CN104665875A - Ultrasonic Doppler envelope and heart rate detection method - Google Patents

Abstract

The invention relates to an ultrasonic Doppler envelope and heart rate detection method which comprises the following steps: constructing an adaptive model for an acquired ultrasonic signal; carrying out Kalman tracking filtering on the ultrasonic signal in the constructed adaptive model; carrying out wavelet transformation on the ultrasonic signal subjected to kalman tracking filtering to finish detection of the Doppler envelope and the heart rate in the ultrasonic signal. According to the method, the Doppler envelope and the heart rate can be directly extracted from an original signal; double frequency interference can be overcome, the robustness is high, and the speed is high.

Description

Ultrasonic doppler envelope and heart rate detection method

Technical field

The present invention relates to a kind of ultrasonic doppler envelope and heart rate detection method.

Background technology

At present, the method realizing envelope detected and heart rate calculating mainly contains Hilbert-Huang transform method, mathematical mor-phology method, average shannon energy method.

But Hilbert-Huang transform method extracts envelope can produce a large amount of burr, the envelope curve obtained is level and smooth not, thus impact treatment and analysis below.Although adopt mathematical mor-phology to extract the method for envelope simply, but the quality of its performance is relevant to structural element, and the experience of the selective dependency algorithm author of structural element, and the envelope curve peak value that profit obtains in this way is not obvious, cannot reflect accurate location.Average shannon energy method then needs the average energy of signal calculated within a period of time, and this can cause the time-domain position of signal to offset, and therefore utilizes shannon energy method to extract temporal signatures and can there is larger error.

Summary of the invention

In view of this, be necessary to provide a kind of ultrasonic doppler envelope and heart rate detection method.

The invention provides a kind of ultrasonic doppler envelope and heart rate detection method, comprise the steps: that a. is that the ultrasonic signal obtained sets up adaptive model; B. in the adaptive model of above-mentioned foundation, Kalman tracking filter is carried out to described ultrasonic signal; C. wavelet transformation is carried out to the ultrasonic signal after described Kalman tracking filter, complete the detection to the Doppler's envelope in described ultrasonic signal and heart rate.

Adaptive model in described step a is: $\left\{\begin{array}{c}{y}_{k+1}=A_k{y}_k+B_k{u}_k+{\mathrm{\Γ}}_k{\mathrm{\ξ}}_k\\ w_k=C_k{y}_k+D_k{u}_k+{\mathrm{\η}}_k\end{array}\right.,$

Wherein, A _k, B _k, Γ _k, C _k, D _kknown constant value matrix respectively, ξ _k, η _ksystem and the observation noise sequence of the statistical information such as known average, variance and covariance respectively.

Described step b specifically comprises: adopt formula $\left\{\begin{array}{c}\widehat{x_{k|k}}=\widehat{x_{k|k-1}}+G_k(v_k-C_k\widehat{x_{k|k-1}})\\ \widehat{x_{k|k-1}}=A_{k-1}\widehat{x_{k-1|k-1}}\end{array}\right.$ Real-time tracking calculates x ^k, wherein, G _kfor kalman gain matrix; Adopt formula $\left\{\begin{array}{c}{P}_{\mathrm{0,0}}=\mathrm{Var}\left(x_0\right)\\ P_{k,k-1}=A_{k-1}{P}_{k-1,k-1}{A}_{k-1}^T+{\mathrm{\Γ}}_{k-1}{Q}_{k-1}{\mathrm{\Γ}}_{k-1}^T\\ G_k=P_{k,k-1}{C}_k^T{(C_k{P}_{k,k-1}{C}_k^T+R_k)}^{-1}\\ P_{k,k}=(1-G_k{C}_k)P_{k,k-1}\\ \widehat{x_{0|0}}=E\left(x_0\right)\\ \widehat{x_{k|k-1}}=A_{k-1}\widehat{x_{k-1|k-1}}+B_{k-1}{u}_{k-1}\\ \widehat{x_{k|k}}=\widehat{x_{k|k-1}}+G_k(v_k-D_k{u}_k-C_k\widehat{x_{k|k-1}})\\ k=\mathrm{1,2},L\end{array}\right.$ Carry out Kalman tracking filter, wherein x is the noisy signal of band obtained.

Described step c specifically comprises: utilize formula carry out small echo integral transformation, wherein for wavelet function; Utilize formula $x(i,n)=\underset{k=-\∞}{\overset{\∞}{\mathrm{\Σ}}}h(2k-n)x_L(i-1,k)+\underset{k=-\∞}{\overset{\∞}{\mathrm{\Σ}}}g(2k-n)x_H(i-1,k)$ Carry out wavelet transform.

A kind of ultrasonic doppler envelope of the present invention and heart rate detection method, can Doppler's envelope in extracting directly ultrasonic signal and heart rate.Through a large amount of data tests, find that accuracy of the present invention is very high, in very large noise circumstance, correct result can both be detected, and do not need any priori knowing signal, also not need to set threshold value, the most valuable, the present invention does not disturb by so-called double frequency.And double frequency interference is the difficult point in ultrasonic continuous wave Doppler heart rate detection.And the envelope curve that the present invention obtains is level and smooth, peak point position is given prominence to, for subsequent analysis process provides great convenience.

Accompanying drawing explanation

Fig. 1 is the flow chart of ultrasonic doppler envelope of the present invention and heart rate detection method.

Fig. 2 is the embodiment of the present invention to 20 seconds result schematic diagrams with noisy sample of signal.

Detailed description of the invention

Below in conjunction with drawings and the specific embodiments, the present invention is further detailed explanation.

Consulting shown in Fig. 1, is the operation process chart of ultrasonic doppler envelope of the present invention and heart rate detection method preferred embodiment.

Step S401, for the ultrasonic signal obtained sets up adaptive model.Specifically:

First, set up a model, consider a linear system in state space description:

$\left\{\begin{array}{c}{y}_{k+1}=A_k{y}_k+B_k{u}_k+{\mathrm{\Γ}}_k{\mathrm{\ξ}}_k\\ w_k=C_k{y}_k+D_k{u}_k+{\mathrm{\η}}_k\end{array}\right.---\left(1\right)$

In formula, A _k, B _k, Γ _k, C _k, D _kknown constant value matrix respectively, ξ _k, η _kbeing system and the observation noise sequence of the statistical information such as known average, variance and covariance respectively, is unknown.Wherein, ξ is supposed _k, η _kzero mean Gaussian white noise sequence, at original state x ₀with ξ _k, η _kindependent.

Determining x ^koptimal estimation time, optimality is that the minimum variance estimate under the method for least square meaning by selecting best initial weights matrix to provide obtains.As j=k, definition as j<k, definition for x _koptimal estimation; As j>k, definition for x _klevel and smooth estimation.Wherein j=0,1, L k.The present embodiment discusses the situation of j=k.

Step S402, carries out Kalman tracking filter to described ultrasonic signal in the adaptive model of above-mentioned foundation.

Following formula real-time tracking is adopted to calculate x ^k:

$\left\{\begin{array}{c}\widehat{x_{k|k}}=\widehat{x_{k|k-1}}+G_k(v_k-C_k\widehat{x_{k|k-1}})\\ \widehat{x_{k|k-1}}=A_{k-1}\widehat{x_{k-1|k-1}}\end{array}\right.---\left(2\right)$

In formula, G _kfor kalman gain matrix.

Starting point initial estimation is because original state x ⁰unbiased esti-mator, can use constant value vector and in the Kalman filtering of reality, G _kalso necessary recurrence calculation, these two processes are exactly Kalman filtering process altogether.Be summarized as follows:

$\left\{\begin{array}{c}{P}_{\mathrm{0,0}}=\mathrm{Var}\left(x_0\right)\\ P_{k,k-1}=A_{k-1}{P}_{k-1,k-1}{A}_{k-1}^T+{\mathrm{\&Gamma;}}_{k-1}{Q}_{k-1}{\mathrm{\&Gamma;}}_{k-1}^T\\ G_k=P_{k,k-1}{C}_k^T{(C_k{P}_{k,k-1}{C}_k^T+R_k)}^{-1}\\ P_{k,k}=(1-G_k{C}_k)P_{k,k-1}\\ \widehat{x_{0|0}}=E\left(x_0\right)\\ \widehat{x_{k|k-1}}=A_{k-1}\widehat{x_{k-1|k-1}}+B_{k-1}{u}_{k-1}\\ \widehat{x_{k|k}}=\widehat{x_{k|k-1}}+G_k(v_k-D_k{u}_k-C_k\widehat{x_{k|k-1}})\\ k=\mathrm{1,2},L\end{array}\right.---\left(3\right)$

Wherein, x is the noisy signal of band obtained.

Step S403, carries out wavelet transformation to the ultrasonic signal after described Kalman tracking filter, completes the detection to the Doppler's envelope in described ultrasonic signal and heart rate.

