A kind of method for detecting blood oxygen saturation
Technical field
The present invention relates to technical field of medical instruments, be specifically related to blood oxygen saturation detection technique, particularly a kind of method for detecting blood oxygen saturation.
Background technology
It is the principle moved with arteriogram the absorbtivity of light based on arterial blood and change that noinvasive blood oxygen saturation detects.During transmission-type blood oxygen saturation detects, when transmission region arteries is beaten, arterial blood will change the absorbtivity of light thereupon, be called flutter component or of ac (AC); And the absorptions of its hetero-organization to light such as skin, muscle, skeleton and venous blood are invariable, be called DC quantity (DC).If ignore the decay because the factor such as scattering, reflection causes.Can know according to Beer-Lambert law (Beer-Lambert Law),
Wherein, I
outthe output current of photodetector, I
inbe the input current (corresponding different HONGGUANG and infrared light) of diode, C is the concentration of blood, and L is two distances between LED and detector, and S is saturation, β
oand β
rrepresent material coefficient respectively.
Usually first calculating ratio value R (Ratio of Ratios) during blood oxygen saturation calculates, then calculate blood oxygen saturation according to R, the key of oxygen saturation measurement is also just the calculating of R value.If the computational accuracy of R value is improved, so the certainty of measurement of blood oxygen saturation is just corresponding improves.
At present, widely used to the calculating of R value is peak-to-peak value method.Between each sphygmic period, the waveform of red/infrared light has a maximum and minima, is respectively R
h, R
l.Wherein, R
lwhen corresponding is heart contraction, value when namely arterial blood volume is maximum; R
hwhen corresponding is diastole, value when namely arterial blood volume is minimum.Consider the time delay coefficient of light by uniform dielectric, can be expressed as:
Wherein, I
orepresent incident intensity, α represents and organizes absorptance, α
arepresent arterial blood absorptance, λ
rrepresent specific absorbance, d represents penetration range, and Δ d represents the variable quantity that heart contraction and diastole cause.
Similarly,
Above two equatioies are made a ratio, obtains
Taking the logarithm in both sides, obtains
Similarly, infrared light can obtain
Wherein, IR
lrepresent infrared light minima; IR
hrepresent infrared light maximum.
By above formula, can obtain
In describing as can be seen from above formula, the rate value adopting peak-to-peak value method to calculate only with peak point and the valley point signal correction of pulse signal, therefore the signal to noise ratio of this algorithm is not high, and rate value is easily interfered the impact of signal.
In order to improve signal to noise ratio, there is again a kind of method being called differentiation recently, with peak-to-peak value method unlike, differentiation does not select peak point and valley point to calculate, but between peak value and valley, choose one section of sample point carry out differential calculation, namely calculate the difference of red/infrared light of ac and the ratio of its D. C. value between adjacent 2 respectively, two ratios ask ratio again.Finally, the average ratio value of this segment data is obtained after all ratios value obtained is average.Differentiation computing formula is as follows:
Wherein, I
outthe output current of photodetector, I
inbe the input current (corresponding different HONGGUANG and infrared light) of diode, C is haemoconcentration, and L is the distance between LED and detector, and S is saturation, β
0and β
rrepresent material coefficient respectively.
Again by
Obtain
Two waveforms are divided by, can ratio be obtained
Namely
Obtain after being averaging
From can know the computing formula of codifferentiation, rate value account form is relevant to the of ac of each sample point chosen, and namely by increasing the mode of sample point, improves the signal to noise ratio of algorithm.But because any one piece of data in pulse wave chosen by differentiation, such mode is not considered and the weight that each section of pulse wave is shared when calculating blood oxygen saturation is reduced the precision that blood oxygen saturation calculates.
In addition, from the pulse signal that human body collects, when analog channel enters processor, just along with various Noise and Interference, as Hz noise, baseline drift, motion artifacts, Low perfusion etc., these all can have influence on R value, finally affect the accuracy of oxygen saturation measurement.
In sum, at present in order to the precision of oxygen saturation measurement can be improved, mainly have employed peak-to-peak value method and differentiation carrys out calculating ratio value.For peak-to-peak value method, when human body blood oxygen saturation lower than 80% time, the blood oxygen saturation peak of curve that peak-to-peak value method obtains can not represent blood oxygen saturation exactly, therefore can not solve the problem of low saturation well.For differentiation, owing to being tried to achieve the rate value of this section of sample point by the rate value calculated often between adjacent 2, although the impact of noise on signal can be reduced, but because its rate value is relevant to the rate value of often adjacent sample point, when appearance one or one group of noise signal, will be greatly affected finally by the average ratio value calculated.
Also there is so not attractive problem in prior art in fact, and that is exactly prior art is all generally from time-domain calculation R value, then just calculates blood oxygen saturation according to R value.This just causes computational process comparatively complicated, and can not give full play to the advantage of time domain and frequency domain, finally causes the result of calculation of R value accurate not.So the method for existing detection blood oxygen saturation all exists the not high problem of accuracy of detection, need to improve.
Summary of the invention
Technical problem to be solved by this invention is to provide a kind of method for detecting blood oxygen saturation, solves the problem that in prior art, blood oxygen saturation detection technique is complicated and precision is not high enough.
For solving the problems of the technologies described above, the present invention adopts following technical scheme:
A kind of method for detecting blood oxygen saturation, comprises the steps:
S1), signals collecting, gathered the HONGGUANG detection signal Red and infrared light detection signal Ir that carry blood oxygen signal by signal pickup assembly;
S2), pretreatment, namely remove HONGGUANG detection signal Red and infrared light detection signal Ir ambient light and High-frequency Interference by pretreatment unit;
S3), self adaptation normalized, mainly utilize self adaptation normalized device, pretreated detection data be normalized;
S4), FFT process, by FFT device, the data V1 after normalization and V2 are adopted fast Fourier transform, transform from the time domain to frequency domain;
S5), R value calculates, and the data after processing according to FFT process after the frequency domain normalization that obtains, calculate R value;
S6), blood oxygen saturation calculate, utilize the curve of R value, according to look-up table obtain correspondence oximetry value;
Step S3) described in normalized time, calculate the AC compounent R of HONGGUANG respectively
acwith DC component R
dcwith the AC compounent IR of infrared light
acwith DC component IR
dc, then obtain the value V1 after normalization and V2 by following formula,
V2=IR
ac,
In formula, R
acit is the AC compounent of HONGGUANG; R
dcit is the DC component of HONGGUANG; IR
acit is the AC compounent of infrared light; IR
dcit is the DC component of infrared light.
The invention has the beneficial effects as follows:
Have employed a kind of method for detecting blood oxygen saturation of technical solution of the present invention, owing to calculating blood oxygen saturation in the frequency domain from HONGGUANG and infrared light, simpler from time-domain calculation blood oxygen relative to prior art, and can random disturbances be reduced, thus accuracy of detection is higher.Experiment proves, adopts the blood oxygen estimation technique that the present invention is based on frequency domain, and can calculate perfusion intensity is the pulse signal of 0.3%, and oximetry value can obtain the oximetry value of precision within 1% at 70%-100%.
Accompanying drawing explanation
Fig. 1 is the theory of constitution schematic block diagram of blood oxygen saturation detection system in the specific embodiment of the invention.
Fig. 2 is the flow chart of method for detecting blood oxygen saturation in the specific embodiment of the invention.
Fig. 3 is the amplitude spectrum of detection signal after FFT in the specific embodiment of the invention.
Below in conjunction with accompanying drawing, the invention will be further described.
Detailed description of the invention
As shown in Figure 1, a kind of blood oxygen saturation detection system that this detailed description of the invention provides, comprise and being linked in sequence: signal pickup assembly, wave trap, low pass filter, self adaptation normalized device, FFT (Fast Fourier Transform, fast Fourier transform) device, R value calculation apparatus, Kalman filtering device and blood oxygen saturation calculate device; Wherein wave trap, low pass filter can be collectively referred to as pre-procesor.
As shown in Figure 2, a kind of method for detecting blood oxygen saturation that this detailed description of the invention provides, comprises the steps:
1, signals collecting, gathers the HONGGUANG detection signal Red and infrared light detection signal Ir that carry blood oxygen signal by signal pickup assembly.
2, pretreatment, namely by wave trap and low pass filter, carries out pretreatment to HONGGUANG detection signal Red and infrared light detection signal Ir, mainly removes surround lighting and High-frequency Interference.
