CN114795253A - A method and system for monitoring uterine contractions based on multi-dimensional information fusion - Google Patents

Abstract

The invention discloses a uterine contraction monitoring method and system based on multi-dimensional information fusion, on one hand, a six-axis sensor is utilized to analyze the motion state of the abdomen of a pregnant woman in the uterine contraction process and draw a uterine contraction motion curve, can avoid the gravity acceleration interference caused by the actions of the pregnant women and the interference of different postures on the extraction of the uterine electromyographic signal characteristics, the uterine contraction movement can be detected in a dynamic environment, the dimension of monitoring uterine contraction is increased, the use scene is effectively expanded, on the other hand, the MECG signal separation is firstly carried out on the mixed bioelectric signal, and then, EHG characteristic signal segment marking is carried out, EHG signals are extracted, interference of MECG signals and interference caused by motion artifacts of sensors can be effectively inhibited, and the technical problems that the existing uterine contraction motion monitoring method is difficult to inhibit noise interference of electrical signals of uterine muscles and can effectively monitor uterine contraction motion in static environment and dynamic environment are solved.

Description

A method and system for monitoring uterine contractions based on multi-dimensional information fusion

technical field

The invention relates to the technical field of uterine contraction monitoring, in particular to a uterine contraction monitoring method and system based on multi-dimensional information fusion.

Background technique

Uterine contractions are the regular contractions of the uterus, which are an important monitoring index in the clinical obstetric examination of pregnant women. Electrohysterogram (EHG) during pregnancy can be collected through electrodes placed on the abdomen of pregnant women, but this signal will be disturbed by noises such as Maternal Eletrocardiogram (MECG) and sensor motion artifacts. Influence, in the process of dynamic monitoring, at the same time, it will be interfered by the electromyographic signal generated by the movements of pregnant women, so it is difficult to effectively monitor the uterine contractions.

The existing method for monitoring uterine contractions is to use a triaxial acceleration sensor with an RMS algorithm for monitoring. The RMS algorithm is one of the most commonly used methods for extracting uterine EMG signals. Its principle is to reflect the energy change trend of the burst and non-burst EHG signals of the entire EHG signal by calculating the standard deviation of the uterine EMG. Although this method can obtain a relatively smooth EHG waveform, which is helpful for the identification of uterine contraction waves, it is easily interfered by the motion artifact of the sensor, which affects the monitoring effect of uterine contraction movement, and the three-axis acceleration sensor cannot be obtained under dynamic conditions. Body position signal information, it is difficult to solve the problem of interference by the electromyographic signals generated by the movements of pregnant women.

In the current uterine contraction movement monitoring technology, how to suppress the noise interference of the uterine EMG signal and effectively monitor the uterine contraction movement in a static environment and a dynamic environment is still a technical problem to be solved urgently by those skilled in the art.

SUMMARY OF THE INVENTION

The invention provides a uterine contraction monitoring method and system based on multi-dimensional information fusion, which is used to solve the problem that the existing uterine contraction movement monitoring method is difficult to suppress the noise interference of the uterine EMG signal and can be used in both static and dynamic environments. Technical aspects of effective monitoring of uterine contractions.

In view of this, a first aspect of the present invention provides a method for monitoring uterine contractions based on multi-dimensional information fusion, including:

The information fusion of the three-axis acceleration signal and the three-axis angular velocity signal of the pregnant woman's abdomen obtained by the six-axis sensor is performed to obtain the three-axis linear acceleration signal;

Mark the uterine contraction characteristic segment of the three-axis linear acceleration signal, and determine the set of uterine contraction time points;

Construct the contraction motion curve according to the triaxial linear acceleration signal and contraction time point set;

Perform MECG signal separation on the obtained mixed bioelectrical signals of the abdomen of pregnant women to obtain intermediate state signals;

Mark the EHG characteristic signal segment for the intermediate state signal, and draw the EHG curve according to the marked EHG characteristic signal segment;

Curve fitting was performed on the contraction motion curve and the EHG curve to generate the target contraction curve.

Optionally, the target contraction curve is:

Among them, C _TOCO is the target contraction curve, C _EHG is the EHG curve, V is the column vector representing the intensity of the EHG signal, C _ACC is the contraction motion curve, and δ is the displacement change of the abdominal surface skin relative to the trunk in the resting state during contractions the maximum value of .

