阅读说明：本技术 Ppg心搏信号的预处理方法、装置及房颤检测设备 (PPG heart beat signal preprocessing method and device and atrial fibrillation detection equipment ) 是由 杨海波 黄韵竹 薛奋 于 2019-01-09 设计创作，主要内容包括：本发明公开了PPG心搏信号预处理方法、装置及房颤检测系统。其中PPG心搏信号预处理方法包括如下步骤：采集被检测心搏信号和被检测者运动信号,对采集的心搏信号进行频域滤波；在已经完成频域滤波的心搏信号片段上,获得全部波谷位置,并依次将相邻两个波谷之间的心搏信号作为一拍,得到连续的多拍心搏信号；判断每拍上的心搏信号是否同时满足运动信号阈值和信噪比阈值,当至少连续N拍均同时满足运动阈值和信噪比阈值,则保留该至少连续N拍心搏信号并输出,信噪比是指统计一拍内的噪音包络和心搏信号包络的平均值,然后得到的比值。本发明通过上述原理,能够准确筛选高质量PPG心搏信号,进行后续房颤检测,保证房颤检测数据的准确可靠。(The invention discloses a PPG heart beat signal preprocessing method, a PPG heart beat signal preprocessing device and an atrial fibrillation detection system. The PPG heart beat signal preprocessing method comprises the following steps: acquiring a heart beat signal to be detected and a motion signal of a detected person, and performing frequency domain filtering on the acquired heart beat signal; acquiring all trough positions on the heart beat signal segments after frequency domain filtering is finished, and sequentially taking the heart beat signals between two adjacent troughs as one beat to obtain continuous multi-beat heart beat signals; judging whether the heartbeat signal of each beat simultaneously meets a motion signal threshold and a signal-to-noise ratio threshold, and when at least N continuous beats simultaneously meet the motion threshold and the signal-to-noise ratio threshold, retaining the heartbeat signal of the at least N continuous beats and outputting the heartbeat signal, wherein the signal-to-noise ratio is the average value of noise envelope and heartbeat signal envelope in one beat, and then obtaining the ratio. According to the principles, the high-quality PPG heart beat signals can be accurately screened, the subsequent atrial fibrillation detection is carried out, and the accuracy and reliability of atrial fibrillation detection data are guaranteed.)

A PPG heart beat signal preprocessing method is characterized by comprising the following steps:

Acquiring a heart beat signal to be detected and a motion signal of a detected person, and performing frequency domain filtering on the acquired heart beat signal;

Acquiring all trough positions on the heart beat signal segments after frequency domain filtering is finished, and sequentially taking the heart beat signals between two adjacent troughs as one beat to obtain continuous multi-beat heart beat signals;

Judging whether the heartbeat signal on each beat simultaneously meets a motion signal threshold and a signal-to-noise ratio threshold, and when at least N continuous beats simultaneously meet the motion threshold and the signal-to-noise ratio threshold, retaining the heartbeat signal of the at least N continuous beats and outputting the heartbeat signal, wherein N is a positive integer greater than or equal to 6, and the signal-to-noise ratio is the average value of noise envelopes and heartbeat signal envelopes in one beat respectively counted by taking beats as a unit, and then obtaining the ratio.

2. The method for pre-processing the PPG heartbeat signal according to claim 1, wherein the motion signal threshold comprises an acceleration threshold and an angular velocity threshold, which need to be met simultaneously.

3. The PPG heart beat signal pre-processing method of claim 1 or 2, wherein the acceleration threshold and/or the angular velocity threshold are averaged over an interval of acceleration and/or angular velocity indicators using respective time windows.

4. The method for preprocessing PPG heart beat signals according to claim 1, wherein after the frequency-domain filtering is performed on the detected heart beat signal segment, the baseline interference detection is performed on the heart beat signal segment which has been subjected to the frequency-domain filtering, and the heart beat signals in the baseline morphological mutation interval are excluded.

PPG heart beat signal preprocessing device, characterized by comprising:

The signal acquisition filtering module: the device is used for acquiring a heart beat signal to be detected and a motion signal of a detected person and filtering the acquired heart beat signal in a frequency domain;

The signal sub-shooting module: the method is used for obtaining all trough positions on a heart beat signal segment which is subjected to frequency domain filtering, and taking heart beat signals between two adjacent troughs as one beat in sequence to obtain continuous multi-beat heart beat signals;

A signal screening module: judging whether the heartbeat signal on each beat simultaneously meets a motion signal threshold and a signal-to-noise ratio threshold, and when at least N continuous beats simultaneously meet the motion threshold and the signal-to-noise ratio threshold, retaining the heartbeat signal of the at least N continuous beats and outputting the heartbeat signal, wherein N is a positive integer greater than or equal to 6, and the signal-to-noise ratio is the average value of noise envelopes and heartbeat signal envelopes in one beat respectively counted by taking beats as a unit, and then obtaining the ratio.

