CN111904406A - Physiological signal motion artifact suppression device and method - Google Patents

Physiological signal motion artifact suppression device and method Download PDF

Info

Publication number
CN111904406A
CN111904406A CN202010864009.2A CN202010864009A CN111904406A CN 111904406 A CN111904406 A CN 111904406A CN 202010864009 A CN202010864009 A CN 202010864009A CN 111904406 A CN111904406 A CN 111904406A
Authority
CN
China
Prior art keywords
frequency
cur
heart rate
result
physiological signal
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202010864009.2A
Other languages
Chinese (zh)
Inventor
王国兴
王敏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Jiaotong University
Original Assignee
Shanghai Jiaotong University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shanghai Jiaotong University filed Critical Shanghai Jiaotong University
Priority to CN202010864009.2A priority Critical patent/CN111904406A/en
Publication of CN111904406A publication Critical patent/CN111904406A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/024—Detecting, measuring or recording pulse rate or heart rate
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • A61B5/7207—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal of noise induced by motion artifacts
    • A61B5/721—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal of noise induced by motion artifacts using a separate sensor to detect motion or using motion information derived from signals other than the physiological signal to be measured
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7253—Details of waveform analysis characterised by using transforms
    • A61B5/7257—Details of waveform analysis characterised by using transforms using Fourier transforms

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Surgery (AREA)
  • Molecular Biology (AREA)
  • Physiology (AREA)
  • Veterinary Medicine (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • General Health & Medical Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Artificial Intelligence (AREA)
  • Psychiatry (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Cardiology (AREA)
  • Mathematical Physics (AREA)
  • Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)

Abstract

The invention discloses a physiological signal motion artifact suppression device, which comprises: a band-pass filtering module; an adaptive noise cancellation module; a heart rate frequency correction module; and a notch filter. The method combines accurate estimation of heart rate frequency with notch filtering, and utilizes a notch filter constructed by the heart rate frequency and second harmonic thereof to filter motion artifact components in the physiological signal and restore the waveform due to the physiological signal. The invention can effectively inhibit the motion artifact component in the physiological signal and highlight the fundamental frequency component in the frequency domain.