Optionally, marking the uterine contraction characteristic segment of the three-axis linear acceleration signal to determine the uterine contraction time point, including:

The signal segment that satisfies the preset constraint conditions in the three-axis linear acceleration signal is marked as the uterine contraction feature segment;

Among them, the preset constraints are:

的时间点集合，m为T中的元素个数，v_min为最小宫缩速度，t_min为最短宫缩时间，t_max为最长宫缩时间，

为三轴线性加速度，

为宫缩强度，δ为宫缩时静止状态下腹部表面皮肤相对于躯干位移变化的最大值；Among them, t∈T, T is satisfying

set of time points, m is the number of elements in T, v _min is the minimum contraction speed, t _min is the shortest contraction time, t _max is the longest contraction time,

is the three-axis linear acceleration,

is the contraction intensity, and δ is the maximum displacement change of the abdominal surface skin relative to the trunk in the resting state during contractions;

Obtain the contraction time point set corresponding to the contraction feature segment.

Optionally, the contraction motion curve is:

Among them, C _ACC is the contraction motion curve, T _α is the contraction time point set composed of uterine contraction time points, and T _α ∈T.

Optionally, the minimum uterine contraction speed is 0.1 m/s, the maximum displacement change of the abdominal surface skin relative to the trunk in a resting state during uterine contractions is 0.04 m, the shortest uterine contraction time is 20s, and the longest uterine contraction time is 180s.

Optionally, MECG signal separation is performed on the obtained mixed bioelectrical signals of the abdomen of pregnant women to obtain intermediate state signals, including:

Obtain the mixed bioelectric signal S of the abdomen of the pregnant woman;

The R-wave site of MECG in the mixed bioelectrical signal S was detected by the Pan & Tompkins algorithm;

Mark the R wave site and the corresponding signal intensity to generate a single QRS waveform;

The mixed bioelectric signal S and the generated single QRS waveform are linearly subtracted over the entire time period to obtain the intermediate state signal S _f .

Optionally, mark R-wave sites and corresponding signal intensities to generate a single QRS waveform, including:

mark the R wave site;

Taking the R wave point as the center, offset the preset number of sampling points to the left and the preset number of sampling points to the right to determine the QRS wave sampling point, and construct the QRS wave sampling point according to the signal intensity corresponding to the QRS wave sampling point and the QRS wave sampling point. QRS matrix;

Perform singular value decomposition on the QRS matrix to obtain the QRS wave template matrix;

Extract the column vector of the QRS template matrix to obtain a single QRS waveform.

Optionally, mark the EHG characteristic signal segment on the intermediate state signal, and draw an EHG curve according to the marked EHG characteristic signal segment, including:

Use the Hanning window of preset length to perform short-time Fourier transform on the intermediate state signal S _f to obtain a complex signal matrix, and the column vector in the complex signal matrix is the signal strength of each frequency;

Extract the element set C _f at the EHG eigenfrequency from the complex signal matrix;

The eigenfrequency matrix W is obtained by subtracting the element set C _f at the EHG eigenfrequency and the corresponding EHG frequency eigenvalue threshold;

n为采样点总数；The elements of each row in the eigenfrequency matrix W are summed to obtain a column vector V representing the strength of the EHG signal,

n is the total number of sampling points;

Take the set of subscripts whose elements in the column vector V are greater than 0 and whose continuous time is greater than t _min and less than t _max as I, where I is the set of sampling points marked as the EHG feature signal segment;

Plot the EHG curve, the EHG curve is:

Among them, the value of i is [1,n].

Optionally, 2 or 3 or 4 or 5 EHG eigenfrequencies are selected.

