CN104739399A - Method for detecting movement offset in pulse wave heart rate calculation - Google Patents

Abstract

The invention provides a method for detecting movement offset in pulse wave heart rate calculation. The method includes the following steps that a photoelectric volume method is utilized to obtain pulse wave signals x(n) through measurement, and another kind of wave length or an acceleration sensor is utilized to obtain movement signals y(n); signals of a certina time length are intercepted, fast fourier transform is utilized to obtain the fourier transform X(k) and Y(k) of the two types of signals, smoothing is conducted on the fourier spectrum of the movement signals, and estimation Pyk of movement interference with the fourier spectrum is obtained; the following operation is conducted on the fourier spectrum of the pulse signals and the normalized movement fourier spectrum, the fourier spectrum of the pulse signals generated after first weighing is conducted is obtained, secondary power spectrum weighing is conducted through a previous heart rate value, it is assumed that the frequency point value corresponding to the previous heart rate is k0, and a weighing curve is built through a Gaussian function.

Description

A kind of method detecting balancing out motions in the calculating of pulse wave heart rate

Technical field

The present invention relates to one to fight ripple heart rate computational methods, particularly relate to a kind of pulse wave heart rate calculate in the method for balancing out motions.

Background technology

Photoplethymograph utilizes blood to the principle of absorption of light, detects the situation of change of blood volume in blood vessel, obtains the pulse wave of user, thus realizes heart rate and the isoparametric measurement of blood oxygen.But the pulse wave utilizing photoplethymograph to obtain is easy to the interference of being moved, thus cause the significant error of pulse wave measurement.Manyly research and propose employing one independently motor message Measurement channel, realize the counteracting of motion artifacts in pulse wave.

Utilize the method for dual wavelength to realize the counteracting of motion, this sensor has the light source of green LED and infrared LED two kinds of wavelength.Wherein green LED is used for measuring pulse wave signal, and infrared LED is used for measuring motor message.After fast fourier transform is carried out to pulse wave signal and motor message, standardization is carried out to fourier spectrometer.Then the spectrum deducting motor message from the spectrum of pulse wave signal from but realize the removal of motion artifacts, and calculate the pulse of people.But in the method, the intensity of spectrum must be good fit to allow the motion effects removed completely after the subtraction of the fourier spectrometer of light wave, and this is a kind of defect, when the method is measured in addition, observes high power consumption.Which constitute another kind of defect.

Patent CN102008299B proposes to adopt spectrum coherent function to realize the removal of motion artifacts.But in the method, the fourier spectrometer of calculation requirement to multistage of coherent function is averaged, when requiring higher heart rate precision, the signal of long period is needed to carry out computing.Can not meet and detect changes in heart rate fast.In addition the method requires within a certain period of time, and motor message is comparatively steady, and therefore the method is when the nonstationary interference caused for some random motions is offset, and its effect is poor.

Pertinent literature adopts acceleration transducer to measure motor message, utilizes sef-adapting filter to realize balancing out motions simultaneously.But the motion artifacts in the motor message of acceleration analysis and pulse wave differs greatly, adaptive cancellation method there is no good performance.

Summary of the invention

Target of the present invention is a kind of method by power spectrum weighting, overcomes the defect of above mentioned prior art, better reduces motion artifacts to the impact of heart rate measurement.

Technical scheme of the present invention is as follows: a kind of method of balancing out motions during pulse wave heart rate calculates, comprises the following steps:

1) utilize photoplethymograph to measure pulse wave signal x (n), utilize another wavelength or acceleration transducer to obtain motor message y (n);

2) signal of certain hour length N is intercepted, fast fourier transform is utilized to obtain Fourier transform X (k) and the Y (k) of two kinds of signals respectively, the Fourier transform that wherein X (k)=FFT (X) is pulse wave signal, the Fourier transform that Y (k)=FFT (Y) is motor message, subsequently following computing is carried out to Fourier transform, obtain normalized fourier spectra Pxk, Py (k) of two signals, the formula adopted is as follows:

Pxk=Xk2Xk2，k=0，1…N/2

Pyk=Yk2Yk2，k=0，1…N/2

3) to the smoothing process of the fourier spectra of motor message, the estimation Pyk to motion artifacts fourier spectra is obtained;

Pyk＝12*mn=-mn＝m-1Pyk+n，k=0，1…N/2

4) the motion fourier spectra after the fourier spectra of pulse signal and normalization is carried out computing as follows, obtain the fourier spectra Px_weight_1 of the pulse signal after first time weighting

Pxweight_1(k)=Pyk*Pxk，k=0，1…N/2

5) before utilization, the value of heart rate carries out the weighting of second time power spectrum, and the frequency point value before supposing corresponding to heart rate is k0, utilizes Gaussian function to build a weighted curve Wk:

Wk=e-(k-k0)2δ2，k=0，1…N/2；6

6) the weighted spectral curve finally obtained is:

Pxweight_2(k)=Pxweight_1(k)*Wk，k=0，1…N/2

7) eventually through the peak point selected in Pxweight_2 (k) curve as final heart rate value, after utilizing said method to obtain heart rate value to every one piece of data, the change of heart rate value can be drawn as a curve, show the changes in heart rate in motor process.

Accompanying drawing explanation

Fig. 1 be the method realize schematic diagram.

Fig. 2 is disturbed schematic diagram by pulse wave signal passive movement signal

Fig. 3 is the fourier spectra schematic diagram of pulse wave signal and motor message

Fig. 4 is the fourier spectra after weighting

Fig. 5 is the heart rate curve utilizing this method to obtain and the heart rate curve comparison obtained based on electro-cardiologic methods

Detailed description of the invention

Below in conjunction with accompanying drawing, the present invention is described in further detail.

