WO2014021515A1

WO2014021515A1 - Method and apparatus for evaluating health of fetus

Info

Publication number: WO2014021515A1
Application number: PCT/KR2012/010672
Authority: WO
Inventors: 박인양
Original assignee: 가톨릭대학교 산학협력단
Priority date: 2012-07-30
Filing date: 2012-12-11
Publication date: 2014-02-06
Also published as: KR101554381B1; KR20140016024A

Abstract

The present invention relates to a method for evaluating the health of a fetus, comprising the steps of: detecting a biosignal from the abdomen of a mother; amplifying the detected biosignal and removing noise to output the same; extracting electrocardiogram data of a fetus from the output signal and converting the extracted electrocardiogram data of a fetus into a digital signal; and acquiring the state of health of a fetus on the basis of entropy information obtained by segmentation, by frames, of the heart rate variability (HRV) signal of a fetus read from the converted signal, and nonlinearly analyzing, by the preset period, the HRV signal by frame.

Description

Prenatal Health Assessment Methods and Devices

The present invention relates to fetal health evaluation by using the fetal ECG data extracted from the mother's abdomen and using the entropy value obtained through nonlinear analysis thereof.

In general, the monitoring of the fetal condition is essential for mothers associated with gestational hypertension, gestational diabetes mellitus, and low birth weight infants, and it is necessary to continuously examine the condition of the fetus from prenatal to analgesia.

At present, almost all mothers are given a baby test to assess the health of the fetus before giving birth, and the health of the baby during delivery is also dependent on this test.

However, these tests are difficult to objectify the numerical values by drawing on the paper the heart rate of the fetus at the moment obtained from the testing device, and accordingly, the interpretation of the results varies depending on the examiner or interpreter. . For example, in adults, when an ECG test is performed to evaluate cardiac function, a computer automatically analyzes it and displays a first opinion, and a doctor checks whether there is an additional matter. However, in the case of fetal heartbeat analysis, since the primary analysis by the computer cannot be performed, there is a problem that the accuracy of the test is lowered by displaying the result as an analog picture instead of being digitized.

The present invention provides a method for evaluating fetal health, which is capable of estimating the fetal ECG signal more accurately than the analog method of printing fetal heart rate on paper, and effectively analyzing unwanted noise or improving signal-to-noise ratio. To provide.

Fetal health evaluation method according to an embodiment of the present invention detects the bio-signal from the mother's abdomen, amplifies the detected bio-signal and removes the noise and outputs the ECG data of the fetus from the output signal Extracting and converting the extracted ECG data of the fetus into a digital signal, subdividing the fetal heart rate variability (HRV) signal read out from the converted signal by frame, and fetal heart rate variation signal by frame It includes a process of recognizing the health state of the fetus based on the entropy information (entropy information) obtained by non-linear analysis for each predetermined cycle.

According to another aspect of the present invention, in the fetal health evaluation device, the electrode unit for detecting the bio-signal from the mother abdomen, the ECG data of the fetus from the detected bio-signal separated from the maternal abdominal bio-signal extracted and extracted A biosignal detection unit that reads a fetal heart rate variability (HRV) signal from electrocardiogram data, a frame generation for subdividing the read fetal heart rate variance signal by frame, and spectral normalization of the HRV signal for each frame An analysis unit configured to calculate an entropy value for each frame of the HRV signal normalized by the analysis unit, and a non-linear analysis of the fetal heart rate variability (HRV) signal for each frame by a predetermined period. Health status of the fetus based on the above non-linear analysis results And a controller to recognize.

The present invention can not only estimate the fetal ECG signal more accurately than the conventional method of printing the fetal heart rate on paper, but also effectively reduce unwanted noise or improve the signal-to-noise ratio.

1 is a whole flow diagram of the fetal health evaluation method according to an embodiment of the present invention.

Figure 2 is a flow chart for performing non-linear analysis in the fetal health evaluation method according to an embodiment of the present invention.

Figure 3 (a) is a raw signal (raw signal) for the heartbeat variance signal of the fetus, (b) is a graph showing the signal interpolated through the Fast Fourier Transform (FFT) from the raw signal.

FIG. 4 (a) shows data index signal intervals which are preset and extracted from data indexes 11 to 15 corresponding to signal loss (5) at a size 65 to 80 bpm (bit per minute) of the local region in the fetal heartbeat data region. (B) is a graph which interpolated the said (a).

5 is a cutoff frequency versus Shannon entropy graph quantified by Shannon entropy according to the data index K of the cutoff frequency for the fetal heartbeat variance signal according to the change of the cutoff frequency according to an embodiment of the present invention.

