WO2023226223A1 - Ppg信号质量评估方法及装置以及ppg信号处理方法及*** - Google Patents

Ppg信号质量评估方法及装置以及ppg信号处理方法及*** Download PDF

Info

Publication number
WO2023226223A1
WO2023226223A1 PCT/CN2022/115490 CN2022115490W WO2023226223A1 WO 2023226223 A1 WO2023226223 A1 WO 2023226223A1 CN 2022115490 W CN2022115490 W CN 2022115490W WO 2023226223 A1 WO2023226223 A1 WO 2023226223A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
ppg signal
original
frequency
original ppg
Prior art date
Application number
PCT/CN2022/115490
Other languages
English (en)
French (fr)
Inventor
成世杰
冯谦谨
Original Assignee
广东玖智科技有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 广东玖智科技有限公司 filed Critical 广东玖智科技有限公司
Publication of WO2023226223A1 publication Critical patent/WO2023226223A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/02—Preprocessing
    • G06F2218/04—Denoising
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/021—Measuring pressure in heart or blood vessels
    • A61B5/02108—Measuring pressure in heart or blood vessels from analysis of pulse wave characteristics
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/024—Detecting, measuring or recording pulse rate or heart rate
    • A61B5/02416—Detecting, measuring or recording pulse rate or heart rate using photoplethysmograph signals, e.g. generated by infrared radiation
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/026—Measuring blood flow
    • A61B5/0261—Measuring blood flow using optical means, e.g. infrared light
    • 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
    • A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
    • 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
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7253—Details of waveform analysis characterised by using transforms
    • A61B5/7257—Details of waveform analysis characterised by using transforms using Fourier transforms
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G06F2218/10—Feature extraction by analysing the shape of a waveform, e.g. extracting parameters relating to peaks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching
    • G06F2218/14—Classification; Matching by matching peak patterns

Definitions

  • the present invention relates to the technical field of signal analysis and processing, and more specifically, to a PPG signal quality assessment method and device and a PPG signal processing method and system.
  • Photoplethysmography PPG Photoplethysmograph
  • PPG Photoplethysmograph
  • PPG is a technology that uses photoelectricity to measure changes in blood flow during a heartbeat cycle. It mainly uses photodiodes to shoot to the skin, and then receives the intensity of the reflected light to obtain a waveform that reflects blood flow. . These waveforms contain rich physiological information of the human body. Due to its portability, low cost and non-invasiveness, PPG is welcomed by more and more technology companies and scientific research institutions, and its applications are also very wide: heart rate detection, blood oxygen calculation, blood pressure estimation and calculation of physiological indicators such as sleep staging.
  • PPG is inevitably interfered by power frequency noise, motion artifacts and ambient light during the actual acquisition process.
  • PPG waveform is severely distorted or there is severe noise, it is impossible to obtain usable PPG waveforms for accurate use through pre-processing techniques such as filtering.
  • Calculation of physiological indicators Therefore, before calculating physiological indicators, it is necessary to evaluate the quality of PPG signals, discard severely distorted unusable PPG signals, and use clean PPG signals with good quality for the calculation of physiological indicators. This can effectively improve the accuracy of physiological indicator calculations. performance, reduce the power consumption of wearable devices, and when combined with hardware circuit selection and embedded parameter adjustment, it can provide objective quantitative indicators for PPG signal quality evaluation.
  • the template matching method mainly divides a PPG signal into a single heartbeat cycle, finds a representative heartbeat cycle template and then matches it with all heartbeat cycles, and evaluates the PPG waveform with a high matching ratio as a good quality PPG waveform.
  • This method is too dependent on The algorithm of period segmentation is used because the shape of the PPG waveform is very complex and diverse. Inaccurate period segmentation will mistakenly evaluate good signals as poor signals, and it is very dependent on the template selection criteria. Inaccurate template selection will seriously affect the algorithm. robustness.
  • the technical problem to be solved by the present invention is to provide a PPG signal quality evaluation method and device and a PPG signal processing method and system.
  • the technical solution adopted by the present invention to solve the technical problem is to construct a PPG signal quality assessment method, which includes the steps:
  • step S2 Preprocess the original PPG signal to obtain several waveform segments of preset length, obtain the average amplitude differences of several waveform segments, and determine whether the average amplitude difference is within the first preset range. If so, perform steps S3, otherwise execute step S11;
  • step S4 Obtain the high-frequency signal-to-noise ratio of the original PPG signal based on the useful signal and the high-pass filtering result, obtain the low-frequency signal-to-noise ratio of the original PPG signal based on the useful signal and the low-pass filtering result, and determine the Whether the high-frequency signal-to-noise ratio and the low-frequency signal-to-noise ratio are respectively greater than their corresponding first threshold values, if so, execute step S5, otherwise execute step S11;
  • step S5 Obtain the autocorrelation function of the useful signal, perform Fourier transform on the autocorrelation function to obtain the power spectrum of the useful signal, obtain the kurtosis and skewness of the power spectrum respectively, and determine the power Whether the kurtosis and skewness of the spectrum are respectively greater than their corresponding second threshold values, if so, execute step S6, otherwise execute step S11;
  • step S6 Obtain the average peak period of the autocorrelation function, and confirm whether the average peak period is within the second preset range. If so, execute step S7; otherwise, execute step S11;
  • step S7 Obtain the maximum peak value and peak period standard deviation of the autocorrelation function, and confirm whether the maximum peak value is greater than the third threshold or the peak period standard deviation is less than the fourth threshold. If so, perform step S8 , otherwise execute step S11;
  • step S8 Obtain the standard deviation of the peak-to-peak difference of the autocorrelation function, and confirm whether the standard deviation of the peak-to-peak difference is less than the fifth threshold. If so, execute step S9; otherwise, execute step S10;
  • the preset time length is greater than or equal to 5s and less than or equal to 15s.
  • step S2 the PPG signal quality assessment method of the present invention.
  • the preprocessing of the original PPG signal to obtain several waveform segments of preset length includes:
  • the obtaining the average amplitude difference of the several waveform segments includes:
  • the amplitude difference of each waveform segment is obtained, and the average value of the amplitude differences of all waveform segments is obtained as the average amplitude difference.
  • the first preset range is greater than or equal to 150 and less than or equal to 90000; and/or,
  • the preset length is 1s.
  • the original PPG signal is subjected to high-pass filtering and low-pass filtering respectively, and the obtained value is obtained based on the high-pass filtering result and the low-pass filtering result.
  • Useful signals from the original PPG signal include:
  • step S4 the PPG signal quality assessment method of the present invention.
  • Obtaining the high-frequency signal-to-noise ratio of the original PPG signal based on the useful signal and the high-pass filtering result includes: obtaining the ratio of the high-frequency noise signal to the useful signal as the high-frequency signal-to-noise ratio;
  • Obtaining the low-frequency signal-to-noise ratio of the original PPG signal based on the useful signal and the low-pass filtering result includes: obtaining the ratio of the low-frequency noise signal to the useful signal as the low-frequency signal-to-noise ratio.