A second aspect of the present invention provides a uterine contraction monitoring system based on multi-dimensional information fusion, including:

The linear acceleration acquisition module is used for information fusion of the three-axis acceleration signal and the three-axis angular velocity signal of the pregnant woman's abdomen obtained by the six-axis sensor to obtain the three-axis linear acceleration signal;

The uterine contraction feature marking module is used to mark the uterine contraction feature segment of the three-axis linear acceleration signal, and determine the set of uterine contraction time points;

The contraction movement curve module is used to construct the contraction movement curve according to the three-axis linear acceleration signal and the contraction time point set;

The MECG signal separation module is used to separate the MECG signal from the obtained mixed bioelectrical signal of the abdomen of the pregnant woman to obtain an intermediate state signal;

The EHG curve module is used to mark the EHG characteristic signal segment of the intermediate state signal, and draw the EHG curve according to the marked EHG characteristic signal segment;

The contraction curve module is used to perform curve fitting on the contraction motion curve and the EHG curve to generate the target contraction curve. As can be seen from the above technical solutions, the method and system for monitoring uterine contractions based on multi-dimensional information fusion provided by the present invention have the following advantages:

The method and system for monitoring uterine contractions based on multi-dimensional information fusion provided by the present invention, on the one hand, utilizes a six-axis sensor to analyze the motion state of the abdomen of the pregnant woman during the uterine contraction, and draws the motion curve of the uterine contraction, which can avoid the interference of gravitational acceleration due to the movement of the pregnant woman And the interference of different postures on the feature extraction of uterine EMG signals, uterine contractions can also be detected in a dynamic environment, which increases the dimension of monitoring uterine contractions and effectively expands the usage scenarios. On the other hand, MECG is first performed on mixed bioelectric signals Signal separation, and then marking the EHG characteristic signal segment to extract the EHG signal, which can effectively suppress the interference of the MECG signal and the interference caused by the motion artifact of the sensor, and solve the problem that the existing uterine contraction motion monitoring method is difficult to suppress the uterus. The noise interference of EMG signals and the technical problems of effective monitoring of uterine contractions in both static and dynamic environments.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings that are required to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only These are some embodiments of the present invention. For those of ordinary skill in the art, other related drawings can also be obtained according to these drawings without creative efforts.

1 is a schematic flowchart of a method for monitoring uterine contractions based on multi-dimensional information fusion provided in the present invention;

FIG. 2 is a schematic structural diagram of a uterine contraction monitoring system based on multi-dimensional information fusion provided in the present invention.

Detailed ways

In order to make those skilled in the art better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only These are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

For ease of understanding, please refer to FIG. 1, an embodiment of a method for monitoring uterine contractions based on multi-dimensional information fusion is provided in the present invention, including:

Step 101: Perform information fusion on the triaxial acceleration signal of the pregnant woman's abdomen obtained by the six-axis sensor and the triaxial angular velocity signal to obtain a triaxial linear acceleration signal.

其中，p、r、y分别对应俯仰角、翻滚角和偏航角。在此姿态下，重力加速度作用在加速度传感器上的分量为

g为重力加速度。将加速度传感器测得的三轴加速度信号减去

得到三轴线性加速度信号

即排除了重力加速度干扰的三轴加速度。It should be noted that, in the embodiment of the present application, the six-axis sensor includes an acceleration sensor and a gyro sensor, the three-axis acceleration signal is obtained through the acceleration sensor, and the three-axis angular velocity signal is obtained through the gyro sensor. After fusing the three-axis acceleration signal and the three-axis angular velocity signal, the Euler angle data is obtained as

Among them, p, r, y correspond to pitch angle, roll angle and yaw angle respectively. In this attitude, the component of gravitational acceleration acting on the acceleration sensor is

g is the acceleration of gravity. Subtract the triaxial acceleration signal measured by the accelerometer

Get the three-axis linear acceleration signal

That is, the three-axis acceleration that excludes the interference of gravitational acceleration.

Step 102 , marking the uterine contraction characteristic segment of the three-axis linear acceleration signal, and determining a set of uterine contraction time points.

It should be noted that the main characteristics of uterine contractions on the skin surface are: (1) the speed of uterine contractions is less than the minimum uterine contraction speed; (2) the displacement of the abdominal skin surface relative to the trunk when the pregnant woman is at rest is greater than 0 and less than when the uterine contractions are at rest. The maximum value of the displacement of the abdominal surface skin relative to the trunk in the state; (3) the duration of a single contraction is greater than the shortest contraction time and less than the longest contraction time. The exercise that satisfies the above three conditions is expressed as the contraction intensity, which is marked as the contraction characteristic segment.

