CN113425282A - Respiration rate monitoring method and device based on multispectral PPG blind source separation method - Google Patents
Respiration rate monitoring method and device based on multispectral PPG blind source separation method Download PDFInfo
- Publication number
- CN113425282A CN113425282A CN202010208190.1A CN202010208190A CN113425282A CN 113425282 A CN113425282 A CN 113425282A CN 202010208190 A CN202010208190 A CN 202010208190A CN 113425282 A CN113425282 A CN 113425282A
- Authority
- CN
- China
- Prior art keywords
- signal
- ppg
- respiratory
- source separation
- blind source
- 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
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
- A61B5/08—Detecting, measuring or recording devices for evaluating the respiratory organs
- A61B5/0816—Measuring devices for examining respiratory frequency
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
- A61B5/0075—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence by spectroscopy, i.e. measuring spectra, e.g. Raman spectroscopy, infrared absorption spectroscopy
-
- 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
- A61B5/7235—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
- A61B5/74—Details of notification to user or communication with user or patient ; user input means
- A61B5/746—Alarms related to a physiological condition, e.g. details of setting alarm thresholds or avoiding false alarms
Abstract
The invention relates to a respiratory rate monitoring method and a respiratory rate monitoring device based on a multispectral PPG blind source separation method. The method comprises the following steps: synchronously acquiring PPG signals at least comprising two wave bands of an individual; extracting a respiratory signal from a PPG signal containing at least two wave bands by a multi-channel fusion method, wherein the multi-channel fusion method specifically adopts a blind source separation algorithm; the respiration rate is calculated based on the respiration signal extracted from the multi-spectral PPG signal by a blind source separation method. The device is provided with means for performing the method. According to the invention, the respiratory component is separated from the multi-spectrum PPG signal, namely, the blind source separation algorithm uses the PPG signals of at least two wave bands as input to separate the respiratory component and the noise component in the multi-spectrum PPG signal, and then the separated respiratory component is selected to calculate the respiratory frequency, so that the robustness and stability of respiratory signal extraction can be enhanced, and the purposes of stable, accurate and continuous monitoring are achieved.
Description
Technical Field
The invention relates to the technical field of physiological parameter monitoring and biological signal processing, in particular to a respiratory rate monitoring method and a respiratory rate monitoring device based on a multispectral PPG blind source separation method.
Background
Multi-spectral Photoplethysmography (PPG) has been used in wearable professional medical devices, such as finger oximeters, to monitor heart rate and blood oxygen saturation. Most of the existing multispectral PPG only comprises two wave band signals, namely a red wave band (650nm) and an infrared wave band (940 nm). The existing heart rate extraction uses frequency information of a dual-band PPG between [40,240] Hz; the blood oxygen saturation extraction uses the relative amplitude variation information of the two-band PPG between [40,240] Hz.
The respiratory rate is a sensitive index of acute respiratory dysfunction and is also an important index for measuring the heart function of a human and judging whether the gas exchange is normal or not. Two basic monitoring methods of respiratory rate are: direct monitoring methods and indirect monitoring methods. The direct monitoring method comprises an impedance method, a temperature sensor method, a pressure sensor method, a carbon dioxide method, a breath sound method and an ultrasonic method, and the indirect monitoring method comprises a method for monitoring the respiratory frequency in the pulse waves of electrocardio, blood pressure, myoelectricity and photo-electric volume.
During respiration, the respiration signal is modulated onto the PPG signal in different ways, on the basis of which the respiration rate can be extracted from the PPG signal. However, since such modulation varies with time, the modulation scheme is not constant, and such modulation is easily annihilated by noise interference, it is an urgent problem to measure the respiratory rate accurately, stably, and continuously.
Patent document CN106983501A, publication No. 2017.07.28, discloses a pulse wave and respiratory wave diagnostic apparatus and a diagnostic method, the pulse wave and respiratory wave diagnostic apparatus including: the acquisition unit is used for acquiring pulse wave signals; the analysis and diagnosis unit is connected with the acquisition unit and is used for carrying out feature extraction on the acquired pulse wave signals, determining waveform feature data of the pulse waves and determining waveform feature data of respiratory waves according to the waveform feature data of the pulse waves; and generating diagnosis information according to the waveform characteristic data of the pulse wave and the waveform characteristic data of the respiratory wave. The beneficial effects are that: the analysis and diagnosis unit of the pulse wave and respiratory wave diagnosis device can directly determine the waveform characteristic data of the respiratory wave according to the waveform characteristic data of the pulse wave, and compared with the traditional device for obtaining the respiratory wave by measuring thoracic impedance based on an impedance method, the device simplifies the respiratory wave obtaining process. However, as mentioned above, there is noise interference in the acquired signal, and the method of this document does not have a step of removing noise, so it is difficult to generate an accurate breathing frequency.