  • the first threshold corresponding to the high-frequency signal-to-noise ratio is greater than or equal to 6dB, and the first threshold corresponding to the low-frequency signal-to-noise ratio is The value is greater than or equal to -15dB; and/or,
  • step S5 the second threshold value corresponding to the skewness of the power spectrum is greater than or equal to 12, and the second threshold value corresponding to the kurtosis of the power spectrum is greater than or equal to 150.
  • step S5 obtaining the autocorrelation function of the useful signal includes: performing unilateral normalization of the autocorrelation function on the useful signal. , where the autocorrelation function satisfies the following formula:
  • R[k] is the autocorrelation function
  • k is the index value of the autocorrelation function
  • x s [n] is the useful signal
  • n is the index value of the useful signal
  • N is the useful signal. The number of signal points.
  • the method further includes:
  • the trough detection algorithm based on the second derivative performs peak detection on the autocorrelation function to obtain several peak coordinates
  • the average peak period, maximum peak value, peak period standard deviation and peak-peak difference standard deviation of the autocorrelation function are respectively obtained based on the several peak coordinates.
  • the method includes one or more of the following parameter settings;
  • the second preset range is greater than 0.3s and less than 2s
  • the third threshold value is greater than or equal to 0.6;
  • the fourth threshold value is less than or equal to 5;
  • the fifth threshold value is less than or equal to 0.07.
  • the present invention also constructs a PPG signal processing method, including obtaining the quality evaluation result of the original PPG signal through any one of the above PPG signal quality evaluation methods; and when the original PPG signal is a third-level signal , eliminate the original PPG signal;
  • the original PPG signal When the original PPG signal is a second-level signal, the original PPG signal is used to calculate some preset physiological indicators; when the original PPG signal is a first-level signal, the original PPG signal is used to compare with the PPG signal. All relevant physiological indicators are calculated.
  • the present invention also constructs a PPG signal quality evaluation device, including:
  • the original signal acquisition unit is used to acquire the original PPG signal of a preset duration, wherein the PPG signal is a photoplethysm pulse waveform signal;
  • the first judgment unit is used to preprocess the original PPG signal to obtain several waveform segments of preset length, obtain the average amplitude difference of the several waveform segments, and determine whether the average amplitude difference is within the first preset range, If so, output a positive result, otherwise output a negative result;
  • a useful signal acquisition unit configured to perform high-pass filtering and low-pass filtering on the original PPG signal, and obtain the useful signal of the original PPG signal based on the high-pass filtering result and the low-pass filtering result;
  • the second judgment unit is configured to obtain the high-frequency signal-to-noise ratio of the original PPG signal based on the useful signal and the high-pass filtering result, and obtain the low-frequency signal of the original PPG signal based on the useful signal and the low-pass filtering result.
  • Noise ratio determine whether the high-frequency signal-to-noise ratio and the low-frequency signal-to-noise ratio are respectively greater than their corresponding first threshold values. If so, output a positive result, otherwise output a negative result;
  • the third judgment unit is used to obtain the autocorrelation function of the useful signal, perform Fourier transform on the autocorrelation function to obtain the power spectrum of the useful signal, and obtain the kurtosis and skewness of the power spectrum respectively. , determine whether the kurtosis and skewness of the power spectrum are greater than their corresponding second threshold values, if so, output a positive result, otherwise output a negative result;
  • the fourth judgment unit is used to obtain the average peak period of the autocorrelation function, and judge whether the average peak period is within the second preset range. If so, output a positive result, otherwise output a negative result;
  • the fifth judgment unit is used to obtain the maximum peak value and the standard deviation of the peak period of the autocorrelation function, and confirm whether the maximum peak value is greater than the third threshold value or the standard deviation of the peak period is less than the fourth threshold value. If so, , then output a positive result, otherwise output a negative result;
  • the sixth judgment unit is used to obtain the standard deviation of the peak-to-peak difference of the autocorrelation function, and confirm whether the standard deviation of the peak-to-peak difference is less than the fifth threshold value. If so, output a positive result, otherwise output a negative result;
  • a result confirmation unit configured to determine when any one of the first judgment unit, the second judgment unit, the third judgment unit, the fourth judgment unit and the sixth judgment unit outputs a negative result.
  • the original PPG signal is a third-level signal.
  • the sixth judgment unit outputs a negative result, it is determined that the original PPG signal is a second-level signal.
  • the sixth judgment unit outputs a positive result, it is determined that the original PPG signal is a second-level signal.
  • the original PPG signal is a first-level signal, wherein the third-level signal quality is lower than the second-level signal quality, and the second-level signal quality is lower than the first-level signal quality.
  • the present invention also constructs a PPG signal processing system, including the PPG signal quality evaluation device as described above, and a processing unit;
  • the processing unit is used to obtain the quality evaluation result of the original PPG signal; and when the original PPG signal is a third-level signal, eliminate the original PPG signal;
  • the original PPG signal is a second-level signal
  • the original PPG signal is used for calculation of some preset physiological indicators
  • the original PPG signal is a first-level signal
  • the original PPG signal is used for calculation of all physiological indicators related to the PPG signal.
  • the implementation of the PPG signal quality assessment method and device and the PPG signal processing method and system of the present invention has the following beneficial effects: it can realize the evaluation of multi-level quantitative indicators of the PPG signal to achieve reasonable utilization based on different quality levels of the PPG signal.
  • Figure 1 is a program flow chart of a PPG signal quality assessment method in an embodiment of the present invention
  • Figure 2 is a schematic diagram of the original PPG signal in an embodiment of the present invention.
  • Figure 3 is a schematic diagram of the average amplitude difference of the original PPG signal in an embodiment of the present invention.
  • Figure 4 is a schematic diagram of the low-frequency signal-to-noise ratio of the original PPG signal in an embodiment of the present invention
  • Figure 5 is a schematic diagram of the low-frequency signal-to-noise ratio of the original PPG signal in an embodiment of the present invention
  • Figure 6 is a schematic diagram of the power spectrum index of the original PPG signal in an embodiment of the present invention.
  • Figure 7 is a schematic diagram of the power spectrum index of the original PPG signal in another embodiment of the present invention.
  • Figure 8 is a schematic diagram of the autocorrelation function index of the original PPG signal in an embodiment of the present invention.
  • Figure 9 is a schematic diagram of the autocorrelation function index of the original PPG signal in another embodiment of the present invention.
  • Figure 10 is a schematic diagram of the autocorrelation function index of the original PPG signal in yet another embodiment of the present invention.
  • Figure 11 is a logic block diagram of a PPG signal quality evaluation device in an embodiment of the present invention.
  • Figure 12 is a logical block diagram of a PPG signal quality processing system in an embodiment of the present invention.
  • the method includes: S1. Obtaining an original PPG signal of a preset duration, where the PPG signal is a photoplethysm waveform signal. Specifically, a preset period of PPG information, which corresponds to the original PPG signal, can be obtained through the acquisition device. Among them, the time length of the original PPG signal cannot be set too long, otherwise the signal will fluctuate violently, and the computational complexity will increase as the time length increases.
  • step S2 Preprocess the original PPG signal to obtain several waveform segments of preset length, obtain the average amplitude difference of the several waveform segments, and determine whether the average amplitude difference is within the first preset range. If so, execute step S3, otherwise execute step S11. . Specifically, when the device is in poor contact with the skin or is not worn, the original PPG signal obtained by the collection device will be very weak or even nonexistent. At this time, the average amplitude difference of the original PPG signal is very low. On the contrary, when saturation noise or spikes appear in the hardware circuit, the average amplitude difference of the signals becomes abnormally high.