6. The PPG heart beat signal preprocessing device according to claim 5, wherein the motion signal threshold comprises an acceleration threshold and an angular velocity threshold, which are required to be satisfied simultaneously.

7. The PPG heart beat signal preprocessing device according to claim 5 or 6, further comprising

A baseline disturbance detection module: and carrying out baseline interference detection on the heart beat signal segments subjected to frequency domain filtering, and excluding the heart beat signals in the baseline morphological mutation interval.

8. An atrial fibrillation detection apparatus, comprising

PPG heart beat signal preprocessing device: screening detected heart beat signals with signal quality meeting set requirements, and outputting the detected heart beat signals to an interval variation value calculation module;

An interval variation value calculation module: calculating interval variation values of the heart beat intervals of the detected heart beat signals screened by the PPG heart beat signal preprocessing device;

Atrial fibrillation score determining module: presetting n interval variation value intervals, wherein n is a positive integer, and calculating the cumulative atrial fibrillation score of the interval variation value in each preset interval variation value interval according to a preset variation value-score mapping table to obtain n atrial fibrillation scores;

Atrial fibrillation score conversion module: forming n atrial fibrillation scores into a matrix x with a row and n columns, substituting the matrix x into a logistic regression cost function or an SVM function, and calculating to obtain an output estimated value H (theta) of the model;

Atrial fibrillation judgment module: and comparing the output estimated value H (theta) of the model pair with a preset threshold value, and judging whether atrial fibrillation exists in the detected heartbeat signal segment according to the comparison result.

9. The atrial fibrillation detection apparatus of claim 8, wherein the interval variation value calculation module uses the interval variation value calculation formula Where iRRCur is the interval of the ith interval, iRRCur is the interval of the (i-1) th interval, and f (i) is the variance of the ith interval.

10. The atrial fibrillation detection apparatus of claim 8, wherein the logistic regression cost function is calculated by the formula: h (Θ) is 1/(1+ exp (- (θ) ^TX)) where H (θ) is the estimate of the output of the model, x is a matrix that groups n atrial fibrillation scores into a row and n columns, θ is the model parameter, θ) is the output of the model, and ^TIs the transposition of the model parameters; the calculation formula of the SVM function is as follows: h (theta) ═ theta ₀+θ₁f₁+θ₂f₂+θ₃f₃Wherein theta ₀、θ₁、θ₂And theta ₃Are all training model parameters, H (theta) is the estimate of the model to the output, x is a matrix that groups n atrial fibrillation scores into a row and n columns, For all features and support vectors l in x ⁽¹⁾The sum of the distances between, σ, is a set parameter.

Technical Field

The invention relates to the field of signal preprocessing, in particular to a PPG heart beat signal preprocessing method and device and atrial fibrillation detection equipment.

Background

Atrial fibrillation (hereinafter referred to as atrial fibrillation) is a common symptom of cardiac arrhythmia, and the incidence of atrial fibrillation increases with age. At present, the diagnosis of atrial fibrillation relies primarily on electrocardiography, which detects atrial fibrillation primarily by analyzing ECG signals. However, electrocardiographic devices are expensive and the procedure is professional. Furthermore, in the electrocardiographic detection of atrial fibrillation, a plurality of lead wires are connected to an individual, and the probing electrodes are arranged at fixed positions on the limbs and the anterior chest wall for measurement. For ordinary people, if the people have obvious problems in physical health conditions or do health examination regularly with health consciousness, people do not actively go to medical institutions for electrocardiographic examination, which is not favorable for the discovery of atrial fibrillation. And PPG data acquisition is simple, can realize incessant collection on wearable equipment, and usable data bulk is big, and the coverage time quantum is comprehensive, can be used to the detection of atrial fibrillation, makes things convenient for the discovery of atrial fibrillation. Because PPG data is easily interfered by the outside, the traditional method is not ideal for preprocessing PPG signals, and the subsequent detection result is not accurate.

Disclosure of Invention

The invention aims to solve the technical problem of realizing accurate screening of high-quality PPG heart signals, and aims to provide a PPG heart signal preprocessing method, a PPG heart signal preprocessing device and atrial fibrillation detection equipment, which can accurately screen high-quality PPG heart signals, perform subsequent atrial fibrillation detection and ensure the accuracy and reliability of atrial fibrillation detection data.