Patent document CN109498022A, published japanese patent No. 2019.03.22, discloses a method for extracting respiratory rate based on photoplethysmography, comprising the steps of: 1) collecting photoelectric volume pulse wave signals at finger tips, low-pass filtering the photoelectric volume pulse wave signals, amplifying weak photoelectric volume pulse wave signals, and transmitting the amplified weak photoelectric volume pulse wave signals to an upper computer through A/D (analog/digital) collection; 2) collecting human body breathing signals and sending the signals to an upper computer; 3) the photoelectric volume pulse wave signal is processed again; 4) extracting a valley point of a pure photoplethysmography signal; 5) fitting an envelope line which is the baseline drift of the photoplethysmography signal, wherein the envelope line is the fitted respiration signal; 6) performing fast Fourier transform on the fitted respiratory signal, and extracting corresponding respiratory frequency; 7) performing fast Fourier transform on the respiration signals, and extracting corresponding respiration frequency as a comparison reference; 8) the number of breaths in one minute is the extracted breathing frequency x 60, and the number of breaths in one minute can be calculated from the extracted breathing frequency. The beneficial effects are that: on the premise of ensuring that the human body does not have large-amplitude movement, the baseline drift of the photoplethysmography signals fitted by the interpolation method is the breathing characteristic curve of the human body, so that the breathing frequency error extracted after the fast Fourier transform is small, the calculated amount is small, the algorithm is simple, and the integration of the breathing interruption monitoring equipment in the future is easy. However, the method of the document is only suitable for monitoring the human body in the scene without large-amplitude motion, and the use of the method is limited.
Journal literature, "wavelet transform combines fast fourier transform to extract respiration rate from PPG", a paper published in 2016, volume 33, phase 1 of chinese medical journal of physiology document "collects human respiratory wave and PPG signals obtained by a temperature sensor and a transmission-type photoelectric pulse sensor at the same time, uses wavelet transform to carry out 9-layer decomposition on the PPG signals, reconstructs and adds together the 9 th-layer detail signals and the 8 th-layer detail signals to obtain respiratory wave, extracts respiration rate from the respiratory wave signals by using an improved fast fourier transform frequency estimation method, extracts respiration rate from 30 PPG samples by using the method, and compares the extracted respiration rate with the respiration rate obtained by the temperature sensor by using a Bland-Altman method to obtain a conclusion that the two have good consistency. However, the method is complex, and it can be seen that the error of the respiratory frequency between 0.18Hz and 0.40Hz is small, and the error is large when the respiratory frequency is less than 0.18 Hz.
There is therefore also a need to develop new methods that enable accurate, fast, stable, continuous monitoring of the respiratory rate from the PPG signal.
Disclosure of Invention
The first purpose of the present invention is to provide a respiration rate monitoring method based on the multispectral PPG blind source separation method, which is directed to the deficiencies in the prior art.
The invention also provides a respiratory rate monitoring device based on the multispectral PPG blind source separation method.
In order to achieve the first purpose, the invention adopts the technical scheme that:
a respiratory rate monitoring method based on a multispectral PPG blind source separation method comprises the following steps:
s1, synchronously acquiring multispectral PPG signals of an individual, wherein the multispectral PPG signals at least comprise PPG signals of two wave bands;
s2, simply preprocessing the original multi-spectrum PPG signal, and separating a low-frequency signal containing a respiratory component;
s3, extracting a respiratory signal from the PPG signal containing at least two wave bands by a multichannel fusion method, wherein the multichannel fusion method specifically adopts a blind source separation algorithm;
s4, calculating a respiration rate based on the respiration signal extracted from the multi-spectral PPG signal by blind source separation.