  • the original PPG signal can be segmented to obtain several waveform segments of preset length, and the average amplitude difference is obtained based on the several waveform segments.
  • the average amplitude difference is judged.
  • the average amplitude difference meets the requirements of the first preset range, that is, it can be determined that the amplitude of the original PPG signal is normal, and it can continue to perform step S3 and subsequent actions.
  • the average amplitude difference does not meet the requirements of the first preset range, it is determined that the amplitude of the original PPG signal is abnormal, and at this time, the subsequent determination action can be directly performed. That is, it can be directly determined that the original PPG signal is a third-level signal, and the quality evaluation process of the original PPG signal is ended.
  • This process can be defined as the first-level evaluation process.
  • the average amplitude difference of normal PPG waveforms ranges from hundreds to tens of thousands. However, when the device is in poor contact with the skin or is not worn, and the average amplitude difference of the PPG signal is less than 150, the waveform of the PPG signal is approximately a straight line. As shown in Figure 3, when the current in the hardware circuit is adjusted too large, the average amplitude difference of the PPG signal will be greater than 90,000, and the waveform of the PPG signal will suffer from saturation distortion. If the average amplitude difference of the original PPG signal is abnormal, the original PPG signal will be evaluated as a poor signal, and there is no need to perform the next level of evaluation.
  • the 150 and 9000 here correspond to the number of the discrete signal, which is determined by the set quantization precision.
  • preprocessing the original PPG signal to obtain a number of waveform segments of preset length includes: performing median filtering on the original PPG signal, and filtering the filtered signal with a sliding window of preset length. Intercept without gaps in sequence to obtain several waveform segments, where there is no overlap between the waveform segments. That is, it can first perform median filtering with a window size of 5 on the original PPG signal to remove the influence of outliers in the original PPG signal and obtain the signal x[n]. Among them, during the median filtering process, the middle number is taken among the five numbers, so that the influence of abnormal outliers can be removed.
  • the signal is [1, 2, 3, 4, 25], and the final output is 3, which avoids the influence of the outlier point 25.
  • the signal x[n] is divided into a sliding window with a length of L (corresponding to the preset length). Divide it into M non-overlapping segments, each segment is recorded as: x m [l], 0 ⁇ m ⁇ M-1, 0 ⁇ l ⁇ L-1.
  • x m [l] is the waveform segment
  • the subscript m is the index value of the waveform segment, which represents the number of small signal segments
  • l is the length of the waveform segment.
  • the value of n is [0, N-1], corresponding to the signal of 10s respectively.
  • obtaining the average amplitude difference of several waveform segments includes: obtaining the amplitude difference of each waveform segment, and obtaining the average amplitude difference of all waveform segments as the average amplitude difference. That is, calculate the amplitude difference ⁇ m of each waveform segment for the obtained waveform segments,
  • ⁇ m max(x m [l])-min(x m [l]).
  • max(x m [l]) is the maximum amplitude value in the x m [l] waveform segment
  • min(x m [l]) is the minimum amplitude value in the x m [l] waveform segment, so all
  • the average amplitude difference Amp of the waveform segment is:
  • the determination of the average amplitude difference Amp may be to determine that it satisfies the first preset range when it is greater than or equal to 150 and less than or equal to 90,000.
  • step S3 high-pass filtering and low-pass filtering are performed on the original PPG signal, and a useful signal of the original PPG signal is obtained based on the high-pass filtering result and the low-pass filtering result; including: S31, taking the cutoff frequency as the first frequency The Butterworth high-pass filter performs high-pass filtering on the original PPG signal to obtain the high-frequency noise signal; S32, the Butterworth low-pass filter with the cutoff frequency as the second frequency performs low-pass filtering on the original PPG signal to obtain Low-frequency noise signal; S33. Remove the high-frequency noise signal and low-frequency noise signal from the original PPG signal respectively, and use the remaining signal as the useful signal.
  • the filter can filter the original PPG signal through a high-pass filter to obtain a high-frequency noise signal, and filter the original PPG signal through a low-pass filter to obtain a low-frequency noise signal.
  • Butterworth filters can be used as both high-pass and low-pass filters here.
  • the low-frequency noise power P low is as follows:
  • the useful signal power P s is as follows:
  • step S4 Obtain the high-frequency signal-to-noise ratio of the original PPG signal based on the useful signal and the high-pass filtering result, obtain the low-frequency signal-to-noise ratio of the original PPG signal based on the useful signal and the low-pass filtering result, and determine whether the high-frequency signal-to-noise ratio and the low-frequency signal-to-noise ratio are greater than their corresponding counterparts.
  • the first threshold value if yes, execute step S5, otherwise execute step S11.
  • the high-frequency signal-to-noise ratio and the low-frequency signal-to-noise ratio are determined. When both the high-frequency signal-to-noise ratio and the low-frequency signal-to-noise ratio meet the requirements, it is determined that the noise of the original PPG signal meets the requirements, and the subsequent step S5 can be continued.
  • the noise of the original PPG signal mainly comes from high-frequency burr noise and low-frequency drift noise. As shown in Figure 4, when the low-frequency signal-to-noise ratio is less than -15dB, the overall drift of the PPG signal is very serious, the fluctuation is very violent, and it is completely irregular.
  • step S4 obtaining the high-frequency signal-to-noise ratio of the original PPG signal based on the useful signal and the high-pass filtering result includes: obtaining the ratio of the high-frequency noise signal to the useful signal as the high-frequency signal-to-noise ratio; based on the useful signal and the low-pass filtering result
  • Obtaining the low-frequency signal-to-noise ratio of the original PPG signal includes: obtaining the ratio of the low-frequency noise signal to the useful signal as the low-frequency signal-to-noise ratio.
  • the specific high-frequency signal-to-noise ratio SNR high acquisition process is as follows:
  • the specific low-frequency signal-to-noise ratio SNR low acquisition process is as follows:
  • the first threshold value corresponding to the high-frequency signal-to-noise ratio is greater than or equal to 6dB, and the first threshold value corresponding to the low-frequency signal-to-noise ratio is greater than or equal to -15dB. That is, when the first threshold value corresponding to the high-frequency signal-to-noise ratio is 6dB, that is, when the high-frequency signal-to-noise ratio SNR high is greater than 6dB, it can be determined that the high-frequency signal-to-noise ratio meets the requirements.
  • the first threshold value corresponding to the low-frequency signal-to-noise ratio is -15dB, that is, when the low-frequency signal-to-noise ratio SNR low is greater than -15dB, it can be determined that the low-frequency signal-to-noise ratio meets the requirements.
  • step S5 Obtain the autocorrelation function of the useful signal, perform Fourier transform on the autocorrelation function to obtain the power spectrum of the useful signal, obtain the kurtosis and skewness of the power spectrum respectively, and determine whether the kurtosis and skewness of the power spectrum are greater than If the corresponding second threshold value is yes, step S6 is executed; otherwise, step S11 is executed. Specifically, when the noise index of the original PPG signal meets the requirements, the power spectrum related index is judged. It mainly judges the skewness and kurtosis of the frequency spectrum corresponding to the original PPG signal.
