A device and method for leadless 12-lead ECG acquisition

Abstract

The application provides a device and method for collecting 12-lead ECG without lead wire, and belongs to the technical field of electrocardio monitoring. The device comprises a first collecting terminal, a second collecting terminal and a third collecting terminal, which are used for being placed on the human body trunk to collect electrocardio signals, and the first collecting terminal, the second collecting terminal and the third collecting terminal are used for synchronously collecting and controlling and transmitting data based on a wireless personal area network; a computing module is located in the first collecting terminal, and is used for constructing a characteristic vector of a pseudo electrocardio vector by using data of the first collecting terminal, the second collecting terminal and the third collecting terminal, and deriving 12-lead ECG through a linear regression model. The application removes the lead wire of the electrocardio collecting system, improves the wearing comfort, and further improves the quality of collected data.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of electrocardio monitoring, and particularly relates to a device and method for collecting 12-lead ECG without lead wires. BACKGROUND

[0002] An electrocardiogram (ECG) is the representation of the electrical activity changes produced by the heart during each cardiac cycle on the body surface. In recent years, electrocardiogram examination has become the most widely used method in the diagnosis of myocardial ischemia. Electrocardiogram plays a crucial role in the diagnosis, evaluation, treatment guidance and prognosis prediction of cardiovascular diseases. Real-time electrocardiogram detection can provide more comprehensive electrocardio information, reduce the risk of missed diagnosis and misdiagnosis, help doctors take appropriate treatment measures, thereby reducing the incidence of myocardial infarction and other cardiovascular events and improving the long-term prognosis of patients.

[0003] According to the electric dipole hypothesis, the process of depolarization and repolarization of myocardial cells can be regarded as the change of a dipole in the chest cavity. The dipole has both size and direction, and is accompanied by regular changes in intensity and direction of the heart. The continuous real-time display of the heart electrical vector in three-dimensional space obtains a stereoscopic electrocardiogram. The stereoscopic electrocardiogram is projected on the frontal plane, horizontal plane and lateral plane for the first time to obtain three corresponding plane vector loops. The stereoscopic electrocardiogram is projected on the lead axis for the second time to form a conventional electrocardiogram. In the vector space, three linearly independent vectors can form a basis of the space and represent all vectors in the three-dimensional space.

[0004] The comfort of ECG wearing is of great significance to improve patient experience and ensure the accuracy of test results, especially for ECG devices that need to be monitored for a long time in daily life. High wearing comfort not only improves patient compliance, but also improves data quality. Various electrodes based on polymer materials and electronic textiles, which do not require conductive gel, are developed as substitutes for electrodes used in wearable sensing platforms. Through the method of improving the wearing comfort of the electrode, the comfort needs to be further improved since the lead wire is not removed. In addition, although the millimeter wave radar type non-contact ECG and single-arm ECG do not have lead wires, they can only test single-lead electrocardiogram. In order to solve the problem of wearing comfort caused by lead wires, some scholars have proposed a two-electrode electrocardiogram measurement method using a common electrode instead of a right leg reference electrode: using two electrode pairs, the power supply and ground of the electrode pair are not connected (without lead wire), and the electrode pair is pasted on the skin with the body as the reference potential to detect the electrophysiological signal. The signal is transmitted to an instrument amplifier filter through two isolation amplifiers to generate an electrocardiogram waveform. However, the signal-to-noise ratio of this method on Lead II and Lead III is not good enough for disease detection. SUMMARY

[0005] To solve the above problems, the application provides a device and method for collecting 12-lead ECG without lead wire.

[0006] To achieve the above-mentioned purposes, the application adopts the technical scheme of:

[0007] A device for collecting 12-lead ECG without lead wire comprises a first collection terminal, a second collection terminal and a third collection terminal, which are used to be placed on the human body trunk for collecting electrocardio signals, and the first collection terminal, the second collection terminal and the third collection terminal are based on wireless individual area network synchronous collection control and data transmission; the first collection terminal comprises a calculation module, which is used to construct a pseudo electrocardio vector feature vector by using the data of the first collection terminal, the second collection terminal and the third collection terminal, and derive 12-lead ECG through a linear regression model.