As a preferred example, the step S3 specifically includes: inputting the low-frequency signal containing the respiratory component separated in the step S2 into a blind source separation algorithm for component decomposition, and if the low-frequency signal is an N-dimensional PPG signal, assuming that N independent linear separable source signals exist, and decomposing N orthogonal or uncorrelated component signals; and then selecting a most approximate respiration signal from the N separated component signals for subsequent calculation of the respiration rate.
As another preferred example, the specific algorithm of the respiratory rate monitoring method based on the multispectral PPG blind source separation method is as follows:
representing the observed multi-channel PPG signal in the time domain by P, wherein the dimension of P is n x t, n is the number of channels and n >1, and t is the time length;
firstly, carrying out band-pass filtering on a frequency domain on P, and removing signal components outside a respiratory frequency band: pf ═ bandpass (P, [ X, Y ] HZ), where [ X, Y ] is an artificially set respiratory bandpass range;
then, the model for blind source separation is set as: pf — W × S, where S is the source signal and W is a parameter of the mixed source signal;
and then calculating a mixing parameter W inverse matrix inv (W) according to Pf by a principal component separation method, wherein the method comprises the following steps:
A＝cov(Pf)
orthogonal decomposition of data: [ U, S, U ]T]SVD (a), where a is the covariance matrix of the filtered signal Pf, U is the eigenvector of the SVD decomposition, and S is the corresponding eigenvalue of the decomposition;
by the formula S ═ UTProjecting Pf onto mutually perpendicular subspaces U by XPf, wherein S' is mutually orthogonal projection signals; let U ═ inv (w);
selecting a source of breathing signals from S': resp select (S'), where the source of the breathing signal can be selected using any hypothetical condition related to the characteristics of the breathing signal;
finally, the respiration rate is calculated in the frequency domain of Resp: f ═ fft (resp),
with max _ idx as the current breathing rate.
As another preferred example, the most recent respiration-like signal is selected with the criterion being an assumption on signal quality, frequency or amplitude.
As another preferable example, the blind source separation method is a principal component analysis method or an independent component analysis method.
In order to achieve the second object, the invention adopts the technical scheme that:
a respiratory rate monitoring device based on multispectral PPG blind source separation method is provided with:
multispectral PPG sensor: the multispectral PPG sensor at least supports two wave bands and is used for acquiring PPG signals of at least two channels;
an operation chip: for executing computational tasks;
a preprocessing module: for simple pre-processing of raw multi-spectral PPG signals obtained from the multi-spectral PPG sensor, separating low-frequency signals containing respiratory components;
the respiratory signal extraction module: the method is used for extracting a respiratory signal from the low-frequency signal which is separated by the preprocessing module and contains a respiratory component through a multi-channel fusion method, wherein the multi-channel fusion method specifically adopts a blind source separation method algorithm;
a respiratory rate calculation module: for calculating the respiration rate based on the respiration signal extracted from the multi-spectral PPG signal by a blind source separation method.
As a preferred example, the respiratory signal extraction module inputs a low-frequency signal containing a respiratory component into a blind source separation algorithm for component decomposition, and if the input signal is an N-dimensional PPG signal, N orthogonal or uncorrelated component signals are decomposed assuming that there are N independent linearly separable source signals; and then selecting a most approximate respiration signal from the N separated component signals for subsequent calculation of the respiration rate.
As another preferred example, the most recent respiration-like signal is selected with the criterion being an assumption on signal quality, frequency or amplitude.
As another preferable example, the blind source separation method is a principal component analysis method or an independent component analysis method.
As another preferred example, the respiratory rate monitoring device based on the multispectral PPG blind source separation method further comprises an early warning module and/or a display module; the early warning module is used for comparing the breathing rate calculated by the breathing rate calculation module with a diagnosis standard and giving an alarm when the breathing rate is abnormal; the display module is used for displaying the respiration rate information.
The invention has the advantages that:
1. the method of the invention separates the respiratory component from the multi-spectrum PPG signal, namely, the blind source separation algorithm uses the PPG signals of at least two wave bands as input (multi-channel input) to separate the respiratory component and the noise component (such as motion interference or light interference) in the multi-spectrum PPG signal, then selects the separated respiratory component to calculate the respiratory frequency, can strengthen the robustness and stability of the respiratory signal extraction, especially improves the signal quality under the environment with motion interference, and achieves the purpose of stable, accurate and continuous monitoring.
2. The method has simple algorithm and is convenient for the integration of the monitoring equipment.
3. The invention improves the stability of the respiration rate monitoring, so the invention can be applied to portable wearable PPG equipment.
4. The device of the invention can be used as professional medical equipment and health products to continuously monitor the breathing rate of a user for a long time in different application scenes, such as sleep apnea syndrome (obstructive, central nervous, mixed), cardiovascular diseases, cardiopulmonary dysfunction/rehabilitation, heart disease monitoring, postoperative monitoring, intensive care, the elderly, neonates, and the like; the device of the invention can also be used as a household health care device to provide physiological monitoring for a certain special population or under a specific scene, such as the elderly, the infants, the snorers, the chronic disease monitoring or the population with related requirements.
5. Based on the invention, three physiological signals, namely heart rate, blood oxygen saturation and respiratory rate, can be extracted from a single multispectral PPG device and are taken as input characteristics together to provide basis for intelligent health monitoring and diagnosis.
Drawings
FIG. 1 is a flow chart of a respiratory rate monitoring method based on a multispectral PPG blind source separation method.
Fig. 2 is a structural block diagram of a respiratory rate monitoring device based on a multispectral PPG blind source separation method.
FIG. 3 shows the PPG signal (respiratory frequency [0,30] Hz), the PPG spectrogram, and the respiratory signal calculated by blind source separation (principal component analysis, PCA) generated by experiments (three normal breaths (about 15Hz breathing rate), two breath-holds (0Hz) of the subject).
FIG. 4 shows the PPG signal (respiratory frequency band [0,30] Hz), the PPG spectrogram, and the respiratory signal calculated by blind source separation (principal component analysis, PCA) generated by experiments (three normal breaths (about 15Hz breathing rate) and two rapid breaths (greater than 20Hz breathing rate) of the subject).
FIG. 5 shows the PPG signal (respiratory frequency [0,30] Hz), the PPG spectrogram, and the respiratory signal calculated by the blind source separation method (principal component analysis PCA) generated by the experiment (three times of normal breath, two times of breath holding and two times of rapid breath alternate with each other).
Detailed Description
The following detailed description of the present invention will be made with reference to the accompanying drawings.
The reference numerals and components referred to in the drawings are as follows:
1. multispectral PPG sensor 2. operation chip
3. Preprocessing module 4. respiratory signal extraction module
5. Respiration signal calculation module
Example 1
Referring to fig. 1, fig. 1 is a flow chart of a respiratory rate monitoring method based on a multispectral PPG blind source separation method according to the present invention. The respiratory rate monitoring method based on the multispectral PPG blind source separation method generally comprises the following steps:
s1, acquiring the individual' S multi-spectrum PPG signals synchronously, wherein the multi-spectrum PPG signals include at least two bands of PPG signals, such as a red channel (660nm) and an infrared channel (940 nm).
S2, the raw multi-spectral PPG signal obtained from the optical sensor is simply pre-processed, e.g. filtered, separating the low frequency signal (respiratory component) and the high frequency signal (heart rate signal).
And S3, extracting the respiratory signal from the PPG signal containing at least two wave bands by a multichannel fusion method. The multi-channel fusion method specifically adopts a blind source separation algorithm, namely a respiratory signal source and a non-respiratory signal noise source (such as motion noise) are separated from a plurality of provided observation signal sources (such as PPG signals of a plurality of wave bands), so that the noise reduction effect is achieved and the robustness of respiratory signal extraction is enhanced. Specifically, the separated low-frequency PPG signal is input into a blind source separation algorithm, such as a Principal Component Analysis (PCA) or Independent Component Analysis (ICA), for Component decomposition, and if the input is an N-dimensional PPG signal, N Independent linearly separable source signals are assumed, and N orthogonal or uncorrelated Component signals are decomposed; a most approximate respiration signal is then selected from the N separated component signals for subsequent calculation of the respiration rate, with the selected criteria being an assumption of signal quality (entropy or signal-to-noise ratio), frequency or amplitude.
S4, calculating a respiration rate based on the respiration signal extracted from the multi-spectral PPG signal by blind source separation.