  • the high-frequency signal-to-noise ratio and low-frequency signal-to-noise ratio of such mutation signals may be very high and cannot be detected through signal-to-noise ratio related indicators, but their sparse representation can be measured by calculating the skewness and kurtosis of the power spectrum of the useful signal. It first obtains the autocorrelation function of the useful signal, and performs Fourier transformation based on the autocorrelation function to obtain the power spectrum of the useful signal. Based on the obtained power spectrum, its corresponding skewness and kurtosis are obtained.
  • This process can be defined as a third-level evaluation process that can evaluate power spectrum related indicators. It can be understood that the evaluation process of the first and second levels is to evaluate the signal from the perspective of overall indicators. When discontinuities and sharp pulses appear in local small areas in the signal, these mutations will cause the power spectrum coefficients to be distributed across the entire frequency axis. cannot obtain a good sparse representation. As shown in Figures 6 and 7, at this time, the power spectrum index can be used to distinguish whether there are abnormalities caused by mutation areas in the signal.
  • the length of the autocorrelation function that performs unilateral normalization of the signal is also 1000 points.
  • the autocorrelation function is called R[k], where k is the index value of the autocorrelation function, its value range is 0 to N-1, and n represents the index of the useful signal value, its value range is also 0 to N-1.
  • the Fourier transform of the autocorrelation function corresponds to the power spectrum of the signal.
  • Fourier transform is performed on the autocorrelation function R
  • the kurtosis P SK and skewness P KT of the power spectrum are as follows:
  • ⁇ is the mean value of P xx (f)
  • ⁇ is the variance of P xx (f).
  • the skewness and kurtosis of the power spectrum are judged simultaneously. When they both meet the corresponding threshold requirements, that is, when both are greater than their corresponding second threshold requirements, the signal of the power spectrum is judged to be normal, and the following steps can be continued. Action of step S6. When any of the skewness and kurtosis of the power spectrum does not meet its corresponding second threshold value, it can be determined that there is a mutation area in the original PPG signal, and the subsequent judgment action is directly performed. That is, it can be directly determined that the original PPG signal is a third-level signal, and the quality evaluation process of the original PPG signal is ended.
  • step S5 the second threshold value corresponding to the skewness of the power spectrum is greater than or equal to 12, and the second threshold value corresponding to the kurtosis of the power spectrum is greater than or equal to 150. That is, the second threshold value corresponding to the skewness of the power spectrum and the second threshold value corresponding to the kurtosis of the power spectrum may be set respectively.
  • step S6 Obtain the average peak period of the autocorrelation function and confirm whether the average peak period is within the second preset range. If so, execute step S7. Otherwise, execute step S11. Specifically, due to the time delay between the signal calculated by the autocorrelation function and its When there is a periodic component in the signal, the autocorrelation function will have a maximum value at an integer multiple of the period. Therefore, the periodicity of the autocorrelation function can be tested by detecting its peak value. That is, the average peak period of the autocorrelation function is determined. When it is within the second preset range, it is determined that the original PPG signal is within the normal range of the sampled signal, such as the heart rate signal, and the action of step S7 can be continued.
  • the original PPG signal is not within the normal range of the heart rate.
  • the following judgment action can be directly performed, that is, the original PPG signal can be directly determined to be a third-level signal. , and ends the quality assessment process of the original PPG signal.
  • the second preset range is greater than 0.3s and less than 2s. That is, in a specific embodiment, it is set that when the average peak period is greater than 0.3s and less than 2s, it is determined that it is within the second preset range.
  • step S7 Obtain the maximum peak value and the standard deviation of the peak period of the autocorrelation function, and confirm whether the maximum peak value is greater than the third threshold or the standard deviation of the peak period is less than the fourth threshold. If so, perform step S8; otherwise, perform step S11; Specifically, when the average peak period of the autocorrelation function meets the requirements, that is, it is judged that the original PPG signal is within the normal range of the detection signal, the maximum peak value and the standard deviation of the peak period of the autocorrelation function are judged, that is, the periodicity of the original PPG signal is determined. judge. When it is determined that the original PPG signal has periodicity, it can continue the action of step S8.
  • the subsequent judgment action can be directly performed, that is, the original PPG signal can be directly determined to be a third-level signal, and the quality evaluation process of the original PPG signal is ended.
  • the specific determination process is that when the maximum peak value of the autocorrelation function is greater than the third threshold value and the standard deviation of the peak period is less than the fourth threshold value at the same time, it is determined that the original PPG signal does not have periodicity, otherwise it can be determined that the original PPG signal has periodicity. sex.
  • the third threshold value is greater than or equal to 0.6
  • the fourth threshold value is less than or equal to 5; specifically, it can be determined that the original PPG signal has periodicity when the maximum peak value of the correlation function is greater than 0.6, which is also It can be determined that the original PPG signal is periodic when the standard deviation of the peak period of the autocorrelation function is less than 5.
  • step S8 Obtain the standard deviation of the peak-to-peak difference of the autocorrelation function, and confirm whether the standard deviation of the peak-to-peak difference is less than the fifth threshold. If so, execute step S9. Otherwise, execute step S10; S9. Determine whether the original PPG signal is the fifth threshold. First-level signal, and end; S10, determine the original PPG signal is a second-level signal, and the second-level signal quality is lower than the first-level signal quality, and end; S11, determine the original PPG signal is a third-level signal, and the third-level signal The signal quality is lower than the second level signal quality and ends.
  • the standard deviation of the peak-to-peak difference of the autocorrelation function is judged.
  • the difference determines whether the signal shapes of each period of the original PPG signal are similar.
  • the standard deviation of the peak-to-peak difference value of the autocorrelation function is less than the fifth threshold value, it can be judged that the signal shape is similar.
  • the original PPG signal can be judged. is a first-level signal, otherwise the original PPG signal is determined to be a second-level signal.
  • the quality of the first-level signal is better than that of the second-level signal, and the quality of the second-level signal is better than that of the third-level signal.
  • the PPG signal quality evaluation method of the present invention also includes: performing peak detection on the autocorrelation function based on the second-order derivative trough detection algorithm to obtain several peak coordinates; obtaining the average peak period, respectively, of the autocorrelation function based on the several peak coordinates. Maximum peak value, peak period standard deviation and peak-to-peak difference standard deviation. Specifically, it can first use the trough detection algorithm based on the second-order derivative to perform peak detection on the autocorrelation function, and obtain w all peak coordinates:
  • the process of determining indicators related to the autocorrelation function can be defined as the fourth level of evaluation. It directly tests the periodicity of useful signals by comparison and calculates the autocorrelation function, which can eliminate the impact of additive noise on the periodicity of the test signal.
  • the average peak period of the autocorrelation function corresponds to the period of the original PPG signal. According to the normal heart rate of 30 to 200 times per minute, the average peak period should be between 0.3s and 2s. Signals exceeding the normal heart rate range are evaluated as poor signals (as shown in the figure) shown in 8). When the maximum peak value of the autocorrelation function is greater than 0.6 or the standard deviation of the peak period is less than 5, it can be judged that the signal is periodic.
  • the autocorrelation function should be a gradually attenuated sine wave signal.
  • the waveform shape of individual periodic signals may be affected by noise in the passband.