6 is a block diagram of a fetal health evaluation apparatus according to an embodiment of the present invention.

7 is a graph comparing the results obtained by another nonlinear method (SampEn; Sample Entropy) with the results obtained by the ApEn calculation method in order to confirm the ApEn calculation method.

8 is a table summarizing the maternal and fetal characteristics according to the delivery method.

Fetal health evaluation apparatus according to an embodiment of the present invention is an electrode unit for detecting a bio-signal from the mother abdomen, the ECG data of the fetus from the detected bio-signal separated from the maternal abdominal bio-signal extracted and the extracted ECG of the fetus A biosignal detection unit for reading a fetal heart rate variability (HRV) signal from the data, a frame generation for subdividing the read fetal heart rate variance signal by frame, and an analysis for performing spectral normalization of the HRV signal for each frame And an entropy calculating unit calculating a frame-specific entropy value for the HRV signal normalized by the analyzer, and controlling the frame rate fetal heart rate variability (HRV) signal to be linearly analyzed for each predetermined period. To recognize the health of the fetus based on the results of nonlinear analysis It includes a control unit.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, specific details such as specific components are shown, which are provided to help a more general understanding of the present invention, and the specific details may be changed or changed within the scope of the present invention. It is self-evident to those of ordinary knowledge in Esau.

1 is an overall flowchart of a fetal health evaluation method according to an embodiment of the present invention.

Referring to FIG. 1, first, a biosignal is detected from a mother's abdomen in step 110. In step 112, the detected biosignal is amplified by an amplifying unit and a filter unit, and the noise is removed and output.

In this case, the biosignal detection is performed through at least two electrodes attached to the mother's abdomen.

In operation 114, the ECG data of the fetus is extracted from the output signal, and the extracted ECG data of the fetus is converted into an analog time series data (time series date) through an A / D converter in step 116.

In operation 118, the fetal heartbeat variation signal read from the converted signal is subdivided into frames, and in step 120, the non-linear analysis of the fetal heartbeat variation signal for each frame is performed at predetermined intervals. The heartbeat transition signal is obtained from a signal measured by an M-mode ultrasonic Doppler method.

Referring to FIG. 3, FIG. 3A is a raw signal of a fetal heartbeat variance signal, and (B) is interpolated from the original signal through a Fast Fourier Transform (FFT). By using the FFT, nonlinear analysis of the fetal heartbeat variability signal for each frame is performed at predetermined intervals.

At this time, the non-linear analysis will be described in more detail with reference to Figure 2, Figure 2 is a flow chart for performing the non-linear analysis in the fetal health evaluation method according to an embodiment of the present invention, in step 210 to the fetal heartbeat variance signal From the time series signal, information about a predetermined data index signal interval is extracted from the remaining sections except for the zero subsection in step 212 except for a zero subsection in the data index.

Referring to FIG. 4, the operation process of 210 and 212 is shown in FIG. 4A. In FIG. 4A, signal loss (5) is performed at a size of 65 to 80 bpm (local per region) in the fetal heartbeat data region of FIG. 3. ) Shows a data index signal interval that is preset and extracted from data indexes 11 to 15 corresponding to), and (b) of FIG. 4 is interpolated from (a).

Thereafter, in step 214, spectral normalization is performed on the extracted information, and in step 216, an entropy value for each frame of the heartbeat variance signal is calculated from the normalization distribution.

In this case, performing spectral normalization in step 214, fast Fourier transform the fetal heartbeat variance signal, obtain a spectrum from the fast Fourier transform, and then determine a cutoff frequency of the spectrum. The cutoff frequency determination of the spectrum is performed by quantifying a distribution of a predetermined data index signal interval difference signal by using Shannon entropy (COF-SE), where the slope of the trajectory of the Shannon entropy exceeds a preset threshold. The frequency is determined as the cutoff frequency for the high frequency noise of the heartbeat transition signal.

5 is quantified by Shannon entropy according to the data index K of the cutoff frequency for the fetal heartbeat variance signal according to the change of the cutoff frequency according to an embodiment of the present invention and expressed as a cutoff frequency vs. Shannon entropy graph. Is a signal containing linear noise, and (b) shows a linearized signal digitized and converted, and it can be seen that the point where the slope changes rapidly is determined as an accurate cutoff frequency.

Returning to the description of FIG. 1, in step 122, entropy information of the fetal heartbeat signal for each frame is obtained by performing the operation of step 120.

In operation 124, the health status of the fetus is evaluated based on the obtained entropy information.

Meanwhile, as the nonlinear analysis method used in the present invention, fetal heartbeat variation is evaluated by using approximate entropy.

