도를 학습이 완료된 복수의 학습 모델에 입력하여 가상 심전도를 생성하는 심전도 생성부, 그리고상기 기준 심전도와 생성된 가상 심전도를 상호 동기화하며, 동기화된 기준 심전도 및 가상 심전도에 대한 파형을 출력하는 제어부를 포함하는 심전도 생성 시스템.
- 제1항에 있어서,상기 환자 정보는,성별, 나이, 심장 질환 여부, 측정된 심전도의 전위 벡터 중에서 적어도 하나를 포함하는 심전도 생성 시스템.
- 제2항에 있어서,상기 학습부는,사지 6유도 심전도와 흉부 6유도 심전도를 제1 학습모델에 입력하여 상기 제 1 학습모델로 하여금 입력된 심전도에 대한 전위 벡터를 판별하도록 학습시키는 심전도 생성 시스템.
- 제3항에 있어서,상기 학습부는,사지 6유도 심전도와 흉부 6유도 심전도에 대한 제1 학습모델과 제2 학습모델을 구축하고,상기 구축된 제 1 학습모델과 제2 학습모델에 기준 유도 심전도를 입력하면, 상기 입력된 기준 유도 심전도를 해당되는 전위 벡터를 가진 스타일로 변환하여 가상 심전도로 출력하도록 학습시키는 심전도 생성 시스템.
- 제3항 또는 제4항에 있어서,상기 제1 학습모델 및 제2 학습모델은,적대적 생성망과 오토인코더 방법을 혼합하거나 각각 사용하여 구축되는 심전도 생성 시스템.
- 제5항에 있어서,상기 심전도 생성부는,입력된 기준 심전도를 제1 학습모델에 입력하여 상기 기준 심전도에 대한 전위 벡터를 추출하고 제 1 가상 심전도를 생성하며,상기 제 1 가상 심전도를 제2 학습모델에 입력하여 제 2 가상 심전도를 생성하는 심전도 생성 시스템.
- 제6항에 있어서,상기 제어부는,상기 기준 심전도와 상기 복수의 가상 심전도를 매칭하여 동기화하고, 동기화된 복수개의 심전도를 모니터를 통해 출력시키며,상기 출력된 기준 심전도 및 가상 심전도의 파형의 진폭, 기울기, 전극 위치 중에서 적어도 하나를 이용하여 측정 대상자의 건강 이상 여부를 판단하는 심전도 생성 시스템.
- 심전도 생성 시스템을 이용한 심전도 생성 방법에 있어서,복수의 환자에 의해 측정된 12 유도 심전도와 상기 12 유도 심전도에 대응되는 환자 정보를 입력받는 단계,상기 입력된 12 유도 심전도를 상기 환자 정보에 따라 분류하여 저장하고, 저장된 12 유도 심전도로부터 학습 데이터를 추출하는 단계,상기 추출된 학습 데이터를 복수의 학습 모델에 입력하여 심전도에 대한 특성을 학습시키는 단계,상기 측정대상자로부터 1개 이상의 기준 심전도를 입력받고, 상기 입력된 기준 심전도를 학습이 완료된 복수의 학습 모델에 입력하여 가상 심전도를 생성하는 단계, 그리고상기 기준 심전도와 생성된 가상 심전도를 상호 동기화하며, 동기화된 기준 심전도 및 가상 심전도에 대한 파형과 상기 측정대상자에 대한 건강 이상 발생 여부를 출력하는 단계를 포함하는 심전도 생성 방법.
- 제8항에 있어서,상기 환자 정보는,성별, 나이, 심장 질환 여부, 측정된 심전도의 전위 벡터 중에서 적어도 하나를 포함하는 심전도 생성 방법.
- 제8항에 있어서,상기 심전도에 대한 특성을 학습시키는 단계는,사지 6유도 심전도와 흉부 6유도 심전도를 제1 학습모델에 입력하여 상기 제 1 학습모델로 하여금 입력된 심전도에 대한 전위 벡터를 판별하고 제 1 가상 심전도를 생성하도록 학습시키는 심전도 생성 방법.
- 제10항에 있어서,상기 심전도에 대한 특성을 학습시키는 단계는,제 1 가상심전도를 기반으로 스타일을 학습하는 제2 학습모델을 구축하고,상기 구축된 제2 학습모델에 제1의 가상심전도를 입력하면, 상기 입력된 유도 심전도를 해당되는 전위 벡터를 가진 스타일로 변환하여 제2이 가상 심전도로 출력하도록 학습시키는 심전도 생성 방법.
- 제10항 또는 제11항에 있어서,상기 제1 학습모델 및 제2 학습모델은,적대적 생성망과 오토인코더 방법을 혼합하거나 각각 사용하여 구축되는 심전도 생성 방법.