  • the waveform shapes of periodic PPG signals are different.
  • the peak value of its autocorrelation function fluctuates. Therefore, by calculating the standard deviation of the peak-to-peak difference of the autocorrelation function, the fluctuation of the peak value of the autocorrelation function can be judged.
  • the standard deviation of the peak-to-peak difference of the correlation function is less than 0.07, it is evaluated as a good signal (as shown in Figure 9), otherwise it is evaluated as a medium signal (as shown in Figure 10).
  • a PPG signal processing method of the present invention includes: obtaining the quality evaluation result of the original PPG signal through any one of the above PPG signal quality evaluation methods; and when the original PPG signal is a third-level signal, eliminating the original PPG signal ; When the original PPG signal is a second-level signal, the original PPG signal is used for the calculation of some preset physiological indicators; when the original PPG signal is a first-level signal, the original PPG signal is used for all physiological indicators related to the PPG signal. calculate. That is, during the PPG signal processing process, based on the above PPG signal quality evaluation method, different operations are performed on the obtained PPG signals of different levels.
  • the third-level signals are directly eliminated, and the second-level signals are used in the calculation process of some indicators such as heart rate, which are preset indicators.
  • the first-level signal can be used normally, that is, it can be used for the calculation of all physiological indicators such as heart rate and blood pressure. All physiological indicators here are physiological indicators related to PPG.
  • a PPG signal quality evaluation device 100 of the present invention includes:
  • the original signal acquisition unit 110 is used to acquire the original PPG signal of a preset duration, where the PPG signal is a photoplethysmogram signal;
  • the first judgment unit 121 is used to preprocess the original PPG signal to obtain several waveform segments of preset length, obtain the average amplitude difference of the several waveform segments, and determine whether the average amplitude difference is within the first preset range. If so, output affirmative result, otherwise output a negative result;
  • the useful signal acquisition unit 130 is used to perform high-pass filtering and low-pass filtering on the original PPG signal, and obtain the useful signal of the original PPG signal based on the high-pass filtering result and the low-pass filtering result;
  • the second judgment unit 122 is used to obtain the high-frequency signal-to-noise ratio of the original PPG signal based on the useful signal and the high-pass filtering result, obtain the low-frequency signal-to-noise ratio of the original PPG signal based on the useful signal and the low-pass filtering result, and determine the high-frequency signal-to-noise ratio and the low-frequency signal-to-noise ratio. Whether the ratio is greater than its corresponding first threshold value respectively, if so, output a positive result, otherwise output a negative result;
  • the third judgment unit 123 is used to obtain the autocorrelation function of the useful signal, perform Fourier transform on the autocorrelation function to obtain the power spectrum of the useful signal, obtain the kurtosis and skewness of the power spectrum respectively, and determine the kurtosis of the power spectrum. Whether and skewness are respectively greater than their corresponding second threshold values, if so, output a positive result, otherwise output a negative result;
  • the fourth judgment unit 124 is used to obtain the average peak period of the autocorrelation function, and judge whether the average peak period is within the second preset range. If so, output a positive result, otherwise output a negative result;