Examples of the calculation method include: let P be the observed multi-channel PPG signal in the time domain, which can be acquired from the skin side of the individual by a contact PPG device. The dimension of P is n x t, wherein n is the number of channels (n >1), and t is the time length. Firstly, carrying out band-pass filtering on a frequency domain on P, and removing signal components outside a respiratory frequency band:
Pf＝bandpass(P，[10，50]HZ)
wherein [10,50] Hz is the artificially set respiratory band-pass range, and different frequency ranges can be set for different individuals (such as the old and the baby) and application scenes. We then assume that the model for blind source separation is:
Pf＝W×S
where S is the source signal (including the respiratory signal and the noise signal); w is a parameter of the mixed source signal. The model assumes that the source signals are linearly combined to the observed signal Pf. Thus, the core step in the restoration of the source signal is to calculate the inverse matrix inv (W) of the mixing parameters W. Whereas the calculation of inv (w) can be done by blind source separation (principal component separation (PCA) or independent component separation (ICA)). How inv (w) is calculated from Pf is described below using PCA as an example.
The algorithm core of PCA consists in orthogonal decomposition of the data, i.e. decomposition of the signal into vertically orthogonal subspaces in its data space (N-dimensional PPG signal has N-dimensional space), making the decomposed (projected) data uncorrelated. This step can be done by Singular Value Decomposition (SVD):
A＝cov(Pf)
[U，S，UTl＝svd(A)
where A is the covariance matrix of the filtered signal Pf; u is the eigenvector (mutually perpendicular) of the SVD decomposition; and S is the corresponding characteristic value of the decomposition. Then, Pf can be projected onto the mutually perpendicular subspaces U by the following formula:
S′＝UT×Pf
where S' are mutually orthogonal (uncorrelated) projection signals. The main purpose of signal decomposition for P is to separate the respiratory signal and the noise signal therein, so that it can be assumed that the mutually independent signal source in S' is composed of a respiratory component and a noise component. This step assumes that U ═ inv (w). The next step is to select the source of the breathing signal (target) from S':
Resp＝select(S′)
wherein different hypothetical conditions may be used to select the source of the breathing signal. The assumed conditions must be related to the characteristics of the respiration signal, such as by frequency selection (assuming respiration as a periodic signal), amplitude selection (assuming a range of respiration intensities), signal-to-noise ratio selection (assuming the cleanest signal), etc. Finally, the respiration rate is calculated in the frequency domain of Resp:
F＝fft(Resp)
where max _ idx may be output to the system as the current breathing rate.
Example 2
Referring to fig. 2, fig. 2 is a block diagram of a respiration rate monitoring device based on the multispectral PPG blind source separation method according to the present invention. The respiratory rate monitoring device based on the multispectral PPG blind source separation method is provided with:
multispectral PPG sensor 1: the multi-spectral PPG sensor 1 supports at least two bands (e.g. red and infrared bands) for acquiring at least two channel numbers of PPG signals.
And (3) an operation chip 2: for performing computational tasks.
The pretreatment module 3: for simple pre-processing, e.g. filtering, of the raw multi-spectral PPG signal obtained from the multi-spectral PPG sensor 1, separating the low frequency signal (respiratory component) and the high frequency signal (heart rate signal).
The respiratory signal extraction module 4: for extracting a respiratory signal from the low-frequency respiratory component separated by the preprocessing module 3 by a multi-channel fusion method. The multi-channel fusion method specifically adopts a blind source separation algorithm. Specifically, the respiratory signal extraction module 4 receives the low-frequency PPG signal preprocessed by the preprocessing module 3, inputs the low-frequency PPG signal into a blind source separation algorithm, such as PCA or ICA, to perform component decomposition, and if the input is an N-dimensional PPG signal, it is assumed that there are N independent linearly separable source signals to decompose N orthogonal or uncorrelated component signals; a most approximate respiration signal is then selected from the N separated component signals for subsequent calculation of the respiration rate, with the selected criteria being an assumption of signal quality (entropy or signal-to-noise ratio), frequency or amplitude.
The respiration rate calculation module 5: for calculating a respiration rate based on the respiration signal extracted from the multi-spectral PPG signal by blind source separation.
Optionally, the respiratory rate monitoring device based on the multispectral PPG blind source separation method of the invention is further provided with an early warning module, wherein the early warning module is used for comparing the respiratory rate value calculated by the respiratory rate calculation module 4 with a diagnostic standard, and giving an alarm when the respiratory rate is abnormal (for example, the respiratory rate of an adult in a resting state is less than 10 times/minute or more than 20 times minute).
