WO2020111464A1 - Procédé d'identification de risque de naissance prématurée sur la base d'électromyogramme utérin - Google Patents

Procédé d'identification de risque de naissance prématurée sur la base d'électromyogramme utérin Download PDF

Info

Publication number
WO2020111464A1
WO2020111464A1 PCT/KR2019/011528 KR2019011528W WO2020111464A1 WO 2020111464 A1 WO2020111464 A1 WO 2020111464A1 KR 2019011528 W KR2019011528 W KR 2019011528W WO 2020111464 A1 WO2020111464 A1 WO 2020111464A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
uterine
uterine contraction
contraction
image data
Prior art date
Application number
PCT/KR2019/011528
Other languages
English (en)
Korean (ko)
Inventor
조영창
홍혁기
최연식
정석원
Original Assignee
전자부품연구원
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 전자부품연구원 filed Critical 전자부품연구원
Publication of WO2020111464A1 publication Critical patent/WO2020111464A1/fr

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/43—Detecting, measuring or recording for evaluating the reproductive systems
    • A61B5/4306—Detecting, measuring or recording for evaluating the reproductive systems for evaluating the female reproductive systems, e.g. gynaecological evaluations
    • A61B5/4343—Pregnancy and labour monitoring, e.g. for labour onset detection
    • A61B5/4356—Assessing uterine contractions
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • 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
    • A61B5/369—Electroencephalography [EEG]
    • 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
    • A61B5/389—Electromyography [EMG]
    • A61B5/391—Electromyography [EMG] of genito-urinary organs
    • 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
    • A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
    • 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
    • A61B5/7275—Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor

Definitions

  • the present invention relates to a method for identifying a preterm birth risk based on uterine electromyography, and more particularly, to a method for identifying a preterm birth risk based on uterine electromyography that determines whether a uterine contraction is performed using a uterine electromyography signal and identifies a preterm birth risk using the uterine electromyography signal will be.
  • Preterm birth usually means giving birth before the 37th week after 20 weeks of pregnancy. It is estimated that around 15 million newborns are born prematurely worldwide each year, and in Korea, the premature birth rate is already over 10%, and it is estimated that about 40,000 premature babies are born each year. Not only do these premature infants not receive sufficient nutrients from their mothers, but there is a great risk of death because the organs of the body are born immature. In addition, premature babies have weak immunity, so there are many residual diseases, which can cause problems with growth and development.
  • premature birth is the most important cause of neonatal death, and its incidence has recently increased, but there is no way to predict and prevent premature birth 100%. That is, in order to prevent the premature birth of pregnant women and lower the mortality rate due to the premature birth of the fetus, accurate prediction of preterm birth is essential.
  • TOCO Tocodynamometry
  • the present invention has been devised to solve the above problems, and an object of the present invention is to label the uterine electromyography signal for uterine contraction, to separate the uterine electromyography signal into signals for a plurality of frequency bands, and to separate signals.
  • an object of the present invention is to label the uterine electromyography signal for uterine contraction, to separate the uterine electromyography signal into signals for a plurality of frequency bands, and to separate signals.
  • Whether to uterine contraction based on the uterine electromyography signal by dividing into a certain time unit, normalizing the segmented signal, converting the segmented and normalized signal into image data in the frequency domain, and deep learning using the converted image data It is to provide a method for determining a uterine contraction, a device for determining a uterine contraction, and a method for identifying a preterm birth risk.
  • a method of determining a uterine contraction by a uterine contraction determination device includes: receiving an uterine electromyography signal; Labeling the uterine electromyography signal for uterine contraction; Separating the EMG signal into a plurality of signals for each frequency band; A segmentation step of dividing the separated signal into predetermined time units; Normalizing the divided signal; Converting the segmented and normalized signal into image data in the frequency domain; And generating a uterine contraction determination model for determining whether uterine contraction is based on the uterine electromyography signal through deep learning learning using the converted image data.
  • the uterine contraction signal may be labeled with the uterine electromyography signal using a signal measured by a tocodynamometer at the same time.
  • the uterine electromyography signal may be separated into four frequency band-specific signals.
  • the step of separating, the first separation signal corresponding to the intrauterine EMG signal, the second separation signal corresponding to the uterine signal in the low frequency band, the third separation signal corresponding to the uterine signal in the high frequency band, and others can also be separated into a fourth separation signal corresponding to the signal.
  • the first separation signal is a signal obtained by separating the frequency band of 0.001 Hz to 0.005 Hz from the EMG signal
  • the second separation signal is a signal obtained by separating the frequency band of 0.02 Hz to 0.45 Hz from the EMG signal.
  • the separation signal is a signal obtained by separating a frequency band of 0.8 Hz to 3 Hz from the EMG signal
  • the fourth separation signal may be a signal obtained by separating a frequency band exceeding 3 Hz from the EMG signal.
  • the segmentation step when the separated signal is divided into units of a predetermined time, some time may be overlapped.
  • the divided and normalized signal may be converted into image data in the frequency domain by using a short-time Fourier transform (STFT).
  • STFT short-time Fourier transform
  • the step of generating a predictive model of uterine contraction is through deep learning learning by a convolutional neural network (CNN) using the converted image data as learning data, and determining uterine contraction to determine whether uterine contraction is based on the EMG signal.
  • CNN convolutional neural network
  • determining the number of uterine contractions for a predetermined time based on the input uterine electromyography signal may further include.
  • the computer-readable recording medium receiving the EMG signal of the womb; Labeling the uterine electromyography signal for uterine contraction; Separating the EMG signal into a plurality of signals for each frequency band; A segmentation step of dividing the separated signal into predetermined time units; Normalizing the segmented signal; Converting the segmented and normalized signal into image data in the frequency domain; And generating a uterine contraction determination model for determining whether uterine contraction is based on the uterine electromyography signal through deep learning learning using the converted image data.
  • a computer program for executing a uterine contraction determination method is included.