  • the fifth judgment unit 125 is used to obtain the maximum peak value and the standard deviation of the peak period of the autocorrelation function, and confirm whether the maximum peak value is greater than the third threshold or the standard deviation of the peak period is less than the fourth threshold. If so, output a positive result. , otherwise output a negative result;
  • the sixth judgment unit 126 is used to obtain the standard deviation of the peak-to-peak difference of the autocorrelation function, and confirm whether the standard deviation of the peak-to-peak difference is less than the fifth threshold value. If so, output a positive result, otherwise output a negative result;
  • the result confirmation unit 140 is configured to determine that the original PPG signal is For third-level signals, when the sixth judgment unit 126 outputs a negative result, it is determined that the original PPG signal is a second-level signal. When the sixth judgment unit 127 outputs a positive result, it is determined that the original PPG signal is a first-level signal, where, Level 3 signal quality is lower than level 2 signal quality, and level 2 signal quality is lower than level 1 signal quality.
  • the specific cooperation operation process between the units of the PPG signal quality evaluation device here can refer to the above-mentioned PPG signal quality evaluation method, which will not be described again here.
  • a PPG signal processing system of the present invention includes the above PPG signal quality evaluation device 100, and a processing unit 200; the processing unit is used to obtain the quality evaluation result of the original PPG signal; and in the original When the PPG signal is the third level signal, the original PPG signal is eliminated; when the original PPG signal is the second level signal, the original PPG signal is used for the calculation of some preset physiological indicators; when the original PPG signal is the first level signal, the original PPG signal is used for the calculation of some preset physiological indicators. The signal is used for all physiological indicator calculations related to the PPG signal.
  • the third-level signals are directly eliminated, and the second-level signals are used in the calculation process of some indicators such as heart rate, which are preset indicators.
  • the first-level signal can be used normally, that is, it can be used for the calculation of all physiological indicators such as heart rate and blood pressure. All physiological indicators here are physiological indicators related to PPG.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Signal Processing (AREA)
  • Animal Behavior & Ethology (AREA)
  • Surgery (AREA)
  • Physiology (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Cardiology (AREA)
  • Artificial Intelligence (AREA)
  • Psychiatry (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Hematology (AREA)
  • Vascular Medicine (AREA)
  • Power Engineering (AREA)
  • Mathematical Physics (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)

Abstract

本发明涉及PPG信号质量评估方法及装置以及PPG信号处理方法及***，包括：S1、获取原始信号，S2、当其平均幅度差在第一预设范围时执行S3，否则执行S11；S3、获取有用信号；S4、当其高频和低频信噪比分别大于对应的第一门限值时执行S5，否则执行S11；S5、当其功率谱的额峰度和偏度分别大于第二门限值时执行S6，否则执行S11；S6、当相关函数的平均峰值周期在第二预设范围内时执行S7，否则执行S11；S7、当出现最大峰值大于第三门限值或峰值周期标准差小于第四门限值时执行S8，否则执行S11；S8、当自相关函数的峰峰差值标准差小于第五门限值时执行S9，否则执行S10；S9、判定为第一等级；S10、判定为第二等级；S11、判定为第三等级。

Description

PPG信号质量评估方法及装置以及PPG信号处理方法及*** 技术领域
本发明涉及信号分析及处理技术领域，更具体地说，涉及PPG信号质量评估方法及装置以及PPG信号处理方法及***。
背景技术
光电容积脉搏波PPG(Photoplethysmograph)技术是一种利用光电的方法测量一次心跳周期血流量变化的技术，其主要利用光电二极管射向皮肤，然后通过接收反射光的强度，获得能够反映血流量的波形。这些波形包含人体丰富的生理信息。由于PPG便携性、低成本以及非侵入性受到越来越多的科技公司、科研机构的欢迎，应用也非常广泛：心率检测、血氧计算、血压估计和睡眠分期等生理指标的计算。
然而PPG在实际采集的过程中难免受到工频噪声、运动伪影和环境光的干扰，当PPG波形严重失真或者存在严重噪声时，是无法通过滤波等预处理技术得到可用的PPG波形用于准确计算生理指标的。因此，在计算生理指标之前需要对PPG信号的质量进行评估，将严重失真的不可用PPG信号舍弃，将质量好的干净的PPG信号用于生理指标的计算，这样可以有效提高生理指标计算的准确性、降低可穿戴设备的功耗，并且在配合硬件电路选型以及嵌入式参数调节时，可以提供客观的PPG信号质量评估的量化指标。
目前已有的方法有模板匹配方法、特征提取加机器学习的方法和深度学习的方法。模板匹配的方法主要是将一段PPG信号分割成单个心跳周期，找到代表性的心跳周期模板然后和所有心跳周期进行匹配，将匹配比例高的评估为质量良好的PPG波形，然而这类方法太依赖周期分割的算法了，因为PPG波形的形态是十分复杂多样的，周期分割不准确会误将好的信号评价为差的信号，而且十分依赖模板的选取标准，模板选择不准确会严重影响算法的鲁棒性。基于机器学习和深度学习的方法也能取得很好的效果，但是需要依赖标签的准确性以及数据集的泛化性，而且给数据打标签的过程十分耗时。而且大部分方法的评估结果只是将信号分为可用或者不可用的二元分类，这样的分类结果会比较粗糙。