Description

Physiological signal motion artifact suppression device and method
Technical Field
The invention relates to the technical field of blood pressure monitoring, in particular to a physiological signal motion artifact suppression device and a physiological signal motion artifact suppression method.
Background
Photoplethysmography (PPG) is a physiological signal acquired by the principle of photoelectricity. It can effectively reflect the heart rate and blood volume change of people, so it has been widely used in clinical monitoring and daily health management, such as measuring heart rate, respiration rate, blood oxygen saturation, blood pressure and blood sugar, etc. To measure the PPG signal, a sensor is usually attached to the skin surface, a beam of light is emitted by the sensor, and the intensity of the reflected or transmitted light is measured. However, when the subject moves, the sensor is displaced relative to the skin, and the blood flow rate changes. These factors in turn cause interference to the physiological signals, such interference noise being commonly referred to as motion artifacts. Conventional methods include basic methods based on Adaptive Noise Cancellation (ANC) and wavelet transform, but have large errors.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a physiological signal motion artifact suppression device and a physiological signal motion artifact suppression method, motion artifacts are primarily suppressed based on an LMS-Newton (Least Mean Square-Newton) algorithm and ANC of an acceleration signal, the heart rate frequency is accurately estimated by further adopting an improved heart rate frequency tracking and heart rate frequency correction mechanism on the basis of fast Fourier transform, and finally, a notch filter constructed based on the heart rate frequency and harmonic waves thereof is used for recovering an original PPG signal waveform, so that the heart rate in violent motion can be accurately estimated.
In order to solve the technical problems, the invention is realized by the following technical scheme.
The invention discloses a physiological signal motion artifact suppression device on one hand, which comprises:
the band-pass filtering module is used for filtering the physiological signals polluted by the motion pseudo-track and the three-axis acceleration signals and transmitting the physiological signals and the three-axis acceleration signals to the adaptive noise elimination module;
-an adaptive noise cancellation module for initially suppressing motion artifact components in the physiological signal and for initially estimating the heart rate frequency from the signal output by the band-pass filtering module;
a heart rate frequency correction module, which determines whether the preliminary estimation result is likely to be erroneous, corrects the likely erroneous result according to the estimation results at the current and previous time instants, and outputs a final estimation result of the heart rate frequency to the notch filter;
a notch filter for performing notch filtering on the estimation result of the physiological signal and subtracting the notch filtering result from the triaxial acceleration signal to obtain the final physiological signal with the motion artifact suppressed.
The triaxial acceleration signal is an ANC noise reference signal.
The adaptive noise elimination module takes the mean value of one path of normalized physiological signal window or a plurality of paths of normalized physiological signal windows as the input of the polluted signal, and uses the normalized triaxial acceleration signal to carry out cascade type adaptive noise elimination processing on the physiological signal.
The band-pass filtering module is a Butterworth band-pass filter with a pass band of 0.4 Hz-4.0 Hz
The invention also discloses a method for inhibiting the physiological signal motion artifact, which comprises the following steps:
s1, filtering physiological signals and three-axis acceleration signals polluted by motion pseudo-tracks;
s2, carrying out cascade type self-adaptive noise elimination processing on the physiological signal by using the normalized triaxial acceleration signal;
s3, analyzing the frequency spectrum of a sub-window with the length L at the rear part of a PPG output signal window by using FFT (fast Fourier transform), and preliminarily estimating the value of heart rate frequency by using a heart rate tracking mechanism;
s4, after the preliminary estimation result of the heart rate frequency is obtained, judging whether the current result is possibly wrong or not by using a heart rate frequency correction mechanism, and correcting the result which is judged to be wrong;
s5, constructing two notch filters: the first notch filter having a notch frequency fcurThe notch frequency of the second filter is fcurSecond harmonic of (f)har(ii) a In the window spectrum of the physiological signal 2fcurFind the frequency corresponding to the peak with the maximum amplitude as fhar；
And subtracting the result of notch filtering from the physiological signal input to obtain the final physiological signal with the motion artifact restrained.
In step S2, the normalized triaxial acceleration signal is used to perform a cascade-type adaptive noise cancellation process on the physiological signal, and an LMS-Newton algorithm is adopted, specifically as follows:
R-1(0)＝I (3-1)
W(0)＝X(0)＝[0,0,...,0]T (3-2)
X(k)＝[x(k),x(k-1),...,x(k-M+1)]T (3-3)
e(k)＝d(k)-XT(k)W(k) (3-4)
Figure BDA0002649121980000031
W(k+1)＝W(k)+2μe(k)R-1(k)X(k) (3-6)
the parameter is a positive constant whose magnitude is generally inversely proportional to the power of the input signal, the parameters α and μ are parameters determining the convergence rate of the algorithm, α is generally a positive number far less than 1, μ has a value generally around one-half of α, d (k) represents the PPG input signal contaminated by motion artifacts, x (k) represents the acceleration signal, e (k) is the noise-reduced physiological signal output, the vector w (k) is the tap coefficient of the FIR filter, M is the order of the filter, and the matrix r (k) is an intermediate variable in the iteration of the algorithm.