  • the apparatus for determining uterine contraction comprises: an input unit that receives an uterine electromyography signal; Labeling the uterine electromyography signal for uterine contraction, separating the uterine electromyography signal into signals for a plurality of frequency bands, segmenting the separated signals into units of a predetermined time, normalizing the segmented signals, and segmenting and normalizing signals It includes; a controller for converting to image data in the frequency domain, and through deep learning learning using the converted image data, to generate a uterine contraction determination model for determining whether uterine contraction is based on the uterus electromyography signal.
  • a method of identifying a preterm birth risk by a uterine contraction determination device includes: receiving an EMG signal; Labeling the uterine electromyography signal for uterine contraction; Separating the EMG signal into a plurality of signals for each frequency band; A segmentation step of dividing the separated signal into predetermined time units; Normalizing the segmented signal; Converting the segmented and normalized signal into image data in the frequency domain; Generating a uterine contraction determination model for determining whether uterine contraction is based on the uterine electromyography signal through deep learning learning using the converted image data; Determining the number of uterine contractions for a predetermined period of time based on the input uterine electromyography signal using the generated uterine contraction determination model; And identifying the risk of preterm birth based on the number of cervical contractions for a period of time.
  • the uterine electromyography signal is labeled with uterine contraction
  • the uterine electromyography signal is divided into signals for a plurality of frequency bands
  • the separated signals are divided into units of a predetermined time
  • the segmented signals are normalized.
  • Uterine contraction which converts segmentation and normalized signals into image data in the frequency domain, and through deep learning learning using the converted image data, generates a uterine contraction judgment model that determines whether uterine contraction is based on the uterine electromyography signal. It is possible to provide a judgment method, a uterine contraction judgment device, and a method for identifying a preterm birth risk, so that it is possible to more easily and accurately predict whether a uterine contraction and a preterm birth risk.
  • FIG. 1 is a block diagram showing the configuration of a device for determining uterine contraction according to an embodiment of the present invention
  • FIG. 2 is a flowchart provided to explain a method of determining uterine contraction and identifying a preterm birth risk according to an embodiment of the present invention
  • FIG. 3 is a view showing a signal measured by a uterine contraction force meter (tocodynamometer) and an EMG signal according to an embodiment of the present invention
  • FIG. 4 is a diagram illustrating a signal in which the uterine electromyography signal is separated into four frequency bands according to an embodiment of the present invention
  • FIG. 5 is a diagram illustrating a process of dividing a separated signal into predetermined time units according to an embodiment of the present invention
  • FIG. 6 is a view showing a short-time Fourier transform process, according to an embodiment of the present invention.
  • FIG. 7 is a diagram illustrating image data obtained as a result of a short-time Fourier transform when a user is in a normal state, according to an embodiment of the present invention.
  • FIG. 8 is a diagram illustrating image data obtained as a result of a short-time Fourier transform when a user is in a uterine contraction state according to an embodiment of the present invention
  • FIG. 9 is a diagram illustrating a process of generating a uterine contraction judgment model by performing deep learning learning using transformed image data according to an embodiment of the present invention.
  • FIG. 10 is a view showing a test result of the apparatus for determining uterine contraction according to an embodiment of the present invention.
  • first and second are used to describe various components, and are used only to distinguish one component from other components, and to limit the components It is not used.
  • the second component may be referred to as a first component without departing from the scope of the present invention, and similarly, the first component may also be referred to as a second component.
  • 1 is a block diagram showing the configuration of a uterine contraction determining device 100 according to an embodiment of the present invention. 1, the input unit 110 and the control unit 120 are included.
  • the uterine electromyography signal is an electrical signal having characteristics of 0.1 to 5 Hz and 0.04 to 0.5 mV generated in the myometrium of the uterus due to uterine electrical activity such as uterine contraction.
  • the uterine electromyography device 10 is composed of an abdominal attachment-type electromyography measurement patch that attaches an electrode to an abdomen of a pregnant woman in which a fetus is located.
  • the abdominal attachment type EMG measurement patch is designed to transmit the measured uterine EMG signal to the input unit 110.
  • the uterus electromyography sensor may be configured to be able to integrally measure bio signals such as fetal heart rate (THR) and body temperature in addition to the uterine electromyogram signal.
  • the input unit 110 receives an uterine electromyogram signal (EHG:electrohysterogram) from the uterine electromyography meter 10.
  • the input unit 110 is connected to the uterine electromyography device 10 so as to be able to communicate by wire or wirelessly, for example, may be connected through a wired cable such as USB, or may be connected through wireless communication such as Wi-Fi or Bluetooth.
  • the input unit 110 receives the EMG signal through wired or wireless communication.