发明内容
本发明要解决的技术问题在于，提供PPG信号质量评估方法及装置以及PPG信号处理方法及***。
本发明解决其技术问题所采用的技术方案是：构造一种PPG信号质量评估方法，包括步骤：
S1、获取预设时长的原始PPG信号，其中，所述PPG信号为光电容积脉搏波形信号；
S2、预处理所述原始PPG信号以得到若干预设长度的波形片段，获取若干所述波形片段的平均幅度差，判断所述平均幅度差是否在第一预设范围内，若是，则执行步骤S3，否则执行步骤S11；
S3、对所述原始PPG信号分别进行高通滤波和低通滤波，基于高通滤波结果和低通滤波结果获取所述原始PPG信号的有用信号；
S4、基于所述有用信号和所述高通滤波结果获取所述原始PPG信号的高频信噪比，基于所述有用信号和所述低通滤波结果获取所述原始PPG信号的低频信噪比，判断所述高频信噪比和所述低频信噪比是否分别大于其对应的第一门限值，若是，执行步骤S5，否则执行步骤S11；
S5、获取所述有用信号的自相关函数，对所述自相关函数进行傅里叶变换以得到所述有用信号的功率谱，分别获取所述功率谱的峰度和偏度，判断所述功率谱的峰度和偏度是否分别大于其对应的第二门限值，若是，执行步骤S6，否则执行步骤S11；
S6、获取所述自相关函数的平均峰值周期，确认所述平均峰值周期是否在第二预设范围内，若是，则执行步骤S7，否则执行步骤S11；
S7、获取所述自相关函数的最大峰值和峰值周期标准差，确认是否出现所述最大峰值大于第三门限值或所述峰值周期标准差小于第四门限值，若是，则执行步骤S8，否则执行步骤S11；
S8、获取所述自相关函数的峰峰差值标准差，确认所述峰峰差值标准差是否小于第五门限值，若是，则执行步骤S9，否则，执行步骤S10；
S9、判定所述原始PPG信号为第一等级信号，并结束；
S10、判定所述原始PPG信号为第二等级信号，所述第二等级信号质量低于所述第一等级信号质量，并结束；
S11、判定所述原始PPG信号为第三等级信号，所述第三等级信号质量低于所述第二等级信号质量，并结束。
优选地，在本发明的PPG信号质量评估方法中，在所述步骤S1中，所述预设时长为大于或等于5s小于或等于15s。
优选地，在本发明的PPG信号质量评估方法中，在所述步骤S2中，
所述预处理所述原始PPG信号以得到若干预设长度的波形片段，包括：
对所述原始PPG信号进行中值滤波，对滤波后的信号以所述预设长度的滑动窗口依次进行无间隔截取以获取若干所述波形片段，其中所述波形片段之间不重叠；和/或，
所述获取所述若干波形片段的平均幅度差，包括：
获取每一波形片段的幅度差，并获取所有波形片段的幅度差的平均值为所述平均幅度差。
优选地，在本发明的PPG信号质量评估方法中，所述第一预设范围大于或等于150且小于或等于90000；和/或，
所述预设长度为1s。
优选地，在本发明的PPG信号质量评估方法中，在所述步骤S3中，所述对所述原始PPG信号分别进行高通滤波和低通滤波，基于高通滤波结果和低通滤波结果获取所述原始PPG信号的有用信号；包括：
S31、以截止频率为第一频率的巴特沃斯高通滤波器对所述原始PPG信号进行高通滤波，以得到高频噪声信号；
S32、以截止频率为第二频率的巴特沃斯低通滤波器对所述原始PPG信号进行低通滤波，以得到低频噪声信号；
S33、对所述原始PPG信号分别剔除所述高频噪声信号和所述低频噪声信号，以剩余信号为所述有用信号。
优选地，在本发明的PPG信号质量评估方法中，在所述步骤S4中，
所述基于所述有用信号和所述高通滤波结果获取所述原始PPG信号的高频信噪比，包括：获取所述高频噪声信号与所述有用信号的比值为所述高频信噪比；
所述基于所述有用信号和所述低通滤波结果获取所述原始PPG信号的低频信噪比，包括：获取所述低频噪声信号与所述有用信号的比值为所述低频信噪比。
优选地，在本发明的PPG信号质量评估方法中，在所述步骤S4中，所述高频信噪比对应的第一门限值大于或等于6dB，所述低频信噪比对应的第一门限值大于或等于-15dB；和/或，
在所述步骤S5中，所述功率谱的偏度对应的第二门限值大于或等于12，所述功率谱的峰度对应的第二门限值大于或等于150。
优选地，在本发明的PPG信号质量评估方法中，在所述步骤S5中，所述获取所述有用信号的自相关函数，包括：对所述有用信号做单边归一化的自相关函数，其中所述自相关函数满足以下公式：
Figure PCTCN2022115490-appb-000001
其中，R[k]为所述自相关函数，k为所述自相关函数的索引值，x s[n]为所述有用信号，n为 所述有用信号的索引值，N为所述有用信号的点数。
优选地，在本发明的PPG信号质量评估方法中，所述方法还包括：
基于二阶导数的波谷检测算法对自相关函数进行峰值检测以获取若干峰值坐标；
基于所述若干峰值坐标分别获取所述自相关函数的平均峰值周期、最大峰值、峰值周期标准差和峰峰差值标准差。
优选地，在本发明的PPG信号质量评估方法中，所述方法包括以下参数设置中的一个或多个；
所述第二预设范围为大于0.3s且小于2s；
所述第三门限值大于或等于0.6；
所述第四门限值小于或等于5；
所述第五门限值小于或等于0.07。
另，本发明还构造一种PPG信号处理方法，包括，通过上面任意一项所述的PPG信号质量评估方法获取原始PPG信号的质量评估结果；并在所述原始PPG信号为第三等级信号时，剔除所述原始PPG信号；
所述原始PPG信号为第二等级信号时，将所述原始PPG信号用于部分预设生理指标计算；所述原始PPG信号为第一等级信号时，将原始PPG信号用于与所述PPG信号相关的所有生理指标计算。
另，本发明还构造一种PPG信号质量评估装置，包括：
原始信号获取单元，用于获取预设时长的原始PPG信号，其中，所述PPG信号为光电容积脉搏波形信号；
第一判断单元，用于预处理所述原始PPG信号以得到若干预设长度的波形片段，获取所述若干波形片段的平均幅度差，判断所述平均幅度差是否在第一预设范围内，若是，则输出肯定结果，否则输出否定结果；
有用信号获取单元，用于对所述原始PPG信号分别进行高通滤波和低通滤波，基于高通滤波结果和低通滤波结果获取所述原始PPG信号的有用信号；
第二判断单元，用于基于所述有用信号和所述高通滤波结果获取所述原始PPG信号的高频信噪比，基于所述有用信号和所述低通滤波结果获取所述原始PPG信号的低频信噪比，判断所述高频信噪比和所述低频信噪比是否分别大于其对应的第一门限值，若是，则输出肯定结果，否则输出否定结果；
第三判断单元，用于获取所述有用信号的自相关函数，对所述自相关函数进行傅里叶变换以 得到所述有用信号的功率谱，分别获取所述功率谱的峰度和偏度，判断所述功率谱的峰度和偏度是否分别大于其对应的第二门限值，若是，则输出肯定结果，否则输出否定结果；
第四判断单元，用于获取所述自相关函数的平均峰值周期，判断所述平均峰值周期是否在第二预设范围内，若是，则输出肯定结果，否则输出否定结果；
第五判断单元，用于获取所述自相关函数的最大峰值和峰值周期标准差，确认是否出现所述最大峰值大于第三门限值或所述峰值周期标准差小于第四门限值，若是，则输出肯定结果，否则输出否定结果；
第六判断单元，用于获取所述自相关函数的峰峰差值标准差，确认所述峰峰差值标准差是否小于第五门限值，若是，则输出肯定结果，否则输出否定结果；
结果确认单元，用于在所述第一判断单元、所述第二判断单元、所述第三判断单元、所述第四判断单元和所述第六判断单元中任意一个输出否定结果时，判定所述原始PPG信号为第三等级信号，在所述第六判断单元输出否定结果时，判定所述原始PPG信号为第二等级信号，在所述第六判断单元输出肯定结果时，判定所述原始PPG信号为第一等级信号，其中，所述第三等级信号质量低于所述第二等级信号质量，所述第二等级信号质量低于所述第一等级信号质量。
另，本发明还构造一种PPG信号处理***，包括如上面所述的PPG信号质量评估装置，以及处理单元；
所述处理单元用于获取原始PPG信号的质量评估结果；并在所述原始PPG信号为第三等级信号时，剔除所述原始PPG信号；
所述原始PPG信号为第二等级信号时，将所述原始PPG信号用于部分预设生理指标计算；
所述原始PPG信号为第一等级信号时，将原始PPG信号用于与所述PPG信号相关的所有生理指标计算。
实施本发明的PPG信号质量评估方法及装置以及PPG信号处理方法及***，具有以下有益效果：能够实现PPG信号的多层级量化指标的评价，以实现基于PPG信号的不同质量等级进行合理的利用。
附图说明
下面将结合附图及实施例对本发明作进一步说明，附图中：
图1是本发明一实施例中的一种PPG信号质量评估方法的程序流程图；
图2是本发明一实施例中的原始PPG信号的示意图；
图3是本发明一实施例中的原始PPG信号的平均幅度差的示意图；
图4是本发明一实施例中的原始PPG信号的低频信噪比的示意图；
图5是本发明一实施例中的原始PPG信号的低频信噪比的示意图；
图6是本发明一实施例中的原始PPG信号的功率谱指标的示意图；
图7是本发明另一实施例中的原始PPG信号的功率谱指标的示意图；
图8是本发明一实施例中的原始PPG信号的自相关函数指标的示意图；
图9是本发明另一实施例中的原始PPG信号的自相关函数指标的示意图；
图10是本发明再一实施例中的原始PPG信号的自相关函数指标的示意图；
图11是本发明一实施例中的一种PPG信号质量评估装置的逻辑框图；
图12是本发明一实施例中的一种PPG信号质量处理***的逻辑框图。
具体实施方式