In the step s4, after the preliminary estimation result of the heart rate frequency is obtained, whether the current result is possibly erroneous is judged by using a heart rate frequency correction mechanism, and the result judged as the error is corrected, which is specifically as follows:
step 4.1, judging whether the difference between the current preliminary estimation result and the estimation result of the previous time window is too large: if the difference is greater than Th0, go to step 4.2, otherwise, keep the current estimation result unchanged;
step 4.2 if fcurGreater than fpreThen the current estimation result is likely to be an excessively large erroneous result, and the correct result is likely to be a result smaller than fcurThe frequency of (d); while the correct result is likely to be greater than f according to the trend of the changepre(ii) a Otherwise if fcurIs less than fpreThe correct result is likely to be a value less than fpreAnd is greater than fcurA value of (d);
step 4.3 in the spectrum of the current window, at fcurAnd fpreFind the frequency f corresponding to the maximum value of the frequency domain peakNThe search range will generally be fpreIn a direction slightly extending if f is not presentNThen the current estimation result is kept unchanged, otherwise step 4.4 is performed.
Step 4.4 judging fcurWhether or not it is possible that one is caused by motion artifactsFrequency domain peak from: averaging three paths of acceleration signal windows, calculating frequency spectrums of the three paths of acceleration signal windows by using FFT (fast Fourier transform), and finding out frequency f corresponding to the maximum peak in the frequency spectrumsaccThis frequency is taken as a representation of the dominant frequency of the motion artifact, if faccAt fcurWithin a range of about + - Δ, and fcurAmplitude Amp (f) in the window frequency domain of the current physiological signalcur) Compared with Amp (f)N) Is not particularly large (Th1 × Amp (f)cur)<Amp(fN)，Th1<0) Then f iscurMost likely caused by motion artifacts, in which case f will beNSetting as a new current heart rate frequency estimate; if f cannot be judgedcurWhether or not it is associated with motion artifacts, which will be further judged in a similar way as fNWhether or not it is possible to be caused by motion artifacts, if fNNearby occurrence of motion artifact frequency and Amp (f)N) Relatively small (Th2 × Amp (f)cur)>Amp(fN)，Th2<0) If f is considered to beNIs the result caused by motion artifacts and keeps the current estimate unchanged;
step 4.5 comparison of Amp (f)N) And Amp (f)cur) If Amp (f)N) Sufficiently large (Th3 × Amp (f)cur)<Amp(fN)，Th3<0) If f is considered to beNIs a new current heart rate frequency estimate.
Compared with the prior art, the invention has the beneficial effects that:
1) aiming at the application scene of the physiological signal with serious distortion under severe movement, the aim is to recover the waveform of the polluted physiological signal as much as possible.
2) The accurate estimation of the heart rate frequency is combined with the notch filtering, and a notch filter constructed by the heart rate frequency and the second harmonic thereof is used for filtering motion artifact components in the physiological signal and restoring the waveform due to the physiological signal.
Drawings
Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of non-limiting embodiments with reference to the following drawings:
FIG. 1 is a schematic diagram of the physiological signal motion artifact suppression device of the present invention;
FIG. 2 is a block diagram of an adaptive noise cancellation module of the present invention;
FIG. 3 is a flow chart of heart rate frequency correction of the present invention;
fig. 4 is an example of heart rate frequency correction of the present invention, wherein (a) the frequency spectrum of the previous time physiological signal window (b) the frequency spectrum of the current time physiological signal window (c) the frequency spectrum of the mean of the three acceleration signal windows.
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 the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
Examples
Referring to fig. 1, fig. 1 is a schematic diagram of the physiological signal motion artifact suppression device according to the present invention, and as shown in the figure, the physiological signal motion artifact suppression device includes:
the band-pass filtering module is used for filtering the physiological signals polluted by the motion pseudo-track and the three-axis acceleration signals and transmitting the physiological signals and the three-axis acceleration signals to the adaptive noise elimination module;
-an adaptive noise cancellation module for initially suppressing motion artifact components in the physiological signal and for initially estimating the heart rate frequency from the signal output by the band-pass filtering module;
a heart rate frequency correction module, which determines whether the preliminary estimation result is likely to be erroneous, corrects the likely erroneous result according to the estimation results at the current and previous time instants, and outputs a final estimation result of the heart rate frequency to the notch filter;
a notch filter for performing notch filtering on the estimation result of the physiological signal and subtracting the notch filtering result from the triaxial acceleration signal to obtain the final physiological signal with the motion artifact suppressed.
Specifically, the input is one or more paths of physiological signals (collected from sensors with similar positions) and a three-axis acceleration signal collected by a wearing part of the PPG sensor. The target is a time window with L sampling points, and L is generally an integral multiple of the signal sampling frequency FS. At regular intervals Itv, the algorithm performs heart rate frequency estimation and motion artifact suppression, Itv is typically an integer multiple of one second and is smaller than the time duration of the time window. In the actual operation of the algorithm, in order to ensure that the signal in the target window after the band-pass filtering is not distorted and the physiological signal in the target window after the self-adaptive filtering is completely converged, the signal with the number of sampling points LP before the target window is spliced with the target window and then used as the input of the algorithm, and the size of LP is close to that of L. When the algorithm is just started, zero padding is used for parts which do not exist in the segment signal. The user initially undergoes an initialization phase of 20 to 30 seconds during which the user should remain as still as possible without vigorous movements.