  • the control unit 120 controls the overall operation of the apparatus 100 for determining the contraction of the uterus. Specifically, the control unit 120 labels the uterine electromyography signal whether or not uterine contraction, separates the uterine electromyography signal into signals for a plurality of frequency bands, segmentes the divided signals into units of a predetermined time, and segments the signals. Normalization, segmentation and normalized signals are converted into image data in the frequency domain, and deep learning learning using the converted image data generates a uterine contraction determination model that determines whether uterine contraction is based on the uterus electromyography signal.
  • the controller 120 determines the number of uterine contractions for a certain time based on the input uterine electromyography signal using the generated uterine contraction prediction model, and identifies the risk of preterm birth based on the number of uterine contractions for a predetermined time. .
  • the apparatus 100 for determining uterine contraction having such a configuration may determine whether the user has contracted the uterus by using the EMG signal measured by the user, and use this to identify the risk of preterm birth.
  • the uterine contraction determination device 100 may include a uterine electromyography measurement module and be configured. .
  • FIG. 2 is a flowchart provided to explain a method of determining uterine contraction and identifying a preterm birth risk according to an embodiment of the present invention.
  • the uterine contraction determination device 100 receives the user's uterine electromyography signal from the uterine electromyography meter 10 (S210).
  • the uterine contraction determining device 100 labels the uterine contraction signal on the uterus electromyography signal (S220). Specifically, the uterine contraction determining device 100 performs labeling by indicating whether the uterine contraction is in the uterine electromyography signal using the uterine contraction signal measured by the tocodynamometer at the same time.
  • FIG. 3 is a view showing a signal measured by a uterine contraction force meter (tocodynamometer) and an uterine electromyogram signal according to an embodiment of the present invention.
  • the graph displayed at the top in FIG. 3 represents a uterine contraction signal (TOCO signal) measured by a tocodynamometer. And, the graph displayed at the bottom in FIG. 3 is a diagram showing the EMG signal of the same time zone.
  • TOCO signal uterine contraction signal
  • the uterine contraction signal indicates that the time zone in which the peak is present indicates the time zone in which uterine contraction occurred. Therefore, the uterine contraction determination device 100 can check the time zone in which uterine contraction occurred using the uterine contraction signal, and label the uterine contraction signal in the uterine electromyography signal using the confirmed time zone in which uterine contraction occurred.
  • the uterine electromyography signal displayed at the bottom of FIG. 3 is labeled with red uterine contraction, a period in which a red graph corresponds to a high value indicates a period of uterine contraction, and a period in which a red graph corresponds to a low value indicates a steady state period. Shows.
  • the uterine contraction determination device 100 can display the uterine contraction signal in the uterine electromyography signal using the uterine contraction signal measured by the tocodynamometer at the same time to perform labeling. .
  • the apparatus for determining uterine contraction 100 separates the uterine electromyography signal into signals for a plurality of frequency bands (S230 ).
  • the uterine contraction determining device 100 may separate the uterine EMG signal into a plurality of frequency band signals by using a band pass filter for each band.
  • the uterine contraction determination device 100 may separate the uterine electromyography signal into four frequency band-specific signals.
  • the uterine contraction determination device 100 is a first separation signal corresponding to the intrauterine pressure signal, the second separation signal corresponding to the uterine signal in the low frequency band, the third separation corresponding to the uterine signal in the high frequency band
  • the signal may be separated into a fourth separation signal corresponding to signals and other signals.
  • the first separation signal is a signal obtained by separating the frequency band of 0.001 Hz to 0.005 Hz from the EMG signal
  • the second separation signal is a signal obtained by separating the frequency band of 0.02 Hz to 0.45 Hz from the EMG signal.
  • the third separation signal is a signal obtained by separating a frequency band of 0.8 Hz to 3 Hz from the EMG signal
  • the fourth separation signal may be a signal obtained by separating a frequency band exceeding 3 Hz from the EMG signal.
  • the uterine contraction determining device 100 separates the uterine electromyography signal into four frequency band-specific signals and processes the signal to determine whether uterine contraction is performed. It can be seen that the accuracy of judging whether the uterus contracted is improved.
  • the uterine contraction determining apparatus 100 performs a segmentation step of dividing the signal separated for each frequency band into predetermined time units (S240).
  • the cervical contraction determining device 100 is divided so that some time overlaps when the separated signal is divided into a certain time unit. That is, the uterine contraction determining device 100 is configured to set the time domain so that the rear part of the segment leading ahead in time overlaps the part of the segment leading backward in time when the separated signal is divided into predetermined time units.
  • the apparatus for determining uterine contraction 100 can more accurately process the characteristics of the boundary portion to be divided.
  • the uterine contraction determining device 100 may divide the signal separated for each frequency band into 2 second units, and at this time, 25% (ie, 0.5 seconds) of the signal may be divided so as to overlap with the previous time domain. have. This will be described in more detail with reference to FIG. 5.