- 제12항에 있어서,상기 가상 심전도를 생성하는 단계는,입력된 기준 심전도를 제1 학습모델에 입력하여 상기 기준 심전도에 대한 전위 벡터를 추출하고 제1 가상 심전도를 생성하고,상기 추출된 전위 벡터로 학습된 제2 학습모델에 상기 제1 가상 심전도를 입력하여 제 2가상 심전도를 생성하는 심전도 생성 방법.
- 제13항에 있어서,상기 건강 이상 발생 여부를 출력하는 단계는,상기 기준 심전도와 상기 복수의 가상 심전도를 매칭하여 동기화하고, 동기화된 복수개의 심전도를 모니터를 통해 출력시키며,상기 출력된 기준 심전도 및 가상 심전도의 파형의 진폭, 기울기, 전극 위치 중에서 적어도 하나를 이용하여 측정 대상자의 건강 이상 여부를 판단하는 심전도 생성 방법.
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP22760098.8A EP4248867A1 (en) | 2021-02-24 | 2022-02-24 | System and method for generating electrocardiogram on basis of deep learning algorithm |
US18/258,859 US20240047075A1 (en) | 2021-02-24 | 2022-02-24 | System and method for generating electrocardiogram on basis of deep learning algorithm |
JP2023545852A JP2024505217A (ja) | 2021-02-24 | 2022-02-24 | ディープラーニングアルゴリズムに基づく心電図生成システム及びその方法 |
CN202280008177.9A CN116744850A (zh) | 2021-02-24 | 2022-02-24 | 基于深度学习算法的心电图生成系统及方法 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR10-2021-0024707 | 2021-02-24 | ||
KR1020210024707A KR20220120922A (ko) | 2021-02-24 | 2021-02-24 | 딥러닝 알고리즘을 기반으로 하는 심전도 생성 시스템 및 그 방법 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022182182A1 true WO2022182182A1 (ko) | 2022-09-01 |
Family
ID=83049528
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2022/002747 WO2022182182A1 (ko) | 2021-02-24 | 2022-02-24 | 딥러닝 알고리즘을 기반으로 하는 심전도 생성 시스템 및 그 방법 |
Country Status (6)
Country | Link |
---|---|
US (1) | US20240047075A1 (ko) |
EP (1) | EP4248867A1 (ko) |
JP (1) | JP2024505217A (ko) |
KR (1) | KR20220120922A (ko) |
CN (1) | CN116744850A (ko) |
WO (1) | WO2022182182A1 (ko) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20060117546A (ko) * | 2005-05-11 | 2006-11-17 | 인하대학교 산학협력단 | 신경망을 이용한 심전도 기반의 심장질환 진단장치 및방법 |
KR20190141326A (ko) * | 2018-06-14 | 2019-12-24 | 한국과학기술원 | 심층 컨볼루션 신경망을 이용한 심전도 부정맥 분류 방법 및 장치 |
US20200121255A1 (en) * | 2017-11-27 | 2020-04-23 | Lepu Medical Technology (Bejing) Co., Ltd. | Artificial intelligence-based interference recognition method for electrocardiogram |
KR20200068161A (ko) * | 2018-12-04 | 2020-06-15 | 건양대학교산학협력단 | 기계학습 모델을 이용한 심장질환예측 시스템, 및 방법 |
KR102142841B1 (ko) * | 2019-11-06 | 2020-08-10 | 메디팜소프트(주) | Ai 기반 심전도 판독 시스템 |
KR102180135B1 (ko) | 2018-10-12 | 2020-11-17 | 계명대학교 산학협력단 | 심혈관 질환 종류에 따른 심전도 패턴 시뮬레이션 생체신호 구현 시스템 및 방법 |
-
2021
- 2021-02-24 KR KR1020210024707A patent/KR20220120922A/ko not_active Application Discontinuation
-
2022
- 2022-02-24 EP EP22760098.8A patent/EP4248867A1/en active Pending
- 2022-02-24 WO PCT/KR2022/002747 patent/WO2022182182A1/ko active Application Filing
- 2022-02-24 JP JP2023545852A patent/JP2024505217A/ja active Pending
- 2022-02-24 US US18/258,859 patent/US20240047075A1/en active Pending
- 2022-02-24 CN CN202280008177.9A patent/CN116744850A/zh active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20060117546A (ko) * | 2005-05-11 | 2006-11-17 | 인하대학교 산학협력단 | 신경망을 이용한 심전도 기반의 심장질환 진단장치 및방법 |