Optionally, the respiration rate monitoring device based on the multispectral PPG blind source separation method is further provided with a display module, and the display module is connected with the respiration rate calculation module 5 and used for displaying information such as respiration rate.
For the above embodiments, it should be noted that:
the principle of extracting the respiratory frequency based on the PPG signal is: the respiration-induced thoracic/abdominal motion (dilation, constriction) compresses the vascular bed and walls of the blood vessels beneath the skin tissue, changing the arterial blood volume in the vessels, forming a low frequency carrier (e.g. respiration rate in the [10,20] Hz band at rest for healthy adults) for the detected PPG signal, and therefore the respiratory signal can be extracted from the low frequency band of the PPG signal. On the basis, the invention further provides a respiration rate extraction method based on the multi-spectral PPG signal, and particularly adopts a multi-channel blind source separation method. The significance of extracting the respiratory rate by using the multispectral PPG signal is that a multichannel blind source separation method is used for separating respiratory components and noise components (such as motion interference or light interference) in an optical signal, then the separated respiratory components are selected to calculate the respiratory frequency, the robustness and the stability of respiratory signal extraction can be enhanced by the strategy, the signal quality is particularly improved in the environment with the motion interference, and the purpose of stable and accurate monitoring is achieved. The principle is as follows: the PPG signal intensities of different spectral bands are different (for example, green light is strongest, red light is weakest), so the modulation carrier intensities of the respiratory signals are also different, but the noise signal (for example, motion interference) does not have the physiological characteristic, and the influence intensity of the noise signal on all bands is the same, so the influence of noise can be removed by a multi-channel fusion mode, and PCA and ICA can be used as one of multi-channel fusion denoising modes.
Example 3
1 healthy adult volunteers were recruited and fixed to the volunteer's chest using a multi-spectral PPG sensing device containing two band signals, the red band (650nm) and the infrared band (940 nm). Three different experimental protocols (experimental protocols) were used to verify the feasibility and sensitivity of multi-spectral PPG monitoring respiratory signals. The results of the experiments are shown in FIGS. 3-5. Each graph contains the experimentally generated PPG signal (respiratory frequency band [0,30] Hz), the PPG spectrogram, and the respiratory signal calculated by blind source separation (principal component analysis PCA).
In the experiment shown in FIG. 3, the subject breathed normally three times (approximately 15Hz breath rate) and held twice (0 Hz). In the experiment shown in FIG. 4, the subject breathed normally three times (approximately 15Hz breath rate) and breathed rapidly two times (greater than 20Hz breath rate). In the experiment shown in fig. 5, the subject breathed normally three times, breathed with two breath holds and breathed with two breaths with a rapid pace. The results of three experiments (signal and spectrogram) indicate that respiratory components can be observed from the PPG signal and this respiratory signal is separated from the heart rate/noise signal by blind source separation (principal component separation, PCA).
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and additions can be made without departing from the method of the present invention, and these modifications and additions should also be regarded as the protection scope of the present invention.
Claims (10)
1. A respiratory rate monitoring method based on a multispectral PPG blind source separation method is characterized by comprising the following steps:
s1, synchronously acquiring multispectral PPG signals of an individual, wherein the multispectral PPG signals at least comprise PPG signals of two wave bands;
s2, simply preprocessing the original multi-spectrum PPG signal, and separating a low-frequency signal containing a respiratory component;
s3, extracting a respiratory signal from the PPG signal containing at least two wave bands by a multichannel fusion method, wherein the multichannel fusion method specifically adopts a blind source separation algorithm;
s4, calculating a respiration rate based on the respiration signal extracted from the multi-spectral PPG signal by blind source separation.
2. The method for respiratory rate monitoring based on multispectral PPG blind source separation according to claim 1, wherein the step S3 specifically comprises: inputting the low-frequency signal containing the respiratory component separated in the step S2 into a blind source separation algorithm for component decomposition, and if the low-frequency signal is an N-dimensional PPG signal, assuming that N independent linear separable source signals exist, and decomposing N orthogonal or uncorrelated component signals; and then selecting a most approximate respiration signal from the N separated component signals for subsequent calculation of the respiration rate.