[0008] A method for collecting 12-lead ECG without lead wire is based on the device for collecting 12-lead ECG without lead wire, and comprises the following steps:

[0009] S110, obtaining electrocardio feature vectors through the first collection terminal, the second collection terminal and the third collection terminal :

[0010] ,

[0011] Among them, is the electrocardio data collected by the first collection terminal (110), is the electrocardio data collected by the second collection terminal (120), is the electrocardio data collected by the third collection terminal (130), is the electrocardio data after difference, is the electrocardio data after difference, is the electrocardio data after difference, is data after difference, is data after difference, is data after difference;

[0012] S120, the calculation module in the first collection terminal obtains derived ECG data according to a linear regression model, and the model parameters include a coefficient matrix and an intercept :

[0013] ,

[0014] Wherein, the matrix is a heart electrical characteristic vector 9 elements of the coefficient matrix is an intercept;

[0015] derived ECG data :

[0016] .

[0017] Compared with the prior art, the present application has the beneficial effects of:

[0018] Removing the lead wire of ECG collection, improving the comfort during ECG collection. The present application forms a wireless personal area network by UWB for the three collection terminals, and synchronizes the collection signals and transmits data in the wireless personal area network. Based on the heart electrical vector and vector space theory, the 12-lead ECG data is generated by using the machine learning linear regression method.

[0019] Suitable for long-term daily wearing to collect ECG signals. The collection terminal has the characteristics of miniaturization, and the collection system contains three collection terminals.

[0020] Suitable for embedded real-time processing system. The data conversion model is a linear regression model, and the calculation amount is small.

[0021] The present application is suitable for long-term daily life heart electrical signal collection and real-time monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments will be briefly introduced below.

[0023] Figure 1 It is a schematic diagram of a device for collecting 12-lead ECG without lead wire proposed by the present application;

[0024] Figure 2 It is a hardware block diagram of the collection terminal of the present application;

[0025] Figure 3 It is a flow chart of the collection terminal of the present application, wherein (a) is the flow chart of the first collection terminal 110, and (b) is the flow chart of the second collection terminal 120 and the third collection terminal 130;

[0026] Figure 4 It is a flow chart of the machine learning linear regression model constructed by the present application;

[0027] Figure 5 It is a flow chart of data processing and generation of 12-lead ECG in the first collection terminal 110 of the present application;

[0028] Figure 6Data collection terminal for the present application

[0029] Figure 7 The present application generates a 12-lead and standard 12-lead comparison chart, wherein (a) is a conventional 12-lead electrocardiogram, and (b) is a derived 12-lead electrocardiogram. DETAILED DESCRIPTION

[0030] In order to make the purposes, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions of the embodiments of the present disclosure will be described clearly and completely below with reference to the drawings of the embodiments of the present disclosure. The described embodiments are part of the embodiments of the present disclosure, rather than all the embodiments. Based on the described embodiments of the present disclosure, all other embodiments obtained by those of ordinary skill in the art without any creative effort fall within the protection scope of the present disclosure.

[0031] Unless otherwise defined, the technical terms or scientific terms used in the present disclosure shall have the meanings commonly understood by those skilled in the art to which the present disclosure belongs.

[0032] In order to remove the lead wires of multi-lead electrocardio collection device, and further improve the comfort of long-time daily multi-lead electrocardio signal collection, the present application provides a device for collecting 12-lead ECG without lead wires. As shown in the figure, the device specifically comprises: a first collection terminal 110, a second collection terminal 120 and a third collection terminal 130, the first collection terminal 110, the second collection terminal 120 and the third collection terminal 130 are used to be placed on the human body trunk for electrocardio signal collection, the first collection terminal 110, the second collection terminal 120 and the third collection terminal 130 are based on wireless individual area network synchronous collection control and data transmission; the first collection terminal 110 contains a calculation module, which is used to construct a pseudo electrocardio vector feature vector by using the data of the first collection terminal 110, the second collection terminal 120 and the third collection terminal 130, and derive 12-lead ECG through a linear regression model. Figure 1