Converted to mathematical expression, it is:

的时间点集合，m为T中的元素个数，v_min为最小宫缩速度，t_min为最短宫缩时间，t_max为最长宫缩时间，

为三轴线性加速度，

为宫缩强度，δ为宫缩时静止状态下腹部表面皮肤相对于躯干位移变化的最大值。Among them, t∈T, T is satisfying

set of time points, m is the number of elements in T, v _min is the minimum contraction speed, t _min is the shortest contraction time, t _max is the longest contraction time,

is the three-axis linear acceleration,

is the contraction intensity, and δ is the maximum displacement change of the abdominal surface skin relative to the trunk in the resting state during contractions.

In the embodiment of the present invention, the minimum uterine contraction speed is preferably 0.1m/s, the maximum value of the displacement change of the abdominal surface skin relative to the trunk in a static state during uterine contractions is preferably 0.04m, the shortest uterine contraction time is preferably 20s, and the longest uterine contraction time is preferably 180s . Algorithmically, the three-axis linear acceleration is expressed as:

Among them, x, y, and z represent the linear accelerations in the x-axis, y-axis, and z-axis directions, respectively.

The above three conditions can be transformed into:

在时间上的积分<0.1m/s；(1) Strength of contractions

Integral over time <0.1m/s;

在时间上从0开始二次积分回到0的时间长度>20s且<180s；(2) Intensity of uterine contractions

The time length for the second integration from 0 to return to 0 in time is >20s and <180s;

在时间上的二次积分小于4cm。(3) Strength of contractions

The quadratic integral over time is less than 4 cm.

The specific signal processing flow can be described as:

There exists a set T, t ∈ T, where t is time such that

Then there is T={t ₁ ,t ₂ ,...,t _l }, where l is the number of elements in T, and t _l is the lth time point.

There exists a set T _α , T _α ∈ T such that the elements in T _α satisfy:

Then, the elements in the set T _α ={t _1α ,t _2α ,...,t _kα } are the contraction time points, and k is the number of elements in T _α .

和由宫缩时间点构成的宫缩时间点集合T_α构造出来：Step 103 , construct a contraction motion curve according to the three-axis linear acceleration signal and the set of contraction time points. It should be noted that the contraction motion curve C _ACC can be based on the three-axis linear acceleration signal

And the contraction time point set T _α composed of contraction time points is constructed:

Step 104: Perform MECG signal separation on the acquired mixed bioelectrical signals of the abdomen of the pregnant woman to obtain an intermediate state signal.

It should be noted that the mixed bioelectrical signal S of the abdomen of the pregnant woman is obtained by collecting electrodes, and the MECG signal S _t is separated from the mixed bioelectrical signal S to obtain the mixed bioelectrical signal minus the MECG signal S _t , that is, the intermediate state. signal S _f .

Specifically, the Pan & Tompkins algorithm is used to detect the R-wave site of MECG in the mixed bioelectrical signal S, mark the R-wave site and the corresponding signal intensity, generate a single QRS waveform, and compare the mixed bioelectrical signal S with the generated single QRS The waveforms are linearly subtracted over the entire time period to obtain the intermediate state signal S _f . The signal processing process is:

(1) Mark the R wave. The mixed bioelectrical signal S is sampled using an analog-to-digital converter to generate a discrete time series signal, denoted as:

Among them, n is the total number of sampling points, the sampling frequency is f _s , and the unit is Hz.

Use Pan & Tompkins to generate vectors:

Among them, K is the number of R waves in the mixed bioelectric signal S, t ₁ to t _K are the time points of the R wave peaks, and K<n.

Let α be the left offset sampling point of the QRS wave centered on the R wave, β be the right offset sampling point of the QRS wave centered on the R wave, and the matrix I _A is the sampling point matrix of the QRS wave, and I _A is expressed as:

(2) Generate the QRS matrix. Generate the QRS matrix A _S from the vector _TR :

Let element a satisfy a _q =s _q if _q∈IA and q∈{1,2,...,n}.

Among them, the element a∈S, A _S is a matrix of size (α+β+1)×K, and a single column vector of A _S is the sampling signal of a single QRS wave.

α and β take empirical values respectively: α=0.25f _s , β=0.45f _s .

(3) Construct QRS template matrix. From A _S as a matrix of size (α+β+1)×K, UYV ^T ≈A _S can be found, and U is the left singular value vector of A _S , which is a matrix of size (α+β+1)×J , J is the number of singular value vectors, and the empirical value is 2. Y is the singular value matrix of A _S , the size is J × J, and V is the right singular value vector of A _S , which is a matrix of size K × J.