The invention is realized by the following technical scheme:

In a first aspect, the invention discloses a PPG heart beat signal preprocessing method, which comprises the following steps:

Acquiring a heart beat signal to be detected and a motion signal of a detected person, and performing frequency domain filtering on the acquired heart beat signal; acquiring all trough positions on the heart beat signal segments after frequency domain filtering is finished, and sequentially taking the heart beat signals between two adjacent troughs as one beat to obtain continuous multi-beat heart beat signals; judging whether the heartbeat signal on each beat simultaneously meets a motion signal threshold and a signal-to-noise ratio threshold, and when at least N continuous beats simultaneously meet the motion threshold and the signal-to-noise ratio threshold, retaining the heartbeat signal of the at least N continuous beats and outputting the heartbeat signal, wherein N is a positive integer greater than or equal to 6, and the signal-to-noise ratio is the average value of noise envelopes and heartbeat signal envelopes in one beat respectively counted by taking beats as a unit, and then obtaining the ratio.

Corresponding to the first aspect, the motion signal threshold in the present invention includes an acceleration threshold and an angular velocity threshold, which need to be satisfied simultaneously.

In accordance with a first aspect of the present invention, the acceleration threshold and/or the angular velocity threshold are the result of averaging the acceleration and/or angular velocity indicators over a certain interval using respective time windows.

Corresponding to the first aspect, after the frequency domain filtering is carried out on the heart beat signal segment to be detected, the base line interference detection is carried out on the heart beat signal segment which is already subjected to the frequency domain filtering, and the heart beat signal in the base line morphological mutation interval is excluded.

In a second aspect, the invention discloses a PPG heart beat signal preprocessing device, comprising:

The signal acquisition filtering module: the device is used for acquiring a heart beat signal to be detected and a motion signal of a detected person and filtering the acquired heart beat signal in a frequency domain; the signal sub-shooting module: the method is used for obtaining all trough positions on a heart beat signal segment which is subjected to frequency domain filtering, and taking heart beat signals between two adjacent troughs as one beat in sequence to obtain continuous multi-beat heart beat signals; a signal screening module: judging whether the heartbeat signal on each beat simultaneously meets a motion signal threshold and a signal-to-noise ratio threshold, and when at least N continuous beats simultaneously meet the motion threshold and the signal-to-noise ratio threshold, retaining the heartbeat signal of the at least N continuous beats and outputting the heartbeat signal, wherein N is a positive integer greater than or equal to 6, and the signal-to-noise ratio is the average value of noise envelopes and heartbeat signal envelopes in one beat respectively counted by taking beats as a unit, and then obtaining the ratio.

Corresponding to the second aspect, the motion signal threshold in the present invention includes an acceleration threshold and an angular velocity threshold, which need to be satisfied simultaneously.

Corresponding to the second aspect, the invention further comprises a baseline disturbance detection module: and carrying out baseline interference detection on the heart beat signal segments subjected to frequency domain filtering, and excluding the heart beat signals in the baseline morphological mutation interval.

In a third aspect the invention discloses an atrial fibrillation detection apparatus comprising

PPG heart beat signal preprocessing device: screening detected heart beat signals with signal quality meeting set requirements, and outputting the detected heart beat signals to an interval variation value calculation module; an interval variation value calculation module: calculating interval variation values of the heart beat intervals of the detected heart beat signals screened by the PPG heart beat signal preprocessing device; atrial fibrillation score determining module: presetting n interval variation value intervals, wherein n is a positive integer, and calculating the cumulative atrial fibrillation score of the interval variation value in each preset interval variation value interval according to a preset variation value-score mapping table to obtain n atrial fibrillation scores; atrial fibrillation score conversion module: forming n atrial fibrillation scores into a matrix x with a row and n columns, substituting the matrix x into a logistic regression cost function or an SVM function, and calculating to obtain an output estimated value H (theta) of the model;

Atrial fibrillation judgment module: and comparing the output estimated value H (theta) of the model pair with a preset threshold value, and judging whether atrial fibrillation exists in the detected heartbeat signal segment according to the comparison result.

In accordance with a third aspect of the present invention, the interval variance value calculation module uses an interval variance value calculation formula of Where iRRCur is the interval of the ith interval, iRRCur is the interval of the (i-1) th interval, and f (i) is the variance of the ith interval.