[0033] Figure 2 The hardware block diagram of the collection terminal for the present application is shown in the figure. Figure 2 ​As shown, the first acquisition terminal 110, the second acquisition terminal 120 and the third acquisition terminal 130 each include: a battery for powering the acquisition terminal; a charging chip for charging the battery in each acquisition terminal; a power supply chip for converting the battery voltage into a fixed voltage to power the acquisition terminal; an electrocardio acquisition circuit (ECG acquisition circuit) for electrocardio signal acquisition; a Bluetooth MCU for system control, data processing and data communication between the acquisition terminal and external devices such as mobile phones, computers and the like; a storage unit for storing data; and a UWB module, by which the acquisition terminal forms a personal wireless network. The storage unit can be GD5F1GQ5UEYIGR. The UWB module can be DW3000. The electrocardio acquisition circuit, the UWB module and the storage unit are each connected to the Bluetooth MCU through an SPI bus.

[0034] The charging chip can be LGS4056HDA-4.35 for charging the battery. The power supply chip includes: TPS62203DBVR for generating a 3.3V power supply to power the Bluetooth MCU, the UWB module and the storage unit; and LM27762DSSR for generating a ±2.5V power supply to power the electrocardio acquisition circuit.

[0035] Figure 3 The working flowchart of the acquisition terminal of the application is shown in the figure, wherein (a) is the flowchart of the first acquisition terminal 110, and (b) is the flowchart of the second acquisition terminal 120 and the third acquisition terminal 130. The first acquisition terminal 110, the second acquisition terminal 120 and the third acquisition terminal 130 form a wireless personal area network through the UWB module. The first acquisition terminal 110 sends an acquisition start instruction through a broadcast signal, and the second acquisition terminal 120 and the third acquisition terminal 130 perform single electrocardio signal acquisition after receiving the start instruction. After the electrocardio signal acquisition is completed, the second acquisition terminal 120 and the third acquisition terminal 130 transmit the electrocardio signal to the first acquisition terminal 110 through the wireless personal area network.

[0036] As shown in the figure, Figure 3 the first acquisition terminal 110, the second acquisition terminal 120 and the third acquisition terminal 130 first complete the initialization of the ECG acquisition circuit when the system is initialized, and then write a pre-set personal area network PAN ID in the UWB module. The acquisition terminals only receive broadcast packets and data packets with the same PAN ID as themselves. The Bluetooth MCU of the first acquisition terminal 110 drives a 4ms timer, and the corresponding ECG sampling frequency is 250Hz. When the timer time is up, the UWB module is controlled to issue an acquisition instruction through an SPI instruction. After judging that the sending is successful, the Bluetooth MCU starts ECG acquisition through a control pin, reads ECG data through SPI and records the data as , as shown in (a) of the figure, Figure 3 . As shown in the figure, Figure 3As shown in (b), after the second acquisition terminal 120 and the third acquisition terminal 130 receive the acquisition instruction through the UWB module, the Bluetooth MCU starts single ECG acquisition through the control pin, reads the ECG data through SPI, and then the Bluetooth MCU controls the UWB module to publish the collected data in the wireless personal area network through SPI, that is, the data is transmitted to the first acquisition terminal 110. It should be noted that, in order to reduce the radio frequency transceiver conflict of the UWB module, the second acquisition terminal 120 transmits immediately after the data acquisition is completed, and the third acquisition terminal 130 transmits after a delay of 1 ms after the data acquisition is completed. After the UWB module receives the data of the second acquisition terminal 120 and the third acquisition terminal 130, the Bluetooth MCU reads the acquisition terminal data through SPI and records them as 、 . At this point, the system completes the acquisition and transmission of single ECG signals. After receiving the data, the first acquisition terminal 110 generates ECG vectors 、 、 using , and the calculation module derives 12-lead data using by a linear regression model. Finally, the first acquisition terminal determines whether there is a shutdown instruction. If there is, it shuts down, otherwise it continues to wait for the timer to arrive. After determining that the data transmission is successful, the second acquisition terminal and the third acquisition terminal determine whether there is a shutdown instruction. If there is, it shuts down, otherwise the UWB module continues to wait for the acquisition instruction.