This is accompanied by an understanding of the normal entropy index pattern for the fetal heart rate variability, so that the present invention is applied and not applied, that is, fetal heart rate for use of the entropy index for fetal monitoring before and during fetal delivery To help understand the normal entropy index pattern for variation, experiments are conducted to compare the fetal heart rate entropy index with the progression of delivery.

Fetal heart rate variation FHR The normal entropy index pattern

To this end, each of the labors was delivered using the FHR records immediately before delivery of the fetus in three different ways of delivery: vaginal delivery in the second analgesia, planned cesarean delivery before analgesia, and emergency cesarean delivery during the first analgesia. The fetal heart rate variability entropy index at different stages was compared.

Thus, the results of this experiment are expected to understand the characteristics of the cortical nervous controlling system of FHR during labor and to improve the clinical application of the entropy index for fetal monitoring during labor.

Signal Acquisition and Preprocessing

Corometrics 150 (Corometrics, CT, USA), Doppler ultrasound cardiotocography and autocorrelation function were used for FHR signal acquisition. The pulse repetition frequency was 2 Hz, the pulse duration was 92 Hz, and the heart rate counting range was 50-210 beats per minute. All records were saved to a connected PC and further offline analysis was performed. We dataate FHR records from the study group's final 30 minutes before delivery through the Catholic computer-based maternity and care system (CCAOD; DoBe Tech, Seoul, Korea).

FHR signals measured during the last few minutes of delivery may be lost or contaminated. Known preprocessing algorithms were used for the preprocessing of the signals. We analyzed and filtered the difference between heartbeats below 60 bpm (beats per minute) and beat-to-beats above 25 bpm. Spline interpolation was used to replace the filtered heartbeat during signal loss periods of 2 seconds or less. For longer periods, we replaced the most recent segment of the same length with no signal loss. Finally, all heart rate data in bpm were converted into RR intervals for entropy calculations. Heart rate (bpm) can be expressed in 60 / RR interval (seconds).

Approximate Entropy Analysis

Nonlinear assays for FHR variability have been important in fetal cardiovascular and endocrine function. In particular, FHR variability was examined using Approximate Entropy, a mathematical nonlinear indicator. ApEn quantifies the complexity of FHR variability. Lower ApEn values indicate lower complexity, while higher ApEn values indicate higher complexity. Summarize the algorithm as follows:

Size time series (ie, the number of RRs in the input sequence) N,

Choose the embedding dimension (m) (sometimes referred to as pattern length in the literature) and construct a new vector sequence.

here,

The size of the new time series (

) Is N-m + 1, and the size of each vector is m. Also select the threshold distance or the comparison length r. Two vectors of size m

And

) Has a distance less than r:

In this case the vectors are considered similar. Vector of distance (r), size (m)

) Is a vector of the same size (

) Is the same as

It is defined as follows.

Approximate entropy is defined as

As described in the above algorithm, three parameters were used to calculate ApEn values: embedding dimension (m), comparison length (r), and number of RRs in the input sequence (N). One of the disadvantages of ApEn is the fact that ApEn depends on the input sequence. It has also been suggested that 1000 input sequence values are sufficient for statistical significance, where m = 2 and r ranges from 0.1 to 0.2 STD in the input sequence. The number of RR intervals as the input sequence in this study is 2000. The embedding dimension (m) was set to 2 through experiments, and the comparison length (r) was calculated as 15% of the standard deviation (STD) of the RR interval for each data.

In order to confirm the ApEn calculation method, as shown in FIG. 7, ApEn results were compared with results obtained using another nonlinear method, SampEn (Sample Entropy). The SampEn algorithm is described in detail in other documents. As mentioned above, in the ApEn calculation, a preprocessing algorithm based on previous studies was applied to the FHR signal, accompanied by conversion to RR intervals. Examination of the last 2000 consecutive RR intervals (16.7 minutes) and the first 600 consecutive RR intervals (5 minutes) was performed in the FHR variability analysis. The calculations were all done with MATLAB 7.10 (R2010a, The Mathworks, Natick, MA).

결과result

8 summarizes the maternal and fetal characteristics according to the delivery method. The mean (± standard deviation (STD)) maternal age of the asphyxia delivery group, emergency cesarean section group, and selective cesarean section group was 31 (± 3), 32 (± 3) and 33 (± 4) years, respectively. Significant differences were observed in maternal age between these three groups ( P = 0.013), and the mean maternal age of the asphyxia delivery group was younger than the other two groups. There were 108 boys and 154 girls, and no significant gender differences were observed. No significant differences were detected in the 1- and 5-minute Apgar scores, mean umbilical artery pH, and hyperbase at delivery.