US20200121255A1 (en) * | 2017-11-27 | 2020-04-23 | Lepu Medical Technology (Bejing) Co., Ltd. | Artificial intelligence-based interference recognition method for electrocardiogram |
KR20190141326A (ko) * | 2018-06-14 | 2019-12-24 | 한국과학기술원 | 심층 컨볼루션 신경망을 이용한 심전도 부정맥 분류 방법 및 장치 |
KR102180135B1 (ko) | 2018-10-12 | 2020-11-17 | 계명대학교 산학협력단 | 심혈관 질환 종류에 따른 심전도 패턴 시뮬레이션 생체신호 구현 시스템 및 방법 |
KR20200068161A (ko) * | 2018-12-04 | 2020-06-15 | 건양대학교산학협력단 | 기계학습 모델을 이용한 심장질환예측 시스템, 및 방법 |
KR102142841B1 (ko) * | 2019-11-06 | 2020-08-10 | 메디팜소프트(주) | Ai 기반 심전도 판독 시스템 |
Also Published As
Publication number | Publication date |
---|---|
KR20220120922A (ko) | 2022-08-31 |
JP2024505217A (ja) | 2024-02-05 |
CN116744850A (zh) | 2023-09-12 |
EP4248867A1 (en) | 2023-09-27 |
US20240047075A1 (en) | 2024-02-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2019088462A1 (ko) | 혈압 추정 모델 생성 시스템 및 방법과 혈압 추정 시스템 및 방법 | |
AU2016281828B2 (en) | Electronic system to control the acquisition of an electrocardiogram | |
WO2021015570A1 (ko) | 싱글리드 심전도 데이터를 이용하여 심장의 질병 유무를 판단하는 심전도 측정 시스템 및 그 방법 | |
WO2022014943A1 (ko) | 생성적 적대 신경망 알고리즘을 기반으로 하는 심전도 생성 장치 및 그 방법 | |
US20210219901A1 (en) | Method and measuring arrangement for monitoring specific activity parameters of the human heart | |
WO2023120775A1 (ko) | 심전도 판독을 수정하기 위한 방법 및 장치 | |
WO2022173103A1 (ko) | 웨어러블 다중 생체 신호 측정장치 및 이를 이용한 인공지능 기반의 원격 모니터링 시스템 | |
WO2022182182A1 (ko) | 딥러닝 알고리즘을 기반으로 하는 심전도 생성 시스템 및 그 방법 | |
WO2023219319A1 (ko) | 웨어러블 심전도 패치를 이용한 재택 심장 재활 시스템 | |
WO2023022485A1 (ko) | 비동기 심전도를 이용한 건강상태 예측 시스템 | |
WO2023022507A1 (ko) | 2개 유도의 비동시적 심전도를 기반으로 동시적 심전도를 생성하는 방법 | |
WO2016200243A1 (ko) | 미병 분류를 보조하는 컴퓨팅 장치 및 방법 | |
WO2018097541A1 (ko) | 순환계질환 발생잠재도를 판단하는 장치 및 그 방법 | |
EP3949850A1 (en) | Mobile electrocardiography recording device | |
WO2023022484A1 (ko) | 단일유도 심전도기기를 활용한 건강상태 예측 시스템 | |
WO2023022516A1 (ko) | 딥러닝 알고리즘을 기반으로 복수개의 표준 심전도 데이터를 생성하는 방법 | |
WO2023022519A1 (ko) | 2유도 심전도 데이터를 이용한 복수개의 표준 심전도 데이터 생성 시스템 | |
WO2023022511A9 (ko) | 2개 유도의 비동시적 심전도를 기반으로 하는 동시적 심전도 생성 방법 | |
WO2023075010A1 (ko) | 생체 신호 처리 장치 및 방법 | |
WO2024014838A1 (ko) | 심전도 판독에 기반한 시각화 콘텐츠를 제공하는 방법, 프로그램 및 장치 | |
WO2023101090A1 (ko) | 딥 러닝을 이용하여 심방세동을 감지하는 방법 및 장치 | |
WO2022169020A2 (ko) | Ppg 기반 스펙트로그램 및 cnn에 기반하는 통증 평가 방법 및 장치 | |
WO2024014844A1 (ko) | 심전도에 기반하여 체성분을 측정하는 방법 및 그 장치 | |
WO2023022521A1 (ko) | 딥러닝기반 모델 및 원칙기반 모델 통합 심전도 판독 시스템 | |
WO2022255852A1 (ko) | 심전도를 통한 고칼륨 혈증 예측 알고리즘 구축 시스템 및 이를 이용한 심전도를 통한 고칼륨 혈증 예측 알고리즘 구축 방법 및 심전도를 이용한 고칼륨 혈증 예측 시스템 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22760098 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 202280008177.9 Country of ref document: CN |
|
WWE | Wipo information: entry into national phase |
Ref document number: 18258859 Country of ref document: US |
|
ENP | Entry into the national phase |
Ref document number: 2022760098 Country of ref document: EP Effective date: 20230622 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2023545852 Country of ref document: JP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |