CN100589759C - Saturation oxygen detection method - Google Patents

Abstract

The invention discloses a method for detecting blood oxygen saturation. The method comprises the following steps: pulse wave signal is acquired, and red light DC signal(rdc), red light AC signal(rac),near infrared light DC signal(irdc) and near infrared light AC signal(irac) are separated from the acquired pulse wave signal; a plurality of segments of local pulse wave having the same initial andending phases are separated from the rdc, rac, irdc and irac; according to the pulse wave data on two segments or multiple segments of local pulse wave adjacent to the local pulse wave, the local pulse wave is corrected; and according to the corrected local pulse waves of the pulse wave signals, the blood oxygen saturation is worked out. Due to the adoption of the method, the interference of moving interfering signal which is coincided with the pulse wave signal frequency band can be eliminated, and no hardware needs to be added.

Description

A kind of method for detecting blood oxygen saturation

Technical field

The present invention relates to biomedical engineering field, relate in particular to a kind of method for detecting blood oxygen saturation.

Background technology

Blood oxygen saturation is that the white concentration of Oxygenated blood red eggs accounts for total ratio in the human body, and it has reflected the cardio-pulmonary function situation of human body.Yet, at blood oxygen detection device in actual use,, thereby influence the pulse wave signal quality that instrument records, the certainty of measurement of reduction blood oxygen saturation because physiology actions such as tester's breathings, cough can make and be tried the position and move.Eliminate during at present common detection blood oxygen saturation and tried the position motion interferential method that produces and be:

Adopt pick off to detect the motion artifacts signal, offset motion artifacts by minimum mean square error criterion (LMS) or minimum error sum-of-squares criterion adaptive filter algorithms such as (RLS) again; The real-time of the method counteracting motion artifacts is better, but needs extra increase pick off, and calculation of complex, needs hardware to have stronger data-handling capacity, has increased the checkout equipment cost.

Summary of the invention

In view of this, main purpose of the present invention is to provide a kind of method for detecting blood oxygen saturation, can eliminate the interference with the motion artifacts signal of pulse wave signal band overlapping part, and not need additionally to increase hardware.

For achieving the above object, technical scheme of the present invention is achieved in that

A kind of method for detecting blood oxygen saturation, this method comprises:

A, collection pulse wave signal, and from the pulse wave signal of being gathered, isolate HONGGUANG direct current signal, HONGGUANG AC signal, near infrared light direct current signal and near infrared light AC signal;

B, from HONGGUANG direct current signal, HONGGUANG AC signal, near infrared light direct current signal and near infrared light AC signal, respectively isolate the local pulse wave of multistage; The start-phase of the described local pulse wave of each section of each road pulse wave signal is identical, and termination phase is also identical;

C, with the local pulse wave of each section of each road pulse wave signal respectively as pulse wave to be revised, according to waiting to revise pulse wave data on pulse wave two neighboring sections or the local pulse wave of multistage, revise the pulse wave data that this waits to revise pulse wave with this;

D, calculate blood oxygen saturation according to the revised local pulse wave of each road pulse wave signal.

Further, step B specifically comprises:

Intercept a segment signal respectively from HONGGUANG direct current signal, HONGGUANG AC signal, near infrared light direct current signal, near infrared light AC signal, each segment signal that is intercepted is the same segment signal of the corresponding pulse wave signal of being gathered all;

Determine the pairing collection point of each trough of institute's intercept signal from the near infrared light AC signal; The institute pairing collection point of each trough of intercept signal is the institute pairing collection point of each trough of intercept signal from HONGGUANG direct current signal, HONGGUANG AC signal, near infrared light direct current signal from the near infrared light AC signal;

With the pairing sampled point of each trough as cut-point, signal between every adjacent two cut-points will become the local pulse wave of multistage from the signal segmentation that near infrared light AC signal, HONGGUANG direct current signal, HONGGUANG AC signal and near infrared light direct current signal are intercepted as one section local pulse wave.

Further, describedly determine that the pairing collection point of each trough of institute's intercept signal comprises from the near infrared light AC signal:

If x (n) is the pulse wave data of this section institute's intercept signal from the near infrared light AC signal of sampled point n correspondence, and x (n) k the sampled point of being separated by got first difference and obtain:

By as minor function:

Obtain each turning point from the upstroke to the decent of signal of being intercepted, i.e. the S of non-zero ₂(n); Wherein, H is S ₁The maximum of L sampled point institute corresponding data (n) before the current sampling point n, and the value of L should guarantee S ₁(i) | _{I=n-L, n-L+1 ..., n-1}In comprise the trailing edge of the pulse wave in a heart rate cycle at least;

From the signal that is intercepted, get the minima between adjacent two turning points, promptly obtain the trough of the signal that intercepted.

Further, step C specifically comprises:

C1, get the local pulse wave of each section of one road pulse wave signal;

C2, with one section local pulse wave of this road pulse wave signal as pulse wave to be revised;

C3, be taken at pulse wave to be revised that a sampled point gathers and wait to revise the pulse wave data of the local pulse wave of the adjacent 2m section of pulse wave with this, the pulse wave data of being taken out are got intermediate value, substitute the pulse wave data of waiting to revise pulse wave that this sampled point is gathered with this intermediate value; Wherein, m is a numerical value;

C4, repeated execution of steps c3 finish substituting of all the other sampled values of waiting to revise pulse wave;

C5, return step c2, up to the correction of the local pulse wave of each section of finishing this road pulse wave signal;

C6, return step c1, revise the local pulse wave of each section of its excess-three road pulse wave signal respectively.

Further, step B further comprises:

B1, the average sample of getting the local pulse wave of described multistage of one road pulse wave signal are counted, and adopt the match interpolation to come the local pulse wave of each section is carried out the data polishing, make that the sampling number of the local pulse wave of each section is identical;

B2, with the pulse wave data matrix composed as follows of the local pulse wave of each section:

$P_{N\×M}=\left[\begin{array}{c}{\mathrm{Seg}}_1\left(1\right),{\mathrm{Seg}}_2\left(1\right)........{\mathrm{Seg}}_i\left(1\right).....{\mathrm{Seg}}_M\left(1\right)\\ {\mathrm{Seg}}_1\left(2\right),{\mathrm{Seg}}_2\left(2\right)........{\mathrm{Seg}}_i\left(2\right).....{\mathrm{Seg}}_M\left(2\right)\\ {\mathrm{Seg}}_1\left(3\right),{\mathrm{Seg}}_2\left(3\right)........{\mathrm{Seg}}_i\left(3\right).....{\mathrm{Seg}}_M\left(3\right)\\ .\\ .\\ {\mathrm{Seg}}_1\left(N\right),{\mathrm{Seg}}_2\left(N\right)........{\mathrm{Seg}}_i\left(N\right).....{\mathrm{Seg}}_M\left(N\right)\end{array}\right]$

Wherein, from this road pulse wave signal, isolate the local pulse wave of M section, every section corresponding N sampled point of local pulse wave, Seg _M(N) be the pulse wave data on the local pulse wave of M section of N sampled point correspondence;

B3, return step b1, the local pulse wave of its excess-three road pulse wave signal is carried out the data polishing.

Further, the data that the capable j of i is listed as in the P matrix after step C revises are expressed as:

P(i，j)＝Seg′ _j(i)＝Med{Seg _(j-m)(i)，...Seg _j(i)...，Seg _(j+m)(i)}，i＝1，2...N，j＝(m+1)...(M-m)

P (i, j)=Seg ' _j(i)=Seg _j(i), i=1,2...N, j=1,2...m or j=(M-m+1) ... M

Wherein, 2m+1 is the medium filtering window width, Med{Seg _(j-m)(i) ... Seg _j(i) ..., Seg _(j+m)(i) } for getting median operation, seg ' _j(i) be the pulse wave data of i sampled point correspondence of the local pulse wave of j section behind the medium filtering.

Further, step C further comprises: the revised local pulse wave of described each road pulse wave signal is carried out smothing filtering, and the data that obtain the capable j row of i in the P matrix are expressed as:

P(i，j)＝Seg″ _j(i)＝Mean{Seg′ _(j-n)(i)，...Seg′ _j(i)...，Seg′ _(j+n)(i)}，i＝1，2...N，j＝(n+1)...(M-n)

P (i, j)=Seg " _j(i)=Seg ' _j(i), i=1,2...N, j=1,2...n or j=(M-n+1) ... M

Wherein, 2n+1 is the smothing filtering window width, Mean{Seg ' _(j-n)(i) ... Seg ' _j(i) ..., Seg ' _(j+n)(i) } for getting equal Value Operations, Seg " _j(i) be the pulse wave data of i sampled point correspondence of the local pulse wave of j section behind the smothing filtering.

Further, step e comprises:

According to the local pulse wave of the j section behind HONGGUANG direct current signal, HONGGUANG AC signal, near infrared light direct current signal and the near infrared light AC signal smothing filtering separately, calculate the instantaneous dual wavelength modulation ratio R of the local pulse wave correspondence of j section _j

Get the instantaneous dual wavelength modulation ratio meansigma methods R of each local pulse wave correspondence;

With R substitution fitting a straight line equation:

SpO ₂＝A-BR

Try to achieve blood oxygen saturation; Wherein, SpO ₂Be blood oxygen saturation ,-B, A are respectively the slope and the intercept of this fitting a straight line.

Further, described instantaneous dual wavelength modulation ratio R _jComputing formula be:

$R_j=\frac{({\mathrm{Vac}}_j\left(r\right)/{\mathrm{Vdc}}_j\left(r\right))}{\left({\mathrm{Vac}}_j\left(\mathrm{ir}\right)\right)/{\mathrm{Vdc}}_j\left(\mathrm{ir}\right))}$

Wherein, Vac _j(r) maximum of the pulse wave data of carrying out obtaining behind the smothing filtering for the j of HONGGUANG AC signal local pulse wave and minima is poor; Vdc _j(r) be j the pulse wave data meansigma methods that local pulse wave carries out obtaining behind the smothing filtering of HONGGUANG direct current signal; Vac _j(ir) maximum of the pulse wave data of carrying out obtaining behind the smothing filtering for the j of near infrared light AC signal local pulse wave and minima is poor; Vdc _j(ir) be j the pulse wave data meansigma methods that local pulse wave carries out obtaining behind the smothing filtering of near infrared light direct current signal.

Further, this method further comprises: the P matrix is stretching by row, and the pulse wave signal behind the motion artifacts that the is eliminated signal.

Method for detecting blood oxygen saturation provided by the invention, from HONGGUANG direct current signal (rdc), HONGGUANG AC signal (rac), near infrared light direct current signal (irdc) and near infrared light AC signal (irac), respectively isolate the local pulse wave of the multistage initial, that termination phase is identical, and by the pulse wave data of local pulse wave being revised according to the pulse wave data on two neighboring sections or the local pulse wave of multistage, can eliminate the interference with the motion artifacts signal of pulse wave signal band overlapping part, and then improve the precision that blood oxygen saturation detects.

Description of drawings

Fig. 1 is the schematic flow sheet of embodiment of the invention method for detecting blood oxygen saturation;

Fig. 2 is the oscillogram of embodiment of the invention rdc, rac, irdc and irac;

Fig. 3 carries out the local pulse waveform figure of each section of one road pulse wave signal behind the data polishing for the embodiment of the invention;

Fig. 4 is each local pulse waveform figure of one road pulse wave signal after the embodiment of the invention correction;

Fig. 5 eliminates pulse wave signal oscillogram behind the motion artifacts signal for the embodiment of the invention;

Fig. 6 is for existing interferential pulse wave signal of burst shake and the correlated oscillogram of pulse wave signal that adopts after the present invention eliminates interference;

Fig. 7 is for existing continuously interferential pulse wave signal of shake and the correlated oscillogram of pulse wave signal that adopts after the present invention eliminates interference;

Fig. 8 rotates interferential pulse wave signal and the correlated oscillogram of pulse wave signal that adopts after the present invention eliminates interference for existing.

The specific embodiment

Basic thought of the present invention is: respectively isolate the local pulse wave of multistage from rdc, rac, irdc and irac, and the start-phase of the described local pulse wave of each section is identical, termination phase is also identical; And pulse wave signal is a quasi-periodic signal, when not having motion artifacts, can think that the pulse wave data of same phase correspondence are identical, therefore, by the pulse wave data of local pulse wave being revised according to the pulse wave data on two neighboring sections or the local pulse wave of multistage, can eliminate the interference with the motion artifacts signal of pulse wave signal band overlapping part, and then improve the precision that blood oxygen saturation detects.

Concrete, the realization flow of embodiment of the invention method for detecting blood oxygen saturation may further comprise the steps as shown in Figure 1:

Step 101: gather pulse wave signal, and from the pulse wave signal of being gathered, isolate rdc, rac, irdc and irac, rdc, rac, irdc and irac are carried out FIR filtering or analog filtering respectively, to improve Signal-to-Noise, as shown in Figure 2.

Step 102: from rdc, rac, irdc and irac, respectively isolate the local pulse wave of multistage; The start-phase of the described local pulse wave of each section is identical, and termination phase is also identical;

Wherein, isolating the local pulse wave of multistage specifically comprises:

A1, intercepting one segment signal from rdc, rac, irdc, irac respectively, each segment signal that is intercepted is the same segment signal of the corresponding pulse wave signal of being gathered all;

B1, determine the pairing collection point of each trough of institute's intercept signal from irac; The institute pairing collection point of each trough of intercept signal is the institute pairing collection point of each trough of intercept signal from rdc, rac, irdc from irac; Wherein, rdc, rac, irdc, irac are separated by same road pulse wave signal, think that sample frequency, the heart rate cycle of rdc, rac, irdc, irac are all identical, therefore, the described signal of each section of the same segment signal correspondence of the pulse wave signal of the collection that from rdc, rac, irdc, irac, intercepts, its trough corresponding sampling points is identical, therefore determine the pairing collection point of each trough of institute's intercept signal from irac, can learn the institute pairing collection point of each trough of intercept signal from rdc, rac, irdc; And among rdc, rac, irdc, the irac, the signal to noise ratio of irac is the highest, is easier to determine trough;

C1, with the pairing sampled point of each trough as cut-point, the signal between every adjacent two cut-points will become the local pulse wave of multistage from the signal segmentation that irac, rdc, rac and irdc are intercepted as one section local pulse wave.

Among the step b1, determine that the pairing collection point of each trough of institute's intercept signal specifically comprises from irac:

B11, establish the pulse wave data of x (n), and x (n) k the sampled point of being separated by got first difference and obtained for this section institute's intercept signal from irac of sampled point n correspondence:

S ₁(n)＝|x(n)-x(n-k)|

B12, by as minor function:

$S_2\left(n\right)=\left\{\begin{array}{cc}0,& S_1\left(n\right)\≤H/2\\ \left|S_1\left(n\right)\right|,& S_1\left(n\right)>H/2\end{array}\right.$

Obtain each turning point from the upstroke to the decent of signal of being intercepted, i.e. the S of non-zero ₂(n); Wherein, H is S ₁(n) L sampled point in before the current sampling point n the maximum of corresponding pulse wave data, i.e. H=Max{S ₁(i) | _I=n-L～n-1, and the value of L should guarantee S ₁(i) | _{I=n-L, n-L+1, L, n-1}In comprise the signal that a heart rate cycle trailing edge is intercepted at least;

B13, from the signal that is intercepted, get the minima between adjacent two turning points, promptly obtain the trough of the signal that intercepted.

Step 103: the average sample of getting the local pulse wave of described multistage of one road pulse wave signal is counted, and adopts the fitting of a polynomial interpolation to come the local pulse wave of each section is carried out the data polishing, makes that the sampling number of the local pulse wave of each section is identical, as shown in Figure 3; Because heart rate cycle, sample frequency are identical, the sampling number of the local pulse wave of each section should be identical, but because there is deviation in pulse wave signal, therefore, eliminate deviation by this step.

After eliminating deviation, the local pulse wave of the described multistage of one road pulse wave signal can matrix composed as follows:

$P_{N\×M}=\left[\begin{array}{c}{\mathrm{Seg}}_1\left(1\right),{\mathrm{Seg}}_2\left(1\right)........{\mathrm{Seg}}_i\left(1\right).....{\mathrm{Seg}}_M\left(1\right)\\ {\mathrm{Seg}}_1\left(2\right),{\mathrm{Seg}}_2\left(2\right)........{\mathrm{Seg}}_i\left(2\right).....{\mathrm{Seg}}_M\left(2\right)\\ {\mathrm{Seg}}_1\left(3\right),{\mathrm{Seg}}_2\left(3\right)........{\mathrm{Seg}}_i\left(3\right).....{\mathrm{Seg}}_M\left(3\right)\\ .\\ .\\ {\mathrm{Seg}}_1\left(N\right),{\mathrm{Seg}}_2\left(N\right)........{\mathrm{Seg}}_i\left(N\right).....{\mathrm{Seg}}_M\left(N\right)\end{array}\right]$

Wherein, from this road pulse wave signal, isolate the local pulse wave of M section, every section corresponding N sampled point of local pulse wave, Seg _M(N) be the pulse wave data on the local pulse wave of M section of N sampled point correspondence.

Step 104: the local pulse wave of each section of each road pulse wave signal respectively as pulse wave to be revised, according to waiting to revise pulse wave data on pulse wave two neighboring sections or the local pulse wave of multistage with this, is revised the pulse wave data that this waits to revise pulse wave;

Concrete modification method comprises:

Step c1: the local pulse wave of each section of each road pulse wave signal is carried out medium filtering;

Step c2: the local pulse wave to described each the road pulse wave signal behind the medium filtering carries out smothing filtering.

The method of described medium filtering specifically comprises:

C11, get the local pulse wave of each section of one road pulse wave signal;

C12, with one section local pulse wave of this road pulse wave signal as pulse wave to be revised;

C13, be taken at pulse wave to be revised that a sampled point gathers and wait to revise the pulse wave data of the local pulse wave of the adjacent 2m section of pulse wave with this, the pulse wave data of being taken out are got intermediate value, substitute the pulse wave data of waiting to revise pulse wave that this sampled point is gathered with this intermediate value;

C14, repeated execution of steps c3 finish substituting of all the other sampled values of waiting to revise pulse wave.

C15, return step c2, up to the correction of the local pulse wave of each section of finishing this road pulse wave signal;

C16, return step c1, revise the local pulse wave of each section of its excess-three road pulse wave signal respectively.

Each local pulse wave has been eliminated the distorted waveform among Fig. 3 as shown in Figure 4 after step c1 revises.

Here, because each local pulse wave sample frequency is identical, therefore the phase place of same sampled point correspondence is identical, therefore can think that the pulse wave data of same sampled point correspondence are identical.

The data of the capable j row of i are expressed as in the P matrix after step c1 revises:

P(i，j)＝Seg′ _j(i)＝Med{Seg _(j-m)(i)，...Seg _j(i)...，Seg _(j+m)(i)}，i＝1，2...N，j＝(m+1)...(M-m)

Wherein, 2m+1 is the medium filtering window width, and Med{g} is for getting median operation, seg ' _j(i) be the pulse wave data of i sampled point correspondence of the local pulse wave of j section behind the medium filtering.In order to make output signal and input signal isometric, end about signal is expanded, as follows:

Make P (i, j)=Seg ' _j(i)=Seg _j(i), i=1,2...N, j=1,2...m or j=(M-m+1) ... M

Step c1 is used to eliminate sudden disturbance, because sudden disturbance generally can appear on one or several phase place of certain local pulse wave, at this moment on this phase place pairing pulse wave data will be obviously greater than/less than the data in adjacent local pulse wave homophase site, to there being interferential local pulse wave, and the pulse wave data in adjacent local pulse wave homophase site are got intermediate value, substitute the pulse wave data that have interferential local this phase point of pulse wave with getting intermediate value, can eliminate described sudden disturbance.

Step c2 is used to eliminate periodic disturbance to be disturbed, and principle is: under some disturbances fast, at the same phase place place of the local pulse wave of difference, average may occur is that zero periodic disturbance disturbs.At this moment the pairing pulse wave data of this phase point are put pairing pulse wave data with the same phase of the 2m that is adjacent a local pulse wave and get average, and substitute with this average and to wait to revise local pulse wave and treating the pairing pulse wave data of phase point, can eliminate periodically turbulent influence on this phase place.

Behind the execution in step c2, the data of the capable j row of i are expressed as in the P matrix:

P(i，j)＝Seg″ _j(i)＝Mean{Seg′ _(j-n)(i)，...Seg′ _j(i)...，Seg′ _(j+n)(i)}，i＝1，2...N，j＝(n+1)...(M-n)

Wherein, 2n+1 is the smothing filtering window width, and Mean{g} is for getting equal Value Operations, Seg " _j(i) be the pulse wave data of i sampled point correspondence of the local pulse wave of j section behind the smothing filtering.In order to make output signal and input signal isometric, end about signal is expanded, as follows:

Make P (i, j)=Seg " _j(i)=Seg ' _j(i), i=1,2...N, j=1,2...n or j=(M-n+1) ... M

Step 105: calculate blood oxygen saturation according to the revised local pulse wave of each road pulse wave signal.

The concrete computational methods of blood oxygen saturation comprise:

According to the local pulse wave of the j section behind rdc, rac, irdc and the irac smothing filtering separately, calculate the instantaneous dual wavelength modulation ratio R of the local pulse wave correspondence of j section _j

Get the instantaneous dual wavelength modulation ratio meansigma methods R of each local pulse wave correspondence;

With R substitution fitting a straight line equation:

SpO ₂＝A-BR

Try to achieve blood oxygen saturation; Wherein, SpO ₂Be blood oxygen saturation ,-B, A are respectively the slope and the intercept of this fitting a straight line.

Described fitting a straight line equation obtains by the calibration experiment, specifically comprises:

Get general 30 personal accomplishments and measure sample, utilize the blood gas analysis method of standard to measure its oximetry value SpO respectively ₂As actual value;

Utilize the method for the invention to measure its R value;

Utilize the method for least square linear fit to set up the described fitting a straight line of R and SpO2.

Dual wavelength modulation ratio R _jComputing formula be:

$R_j=\frac{({\mathrm{Vac}}_j\left(r\right)/{\mathrm{Vdc}}_j\left(r\right))}{\left({\mathrm{Vac}}_j\left(\mathrm{ir}\right)\right)/{\mathrm{Vdc}}_j\left(\mathrm{ir}\right))}$

Wherein, Vac _j(r) maximum of the pulse wave data of carrying out obtaining behind the smothing filtering for the j of rac local pulse wave and minima is poor; Vdc _j(r) be j the pulse wave data meansigma methods that local pulse wave carries out obtaining behind the smothing filtering of rdc; Vac _j(ir) maximum of the pulse wave data of carrying out obtaining behind the smothing filtering for the j of irac local pulse wave and minima is poor; Vdc _j(ir) be j the pulse wave data meansigma methods that local pulse wave carries out obtaining behind the smothing filtering of irdc.

Step 106: the P matrix is stretching by row, the pulse wave signal behind the motion artifacts that the is eliminated signal, as shown in Figure 5; The arrangement of the pulse wave data of resultant signal is as follows:

S _out＝{P(1，1)，P(2，1)，...P(N，1)，P(1，2)，P(2，2)....P(N，M)}

The present invention can eliminate the interference that multiple motion produces, Fig. 6, Fig. 7, Fig. 8 have shown respectively and exist burst to shake, shake continuously, rotate interferential pulse wave signal, and adopt the present invention to eliminate the pulse wave signal after the interference, by contrasting visible method proposed by the invention multiple motion artifacts all there is anti-interference preferably effect.

The above is preferred embodiment of the present invention only, is not to be used to limit protection scope of the present invention.

Claims (10)

1, a kind of method for detecting blood oxygen saturation is characterized in that, this method comprises:

A, collection pulse wave signal, and from the pulse wave signal of being gathered, isolate HONGGUANG direct current signal, HONGGUANG AC signal, near infrared light direct current signal and near infrared light AC signal;

B, from HONGGUANG direct current signal, HONGGUANG AC signal, near infrared light direct current signal and near infrared light AC signal, respectively isolate the local pulse wave of multistage; The start-phase of the described local pulse wave of each section of each road pulse wave signal is identical, and termination phase is also identical;

C, with the local pulse wave of each section of each road pulse wave signal respectively as pulse wave to be revised, according to waiting to revise pulse wave data on pulse wave two neighboring sections or the local pulse wave of multistage, revise the pulse wave data that this waits to revise pulse wave with this;

D, calculate blood oxygen saturation according to the revised local pulse wave of each road pulse wave signal.