Small echo, by two shirtsleeve operations " translation " and " stretching ", is realized by " wavelet basis " family of functions.If be a wavelet basis function, in conjunction with flexible constant a and translation constant b, obtaining a series of form is wavelet function.With a kind of integral transformation for integral kernel defines of described wavelet function, be called small echo integral transformation:

Wavelet transform is as follows:

$x(i,n)=\underset{k=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}h(2k-n)x_L(i-1,k)+\underset{k=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}g(2k-n)x_H(i-1,k)---\left(5\right)$

In the embodiment of the present invention 20 seconds with the data of random noise after small echo Kalman filtering, just time-domain signal can be converted to frequency-region signal, obtain the large small component of frequency, as shown in Figure 2.

The present invention utilizes modern signal processing, directly can extract Doppler's envelope and heart rate from primary signal, any priori of undesired signal, does not also need to set any threshold value; , adopt Kalman adaptive technique, smooth envelope curve can be obtained, and the interference of double frequency can be overcome well; Adopt adaptive tracing technology, robustness is good, and the segment signal specially picking noise in experiment larger is verified, the envelope curve obtained still unusual light, and heart rate result of calculation is very accurate; Under PC platform (AMD Athlon II X4635), computational speed is quickly in application the present invention, about 0.25 second of the processing time of 20 number of seconds certificates, therefore in real-time embedded system, current MCU can be competent at completely, LCD is coordinated to show again, so just can well by the present invention's marketization.

Although the present invention is described with reference to current better embodiment; but those skilled in the art will be understood that; above-mentioned better embodiment is only used for the present invention is described; not be used for limiting protection scope of the present invention; any within the spirit and principles in the present invention scope; any modification of doing, equivalence replacement, improvement etc., all should be included within the scope of the present invention.

Claims (4)

1. ultrasonic doppler envelope and a heart rate detection method, is characterized in that, the method comprises the steps:

A. the ultrasonic signal for obtaining sets up adaptive model;

B. in the adaptive model of above-mentioned foundation, Kalman tracking filter is carried out to described ultrasonic signal;

C. wavelet transformation is carried out to the ultrasonic signal after described Kalman tracking filter, complete the detection to the Doppler's envelope in described ultrasonic signal and heart rate.

2. the method for claim 1, is characterized in that, the adaptive model in described step a is:

$\left\{\begin{array}{c}{y}_{k+1}=A_k{y}_k+B_k{u}_k+{\mathrm{\&Gamma;}}_k{\mathrm{\&xi;}}_k\\ w_k=C_k{y}_k+D_k{u}_k+{\mathrm{\&eta;}}_k\end{array}\right.,$

Wherein, A _k, B _k, Γ _k, C _k, D _kknown constant value matrix respectively, ξ _k, η _ksystem and the observation noise sequence of the statistical information such as known average, variance and covariance respectively.

3. the method for claim 1, is characterized in that, described step b specifically comprises:

Adopt formula $\left\{\begin{array}{c}\widehat{x_{k|k}}=\widehat{x_{k|k-1}}+G_k(v_k-G_k\widehat{x_{k|k-1}})\\ \widehat{x_{k|k-1}}=A_{k-1}\widehat{x_{k-1|k-1}}\end{array}\right.$ Real-time tracking calculates x _k, wherein, G _kfor kalman gain matrix;

Adopt formula $\left\{\begin{array}{c}{P}_{\mathrm{0,0}}=\mathrm{Var}\left(x_0\right)\\ P_{k,k-1}=A_{k-1}{P}_{k-1,k-1}{A}_{k-1}^T+{\mathrm{\&Gamma;}}_{k-1}{Q}_{k-1}{\mathrm{\&Gamma;}}_{k-1}^T\\ G_k=P_{k,k-1}{C}_k^T{(C_k{P}_{k,k-1}{C}_k^T+R_k)}^{-1}\\ P_{k,k}=(1-G_k{C}_k)P_{k,k-1}\\ \widehat{x_{0|0}}=E\left(x_0\right)\\ \widehat{x_{k|k-1}=A_{k-1}\hat{x_{k-1|k-1}}}+B_{k-1}{u}_{k-1}\\ \widehat{x_{k|k}}=\widehat{x_{k|k-1}}+G_k(v_k-D_k{u}_k-C_k\widehat{x_{k|k-1}})\\ k=\mathrm{1,2},L\end{array}\right.$ Carry out Kalman tracking filter, wherein x is the noisy signal of band obtained.

4. the method for claim 1, is characterized in that, described step c specifically comprises:

Utilize formula carry out small echo integral transformation, wherein for wavelet function;

Utilize formula $x(i,n)=\underset{k=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}h(2k-n)x_L(i-1,k)+\underset{k=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}g(2k-n)x_H(i-1,k)$ Carry out wavelet transform.