将A_t中的元素记作b，有：You can get:

Denote the element in A _t as b, we have:

The size of the matrix _At is (α+β+1)×K.

The column vector in the matrix A _t corresponds to the template generated by the QRS wave of the pregnant woman's electrocardiogram in the signal S, and the vector S _t can be constructed from this template:

(4) Subtract the signal S and the vector S _t to obtain S _f , which _is the electrical signal from which the electrocardiogram MECG of the pregnant woman has been removed, which can be defined as an intermediate state signal.

If the signal acquisition contains multiple channels, the same _TR and S of different channels can be used, and steps (1)-(4) can be repeated.

Step 105: Mark the EHG characteristic signal segment on the intermediate state signal, and draw an EHG curve according to the marked EHG characteristic signal segment.

It should be noted that the electrical signal S _f of the pregnant woman's ECG MECG is mainly mixed with interference caused by sensor motion artifacts, fetal ECG signal interference, uterine EMG signal, and internal noise of the sensor. The most simple and obvious feature of uterine EMG signal and other interference is the frequency feature. Therefore, multiple frequency characteristic points can be selected as reference points to mark the EHG characteristic signal segment.

Specifically, in this embodiment, the short-time Fourier transform is used to extract the signal segment whose frequency characteristic point is greater than the threshold, and mark it as the EHG characteristic signal segment. The signal processing flow is:

Use the Hanning window of preset length N to perform short-time Fourier transform on the intermediate state signal S _f to obtain the complex signal matrix and frequency column vector.

The complex signal matrix is denoted by C:

The frequency column vector is denoted F:

Among them, the frequency column vector F is the frequency vector corresponding to the complex signal C, E is the number of frequencies, which satisfies the Nyquist theorem, and n is the total number of points used.

Take out a preset number (preferably 2 or 3 or 4 or 5) signal segments whose frequency feature points are greater than the frequency feature threshold from the frequency column vector, and denote it as:

Set the frequency feature threshold:

Get the frequency feature threshold matrix:

Find the point in the matrix C corresponding to the frequency eigenvector F _f and take the modulus to get the set C _f :

The frequency characteristic value of the EHG characteristic signal segment is subtracted from the frequency characteristic threshold to obtain the characteristic frequency matrix W:

Summing the elements of each row in the eigenfrequency matrix W yields a column vector V representing the strength of the EHG signal:

Among them, the set of subscripts with elements greater than zero in V satisfying the consecutive number greater than 20f _s and less than 180f _s is I, and I is marked as the set of sampling points of the EHG feature signal segment.

Use the column vector V to draw the EHG curve, zeroing in the unlabeled regions. The EHG curve is obtained as:

Among them, the value of i is [1,n].

It should be noted that, in this embodiment of the present application, steps 101-step 103 may be performed simultaneously with steps 104-step 105, and there is no execution time sequence relationship between steps 101-step 103 and steps 104-step 105, that is, it can be performed in Steps 104 to 105 are executed simultaneously when steps 101 to 103 are executed, or steps 101 to 103 may be executed first and then steps 104 to 105 are executed.

Step 106: Perform curve fitting on the contraction motion curve and the EHG curve to generate a target contraction curve.

It should be noted that, after obtaining the uterine contraction motion curve in step 103 and obtaining the EHG curve in step 105, a linear superposition method is used to perform curve fitting on the uterine contraction motion curve and the EHG curve to generate the target uterine contraction curve:

Among them, C _TOCO is the target contraction curve, C _EHG is the EHG curve, V is the column vector representing the intensity of the EHG signal, C _ACC is the contraction motion curve, and δ is the displacement change of the abdominal surface skin relative to the trunk in the resting state during contractions The maximum value of , and the empirical value is 0.04.

In the method for monitoring uterine contractions based on multi-dimensional information fusion provided by the embodiments of the present invention, on the one hand, six-axis sensors are used to analyze the motion state of the abdomen of the pregnant woman during the uterine contraction, and the motion curve of the uterine contraction can be drawn, which can avoid the interference of gravitational acceleration due to the movement of the pregnant woman And the interference of different postures on the feature extraction of uterine EMG signals, uterine contractions can also be detected in a dynamic environment, which increases the dimension of monitoring uterine contractions and effectively expands the usage scenarios. On the other hand, MECG is first performed on mixed bioelectric signals Signal separation, and then marking the EHG characteristic signal segment to extract the EHG signal, which can effectively suppress the interference of the MECG signal and the interference caused by the motion artifact of the sensor, and solve the problem that the existing uterine contraction motion monitoring method is difficult to suppress the uterus. The noise interference of EMG signals and the technical problems of effective monitoring of uterine contractions in both static and dynamic environments.