The input signal is first passed through a Band-pass Filter (BPF) of order 4 and passband 0.4 Hz-4.0 Hz to Filter out-of-Band noise. The passband of the filter is substantially the frequency range of the active components in the physiological signal. And then, a cascade type adaptive noise elimination is used for preliminarily inhibiting motion artifact components in the physiological signal, and the triaxial acceleration signal is a noise reference signal of ANC. After this step, motion artifacts in the physiological signal have been suppressed to a large extent, and the Heart Rate Frequency components have been highlighted in the spectrum, so fast fourier transform and a special Heart Rate Frequency Tracking (HRFT) mechanism are used to make a preliminary estimate of the Heart Rate Frequency by comparing amplitudes directly in the spectrum. After the preliminary estimation result is obtained, the algorithm judges whether the estimation result is possibly wrong through a Heart Rate Frequency Correction (HRFC) mechanism, corrects the possibly wrong result according to the estimation results of the current time and the previous time, and outputs the final estimation result of the Heart Rate Frequency. With this result, a series of notch filters with the heart rate frequency and its harmonic frequencies as notch frequencies are constructed, and the adaptive noise-canceling output physiological signal is subjected to notch filtering processing. Due to the nature of the notch filter itself, the output of the notch filter is substantially completely free of noise and motion artifacts of the PPG component. Therefore, the algorithm also subtracts the notch filtering result from the signal output by ANC to obtain the final physiological signal after motion artifact suppression.
Fig. 2 is a structural diagram of the adaptive noise canceling module of the present invention, and as shown in the drawing, at this stage, the one-path normalized physiological signal window or the average value of the multiple-path normalized physiological signal windows is used as the input of the contaminated signal, and the normalized triaxial acceleration signal is used to perform the cascade type adaptive noise canceling process on the physiological signal. The length of the signal window is LP+ L, but the final processing goal is a sub-window of length L at the back. The output of the first two stages of adaptive filtering is also input to the next stage after normalization processing. The normalization method used in this step is Z-score normalization (Z-score normalization), i.e., the mean value of the normalized data is 0 and the standard deviation is 1. The purpose of normalization is to eliminate situations where the adaptive filtering does not converge or converges slowly due to too large a difference in absolute amplitude between the physiological signal and the acceleration signal. By using the acceleration signals in the three directions to respectively reduce the noise of the physiological signal, the step can effectively inhibit the motion artifact noise caused by the motion in different directions in the physiological signal.
The LMS-Newton algorithm is used in the adaptive noise cancellation stage.
R-1(0)＝I
W(0)＝X(0)＝[0,0,...,0]T
X(k)＝[x(k),x(k-1),...,x(k-M+1)]T
e(k)＝d(k)-XT(k)W(k)
Figure BDA0002649121980000061
W(k+1)＝W(k)+2μe(k)R-1(k)X(k)
The parameter is a positive constant whose magnitude should generally be inversely proportional to the power of the input signal. The parameters α and μ are parameters that determine the convergence speed of the algorithm, α typically being a positive number much less than 1, and μ typically having a value around one-half of α. For the problem of physiological signal motion artifact suppression, d (k) represents the PPG input signal contaminated by motion artifacts, x (k) represents the acceleration signal, and e (k) is the noise-reduced physiological signal output. Vector w (k) is the tap coefficient of the FIR filter and M is the order of the filter. The matrix r (k) is an intermediate variable in the iteration of the algorithm.
The spectrum of a sub-window of length L behind the PPG output signal window is analyzed using FFT and a heart rate tracking mechanism is used to initially estimate the value of the heart rate frequency. In this process, frequency domain amplitude information is mainly considered. In order to increase the frequency domain resolution to obtain more accurate estimation results, the number of points L of the FFT is usually setFFTIs set to a value greater than the window length L and much greater than the sampling frequency FSThe value of (c).
The heart rate tracking mechanism came mainly from a paper published in 2017 by a.temko. If the inter-window interval Itv is small (e.g. 1-2s), the difference between the heart rates of adjacent windows should be small (below 5-10 bpm), so the main idea of heart rate tracking is to find the heart rate frequency estimate of the current window in the frequency domain of the current PPG time window in the vicinity of the heart rate frequency estimate of the previous time window.
And using Amp (f) to represent the amplitude corresponding to the frequency f in the spectrum of the physiological signal window. For the first time window at the beginning, the estimation result of the previous moment does not exist, and the frequency point with the maximum Amp (f) in the physiological signal spectrum is directly selected as the estimation value of the heart rate frequency. After this, the estimation result f of the last time window is obtainedpreRange of near + - Δ f (i.e., [ f ]pre–Δf,fpre+Δf]) Searching the frequency point with the maximum amplitude in the frequency spectrum of the current time window as the preliminary estimation result of the heart rate frequency at the current moment, and marking the result as fcur. When the algorithm starts to run, the algorithm goes through an initialization phase that the user keeps still, and the window number corresponding to the period of the initialization phase is marked as Ninit. The parameter Δ f is selected as follows:
Figure BDA0002649121980000071
for the ith time window, when i is less than or equal to NinitI.e. the algorithm is still in the initialization phase, Δ f is set to a constant Δ f0. When i is greater than NinitThen, Δ f is set as the sum of the maximum value of the absolute value of the difference between two previous successive estimation results and an offset value b.
After obtaining the preliminary estimation result of the heart rate frequency, a heart rate frequency correction mechanism is used to determine whether the current result is likely to be erroneous, and the result determined to be erroneous is corrected. The algorithm flow of the heart rate frequency correction step is shown in fig. 3. Firstly, whether the difference between the current preliminary estimation result and the estimation result of the previous time window is too large is judged. If the difference is greater than Th0, the next step is started, otherwise, the current estimation result is kept unchanged. If fcurGreater than fpreThen the current estimation result is likely to be an excessively large erroneous result, and the correct result is likely to be a result smaller than fcurOf (c) is detected. While the correct result is likely to be greater than f according to the trend of the changepre. Otherwise if fcurIs less than fpreThe correct result is likely to be a value less than fpreAnd is greater than fcurThe value of (c). In any event, the correct estimation result is likely to be at fcurAnd fpreIn the meantime. So next, in the spectrum of the current window, at fcurAnd fpreFind the frequency f corresponding to the maximum value of the frequency domain peakN. In practice, the range sought will generally be fpreSlightly extending in the direction of (a). If f is not presentNIf not, the current estimation result is kept unchanged, otherwise, the next step is carried out.
fNI.e. is likely to be the correct heart rate frequency estimate, but still requires further judgment. First, f is judgedcurWhether it is likely to be a frequency domain spike caused by motion artifacts. Three paths of acceleration signal windows are averaged, the frequency spectrum of the three paths of acceleration signal windows is calculated by using FFT (fast Fourier transform), and the frequency f corresponding to the maximum peak in the frequency spectrum is foundaccThis frequency is taken as a representation of the dominant frequency of the motion artifact. If faccIs located atfcurWithin a range of about + - Δ, and fcurAmplitude Amp (f) in the window frequency domain of the current physiological signalcur) Compared with Amp (f)N) Is not particularly large (Th1 × Amp (f)cur)<Amp(fN)，Th1<0) Then f iscurMost likely caused by motion artifacts, in which case f will beNSet to a new current heart rate frequency estimate. If f cannot be judgedcurWhether or not it is associated with motion artifacts, which will be further judged in a similar way as fNWhether or not it is likely to be caused by motion artifacts. If fNNearby occurrence of motion artifact frequency and Amp (f)N) Relatively small (Th2 × Amp (f)cur)>Amp(fN)，Th2<0) Will consider fNIs a result of motion artifacts and keeps the current estimate unchanged. After the above steps, the correctness of each frequency point cannot be continuously and effectively judged, and at this moment, Amp (f) is comparedN) And Amp (f)cur) If Amp (f)N) Sufficiently large (Th3 × Amp (f)cur)<Amp(fN)，Th3<0) If f is considered to beNIs a new current heart rate frequency estimate. Generally, there is Th1<Th3<Th2 and Th1 are around 0.5, and Th2 and Th3 are both relatively close to 1.
Obtaining an accurate estimated value f of the window heart rate frequency of the current physiological signalcurThen, using fcurTwo notch filters were constructed. The input of the notch filtering is a physiological signal window after adaptive noise elimination. The first notch filter having a notch frequency fcurThe notch frequency of the second filter is fcurSecond harmonic of (f)har. In the window spectrum of the physiological signal 2fcurFind the frequency corresponding to the peak with the maximum amplitude and consider it as fhar. In this embodiment, the bandwidth of the notch filter is set to about 0.4 Hz. It can be seen that after the physiological signal passes through two cascaded notch filters, almost all PPG original components are actually filtered, and the rest are motion artifacts and other noises, so that the result of notch filtering needs to be subtracted from the physiological signal input finally, and then the final physiological signal with motion artifacts suppressed is obtained.
Experiments show that the invention tries to recover the waveform of the polluted physiological signal as much as possible aiming at the application scene of the physiological signal with more serious distortion under more strenuous exercise. For a time window containing normal physiological signals, its main frequency domain components are distributed around its fundamental frequency (i.e. heart rate frequency) and harmonics of the fundamental frequency. Therefore, in theory, the notch filter or the comb filter can effectively filter the noise of the physiological signal and reserve the original components of the physiological signal. A corresponding notch filter is constructed for the heart rate frequency through a given physiological signal window and motion artifacts in the signal are filtered out. The difficulty of this process is actually an accurate estimate of the heart rate frequency. Adaptive noise cancellation has been shown to be effective in suppressing motion artifact components in physiological signals and emphasizing fundamental frequency components in the frequency domain.
The above description is only an embodiment of the present invention, and not intended to limit the scope of the present invention, and all equivalent structures or equivalent flow transformations that are made by using the contents of the specification and the drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims (7)