2, according to the described method for detecting blood oxygen saturation of claim 1, it is characterized in that step B specifically comprises:

Intercept a segment signal respectively from HONGGUANG direct current signal, HONGGUANG AC signal, near infrared light direct current signal, near infrared light AC signal, each segment signal that is intercepted is the same segment signal of the corresponding pulse wave signal of being gathered all;

Determine the pairing collection point of each trough of institute's intercept signal from the near infrared light AC signal; The institute pairing collection point of each trough of intercept signal is the institute pairing collection point of each trough of intercept signal from HONGGUANG direct current signal, HONGGUANG AC signal, near infrared light direct current signal from the near infrared light AC signal;

With the pairing sampled point of each trough as cut-point, signal between every adjacent two cut-points will become the local pulse wave of multistage from the signal segmentation that near infrared light AC signal, HONGGUANG direct current signal, HONGGUANG AC signal and near infrared light direct current signal are intercepted as one section local pulse wave.

3, according to the described method for detecting blood oxygen saturation of claim 2, it is characterized in that, describedly determine that the pairing collection point of each trough of institute's intercept signal comprises from the near infrared light AC signal:

If x (n) is the pulse wave data of this section institute's intercept signal from the near infrared light AC signal of sampled point n correspondence, and x (n) k the sampled point of being separated by got first difference and obtain:

S ₁(n)＝|x(n)-x(n-k)|

By as minor function:

$S_2\left(n\right)=\left\{\begin{array}{cc}0,& S_1\left(n\right)\≤H/2\\ \left|S_1\left(n\right)\right|,& S_1\left(n\right)>H/2\end{array}\right.$

Obtain each turning point from the upstroke to the decent of signal of being intercepted, i.e. the S of non-zero ₂(n); Wherein, H is S ₁The maximum of L sampled point institute corresponding data (n) before the current sampling point n, and the value of L should guarantee S ₁(i) | _{I=n-L, n-L+1 ..., n-1}In comprise the trailing edge of the pulse wave in a heart rate cycle at least;

From the signal that is intercepted, get the minima between adjacent two turning points, promptly obtain the trough of the signal that intercepted.

4, according to the described method for detecting blood oxygen saturation of claim 1, it is characterized in that step C specifically comprises:

C1, get the local pulse wave of each section of one road pulse wave signal;

C2, with one section local pulse wave of this road pulse wave signal as pulse wave to be revised;

C3, be taken at pulse wave to be revised that a sampled point gathers and wait to revise the pulse wave data of the local pulse wave of the adjacent 2m section of pulse wave with this, the pulse wave data of being taken out are got intermediate value, substitute the pulse wave data of waiting to revise pulse wave that this sampled point is gathered with this intermediate value; Wherein, m is a numerical value;

C4, repeated execution of steps c3 finish substituting of all the other sampled values of waiting to revise pulse wave;

C5, return step c2, up to the correction of the local pulse wave of each section of finishing this road pulse wave signal;

C6, return step c1, revise the local pulse wave of each section of its excess-three road pulse wave signal respectively.

5, according to the described method for detecting blood oxygen saturation of claim 1, it is characterized in that step B further comprises:

B1, the average sample of getting the local pulse wave of described multistage of one road pulse wave signal are counted, and adopt the match interpolation to come the local pulse wave of each section is carried out the data polishing, make that the sampling number of the local pulse wave of each section is identical;

B2, with the pulse wave data matrix composed as follows of the local pulse wave of each section:

$P_{N\×M}=\left[\begin{array}{cccccc}{\mathrm{Seg}}_1\left(1\right),& {\mathrm{Seg}}_2\left(1\right)& ........& {\mathrm{Seg}}_i\left(1\right)& .....& {\mathrm{Seg}}_M\left(1\right)\\ {\mathrm{Seg}}_1\left(2\right),& {\mathrm{Seg}}_2\left(2\right)& .......& {\mathrm{Seg}}_i\left(2\right)& ....& {\mathrm{Seg}}_M\left(2\right)\\ {\mathrm{Seg}}_1\left(3\right),& {\mathrm{Seg}}_2\left(3\right)& ........& {\mathrm{Seg}}_i\left(3\right)& ....& {\mathrm{Seg}}_M\left(3\right)\\ .& .& & .& & .\\ .& .& & .& & .\\ {\mathrm{Seg}}_1\left(N\right),& {\mathrm{Seg}}_2\left(N\right)& ....& {\mathrm{Seg}}_i\left(N\right)& ..& {\mathrm{Seg}}_M\left(N\right)\end{array}\right]$

Wherein, from this road pulse wave signal, isolate the local pulse wave of M section, every section corresponding N sampled point of local pulse wave, Seg _M(N) be the pulse wave data on the local pulse wave of M section of N sampled point correspondence;

B3, return step b1, the local pulse wave of its excess-three road pulse wave signal is carried out the data polishing.