[0037] The acquisition terminal is placed on the human body trunk: the first acquisition terminal 110 is placed parallel to the fourth intercostal space, and the left electrode is located on the left midline of the human body; the second acquisition terminal 120 is placed vertically on the middle of the sternum, and the terminal center is located in the fourth intercostal space of the human body; the third acquisition terminal 130 is placed at an angle of 45° in the third intercostal space, and the right electrode is on the left sternal line. By comparing with the surface electromotive force map of the human body, it can be known that the data collected by the first acquisition terminal 110 mainly includes: ECG vector X-axis data, containing part of the Z-axis data; the data collected by the second acquisition terminal 120 mainly includes: ECG vector Y-axis data, containing part of the Z-axis data; the data collected by the third acquisition terminal 130 mainly includes: ECG vector Z-axis data, containing part of the Y-axis data. The data obtained by the three acquisition terminals has no linear relationship.

[0038] After wearing the acquisition terminal, calibration is required for the first measurement. The calibration process is as follows Figure 4The calibration needs to place the 12-lead ECG device according to the Mason-linker lead system synchronously. The calibration contains the simultaneous acquisition of: the acquisition terminal data and the conventional 12-lead ECG data, and the acquisition terminal data is 2 minutes of acquisition terminal data. After the completion of wearing the acquisition terminal, the 2-minute 12-lead data is acquired synchronously using the conventional 12-lead device according to the Mason-linker lead system, and the linear regression model is established. The model establishment includes: preprocessing the 2-minute 12-lead data; constructing the electrocardiogram feature vector based on the first to third acquisition terminal acquisition ; comparing and aligning the preprocessed 12-lead data and the electrocardiogram feature vector using the RR interval of the first lead of the 12-lead and the RR interval of the first acquisition terminal data . The 12-lead data is divided into data segments according to the RR position of the first lead as label data, and the electrocardiogram feature vector is divided into data as original data according to the RR position of the first acquisition terminal data , , and the calibration data set is constructed. Considering the interpretability and computational complexity of the model, the linear regression algorithm ElasticNetCV is used. ElasticNetCV combines the L1 (Lasso) and L2 (Ridge) regularized linear regression model, and automatically optimizes the hyperparameters through cross-validation (CV). Its loss function is as follows:

[0039] ,

[0040] wherein, is the original data, i.e. the electrocardiogram feature vector , is the regression coefficient, including the coefficient matrix and the intercept , is the label data, is the sample number, is the regularization parameter, is the mixing ratio of L1 regularization and L2 regularization, is the prediction value of the model.

[0041] The training set accounts for 90% of the data set, and the test set accounts for 10% of the data set. The hyperparameters and (l1_ratio) are automatically selected using 5-fold cross-validation (cv=5), and the maximum number of iterations (max_iter) is 5000 times. The termination condition is that the number of iterations reaches 5000 times or the change of the loss function of the adjacent 2 times is less than 0.1.

[0042] In order to further evaluate the performance of the model, the evaluation index of the regression model, the coefficient of determination (R2) is used ) and root mean square error ( An evaluation will be conducted. When and When the linear regression model is effective, it is valid.

[0043] Furthermore, after the model is established, the 12-lead device can be removed, and the linear regression model can be deployed in the calculation module of the first acquisition terminal 110, allowing for long-term continuous ECG measurement using the self-acquisition terminal. The first acquisition terminal 110 can perform data preprocessing and feature vector generation. It constructs, converts, and stores 12-lead ECG data; no external intervention is required during use. Data can be viewed and reviewed via smart terminals or computers.

[0044] The calibration process includes: preprocessing the acquisition terminal data, including: removing the baseline using a median filter, removing high-frequency noise using a low-pass filter (cutoff frequency 40Hz), and removing high-frequency noise in the SP segment using a segmented filter (cutoff frequency 10Hz); performing RR segmentation on the preprocessed acquisition terminal data, and aligning it according to the RR interval and standard leads; constructing a dataset based on the segmented and aligned acquisition terminal data for training and testing the linear regression model. The linear regression model is based on the ElasticNet machine learning linear regression algorithm to obtain the linear regression model, and then derives the 12-lead ECG from the linear regression model. The linear regression model can also be called a regression model. It should be noted that in the conventional 12-lead ECG, Lead III, aVR, aVL, and aVF are directly calculated from Lead I and Lead II. Therefore, the model only performs the conversion and comparison of "Lead I", "Lead II" and the precordial leads "V1", "V2", "V3", "V4", "V5", and "V6".