1. A physiological signal motion artifact suppression device, the device comprising:
the band-pass filtering module is used for filtering the physiological signals polluted by the motion pseudo-track and the three-axis acceleration signals and transmitting the physiological signals and the three-axis acceleration signals to the adaptive noise elimination module;
-an adaptive noise cancellation module for initially suppressing motion artifact components in the physiological signal and for initially estimating the heart rate frequency from the signal output by the band-pass filtering module;
a heart rate frequency correction module, which determines whether the preliminary estimation result is likely to be erroneous, corrects the likely erroneous result according to the estimation results at the current and previous time instants, and outputs a final estimation result of the heart rate frequency to the notch filter;
a notch filter for performing notch filtering on the estimation result of the physiological signal and subtracting the notch filtering result from the triaxial acceleration signal to obtain the final physiological signal with the motion artifact suppressed.
2. The physiological signal motion artifact suppression device of claim 1, wherein the three-axis acceleration signal is a noise reference signal of ANC.
3. The device for suppressing the motion artifacts of the physiological signals according to claim 1, wherein the adaptive noise elimination module takes the mean value of one or more normalized physiological signal windows as the input of the contaminated signals, and performs a cascade-type adaptive noise elimination process on the physiological signals by using the normalized three-axis acceleration signals.
4. The physiological signal motion artifact suppression device of claim 1, wherein the band-pass filtering module is a butterworth band-pass filter with a pass-band of 0.4 Hz-4.0 Hz
5. A method for suppressing motion artifacts of a physiological signal, comprising the steps of:
s1, filtering physiological signals and three-axis acceleration signals polluted by motion pseudo-tracks;
s2, carrying out cascade type self-adaptive noise elimination processing on the physiological signal by using the normalized triaxial acceleration signal;
s3, analyzing the frequency spectrum of a sub-window with the length L at the rear part of a PPG output signal window by using FFT (fast Fourier transform), and preliminarily estimating the value of heart rate frequency by using a heart rate tracking mechanism;
s4, after the preliminary estimation result of the heart rate frequency is obtained, judging whether the current result is possibly wrong or not by using a heart rate frequency correction mechanism, and correcting the result which is judged to be wrong;
s5, constructing two notch filters: the first notch filter having a notch frequency fcurThe notch frequency of the second filter is fcurSecond harmonic of (f)har(ii) a In the window spectrum of the physiological signal 2fcurTo find the peak corresponding to the peak with the largest amplitudeAnd is as fhar；
And subtracting the result of notch filtering from the physiological signal input to obtain the final physiological signal with the motion artifact restrained.
6. The method for suppressing the motion artifacts of the physiological signals according to claim 5, wherein the step S2 uses the normalized triaxial acceleration signals to perform a cascade-type adaptive noise reduction process on the physiological signals by using an LMS-Newton algorithm, specifically as follows:
R-1(0)＝I (3-1)
W(0)＝X(0)＝[0,0,...,0]T (3-2)
X(k)＝[x(k),x(k-1),...,x(k-M+1)]T (3-3)
e(k)＝d(k)-XT(k)W(k) (3-4)
Figure FDA0002649121970000021
W(k+1)＝W(k)+2μe(k)R-1(k)X(k) (3-6)
the parameter is a positive constant whose magnitude is generally inversely proportional to the power of the input signal, the parameters α and μ are parameters determining the convergence rate of the algorithm, α is generally a positive number far less than 1, μ has a value generally around one-half of α, d (k) represents the PPG input signal contaminated by motion artifacts, x (k) represents the acceleration signal, e (k) is the noise-reduced physiological signal output, the vector w (k) is the tap coefficient of the FIR filter, M is the order of the filter, and the matrix r (k) is an intermediate variable in the iteration of the algorithm.
7. The method for suppressing the motion artifacts of physiological signals according to claim 5, wherein in step S4. after obtaining the preliminary estimation result of the heart rate frequency, a heart rate frequency correction mechanism is used to determine whether the current result is possibly erroneous, and correct the result determined as erroneous, specifically as follows:
step 4.1, judging whether the difference between the current preliminary estimation result and the estimation result of the previous time window is too large: if the difference is greater than Th0, go to step 4.2, otherwise, keep the current estimation result unchanged;
step 4.2 if fcurGreater than fpreThen the current estimation result is likely to be an excessively large erroneous result, and the correct result is likely to be a result smaller than fcurThe frequency of (d); while the correct result is likely to be greater than f according to the trend of the changepre(ii) a Otherwise if fcurIs less than fpreThe correct result is likely to be a value less than fpreAnd is greater than fcurA value of (d);
step 4.3 in the spectrum of the current window, at fcurAnd fpreFind the frequency f corresponding to the maximum value of the frequency domain peakNThe search range will generally be fpreIn a direction slightly extending if f is not presentNThen the current estimation result is kept unchanged, otherwise step 4.4 is performed.
Step 4.4 judging fcurWhether it is likely to be a frequency domain spike caused by motion artifacts: averaging three paths of acceleration signal windows, calculating frequency spectrums of the three paths of acceleration signal windows by using FFT (fast Fourier transform), and finding out frequency f corresponding to the maximum peak in the frequency spectrumsaccThis frequency is taken as a representation of the dominant frequency of the motion artifact, if faccAt fcurWithin a range of about + - Δ, and fcurAmplitude Amp (f) in the window frequency domain of the current physiological signalcur) Compared with Amp (f)N) Is not particularly large (Th1 × Amp (f)cur)<Amp(fN)，Th1<0) Then f iscurMost likely caused by motion artifacts, in which case f will beNSetting as a new current heart rate frequency estimate; if f cannot be judgedcurWhether or not it is associated with motion artifacts, which will be further judged in a similar way as fNWhether or not it is possible to be caused by motion artifacts, if fNNearby occurrence of motion artifact frequency and Amp (f)N) Relatively small (Th2 × Amp (f)cur)>Amp(fN)，Th2<0) If f is considered to beNIs the result caused by motion artifacts and keeps the current estimate unchanged;
step 4.5 comparison of Amp (f)N) And Amp (f)cur) If Amp (f)N) Sufficiently large (Th3 × Amp (f)cur)<Amp(fN)，Th3<0) If f is considered to beNIs a new current heart rate frequency estimate.
CN202010864009.2A 2020-08-25 2020-08-25 Physiological signal motion artifact suppression device and method Pending CN111904406A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010864009.2A CN111904406A (en) 2020-08-25 2020-08-25 Physiological signal motion artifact suppression device and method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010864009.2A CN111904406A (en) 2020-08-25 2020-08-25 Physiological signal motion artifact suppression device and method