In the case of FIG. 8, the linear index including the mean FHR, STD, coefficient of variation (CV), and variance index (DI) was found to be significantly different for each delivery method in all time segments. In comparison of the mean and STD of FIG. 8, the RR interval of the asphyxia delivery group was significantly longer and varied than the RR interval of other groups. This result is consistent with the results of other linear measurements such as CV and DI. Pairwise comparisons show that the linear indices of the RR intervals in the vaginal delivery group, the emergency cesarean section group, and the choking delivery group and the optional cesarean section group are significantly different in all time segments (except for the average index of the first five minutes). It is present. In the mean index of the RR intervals, it was found that there was a significant difference in the first 5 minutes only between the asphyxia and selective cesarean sections. In the choking delivery group, the linear index was significantly larger in the last 5 minutes of the 2000 RR interval than in the first 5 minutes of the 2000 RR interval. From this, it can be seen that the RR interval of the last 5 minutes of the choking delivery group is better distributed than the initial RR interval.

Nonlinear exponents including ApEn and SampEn in different time segments of different delivery modes are shown in Table 3. The average ApEn of the vaginal delivery group was the lowest, followed by the emergency cesarean section and selective cesarean section. This was true for all time segments ( P = 0.0001, P = 0.0366, and P = 0.0024. 2000 RR intervals, first 5 minutes and last 5 minutes, respectively).

In addition, in all time segments, the mean SampEn in the choking delivery group showed the lowest value, followed by the emergency cesarean section and the optional cesarean group ( P <0.0001, P = 0.0264, and P = 0.0011. 2000 FHR, first 5 minutes, and last 5 minutes respectively). In addition, the pairwise comparison (after applying the Bonferroni correction) showed that there was a significant difference in mean ApEn and SampEn between the vaginal delivery group and the selective cesarean section group in all time segments. At the 2000 RR interval, significant differences were also detected between the mean ApEn and SampEn intervals between the choking delivery group and the emergency cesarean section.

In the vaginal delivery group, there was a significant difference in mean ApEn and SampEn values at the first and last 5 minutes at the 2000 RR interval. At the last 5 minutes, mean ApEn (0.49 vs. 0.44, p = 0.0007) and mean SampEn (0.34 vs. 0.29, p <0.0001) were significantly lower than the first 5 minutes of the asphyxia delivery group. Differences between the different time segments were not significantly detected between the selective cesarean section and the emergency cesarean section.

7 to 8 show mean ApEn and mean SampEN according to the first 5 minutes, the second 5 minutes and the last 5 minutes of the 2000 RR interval of the asphyxia delivery group. In the vaginal delivery group, mean ApEn and mean SampEN gradually decreased with the progress of delivery.

In conclusion, the FHR entropy index of the second analgesic phase was significantly lower than that of the first analgesic and analgesic phase. In addition, among the last 16 minutes before vaginal delivery, the last 5 minutes of fetus of analgesia were much lower in FHR entropy index than the first 5 minutes of analgesia. These results suggest that the physiological response of the fetus in the high stress environment of analgesia can be understood, and that different judgments should be given according to the analgesia in performing fetal monitoring during delivery using the FHR entropy index.

In the above, the fetal health evaluation method according to an embodiment of the present invention has been described.

Hereinafter, a fetal health evaluation apparatus according to an embodiment of the present invention will be described in detail with reference to FIG. 6.

6 is a block diagram of the fetal health evaluation apparatus according to an embodiment of the present invention.

Referring to FIG. 6, the apparatus 600 to which the present invention is applied includes an electrode unit 610, a signal detector 612, an analyzer 614, a controller 616, an entropy calculator 618, and an A / D converter ( 620 and the frame generator 622.

The electrode unit 610 detects a biosignal from a maternal abdomen.

The biosignal detection unit 612 amplifies the biosignal detected from the electrode unit 610 and removes noise.

The control unit 616 extracts the fetal ECG data from the maternal abdominal biosignal in the signal output from the signal detector 612 and converts the fetal ECG data into a digital signal, thereby reading the fetal heartbeat variation read from the converted signal (Heart). Rate Variability (HRV) signal is controlled to be non-linear analysis and output to recognize the fetal health state based on the obtained entropy information.

In addition, the controller 616 extracts information of a predetermined data index signal section from the remaining sections except for a section in which a data index is less than zero, from the time series signal for the fetal heartbeat variance signal, and extracts the information. Spectral normalization is performed on the received information, and the control is performed to calculate an entropy value for each frame of the heartbeat variance signal from the normalization distribution, thereby performing nonlinear analysis.