  • FIG. 5 is a diagram illustrating a process of dividing a separated signal into predetermined time units according to an embodiment of the present invention. Specifically, FIG. 5 shows that the uterine contraction determining device 100 divides the signal separated for each frequency band into a window of 2 seconds, and divides 25% (ie, 0.5 seconds) of the signal so that it overlaps the previous time domain window. The case is illustrated.
  • the signal graph displayed on the upper part of FIG. 5 indicates the input uterine electromyogram (EHG) (electrohysterogram), and it can be seen that the period of contraction of the uterus is labeled in red.
  • EHG electromyogram
  • the uterine contraction determination device 100 performs all of the above-described segmentation steps for the four separation signals for each frequency band, that is, the first separation signal, the second separation signal, the third separation signal, and the fourth separation signal. In this case, all the separated signals are made into a plurality of divided signals divided into time domains.
  • the apparatus for determining uterine contraction 100 normalizes the divided signal (S250 ). Specifically, the cervical contraction determining apparatus 100 performs normalization processing such that the divided signals have an overall amplitude of between -1 and 1, an average of 0, and a standard deviation between 0 and 1.
  • the uterine contraction determining apparatus 100 converts the segmented and normalized signal into image data in the frequency domain (S260). Specifically, the cervical contraction determining apparatus 100 converts the segmented and normalized signal into image data in the frequency domain by using a short-time Fourier transform (STFT).
  • STFT short-time Fourier transform
  • FIG. 6 is a diagram illustrating a short-time Fourier transform process according to an embodiment of the present invention.
  • the short-time Fourier transform is a widely used transform method for expressing the frequency change over time contained in a time series signal.
  • the uterine contraction determining device 100 multiplies the EMG signal of a specific range and a window function having a specific characteristic, and then performs frequency analysis using FFT (Fast Fourier Transform), and performs the splitting of the signals along the time axis. If FFT (Fast Fourier Transform) is continuously performed, it is possible to obtain image data in the frequency domain that is a result of the short-time Fourier transform.
  • FFT Fast Fourier Transform
  • the uterine contraction determining device 100 performs a short-time Fourier transform through the schematic process illustrated in FIG. 6 and acquires image data for signals separated for each frequency band.
  • the image data corresponds to the result of the short-time Fourier transform of the signals separated for each frequency band
  • the x-axis of the image data represents time
  • the y-axis represents frequency
  • the color represents the amplitude.
  • FIG. 7 is a diagram illustrating image data obtained as a result of a short-time Fourier transform when a user is in a normal state according to an embodiment of the present invention.
  • the graph displayed at the top of FIG. 7 shows the EMG signal 700 when the user is in a normal state.
  • the 1-1 image data 710 which is the result of the short time Fourier transform for the first separation signal (0.001 Hz to 0.005 Hz), and the short time Fourier for the second separation signal (0.02 Hz to 0.45 Hz).
  • the 1-4 image data 740 which is the result of the short-time Fourier transform, is shown.
  • FIG. 8 is a diagram illustrating image data obtained as a result of a short-time Fourier transform when a user is in a uterine contraction according to an embodiment of the present invention.
  • the graph displayed at the top of FIG. 8 shows the uterine electromyography signal 800 when the user is in a uterine contraction state.
  • the 2-1 image data 810 as a result of the short-time Fourier transform for the first separation signal (0.001 Hz to 0.005 Hz), and the short-time Fourier for the second separation signal (0.02 Hz to 0.45 Hz).
  • the 2-4 image data 840 which is the result of the short-time Fourier transform, is shown.
  • the cervical contraction determining apparatus 100 converts the segmented and normalized signal into image data in the frequency domain using a short-time Fourier transform (STFT), and uses the image pattern difference of the corresponding image data. Therefore, it can be used to determine whether or not the uterine contraction.
  • STFT short-time Fourier transform
  • the uterine contraction determination device 100 generates a uterine contraction determination model that determines whether uterine contraction is based on the EMG signal through deep learning learning using the converted image data (S270).
  • the uterine contraction determination device 100 is a uterine contraction determination model that determines whether uterine contraction is based on the EMG signal through deep learning learning using a convolutional neural network (CNN) using the converted image data as learning data.
  • CNN convolutional neural network
  • FIG. 9 is a diagram illustrating a process of generating a uterine contraction judgment model by performing deep learning learning using transformed image data according to an embodiment of the present invention.
  • CNN Convolutional Neural Network
  • the convolutional filter is a feature extraction filter basically used in a deep learning model using an image as an input.
  • the convolutional filter can extract various features as there are more types of filters, and serves to convert data of different patterns to be separated from each other by converting the pixel data of the image to different domains. Since the convolutional filter does not check the relationship of the entire data, the computational amount and computational parameters are less than the existing neural network.
  • the convolutional filter has a filter weights matrix for each channel, and after performing matrix multiplication on the weighting matrix that fits each channel, it adds all of them to extract the characteristics of the corresponding pixel.