3. The method for monitoring the respiration rate based on the multispectral PPG blind source separation method according to claim 2, wherein the specific algorithm of the method for monitoring the respiration rate based on the multispectral PPG blind source separation method is as follows:
representing the observed multi-channel PPG signal in the time domain by P, wherein the dimension of P is n x t, n is the number of channels and n >1, and t is the time length;
firstly, carrying out band-pass filtering on a frequency domain on P, and removing signal components outside a respiratory frequency band: pf ═ bandpass (P, [ X, Y ] HZ), where [ X, Y ] is an artificially set respiratory bandpass range;
then, the model for blind source separation is set as: pf — W × S, where S is the source signal and W is a parameter of the mixed source signal;
and then calculating a mixing parameter W inverse matrix inv (W) according to Pf by a principal component separation method, wherein the method comprises the following steps:
orthogonal decomposition of data:
wherein A is a covariance matrix of the filtered signal Pf, U is an eigenvector of SVD decomposition, and S is a corresponding eigenvalue of the decomposition;
by the formula S ═ UTProjecting Pf onto mutually perpendicular subspaces U by XPf, wherein S' is mutually orthogonal projection signals; let U ═ inv (w);
selecting a source of breathing signals from S': resp select (S'), where the source of the breathing signal can be selected using any hypothetical condition related to the characteristics of the breathing signal;
finally, the respiration rate is calculated in the frequency domain of Resp: f ═ fft (resp),
with max _ idx as the current breathing rate.
4. The method for respiratory rate monitoring based on multi-spectral PPG blind source separation of claim 2, wherein the most similar respiratory signal is selected with the criterion of an assumption on signal quality, frequency or amplitude.
5. The method for respiratory rate monitoring based on multispectral PPG blind source separation according to claim 1, wherein the blind source separation is a principal component analysis or an independent component analysis.
6. A respiratory rate monitoring device based on multispectral PPG blind source separation method is characterized by comprising the following components:
multispectral PPG sensor: the multispectral PPG sensor at least supports two wave bands and is used for acquiring PPG signals of at least two channels;
an operation chip: for executing computational tasks;
a preprocessing module: for simple pre-processing of raw multi-spectral PPG signals obtained from the multi-spectral PPG sensor, separating low-frequency signals containing respiratory components;
the respiratory signal extraction module: the method is used for extracting a respiratory signal from the low-frequency signal which is separated by the preprocessing module and contains a respiratory component through a multi-channel fusion method, wherein the multi-channel fusion method specifically adopts a blind source separation method algorithm;
a respiratory rate calculation module: for calculating the respiration rate based on the respiration signal extracted from the multi-spectral PPG signal by a blind source separation method.
7. The apparatus according to claim 6, wherein the respiratory signal extraction module inputs the low frequency signal containing respiratory component into a blind source separation algorithm for component decomposition, and if the input is an N-dimensional PPG signal, then N orthogonal or uncorrelated component signals are decomposed assuming that there are N independent linearly separable source signals; and then selecting a most approximate respiration signal from the N separated component signals for subsequent calculation of the respiration rate.
8. The multispectral PPG blind source separation based respiration rate monitoring device of claim 7 wherein the most similar respiration signal is selected with the criteria of signal quality, frequency or amplitude assumption.
9. The respiratory rate monitoring device according to claim 6, wherein the blind source separation method is a principal component analysis method or an independent component analysis method.