[0045] After calibration, the standard 12-lead device can be removed, and the linear regression model can be deployed in the calculation module of the first acquisition terminal 110 for long-term daily heart rate monitoring.

[0046] S110, obtain electrocardiogram feature vectors through the first acquisition terminal 110, the second acquisition terminal 120 and the third acquisition terminal 130. :

[0047] ,

[0048] in, These are electrocardiogram (ECG) data collected by the first acquisition terminal 110. These are electrocardiogram (ECG) data collected by the second acquisition terminal 120. These are electrocardiogram (ECG) data collected by the third acquisition terminal 130. It's electrocardiogram data. Differential data, is ECG data Differential data, is ECG data Differential data, is Differential data, is Differential data, is Differential data.

[0049] S120, the calculation module obtains the derived ECG data according to the model parameters of the linear regression model, including the coefficient matrix and the intercept

[0050]

[0051] wherein the coefficient matrix is the coefficient of the 9 parameters in the matrix , and the matrix is the intercept.

[0052] The derived ECG data

[0053]

[0054] Table 1 shows a specific example of the conversion model:

[0055] Table 1 Conversion model example

[0056]

[0057] In the normal wearing collection process, the data processing flow is as shown in Figure 5 . Correspondingly, the data needs to be preprocessed: using a median filter to remove the baseline, removing high-frequency noise through a low-pass filter (cutoff frequency 40Hz), and removing S-P segment high-frequency noise (cutoff frequency 10Hz) by a segment filter. As shown in Figure 6 , it is a data example of the first collection terminal 110, the second collection terminal 120 and the third collection terminal 130 after removing the baseline and filtering. The ECG vector is constructed from the collection terminal data. Linear regression model: using the linear regression model obtained during calibration to derive 12-lead ECG, the linear regression model parameters include the coefficient matrix and the intercept.

[0058] Table 1 shows the data obtained by an example of the collection terminal, and the "Lead I", "Lead II" and precordial leads "V1", "V2", "V3", "V4", "V5", "V6" after coefficient conversion​​​​Figure 7 (b) shows the derived ECG data, Figure 7 (a) shows the conventional ECG corresponding leads, which are used to compare the derived data and the conventional ECG data.

[0059] Further, in order to verify the present application, Table 2 gives the determination coefficient (R2) and the root mean square error (RMSE) of the derived ECG data and the conventional ECG in the long time measurement scenario. The comparison results show that the derived ECG data is effective.

[0060] Table 2 determination coefficient (R2) and the root mean square error (RMSE) of the derived ECG data and the conventional ECG in the long time measurement scenario

[0061]

[0062] The above description is merely preferred embodiments of the present application, but not to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.​​​

Claims

1. A device for 12-lead ECG acquisition without lead wires, characterized in that, Comprise: The first acquisition terminal (110), the second acquisition terminal (120) and the third acquisition terminal (130) are used for being placed on the human body torso to carry out electrocardiogram signal acquisition, and the first acquisition terminal (110), the second acquisition terminal (120) and the third acquisition terminal (130) are based on wireless personal area network synchronous acquisition control and data transmission;The first acquisition terminal (110) comprises a calculation module, which is used for constructing the feature vector of pseudo electrocardiogram vector by using the data of the first acquisition terminal (110), the second acquisition terminal (120) and the third acquisition terminal (130), and deriving 12-lead ECG through a linear regression model; The first acquisition terminal (110) is placed in the fourth intercostal space in parallel, and the left electrode is located on the left midline of the human body;The second acquisition terminal (120) is vertically placed in the middle of the human body sternum, and the terminal center is located in the fourth intercostal space of the human body;The third acquisition terminal (130) is placed in the third intercostal space at an angle of 45°, and the right electrode is on the left sternal line; The data collected by the first acquisition terminal (110) comprises electrocardiogram vector X-axis data and part of Z-axis data;The data collected by the second acquisition terminal (120) comprises electrocardiogram vector Y-axis data and part of Z-axis data;The data collected by the third acquisition terminal (130) comprises electrocardiogram vector Z-axis data and part of Y-axis data; The characteristic vector of the pseudo heart electric vector is obtained by the first acquisition terminal (110), the second acquisition terminal (120) and the third acquisition terminal (130) : ， Wherein, is the electrocardio data collected by the first collection terminal (110), is the electrocardio data collected by the second collection terminal (120), is the electrocardio data collected by the third collection terminal (130), is the electrocardio data is the differential data, is the electrocardio data is the differential data, is the electrocardio data is the differential data, is is the differential data, is is the differential data, is is the differential data.