Wave trap can remove the interference of detection signal ambient light, and low pass filter can remove the High-frequency Interference in detection signal.As shown in Figure 1, Red1 is through the filtered HONGGUANG of wave trap; Ir1 is through the filtered infrared light of wave trap; Red ' is through the HONGGUANG after low pass filter; Ir ' is through the infrared light after low pass filter.
It should be noted that, due to general first-harmonic, second harmonic and the triple-frequency harmonics of only considering pulse signal, so be the pulse signal of 0.5HZ-5HZ for bandwidth, generally use the low pass filter that cut-off frequency domain is 15HZ to reach promising result.Certainly, the exponent number of wave filter can also be determined if desired as required.
3, self adaptation normalized, mainly utilizes self adaptation normalized device, is normalized obtains V1 and V2 to pretreated detection data;
(formula 1)
V2=IR
ac(formula 2)
In formula, R
acit is the AC compounent of HONGGUANG; R
dcit is the DC component of HONGGUANG; IR
acit is the AC compounent of infrared light; IR
dcit is the DC component of infrared light.
4, FFT process, adopts fast Fourier transform by FFT device by the data V1 after normalization and V2, transforms from the time domain to frequency domain.
FFT part by carrying out FFT conversion to V1 and V2, obtain as shown in Figure 3 V1 and V2 at the amplitude spectrum of frequency domain.
V1 and V2 can carry out windowing process before FFT conversion, and as added rectangular window or Blackman window etc., and the length of FFT conversion can be 1024,2048,4096 etc., and these parameters are determined as required.
V1 and V2 is wherein exactly variable corresponding to two expression formulas below, in theory directly with ratio formula below
ask R value, namely respectively to R
ac/ R
dcand IR
ac/ IR
dccarry out FFT conversion, then obtain R value according to them at the ratio at pulse first-harmonic place, but a lot of processor can only process integer, is just not suitable for (because R with this formula
ac/ R
dc<1, IR
ac/ IR
dc<1), just formula can be become
the problem run in such cases is, R
ac *iR
dcthe very large or IR of product
ac *r
dcproduct very large, cause processor to overflow, and the best approach solving them is by right respectively
with IR
accarry out FFT conversion, then obtain R ratio at their frequency domain, this method can realize on more processor.
5, R value calculates, and according to the frequency domain detection data obtained after FFT process process, calculates R value.The calculating of R value belongs to prior art, no longer describes in detail herein to this.This concrete enforcement specifically, first can select 20 maximum peak values separately at the frequency domain of FV1 and FV2 (FV1 and FV2 is the signal of V1 and V2 after FFT conversion respectively), in these peak values, whether the first-harmonic of request signal and harmonic wave place FV1/FV2 be in the scope that certain is very little, if in the scope that certain is very little, first-harmonic is herein pulse frequency, and FV1/FV2 is herein exactly R value.
6, Kalman filtering, can prevent the interference that random noise causes, and anti-stop signal is undergone mutation, and namely plays level and smooth effect to R value.
It is emphasized that, when carrying out Kalman filtering to R value, need to judge whether to proceed Kalman filtering according to residual sum signal quality index FSQI, if residual error is too large or signal quality index is little, abandon Kalman filtering, wherein signal quality index is determined at the amplitude spectrum of frequency domain according to V2, and specific formula for calculation is as follows:
(formula 3)
Wherein, S is the amplitude of V2 at frequency domain pulse frequency place, N1, N2 ... N512 is the amplitude of V2 in frequency domain non-dc.
7, blood oxygen saturation calculates, and utilizes the curve of R value, and obtain corresponding oximetry value according to look-up table, this part belongs to prior art, no longer describes in detail herein.
The blood oxygen saturation of this detailed description of the invention calculates system, adopt the method for detecting blood oxygen saturation of this detailed description of the invention, from HONGGUANG and and infrared light frequency domain calculate blood oxygen saturation, relatively with prior art from time-domain calculation blood oxygen more simply, and can random disturbances be reduced.Experiment proves, adopts the blood oxygen estimation technique that the present invention is based on frequency domain, and can calculate perfusion intensity is the pulse signal of 0.3%, and oximetry value can obtain the oximetry value of precision within 1% at 70%-100%.
Above content is in conjunction with concrete preferred implementation further description made for the present invention, can not assert that specific embodiment of the invention is confined to these explanations.For general technical staff of the technical field of the invention, without departing from the inventive concept of the premise, some simple deduction or replace can also be made, all should be considered as belonging to protection scope of the present invention.