The present invention also provides an embodiment of a uterine contraction monitoring system based on multi-dimensional information fusion, including:

The linear acceleration acquisition module is used for information fusion of the three-axis acceleration signal and the three-axis angular velocity signal of the pregnant woman's abdomen obtained by the six-axis sensor to obtain the three-axis linear acceleration signal;

The uterine contraction feature marking module is used to mark the uterine contraction feature segment of the three-axis linear acceleration signal, and determine the set of uterine contraction time points;

The contraction movement curve module is used to construct the contraction movement curve according to the three-axis linear acceleration signal and the contraction time point set;

The MECG signal separation module is used to separate the MECG signal from the obtained mixed bioelectrical signal of the abdomen of the pregnant woman to obtain an intermediate state signal;

The EHG curve module is used to mark the EHG characteristic signal segment of the intermediate state signal, and draw the EHG curve according to the marked EHG characteristic signal segment;

The contraction curve module is used to perform curve fitting on the contraction motion curve and the EHG curve to generate the target contraction curve.

The target contraction curve is:

Among them, C _TOCO is the target contraction curve, C _EHG is the EHG curve, V is the column vector representing the intensity of the EHG signal, C _ACC is the contraction motion curve, and δ is the displacement change of the abdominal surface skin relative to the trunk in the resting state during contractions the maximum value of .

The contraction signature module is specifically used to:

The signal segment that satisfies the preset constraint conditions in the three-axis linear acceleration signal is marked as the uterine contraction feature segment; wherein, the preset constraint conditions are:

的时间点集合，m为T中的元素个数，v_min为最小宫缩速度，t_min为最短宫缩时间，t_max为最长宫缩时间，

为三轴线性加速度，

为宫缩强度，δ为宫缩时静止状态下腹部表面皮肤相对于躯干位移变化的最大值；Among them, t∈T, T is satisfying

set of time points, m is the number of elements in T, v _min is the minimum contraction speed, t _min is the shortest contraction time, t _max is the longest contraction time,

is the three-axis linear acceleration,

is the contraction intensity, and δ is the maximum displacement change of the abdominal surface skin relative to the trunk in the resting state during contractions;

Obtain the contraction time point corresponding to the contraction feature segment.

The contraction motion curve is:

Among them, C _ACC is the contraction motion curve, T _α is the contraction time point set composed of uterine contraction time points, and T _α ∈T.

The minimum uterine contraction speed was 0.1m/s, the maximum displacement of the abdominal surface skin relative to the trunk was 0.04m in the resting state during uterine contractions, the shortest uterine contraction time was 20s, and the longest uterine contraction time was 180s.

Perform MECG signal separation on the obtained mixed bioelectrical signals of the abdomen of pregnant women to obtain intermediate state signals, including:

Obtain the mixed bioelectric signal S of the abdomen of the pregnant woman;

The R-wave site of MECG in the mixed bioelectrical signal S was detected by the Pan & Tompkins algorithm;

Mark the R wave site and the corresponding signal intensity to generate a single QRS waveform;

The mixed bioelectric signal S and the generated single QRS waveform are linearly subtracted over the entire time period to obtain the intermediate state signal S _f .

Mark the R-wave site and the corresponding signal intensity to generate a single QRS waveform, including:

mark the R wave site;

Taking the R wave point as the center, offset the preset number of sampling points to the left and the preset number of sampling points to the right to determine the QRS wave sampling point, and construct the QRS wave sampling point according to the signal intensity corresponding to the QRS wave sampling point and the QRS wave sampling point. QRS matrix;

Perform singular value decomposition on the QRS matrix to obtain the QRS wave template matrix;

Extract the column vector of the QRS template matrix to obtain a single QRS waveform.

Mark the EHG characteristic signal segment for the intermediate state signal, and draw the EHG curve according to the marked EHG characteristic signal segment, including:

Use the Hanning window of preset length to perform short-time Fourier transform on the intermediate state signal S _f , and obtain the complex signal matrix The column vector in the complex signal matrix is the signal intensity of each frequency;

Extract the element set C _f at the EHG eigenfrequency from the complex signal matrix;

The eigenfrequency matrix W is obtained by subtracting the element set C _f at the EHG eigenfrequency and the corresponding EHG frequency eigenvalue threshold;

n为采样点总数；The elements of each row in the eigenfrequency matrix W are summed to obtain a column vector V representing the strength of the EHG signal,

n is the total number of sampling points;

Take the set of subscripts whose elements in the column vector V are greater than 0 and whose continuous time is greater than t _min and less than t _max as I, where I is the set of sampling points marked as the EHG feature signal segment;

Plot the EHG curve, the EHG curve is:

Among them, the value of i is [1,n].

EHG eigenfrequencies take 2 or 3 or 4 or 5.

The uterine contraction monitoring system based on multi-dimensional information fusion provided by the embodiment of the present invention uses six-axis sensors to analyze the motion state of the abdomen of the pregnant woman during the uterine contraction, and draws the motion curve of the uterine contraction, which can avoid the interference of the acceleration of gravity caused by the movement of the pregnant woman And the interference of different postures on the feature extraction of uterine EMG signals, uterine contractions can also be detected in a dynamic environment, which increases the dimension of monitoring uterine contractions and effectively expands the usage scenarios. Signal separation, and then marking the EHG characteristic signal segment to extract the EHG signal, which can effectively suppress the interference of the MECG signal and the interference caused by the motion artifact of the sensor, and solve the problem that the existing uterine contraction motion monitoring method is difficult to suppress the uterus. The noise interference of EMG signals and the technical problems of effective monitoring of uterine contractions in both static and dynamic environments.

The uterine contraction monitoring system based on multi-dimensional information fusion provided by the embodiment of the present invention is used to execute the uterine contraction monitoring method based on multi-dimensional information fusion in the foregoing embodiment of the uterine contraction monitoring method based on multi-dimensional information fusion. The method for monitoring uterine contractions based on multi-dimensional information fusion in the embodiment of the method for monitoring uterine contractions based on multi-dimensional information fusion is the same, which will not be repeated here.

As mentioned above, the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand: The technical solutions described in the embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions in the embodiments of the present invention.

Claims (10)

1. a uterine contraction monitoring method based on multi-dimensional information fusion, is characterized in that, comprises: The information fusion of the three-axis acceleration signal and the three-axis angular velocity signal of the pregnant woman's abdomen obtained by the six-axis sensor is performed to obtain the three-axis linear acceleration signal; Mark the uterine contraction characteristic segment of the three-axis linear acceleration signal, and determine the set of uterine contraction time points; Construct the contraction motion curve according to the triaxial linear acceleration signal and contraction time point set; Perform MECG signal separation on the obtained mixed bioelectrical signals of the abdomen of pregnant women to obtain intermediate state signals; Mark the EHG characteristic signal segment for the intermediate state signal, and draw the EHG curve according to the marked EHG characteristic signal segment; Curve fitting was performed on the contraction motion curve and the EHG curve to generate the target contraction curve. 2. the uterine contraction monitoring method based on multi-dimensional information fusion according to claim 1, is characterized in that, target uterine contraction curve is:

Among them, C _TOCO is the target contraction curve, C _EHG is the EHG curve, V is the column vector representing the intensity of the EHG signal, C _ACC is the contraction motion curve, and δ is the displacement change of the abdominal surface skin relative to the trunk in the resting state during contractions the maximum value of . 3. The uterine contraction monitoring method based on multi-dimensional information fusion according to claim 1, wherein the uterine contraction feature segment of the three-axis linear acceleration signal is marked to determine the uterine contraction time point, comprising: The signal segment that satisfies the preset constraint conditions in the three-axis linear acceleration signal is marked as the uterine contraction feature segment; wherein, the preset constraint conditions are:

Among them, t∈T, T is satisfying

set of time points, m is the number of elements in T, v _min is the minimum contraction speed, t _min is the shortest contraction time, t _max is the longest contraction time,

is the three-axis linear acceleration,

is the contraction intensity, and δ is the maximum displacement change of the abdominal surface skin relative to the trunk in the resting state during contractions; Obtain the contraction time point set corresponding to the contraction feature segment. 4. the uterine contraction monitoring method based on multi-dimensional information fusion according to claim 3, is characterized in that, uterine contraction motion curve is:

Among them, C _ACC is the contraction motion curve, T _α is the contraction time point set composed of uterine contraction time points, and T _α ∈T. 5. The method for monitoring uterine contractions based on multi-dimensional information fusion according to claim 3, wherein the minimum uterine contraction velocity is 0.1 m/s, and the maximum displacement change of the abdominal surface skin relative to the trunk in a resting state during uterine contractions The value is 0.04m, the shortest contraction time is 20s, and the longest contraction time is 180s. 6. the uterine contraction monitoring method based on multi-dimensional information fusion according to claim 3, is characterized in that, carries out MECG signal separation to the mixed bioelectrical signal of the pregnant woman's abdomen that obtains, obtains intermediate state signal, comprises: Obtain the mixed bioelectric signal S of the abdomen of the pregnant woman; The R-wave site of MECG in the mixed bioelectrical signal S was detected by the Pan & Tompkins algorithm; Mark the R wave site and the corresponding signal intensity to generate a single QRS waveform; The mixed bioelectric signal S and the generated single QRS waveform are linearly subtracted over the entire time period to obtain the intermediate state signal S _f . 7. the uterine contraction monitoring method based on multi-dimensional information fusion according to claim 6, is characterized in that, mark R wave position and corresponding signal intensity, generate single QRS waveform, comprise: mark the R wave site; Taking the R wave point as the center, offset the preset number of sampling points to the left and the preset number of sampling points to the right to determine the QRS wave sampling point, and construct the QRS wave sampling point according to the signal intensity corresponding to the QRS wave sampling point and the QRS wave sampling point. QRS matrix; Perform singular value decomposition on the QRS matrix to obtain the QRS wave template matrix; Extract the column vector of the QRS template matrix to obtain a single QRS waveform. 8. the uterine contraction monitoring method based on multi-dimensional information fusion according to claim 6, is characterized in that, carry out EHG characteristic signal segment mark to intermediate state signal, and draw EHG curve according to the EHG characteristic signal segment of mark, comprising: Use the Hanning window of preset length to perform short-time Fourier transform on the intermediate state signal S _f to obtain a complex signal matrix, and the column vector in the complex signal matrix is the signal strength of each frequency; Extract the element set C _f at the EHG eigenfrequency from the complex signal matrix; The eigenfrequency matrix W is obtained by subtracting the element set C _f at the EHG eigenfrequency and the corresponding EHG frequency eigenvalue threshold; The elements of each row in the eigenfrequency matrix W are summed to obtain a column vector V representing the strength of the EHG signal,

n is the total number of sampling points; Take the set of subscripts whose elements in the column vector V are greater than 0 and whose continuous time is greater than t _min and less than t _max as I, where I is the set of sampling points marked as the EHG feature signal segment; Plot the EHG curve, the EHG curve is:

Among them, the value of i is [1,n]. 9 . The method for monitoring uterine contractions based on multi-dimensional information fusion according to claim 8 , wherein the EHG characteristic frequencies are 2 or 3 or 4 or 5. 10 . 10. A uterine contraction monitoring system based on multi-dimensional information fusion, characterized in that, comprising: The linear acceleration acquisition module is used for information fusion of the three-axis acceleration signal and the three-axis angular velocity signal of the pregnant woman's abdomen obtained by the six-axis sensor to obtain the three-axis linear acceleration signal; The uterine contraction feature marking module is used to mark the uterine contraction feature segment of the three-axis linear acceleration signal, and determine the set of uterine contraction time points; The contraction movement curve module is used to construct the contraction movement curve according to the three-axis linear acceleration signal and the contraction time point set; The MECG signal separation module is used to separate the MECG signal of the obtained mixed bioelectrical signal of the abdomen of the pregnant woman to obtain an intermediate state signal; The EHG curve module is used to mark the EHG characteristic signal segment of the intermediate state signal, and draw the EHG curve according to the marked EHG characteristic signal segment; The contraction curve module is used to perform curve fitting on the contraction motion curve and the EHG curve to generate the target contraction curve.

Images