  • Batch Normalization is a technique that stabilizes learning convergence through normalization when learning deep learning models. Assuming that the mean and standard deviation of the data that are linearly transformed through the filter are biased, batch normalization adjusts the mean and standard deviation in a way to re-adjust it, so that learning is stable and aims to accelerate learning speed and improve performance. There is this. Batch normalization is inserted as a module between deep learning layers, and scale parameters and shift parameters are scaled and shifted to fit the data based on the data when learning a deep learning model. ). After learning the scaler value, and completing the learning, it plays the role of re-adjusting the data passing through the layer using each batch normalization parameter when inferring the deep learning model.
  • Convolutional filters can see a certain portion of the input image according to the size of the filter weight (Filter weights), has a trade-off (trade-off) characteristic that the number of calculations and parameters increases as the filter weight increases.
  • trade-off trade-off
  • pooling is performed, and pooling has the advantage that the entire image can be viewed with the same convolutional filter size through the role of reducing the image size while leaving the typical properties of the image.
  • a Max pooling technique that reduces the size of an image while maintaining sharpness may be used.
  • Dropout is a technique that has the effect of normalizing by not randomly using the weights of the network, and aims to improve learning stability, convergence speed, and performance of deep learning models.
  • the dropout has the effect that all weights in the model do not learn all the imbalanced data by turning off certain weights for specific data in a situation where the data is imbalanced. Because it has a normalization effect, it serves to prevent overfitting of the training data.
  • the fully connected network is called a neural network (Neural Netowrk) or an artificial neural network, and in this embodiment, represents a network that determines boundary functions for data representations that can be separated through a convolutional filter.
  • CNN convolutional neural network
  • the uterine contraction determining device 100 integrates the four transformed image data into one image having 4 channels and performs deep learning learning through a convolutional neural network (CNN) having the structure shown in FIG. 9 as an input. .
  • CNN convolutional neural network
  • the apparatus 100 for determining uterine contraction undergoes deep learning by CNN (Convolutional Neural Network) using four transformed image data as learning data, and uterine contraction to determine whether uterine contraction is based on the EMG signal. Create a judgment model
  • the uterine contraction determination device 100 determines the number of uterine contractions for a predetermined time based on the input uterine electromyography signal using the generated uterine contraction determination model (S280).
  • the uterine contraction determination device 100 generates a uterine contraction determination model through the process described above using the user's existing uterine electromyography signal, inputs the currently input uterine electromyography signal into the corresponding uterine contraction determination model, and is currently input It is possible to determine how many times the uterine contraction is included in the EMG signal.
  • the uterine contraction determining device 100 identifies the risk of preterm birth based on the number of uterine contractions for a certain period of time (S290). For example, the uterine contraction determination device 100 may identify that there is a risk of preterm birth when uterine contraction is detected more than 10 times in 5 hours.
  • the uterine contraction determining device 100 may determine whether the uterine contraction is performed using the user's uterine electromyography signal and identify the risk of preterm birth.
  • FIG. 10 is a diagram illustrating a test result of the uterine contraction determining device 100 according to an embodiment of the present invention.
  • a total of 8 data sets are predicted for the entire data in the uterine contraction determination device 100, and 90% of the defined total data is trained for training the uterine contraction determination model. dataset), and the rest 10% was randomly shuffled with test dataset to test.
  • the apparatus 100 for determining uterine contraction can determine whether the uterine contraction is performed using the uterine EMG signal with a fairly high accuracy.
  • the technical idea of the present invention can be applied to a computer-readable recording medium containing a computer program for performing functions and methods of the apparatus according to the present embodiment.
  • the technical idea according to various embodiments of the present disclosure may be implemented in the form of computer-readable programming language codes recorded on a computer-readable recording medium.
  • the computer-readable recording medium can be any data storage device that can be read by a computer and stores data.
  • the computer-readable recording medium may be a ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, hard disk drive, flash memory, solid state disk (SSD), or the like.
  • computer-readable codes or programs stored on a computer-readable recording medium may be transmitted through a network connected between computers.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Veterinary Medicine (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Psychiatry (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Signal Processing (AREA)
  • Physiology (AREA)
  • Fuzzy Systems (AREA)
  • Pregnancy & Childbirth (AREA)
  • Mathematical Physics (AREA)
  • Evolutionary Computation (AREA)
  • Reproductive Health (AREA)
  • Gynecology & Obstetrics (AREA)
  • Psychology (AREA)
  • Urology & Nephrology (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