Corresponding to the third aspect, the calculation formula of the logistic regression cost function in the present invention is: h (theta) is 1/(1+ exp (- (theta)) ^TX)) where H (θ) is the estimate of the output of the model, x is a matrix that groups n atrial fibrillation scores into a row and n columns, θ is the model parameter, θ) is the output of the model, and ^TIs the transposition of the model parameters; the calculation formula of the SVM function is as follows: h (theta) ═ theta ₀+θ₁f₁+θ₂f₂+θ₃f₃Wherein theta ₀、θ₁、θ₂And theta ₃Are all training model parameters H (theta) is the estimate of the output of the model, x is a matrix that groups n atrial fibrillation scores into a row and n columns, For all features and support vectors l in x ⁽¹⁾The sum of the distances between, σ, is a set parameter.

Corresponding to the third aspect, the variance-score mapping table described in the present invention is derived from the output of a empirical statistical process.

Corresponding to the third aspect, the apparatus for preprocessing the PPG heart beat signals in the invention comprises

The signal acquisition filtering module: the device is used for acquiring a heart beat signal to be detected and a motion signal of a detected person and filtering the acquired heart beat signal in a frequency domain; the signal sub-shooting module: the method is used for obtaining all trough positions on a heart beat signal segment which is subjected to frequency domain filtering, and taking heart beat signals between two adjacent troughs as one beat in sequence to obtain continuous multi-beat heart beat signals; a signal screening module: judging whether the heartbeat signal on each beat simultaneously meets a motion signal threshold and a signal-to-noise ratio threshold, and when at least N continuous beats simultaneously meet the motion threshold and the signal-to-noise ratio threshold, retaining the heartbeat signal of the at least N continuous beats and outputting the heartbeat signal, wherein N is a positive integer greater than or equal to 6, and the signal-to-noise ratio is the average value of noise envelopes and heartbeat signal envelopes in one beat respectively counted by taking beats as a unit, and then obtaining the ratio.

Corresponding to the third aspect, the present invention further includes a baseline disturbance detection module: and carrying out baseline interference detection on the heart beat signal segments subjected to frequency domain filtering, and excluding the heart beat signals in the baseline morphological mutation interval.

Corresponding to the third aspect, the motion signal threshold in the present invention includes an acceleration threshold and an angular velocity threshold, and the acceleration threshold and the angular velocity threshold need to be satisfied simultaneously.

Compared with the prior art, the invention has the following advantages and beneficial effects:

1. This scheme adopts frequency domain filtering to get rid of the baseline interference, isolate effective signal and high frequency noise, utilize the characteristics of effective signal and high frequency noise of isolating, found a method that calculates the SNR more accurately than prior art, be used for directly getting rid of the poor PPG heart beat signal of waveform quality, utilize motion signal threshold value to get rid of the big PPG heart beat signal of motion interference indirectly, the signal of screening out is the steady and high bat of SNR of motion state, high quality signal promptly, carry out subsequent atrial fibrillation and detect, guaranteed the accuracy reliable of atrial fibrillation detection data.

2. The method also needs to perform baseline interference detection on the heartbeat signal segments which are subjected to frequency domain filtering, eliminates heartbeat signals in a baseline morphological mutation interval, and further ensures the accuracy and reliability of atrial fibrillation detection data.

3. The invention has simple acquisition mode of PPG heart beat signals, can realize on-line real-time all-day detection and has convenient acquisition.

Drawings

The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention. In the drawings:

FIG. 1 is a flow chart of the present invention;

Fig. 2 is a schematic block diagram of a PPG heart beat signal preprocessing device;

FIG. 3 is a schematic block diagram of an atrial fibrillation detection apparatus;

Figure 4 shows a schematic diagram of an example PPG heart beat signal segment;

Fig. 5 is a graph comparing PPG signals and noise of a PPG heart beat signal segment after frequency domain filtering.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to examples and accompanying drawings, and the exemplary embodiments and descriptions thereof are only used for explaining the present invention and are not meant to limit the present invention.

As shown in fig. 4, as will be understood by those skilled in the art, the PPG (photoplethysmogram) signal can reflect the regularity of heart beats, and the inter-beat interval described in the present disclosure refers to an interval of each PPG heart beat signal, i.e., a time difference (RR interval) between two adjacent wave troughs; one beat of heart beat signal refers to all heart beat signals between two adjacent wave troughs; the inter-beat interval variation value refers to the difference value between two adjacent RR interval values. The noise envelope refers to the curve of the amplitude of a noise signal changing with time in a random process, and the heart beat signal envelope refers to the curve of the amplitude of a PPG heart beat signal changing with time in a random process.