Publications (1)

Publication Number Publication Date
CN111904406A true CN111904406A (en) 2020-11-10

Family

ID=73279487

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010864009.2A Pending CN111904406A (en) 2020-08-25 2020-08-25 Physiological signal motion artifact suppression device and method

Country Status (1)

Country Link
CN (1) CN111904406A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114869294A (en) * 2022-05-05 2022-08-09 电子科技大学 Particle filter motion artifact suppression method based on VMD decomposition and LET model
CN116304564A (en) * 2023-02-23 2023-06-23 南京理工大学 Signal noise reduction method based on improved EEMD algorithm and autocorrelation noise reduction

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN205144547U (en) * 2015-11-29 2016-04-13 浙江师范大学 Motion noise eliminates system suitable for wearable heart rate measurement equipment
US20170071545A1 (en) * 2015-09-14 2017-03-16 Whoop, Inc. Removing motion artifacts from a physiological signal
CN106691425A (en) * 2016-12-30 2017-05-24 北京工业大学 Wrist heart rate monitoring method of sports bracelet
CN106880351A (en) * 2015-12-15 2017-06-23 德州仪器公司 Reduce the artifact of exercise induced in photo-plethysmographic (PPG) signal
CN110169764A (en) * 2019-05-06 2019-08-27 上海理工大学 A kind of LMS adaptive-filtering PPG signal heart rate extracting method
CN110432863A (en) * 2019-06-09 2019-11-12 中国人民解放军海军特色医学中心 A kind of vital sign parameter signals processing method