2. The device for 12-lead ECG acquisition without leads wires of claim 1, wherein, The first acquisition terminal (110), the second acquisition terminal (120) and the third acquisition terminal (130) each comprise: a battery for supplying power to the acquisition terminal;A charging chip for charging the battery in each acquisition terminal;A power supply chip for converting the battery voltage into a fixed voltage to supply power to the acquisition terminal;An electrocardiogram acquisition circuit, i.e. ECG acquisition circuit, for electrocardiogram signal acquisition;A Bluetooth MCU for system control, data processing and data communication between the acquisition terminal and external equipment;A storage unit for storing data;A UWB module, and the acquisition terminal is composed of a personal wireless network through the UWB module;The electrocardiogram acquisition circuit, the UWB module and the storage unit are connected with the Bluetooth MCU through the SPI bus respectively.

3. The device for 12-lead ECG acquisition without leads wires of claim 2, wherein, The storage unit is GD5F1GQ5UEYIGR, and the UWB module is DW3000.

4. The device for 12-lead ECG acquisition without leads wires of claim 2, wherein, The power supply chip comprises TPS62203DBVR for generating 3.3V power supply to supply power to the Bluetooth MCU, the UWB module and the storage unit, and LM27762DSSR for generating ±2.5V power supply to supply power to the electrocardiogram acquisition circuit.

5. The device for 12-lead ECG acquisition without leads wires of claim 2, wherein, The acquisition process of the first acquisition terminal (110), the second acquisition terminal (120) and the third acquisition terminal (130) is as follows: the first acquisition terminal (110) sends an acquisition start instruction through the broadcast signal of the wireless personal area network, and the second acquisition terminal (120) and the third acquisition terminal (130) perform single electrocardiogram signal acquisition after receiving the start instruction;After the electrocardiogram signal acquisition is completed, the second acquisition terminal (120) and the third acquisition terminal (130) transmit the electrocardiogram signal to the first acquisition terminal (110) through the wireless personal area network.

6. A method of leadless 12-lead ECG acquisition, performed on the basis of the leadless 12-lead ECG acquisition device according to any one of claims 1-5, characterized in that, Comprise: S110, obtaining the electrocardio feature vector by the first acquisition terminal (110), the second acquisition terminal (120) and the third acquisition terminal (130) : ， wherein, is the electrocardio data collected by the first collection terminal (110), is the electrocardio data collected by the second collection terminal (120), is the electrocardio data collected by the third collection terminal (130), is the electrocardio data after difference, is the electrocardio data after difference, is the electrocardio data after difference, is after difference, is after difference, is after difference. S120, the first acquisition terminal (110) computing module according to linear regression model derived ECG data, model parameters include coefficient matrix and intercept : ， wherein the matrix is the electrocardiographic feature vector is the coefficient of the 9th element in the matrix is the intercept; Deriving ECG data : 。 7. The method of acquiring a 12-lead ECG without leads wires of claim 6, wherein, Further comprising a calibration process, and the calibration process comprises: The collected terminal data is preprocessed; the preprocessed collected terminal data is subjected to RR segmentation, and alignment is performed according to the RR interval and the standard lead; a data set is constructed based on the segmented and aligned collected terminal data, which is used for training and testing of a linear regression model, the linear regression model is based on a machine learning linear regression algorithm ElasticNetCV, a linear regression model is obtained, and then 12-lead ECG is derived through the linear regression model.

8. The method of acquiring a 12-lead ECG without leads wires of claim 7, wherein, The preprocessing includes: using a median filter to remove the baseline, using a low-pass filter to remove high-frequency noise, and using a segment filter to remove high-frequency noise in the S-P segment.

Images