L'invention concerne un procédé et un appareil pour déterminer une contraction utérine, et un procédé pour identifier un risque de naissance prématurée. Le procédé, visant à déterminer une contraction utérine, repère dans un signal électromyographique utérin, si l'utérus se contracte, divise le signal électromyographique utérin en une pluralité de signaux spécifiques de bande de fréquence, segmente les signaux divisés sur la base d'une unité de temps prédéterminée, normalise les signaux segmentés, convertit les signaux segmentés et normalisés en données d'image d'un domaine fréquentiel, et génère, par apprentissage profond à l'aide des données d'image converties, un modèle de détermination de contraction utérine pour déterminer si l'utérus se contracte sur la base du signal électromyographique utérin, et permet ainsi de prédire plus facilement et avec précision si l'utérus est contracté et s'il y a un risque de naissance prématurée.
PCT/KR2019/011528 2018-11-27 2019-09-06 Procédé d'identification de risque de naissance prématurée sur la base d'électromyogramme utérin WO2020111464A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR1020180148241A KR102222282B1 (ko) 2018-11-27 2018-11-27 자궁 근전도 기반의 조산 위험도 식별방법
KR10-2018-0148241 2018-11-27

Publications (1)

Publication Number Publication Date
WO2020111464A1 true WO2020111464A1 (fr) 2020-06-04

Family

ID=70852840

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2019/011528 WO2020111464A1 (fr) 2018-11-27 2019-09-06 Procédé d'identification de risque de naissance prématurée sur la base d'électromyogramme utérin

Country Status (2)

Country Link
KR (1) KR102222282B1 (fr)
WO (1) WO2020111464A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113133771A (zh) * 2021-03-18 2021-07-20 浙江工业大学 基于时频域熵特征的子宫肌电信号分析及早产预测方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102603983B1 (ko) * 2021-07-01 2023-11-22 주식회사 인텔리고 근전도 신호를 이용한 딥러닝 기반의 행동 인식 시스템 및 방법
KR102594173B1 (ko) * 2021-11-29 2023-11-16 주식회사 데이터스튜디오 타임 시리즈 데이터 예측을 위한 학습 이미지를 생성하는 방법 및 장치

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003245258A (ja) * 2002-01-15 2003-09-02 General Electric Co <Ge> 腹部表面ｅｍｇ信号の線形予測モデリングを使用する子宮収縮監視の方法及び装置
KR20170130773A (ko) * 2016-05-19 2017-11-29 한국과학기술연구원 Eeg 및 emg 신호에 기반한 자궁 수축도 검사 장치 및 그 방법
KR101858411B1 (ko) * 2017-11-20 2018-05-15 재단법인 구미전자정보기술원 생체 신호를 이용한 조산 예측 방법 및 장치
KR20180087084A (ko) * 2017-01-24 2018-08-01 계명대학교 산학협력단 착용형 자궁 근전도 센서를 이용한 조산 예측 시스템 및 그 방법
KR101890072B1 (ko) * 2017-11-28 2018-08-21 주식회사 멕 아이씨에스 자궁 근전도 신호를 기반으로 임산부의 조산예측기능을 갖는 임산부 모니터링 시스템

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003245258A (ja) * 2002-01-15 2003-09-02 General Electric Co <Ge> 腹部表面ｅｍｇ信号の線形予測モデリングを使用する子宮収縮監視の方法及び装置
KR20170130773A (ko) * 2016-05-19 2017-11-29 한국과학기술연구원 Eeg 및 emg 신호에 기반한 자궁 수축도 검사 장치 및 그 방법
KR20180087084A (ko) * 2017-01-24 2018-08-01 계명대학교 산학협력단 착용형 자궁 근전도 센서를 이용한 조산 예측 시스템 및 그 방법
KR101858411B1 (ko) * 2017-11-20 2018-05-15 재단법인 구미전자정보기술원 생체 신호를 이용한 조산 예측 방법 및 장치
KR101890072B1 (ko) * 2017-11-28 2018-08-21 주식회사 멕 아이씨에스 자궁 근전도 신호를 기반으로 임산부의 조산예측기능을 갖는 임산부 모니터링 시스템