10. The multispectral PPG blind source separation based respiration rate monitoring device according to claim 6, wherein the multispectral PPG blind source separation based respiration rate monitoring device further comprises an early warning module and/or a display module; the early warning module is used for comparing the breathing rate calculated by the breathing rate calculation module with a diagnosis standard and giving an alarm when the breathing rate is abnormal; the display module is used for displaying the respiration rate information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010208190.1A CN113425282A (en) | 2020-03-23 | 2020-03-23 | Respiration rate monitoring method and device based on multispectral PPG blind source separation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010208190.1A CN113425282A (en) | 2020-03-23 | 2020-03-23 | Respiration rate monitoring method and device based on multispectral PPG blind source separation method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113425282A true CN113425282A (en) | 2021-09-24 |
Family
ID=77752594
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010208190.1A Pending CN113425282A (en) | 2020-03-23 | 2020-03-23 | Respiration rate monitoring method and device based on multispectral PPG blind source separation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113425282A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023103769A1 (en) * | 2021-12-08 | 2023-06-15 | 华为技术有限公司 | Wearable device and respiratory tract infection estimation method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140275877A1 (en) * | 2013-03-15 | 2014-09-18 | Jimmy Dripps | Systems and methods for determining respiration information based on principal component analysis |
| CN106983501A (en) * | 2017-03-29 | 2017-07-28 | 汪欣 | Pulse wave and respiratory wave diagnostic device and method |
| CN109498022A (en) * | 2018-12-29 | 2019-03-22 | 西安理工大学 | A kind of respiratory rate extracting method based on photoplethysmographic |
| CN110367950A (en) * | 2019-07-22 | 2019-10-25 | 西安爱特眼动信息科技有限公司 | Contactless physiologic information detection method and system |
-
2020
- 2020-03-23 CN CN202010208190.1A patent/CN113425282A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140275877A1 (en) * | 2013-03-15 | 2014-09-18 | Jimmy Dripps | Systems and methods for determining respiration information based on principal component analysis |
| CN106983501A (en) * | 2017-03-29 | 2017-07-28 | 汪欣 | Pulse wave and respiratory wave diagnostic device and method |
| CN109498022A (en) * | 2018-12-29 | 2019-03-22 | 西安理工大学 | A kind of respiratory rate extracting method based on photoplethysmographic |
| CN110367950A (en) * | 2019-07-22 | 2019-10-25 | 西安爱特眼动信息科技有限公司 | Contactless physiologic information detection method and system |
Non-Patent Citations (2)
| Title |
|---|
| K. VENU MADHAV: "Estimation of respiratory rate from principal components of photoplethysmographic signals", 2010 IEEE EMBS CONFERENCE ON BIOMEDICAL ENGINEERING AND SCIENCES (IECBES), pages 311 - 314 * |
| 赵素文; 高凡; 邓莉: "小波变换结合 快速傅里叶变换从PPG中提取呼吸率", 中国医学物理学杂志, vol. 33, no. 1 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023103769A1 (en) * | 2021-12-08 | 2023-06-15 | 华为技术有限公司 | Wearable device and respiratory tract infection estimation method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6194105B2 (en) | Improved signal selection for acquiring remote photoplethysmographic waveforms | |
| Zhu et al. | Real-time monitoring of respiration rhythm and pulse rate during sleep | |
| EP2465419B1 (en) | Wavelet-based analysis of pulse oximetry signals | |
| US9220460B2 (en) | Method of analyzing and processing signals | |
| US9119597B2 (en) | Systems and methods for determining respiration information from a photoplethysmograph | |
| KR101210828B1 (en) | Apparatus and method improving accuracy of wrist blood pressure by using multiple bio-signal | |
| US20090024044A1 (en) | Data recording for patient status analysis | |
| US9402554B2 (en) | Systems and methods for determining respiration information from a photoplethysmograph | |
| EP3454730B1 (en) | System for determining the frequency of a periodic physiological process of a subject | |
| AU2009265270B2 (en) | Signal processing mirroring technique | |
| WO2017009465A1 (en) | Processing biological data | |
| US20130079606A1 (en) | Systems and methods for determining respiration information from a photoplethysmograph | |
| US20130080489A1 (en) | Systems and methods for determining respiration information from a photoplethysmograph | |
| CN111387959A (en) | Non-contact physiological parameter detection method based on IPPG | |
| WO2013041631A1 (en) | Systems and methods for determining respiration information from a photoplethysmograph | |
| JP7065845B2 (en) | Devices, systems, methods, and computer programs for obtaining vital signs of subjects | |
| CN112263249A (en) | ECG-based method and device for enhancing blood oxygen saturation monitoring | |
| CN113425282A (en) | Respiration rate monitoring method and device based on multispectral PPG blind source separation method | |
| Li et al. | Unobtrusive monitoring of cardiorespiratory signals during sleep based on PVDF sensor and singular spectrum analysis | |
| CN216629572U (en) | Pressure reduction system and equipment based on respiratory training method | |
| Sharma et al. | Non-contact measurement of human heart rate using low cost video camera | |
| Madhav et al. | Extraction of respiratory activity from pulse oximeter's PPG signals using MSICA | |
| CN210228129U (en) | Sleep apnea hypopnea syndrome detection device | |
| Ferroukhi et al. | Robust and reliable PPG and ECG integrated biosensor | |
| Talukdar et al. | The Evaluation of Remote Monitoring Technology Across Participants With Different Skin Tones |
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 |