In this case, the cutoff frequency of the spectrum may be determined by quantifying the distribution of a predetermined data index signal interval difference signal by using Shannon entropy (COF-SE) to determine a frequency at a point where the slope of the Shannon entropy exceeds a preset threshold. The cutoff frequency for the high frequency noise of the heartbeat transition signal is determined.

Subsequently, the A / D converter 620 converts the ECG data of the fetus extracted under the control of the controller 616 into a digital signal.

The frame generator 622 subdivides the HRV signal read out from the converted signal for each frame.

The analyzer 614 performs spectral normalization on the HRV signal for each frame.

The entropy calculator 618 calculates an entropy value for each frame of the HRV signal normalized from the analyzer 614.

As described above, the operation of the fetal health evaluation method and apparatus according to the present invention can be made. Meanwhile, in the above description of the present invention, specific embodiments have been described, but various modifications can be made without departing from the scope of the present invention. have. Therefore, the scope of the present invention should not be defined by the described embodiments, but by the claims and equivalents of the claims.

The present invention not only can estimate the fetal ECG signal more accurately, but also can perform nonlinear analysis according to the rapid change of the input signal using the entropy change information, thereby effectively reducing the unwanted noise or improving the signal-to-noise ratio. To make it possible.

Claims

In the fetal health evaluation method,

Detecting a biological signal from the mother's abdomen,

Extracting ECG data of the fetus from the detected signal;

Reading a fetal heart rate variability (HRV) signal from the extracted ECG data of the fetus,

Subdividing the read fetal heartbeat variance signal by frame and nonlinear analysis of the fetal heartbeat variance signal by frame at predetermined intervals;

Fetal health evaluation method comprising the step of recognizing the health status of the fetus based on the non-linear analysis results
The method of claim 1, wherein the heartbeat variance signal is obtained from a signal measured by a motion-mode ultrasonic Doppler method.
The method of claim 1, wherein the non-linear analysis,

Extracting information of a predetermined data index signal section from a section except for a section in which a data index is less than zero from a time series signal for the fetal heartbeat variance signal;

Performing spectral normalization on the extracted information;

And calculating the frame-specific entropy value of the heartbeat variance signal from the normalized distribution.
The method of claim 2, wherein performing normalization comprises:

Fast Fourier Transform (Fast Fourier Transform) the fetal heartbeat variance signal, and after obtaining a spectrum from the Fast Fourier transform results, fetal health evaluation method characterized in that it is made by determining the cutoff frequency of the spectrum.
The method of claim 3, wherein the predetermined data index signal interval is from data indexes 11 to 15.
The method of claim 4, wherein

Quantifying the distribution of the difference signal interval of a predetermined data index signal by Shannon entropy (COF-SE) to cut off the frequency at the point where the slope of the Shannon entropy exceeds a preset threshold for the high frequency noise of the heartbeat variance signal. Fetal health assessment method characterized in that it is determined by the frequency.
In the fetal health evaluation device,

An electrode unit for detecting a biological signal from the mother's abdomen,

A biosignal detection unit extracting fetal ECG data from the detected biosignal from a maternal abdominal biosignal and reading fetal heart rate variability (HRV) signals from the extracted fetal ECG data;

A frame generator for subdividing the read fetal heartbeat variance signal for each frame;

An analyzer for performing spectral normalization on the frame-specific HRV signal;

An entropy calculator configured to calculate an entropy value for each frame of the HRV signal normalized from the analyzer;

And a control unit configured to non-linearly analyze the fetal heart rate variability (HRV) signal for each frame at predetermined intervals and recognize a fetal health state based on the non-linear analysis result. .
The method of claim 7, wherein the control unit,

From the time series signal for the fetal heartbeat variance signal, information of a predetermined data index signal section is extracted from the remaining sections except for a section in which a data index is less than zero, and spectral normalization is performed on the extracted information. And calculating a frame-by-frame entropy value of the heartbeat variance signal from the normalized distribution to perform nonlinear analysis.
The method of claim 8, wherein performing spectral normalization,

And fast Fourier transform the fetal heartbeat variance signal, obtain a spectrum from the fast Fourier transform, and determine the cutoff frequency of the spectrum.
The method of claim 9,

Quantify the distribution of the data signal difference signal section by using Shannon entropy (COF-SE) to cut off the frequency at which the slope of the Shannon entropy exceeds a preset threshold for the high frequency noise of the heartbeat variance signal. Fetal health evaluation device, characterized in that determined by frequency.