According to the described method for detecting blood oxygen saturation of claim 5, it is characterized in that 6, the data of the capable j row of i are expressed as in the P matrix after step C revises:

P(i，j)＝Seg′ _j(i)＝Med{Seg _(j-m)(i)，...Seg _j(i)...，Seg _(j+m)(i)}，i＝1，2...N，j＝(m+1)...(M-m)

P (i, j)=Seg ' _j(i)=Seg _j(i), i=1,2...N, j=1,2...m or j=(M-m+1) ... M

Wherein, 2m+1 is the medium filtering window width, Med{Seg _(j-m)(i) ... Seg _j(i) ..., Seg _(j+m)(i) } for getting median operation, seg ' _j(i) be the pulse wave data of i sampled point correspondence of the local pulse wave of j section behind the medium filtering.

7, according to the described method for detecting blood oxygen saturation of claim 6, it is characterized in that step C further comprises: the revised local pulse wave of described each road pulse wave signal is carried out smothing filtering, and the data that obtain the capable j row of i in the P matrix are expressed as:

P(i，j)＝Seg _j ^#(i)＝Mean{Seg′ _(j-n)(i)，...Seg′ _j(i)...，Seg′ _(j+n)(i)}，i＝1，2...N，j＝(n+1)...(M-n)

P (i, j)=Seg _j ^#(i)=Seg ' _j(i), i=1,2...N, j=1,2...n or j=(M-n+1) ... M

Wherein, 2n+1 is the smothing filtering window width, Mean{Seg ' _(j-n)(i) ... Seg ' _j(i) ..., Seg ' _(j+n)(i) } for getting equal Value Operations, Seg _j ^#(i) be the pulse wave data of i sampled point correspondence of the local pulse wave of j section behind the smothing filtering.

8, according to the described method for detecting blood oxygen saturation of claim 7, it is characterized in that step e comprises:

According to the local pulse wave of the j section behind HONGGUANG direct current signal, HONGGUANG AC signal, near infrared light direct current signal and the near infrared light AC signal smothing filtering separately, calculate the instantaneous dual wavelength modulation ratio R of the local pulse wave correspondence of j section _j

Get the instantaneous dual wavelength modulation ratio meansigma methods R of each local pulse wave correspondence;

With R substitution fitting a straight line equation:

SpO ₂＝A-BR

Try to achieve blood oxygen saturation; Wherein, SpO ₂Be blood oxygen saturation ,-B, A are respectively the slope and the intercept of this fitting a straight line.

9, described according to Claim 8 method for detecting blood oxygen saturation is characterized in that, described instantaneous dual wavelength modulation ratio R _jComputing formula be:

$R_j=\frac{({\mathrm{Vac}}_j\left(r\right)/{\mathrm{Vdc}}_j\left(r\right))}{\left({\mathrm{Vac}}_j\left(\mathrm{ir}\right)\right)/{\mathrm{Vdc}}_j\left(\mathrm{ir}\right))}$

Wherein, Vac _j(r) maximum of the pulse wave data of carrying out obtaining behind the smothing filtering for the j of HONGGUANG AC signal local pulse wave and minima is poor; Vdc _j(r) be j the pulse wave data meansigma methods that local pulse wave carries out obtaining behind the smothing filtering of HONGGUANG direct current signal; Vac _j(ir) maximum of the pulse wave data of carrying out obtaining behind the smothing filtering for the j of near infrared light AC signal local pulse wave and minima is poor; Vdc _j(ir) be j the pulse wave data meansigma methods that local pulse wave carries out obtaining behind the smothing filtering of near infrared light direct current signal.

According to the described method for detecting blood oxygen saturation of claim 7, it is characterized in that 10, this method further comprises: the P matrix is stretching by row, the pulse wave signal behind the motion artifacts that the is eliminated signal.

Images