What Fig. 1 indicated the method realizes block diagram, there are two input signals, one is pulse wave signal x (n) utilizing photoplethymograph measurement to obtain, and one is motor message y (n) utilizing another wavelength or acceleration transducer to obtain.Fig. 2 shows one piece of data, and solid line is pulse wave signal, and dotted line is motor message.Can see that pulse wave signal passive movement signal disturbed, cause the information of beat pulse to be overshadowed in completely in motion artifacts.

(1) intercept the signal of certain hour length N, utilize fast fourier transform to obtain Fourier transform X (k) and the Y (k) of two kinds of signals respectively.The Fourier transform that wherein X (k)=FFT (X) is pulse wave signal, the Fourier transform that Y (k)=FFT (Y) is motor message; We carry out following computing to Fourier transform subsequently, and obtain normalized fourier spectra Pxk, Py (k) of two signals, the formula adopted is as follows:

Pxk=Xk2Xk2，k=0，1…N/2

Pyk=Yk2Yk2，k=0，1…N/2

Consider the requirement of heart rate precision, general selection data length is 16-32s.The heart rate precision of corresponding 3.75bpm and 1.875bpm respectively.Fig. 2 shows the normalization fourier spectrometer of two signals.As can be seen from the figure, the fourier spectra of pulse wave has two peak values in 80-120 beats/min and 110-130 heart beating place per minute, and the fourier spectra of motor message also has a peak value within the scope of 80-120 beats/min.

(2) we, to the smoothing process of the fourier spectra of motor message, obtain the estimation Pyk to motion artifacts fourier spectra subsequently.

Pyk=12*mn=-mn=m-1Pyk+n，k=0，1…N/2

(3) the motion fourier spectra after the fourier spectra of pulse signal and normalization is carried out computing as follows, obtain the fourier spectra Px_weight_1 of the pulse signal after first time weighting.Fig. 3 shows the first time weighted Fourier spectrum of signal shown in Fig. 1.Can see, in the fourier spectra after computing, motion composition is cancelled, and the spectrum corresponding to heart rate can show preferably.

Pxweight_1(k)=Pyk*Pxk，k=0，1…N/2

(4) based on following practical situation: the change of heart rate value is generally relatively more stable, there will not be unexpected violent change.Therefore we can utilize before the value of heart rate carry out the weighting of second time power spectrum.Frequency point value before supposing corresponding to heart rate is k0.We utilize Gaussian function to build a weighted curve Wk

Wk=e-(k-k0)2δ2，k=0，1…N/2

(5) the weighted spectral curve finally obtained is

Pxweight_2(k)=Pxweight_1(k)*Wk，k=0，1…N/2

(6) eventually through the peak point selected in Pxweight_2 (k) curve as final heart rate value.After utilizing said method to obtain heart rate value to every one piece of data, the change of heart rate value can be drawn as a curve, show the changes in heart rate in motor process.Fig. 4 shows in a certain subject motion's process, and the heart rate curve adopting the method for the invention to obtain and the contrast of the standardized heart rate utilizing electrocardiogram to obtain, can find out that this method can be good at the changes in heart rate in tracing movement process.

Should be understood that, for those of ordinary skills, can be improved according to the above description or convert, and all these improve and convert the protection domain that all should belong to claims of the present invention.Above-described embodiment is only for illustrating technical conceive of the present invention and feature; its object is to allow the personage being familiar with this art can understand content of the present invention and be implemented; can not limit the scope of the invention with this; all equivalences done according to spirit of the present invention change or modify; effect is similar, all should be encompassed in protection scope of the present invention.

Claims (1)

1. detect a method for balancing out motions in the calculating of pulse wave heart rate, it is characterized in that, comprise the following steps:

1) utilize photoplethymograph to measure pulse wave signal x (n), utilize another wavelength or acceleration transducer to obtain motor message y (n);

2) signal of certain hour length N is intercepted, fast fourier transform is utilized to obtain Fourier transform X (k) and the Y (k) of two kinds of signals respectively, the Fourier transform that wherein X (k)=FFT (X) is pulse wave signal, the Fourier transform that Y (k)=FFT (Y) is motor message, subsequently following computing is carried out to Fourier transform, obtain normalized fourier spectra Pxk, Py (k) of two signals, the formula adopted is as follows:

Pxk＝Xk2Xk2，k＝0，1…N/2

Pyk＝Yk2Yk2，k＝0，1…N/2

3) to the smoothing process of the fourier spectra of motor message, the estimation Pyk to motion artifacts fourier spectra is obtained;

Pyk＝12*mn＝-mn＝m-1Pyk+n，k＝0，1…N/2

4) the motion fourier spectra after the fourier spectra of pulse signal and normalization is carried out computing as follows, obtain the fourier spectra Px_weight_1 of the pulse signal after first time weighting

Pxweight_1(k)＝Pyk*Pxk，k＝0，1…N/2

5) before utilization, the value of heart rate carries out the weighting of second time power spectrum, and the frequency point value before supposing corresponding to heart rate is k0, utilizes Gaussian function to build a weighted curve Wk:

Wk＝e-(k-k0)2δ2， k＝0，1…N/2；6

6) the weighted spectral curve finally obtained is:

Pxweight_2(k)＝Pxweight_1(k)*Wk，k＝0，1…N/2

7) eventually through the peak point selected in Pxweight_2 (k) curve as final heart rate value, after utilizing said method to obtain heart rate value to every one piece of data, the change of heart rate value can be drawn as a curve, show the changes in heart rate in motor process.