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170071545A1 (en) * 2015-09-14 2017-03-16 Whoop, Inc. Removing motion artifacts from a physiological signal
CN205144547U (en) * 2015-11-29 2016-04-13 浙江师范大学 Motion noise eliminates system suitable for wearable heart rate measurement equipment
CN106880351A (en) * 2015-12-15 2017-06-23 德州仪器公司 Reduce the artifact of exercise induced in photo-plethysmographic (PPG) signal
CN106691425A (en) * 2016-12-30 2017-05-24 北京工业大学 Wrist heart rate monitoring method of sports bracelet
CN110169764A (en) * 2019-05-06 2019-08-27 上海理工大学 A kind of LMS adaptive-filtering PPG signal heart rate extracting method
CN110432863A (en) * 2019-06-09 2019-11-12 中国人民解放军海军特色医学中心 A kind of vital sign parameter signals processing method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
QIRUI ZHANG ET AL.: "Motion Artifact Removal for PPG Signals based on Accurate Fundamental Frequency Estimation and Notch Filtering", 《 2018 40TH ANNUAL INTERNATIONAL CONFERENCE OF THE IEEE ENGINEERING IN MEDICINE AND BIOLOGY SOCIETY (EMBC)》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114869294A (en) * 2022-05-05 2022-08-09 电子科技大学 Particle filter motion artifact suppression method based on VMD decomposition and LET model
CN114869294B (en) * 2022-05-05 2023-05-30 电子科技大学 Particle filter motion artifact suppression method based on VMD decomposition and LET model
CN116304564A (en) * 2023-02-23 2023-06-23 南京理工大学 Signal noise reduction method based on improved EEMD algorithm and autocorrelation noise reduction
CN116304564B (en) * 2023-02-23 2023-10-31 南京理工大学 Signal noise reduction method based on improved EEMD algorithm and autocorrelation noise reduction

Similar Documents

Publication Publication Date Title
Jane et al. Adaptive baseline wander removal in the ECG: Comparative analysis with cubic spline technique
US6094592A (en) Methods and apparatus for estimating a physiological parameter using transforms
Zidelmal et al. QRS detection using S-Transform and Shannon energy
Sameni et al. A nonlinear Bayesian filtering framework for ECG denoising
US7039538B2 (en) Pulse oximeter with separate ensemble averaging for oxygen saturation and heart rate
Chen et al. Removal of power line interference from ECG signals using adaptive notch filters of sharp resolution
KR101983037B1 (en) An apparatus for detecting r-peak in electrocardiogram signal by using wavelet transform and modified shannon energy envelope and the method thereof
CN111904406A (en) Physiological signal motion artifact suppression device and method
US9986951B1 (en) Low-distortion ECG denoising
JPH0751237A (en) Method and device for raising signal/noise ratio of ecg signal
Majeed et al. Motion artifact removal of photoplethysmogram (PPG) signal
Dai et al. Removal of baseline wander from dynamic electrocardiogram signals
Ram et al. On the performance of AS-LMS based adaptive filter for reduction of motion artifacts from PPG signals
Rahman et al. Baseline wander and power line interference elimination from cardiac signals using error nonlinearity LMS algorithm
CN110269608B (en) Method and device for removing signal interference and readable storage medium
Lin et al. Removal of power line interference in EEG signals with spike noise based on robust adaptive filter
TW201801671A (en) Method of heart rate detection
Mishra et al. ECG signal denoising using time-frequency based filtering approach
KR101833254B1 (en) Device for processing bio-signal
Ahamed et al. An efficient method for heart rate monitoring using wrist-type photoplethysmographic signals during intensive physical exercise
CN113040784B (en) Myoelectric noise filtering method for electrocardiosignal
CN114795241A (en) Electrocardiosignal noise detection method and system
EP2938247B1 (en) Method and apparatus for reducing motion artifacts in ecg signals
Ulukaya et al. Performance comparison of wavelet based denoising methods on discontinuous adventitious lung sounds
Bodile et al. Ecg denoising using cubature kalman filter framework

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20201110

WD01 Invention patent application deemed withdrawn after publication