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113133771A (zh) * 2021-03-18 2021-07-20 浙江工业大学 基于时频域熵特征的子宫肌电信号分析及早产预测方法
CN113133771B (zh) * 2021-03-18 2022-10-28 浙江工业大学 基于时频域熵特征的子宫肌电信号分析及早产预测方法

Also Published As

Publication number Publication date
KR20200062627A (ko) 2020-06-04
KR102222282B1 (ko) 2021-03-04

Similar Documents

Publication Publication Date Title
WO2020111464A1 (fr) Procédé d&#39;identification de risque de naissance prématurée sur la base d&#39;électromyogramme utérin
US10987047B2 (en) System and method for analyzing progress of labor and preterm labor
Jager et al. Characterization and automatic classification of preterm and term uterine records
US8892181B2 (en) Non-invasive fetal monitoring
US7949389B2 (en) Fetal ECG monitoring
Sameshima et al. Unselected low-risk pregnancies and the effect of continuous intrapartum fetal heart rate monitoring on umbilical blood gases and cerebral palsy
US20130102856A1 (en) Non-invasive detection of fetal or maternal illness
Nitzschke et al. Single-channel amplitude integrated EEG recording for the identification of epileptic seizures by nonexpert physicians in the adult acute care setting
JP2003517865A (ja) 人類や哺乳類の陣痛（ｌａｂｏｒ）予測装置
Hao et al. Application of decision tree in determining the importance of surface electrohysterography signal characteristics for recognizing uterine contractions
Mas-Cabo et al. Uterine contractile efficiency indexes for labor prediction: A bivariate approach from multichannel electrohysterographic records
WO2014021515A1 (fr) Procédé et appareil pour l&#39;évaluation de la santé de fœtus
WO2018092986A2 (fr) Procédé de diagnostic du système nerveux autonome fœtal et de l&#39;état de l&#39;activité cardiaque
WO2012057406A1 (fr) Dispositif et procédé non invasifs de mesure de la motilité intestinale à l&#39;aide d&#39;une analyse des sons émis par les intestins
Vairavan et al. A computer-aided approach to detect the fetal behavioral states using multi-sensor Magnetocardiographic recordings
Huber et al. Potential use of electrohysterography in obstetrics: a review article
WO2023136588A1 (fr) Procédé de fourniture d&#39;informations auxiliaires sur la dysphagie à l&#39;aide d&#39;une analyse vocale
Fergus et al. Prediction of intrapartum hypoxia from cardiotocography data using machine learning
Tian et al. A Comprehensive Evaluation of the Predictive Abilities of Fetal Electrocardiogram‐Derived Parameters during Labor in Newborn Acidemia: Our Institutional Experience
KR101958863B1 (ko) 착용형 자궁 근전도 센서를 이용한 조산 예측 시스템 및 그 방법
WO2018236195A1 (fr) Procédé de traitement d&#39;image ultrasonore
Abbasi et al. 2D Wavelet-scalogram deep-learning for seizures pattern identification in the post-hypoxic-ischemic EEG of preterm fetal sheep
WO2018056777A2 (fr) Dispositif d&#39;évaluation du bien-être fœtal et son procédé de commande
WO2023022521A9 (fr) Système de lecture d&#39;électrocardiogramme dans lequel un modèle reposant sur l&#39;apprentissage profond et un modèle reposant sur des règles sont intégrés
Odendaal et al. Association between Increased Uterine Activity, as Recorded Noninvasively from the Anterior Abdominal Wall at 34 Weeks’ Gestation, and Preterm Birth

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: 19889716

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19889716

Country of ref document: EP

Kind code of ref document: A1