为了对本发明的技术特征、目的和效果有更加清楚的理解，现对照附图详细说明本发明的具体实施方式。
如图1所示，在本发明的一种PPG信号质量评估方法第一实施例中，包括：S1、获取预设时长的原始PPG信号，其中，PPG信号为光电容积脉搏波形信号。具体的，可以通过采集设备获取一段预设时长的PPG信息即对应原始PPG信号。其中，原始PPG信号的时间长度不能设置太长，否则信号的波动十分剧烈，而且计算复杂度会随着时长的增加而升高。其原始PPG信号的长度至少是5s，最多不超过15s。根据正常最低心率为30次每分钟，5s时间段的波形至少包含两到三个脉冲周期。如图2所示，其可以选择原始PPG信号的长度为10s，其在对应采样频率f s＝100Hz时，信号的总点数为1000。
S2、预处理原始PPG信号以得到若干预设长度的波形片段，获取若干波形片段的平均幅度差，判断平均幅度差是否在第一预设范围内，若是，则执行步骤S3，否则执行步骤S11。具体的，当设备和皮肤接触不良时或者没有佩戴时，采集设备得到的原始PPG信号会很微弱甚至没有，此时原始PPG信号的平均幅度差很低。相反，当硬件电路中出现饱和噪声或者尖峰脉冲时，信号的平均幅度差会变得异常高。因此，可以对原始PPG信号进行分段，得到预设长度的若干波形片段，基于该若干波形片段得到其平均幅度差。对该平均幅度差进行判断。当平均幅度差满足第一预设范围的要求时，即此时可以判断该原始PPG信号的幅度正常，其可以继续执行步骤S3及其之后的动作。当平均幅度差不满足第一预设范围的要求时，则判断该原始PPG信号的幅度异常，此时可以直接执行后面的判定动作。即可以直接判定原始PPG信号为第三等级信号，并结束该原始PPG信号的质量评估过程。该过程可以定义为第一层级评估过程，正常PPG波形的平均幅度差的范围是几百到几万。但是 当设备和皮肤接触不良或者没有佩戴时，PPG信号的平均幅度差低于150时，PPG信号的波形近似一条直线。如图3所示，当硬件电路中的电流调节得过大时，PPG信号的平均幅度差会大于90000，PPG信号的波形就会出现饱和失真。如果原始PPG信号的平均幅度差出现异常，将评估该原始PPG信号为差的信号，就没有必要进行下一层级的评估了。这里的150和9000对应的就是离散后信号的数字，其由设定的量化精度决定。
在一实施例中，在上述步骤S2中，预处理原始PPG信号以得到若干预设长度的波形片段，包括：对原始PPG信号进行中值滤波，对滤波后的信号以预设长度的滑动窗口依次进行无间隔截取以获取若干波形片段，其中波形片段之间不重叠。即，其可以首先对原始PPG信号进行窗口大小为5的中值滤波，去除原始PPG信号中离群点的影响，得到信号x[n]。其中，在该中值滤波滤波过程中，五个数中取中间那个数，这样就可以去除异常离群点的影响。比如信号为【1，2，3，4，25】，最后输出的是3，避免了离群点25的影响了，以L点长(对应预设长度)的滑动窗口将信号x[n]分成M个不重叠的片段，每个片段记作：x m[l]，0≤m≤M-1，0≤l≤L-1。
其中，x m[l]为波形片段，下标m为该波形片段的索引值，即代表第几个小信号片段，l为该波形片段的长度。如对信号x[n]，其中n取值是【0，N-1】，分别对应10s的信号。在M为10时得到的波形片段x m[l]，其中m取值【0，M-1】，分别对应1s的信号片段。
在一实施例中，当信号的采样频率f s＝100Hz，信号x[n]长度为10s，L取100个点，即对应的预设长度也可以理解为1s，所以M＝10。
在一实施例中，在上述步骤S2中，获取若干波形片段的平均幅度差，包括：获取每一波形片段的幅度差，并获取所有波形片段的幅度差的平均值为平均幅度差。即，对得到的波形片段计算每个波形片段的幅度差Δm，
Δ m＝max(x m[l])-min(x m[l])。
其中，max(x m[l])为x m[l]波形片段中的幅度最大值，min(x m[l])为x m[l]波形片段中的幅度最小值，所以得到所有的波形片段的平均幅度差Amp为：
Figure PCTCN2022115490-appb-000002
其中，对平均幅度差Amp的判断可以为，在其大于或等于150且小于或等于90000判定其满足第一预设范围。
S3、对原始PPG信号分别进行高通滤波和低通滤波，基于高通滤波结果和低通滤波结果获取原始PPG信号的有用信号。具体的，当在原始PPG信号的幅度指标满足要求时继续对原始PPG信号的噪声进行判断。虽然，PPG信号的主要频率是0.5Hz-5Hz，大于5Hz的频率是信号的高频毛刺噪声，能量占比少；小于0.5Hz的是基线漂移相关的低频噪声，占了绝大部分能量。但在信号处理过程中，需要对高频噪声和低频噪声的影响分别进行判断。假设一段信号没有基线漂移，即它的低频噪声很小，但是高频噪声非常大，那么这段信号就是高频噪声太大导致信号无法使用了。因此需要剔除其中的高频噪声和低频噪声，最终得到原始PPG信号的有用信号，并基于有用信号对原始PPG信号进行判定。
可选的，在步骤S3中，对原始PPG信号分别进行高通滤波和低通滤波，基于高通滤波结果和低通滤波结果获取原始PPG信号的有用信号；包括：S31、以截止频率为第一频率的巴特沃斯高通滤波器对原始PPG信号进行高通滤波，以得到高频噪声信号；S32、以截止频率为第二频率的巴特沃斯低通滤波器对原始PPG信号进行低通滤波，以得到低频噪声信号；S33、对原始PPG信号分别剔除高频噪声信号和低频噪声信号，以剩余信号为有用信号。具体的，其可以对原始PPG信号分别通过高通滤波器进行滤波得到高频噪声信号和通过低通滤波器进行滤波得到低频噪声信号。在这里高通滤波器和低通滤波器均可以采用巴特沃斯滤波器。此外，可以设置高通滤波器的截止频率即第一频率f high＝5Hz,得到高频噪声信号x high[n]，则高频噪声功率P high如下：
Figure PCTCN2022115490-appb-000003
可以设置低通滤波器的截止频率即第二频率f low＝0.5Hz,得到低频噪声信号x low[n]，则低频噪声功率P low如下：
Figure PCTCN2022115490-appb-000004
最后，将原始PPG信号分别减去高频和低频噪声信号，可以得到其对应的有用信号x s[n]，则有用信号功率P s如下：
Figure PCTCN2022115490-appb-000005
可以理解，对信号进行滤波后，得到高频噪声信号x high[n]和低频噪声信号x low[n]对应的长度与原始信号x[n]长度一样，例如，N＝1000。
S4、基于有用信号和高通滤波结果获取原始PPG信号的高频信噪比，基于有用信号和低通滤波结果获取原始PPG信号的低频信噪比，判断高频信噪比和低频信噪比是否分别大于其对应的第一门限值，若是，执行步骤S5，否则执行步骤S11。具体的，由于高频噪声和低频噪声对原始PPG信号的影响结果不同，因此需要进行高频信噪比和低频信噪比的区分，其可以根据高通滤波结果和低通滤波结果分别获取原始PPG信号的高频信噪比和低频信噪比，并在其高频信噪比和低频信噪比均满足要求的情况下，判定原始PPG信号的噪声满足要求，可以继续执行后面步骤S5的动作。当出现高频噪声比和低频噪声比中任意一个不满足要求的情况，则可以判断该原始PPG信号的噪声异常，此时可以直接执行后面的判断动作。即可以直接判定原始PPG信号为第三等级信号，并结束该原始PPG信号的质量评估过程。该过程可以定义为第二层级评估过程。PPG信号的噪声主要来源于高频毛刺噪声和低频的漂移噪声。如图4所示，当低频信噪比小于-15dB时，PPG信号的整体漂移十分严重，波动十分剧烈，完全是不规则的状态。如图5所示，当高频信噪比小于6dB时，PPG波形形态完全被毛刺掩盖，即使去除掉高频噪声信号和低频噪声信号也无法观察到其形态和周期。总之，频低频信噪比太低时，信号十分微弱被噪声掩盖，甚至通带内噪声的能量也远比信号的能量高，或者根本没有检测到PPG信号。因此，在信噪比出现异常时可以将此原始PPG信号直接判定为差的信号，没必要进行第三层的评估。
可选的，在步骤S4中，基于有用信号和高通滤波结果获取原始PPG信号的高频信噪比，包括：获取高频噪声信号与有用信号的比值为高频信噪比；基于有用信号和低通滤波结果获取原始PPG信号的低频信噪比，包括：获取低频噪声信号与有用信号的比值为低频信噪比。其具体的高频信噪比SNR high获取过程如下：
Figure PCTCN2022115490-appb-000006
其具体的低频信噪比SNR low获取过程如下：
Figure PCTCN2022115490-appb-000007
在一实施例中，高频信噪比对应的第一门限值大于或等于6dB，低频信噪比对应的第一门限值大于或等于-15dB。即，在高频信噪比对应的第一门限值为6dB的时候，即在高频信噪比SNR high大于6dB时可以判定该高频信噪比满足要求。在低频信噪比对应的第一门限值为-15dB时，即在低频信噪比SNR low大于-15dB时可以判定低频信噪比满足要求。
S5、获取有用信号的自相关函数，对自相关函数进行傅里叶变换以得到有用信号的功率谱，分别获取功率谱的峰度和偏度，判断功率谱的峰度和偏度是否分别大于其对应的第二门限值，若是，执行步骤S6，否则执行步骤S11。具体的，在原始PPG信号的噪声指标满足要求时，对功率谱相关指标进行判断。其主要对原始PPG信号对应的频率普的偏度和峰度进行判断。当原始PPG信号中存在有间断点和尖脉冲时，这些突变会导致功率谱系数将分布整个频率轴上，无法得到好的稀疏表示。这类突变信号的高频信噪比和低频信噪比都可能很高，无法通过信噪比相关指标检测出来，但其可以通过计算有用信号的功率谱的偏度和峰度衡量其稀疏表示。其先获取有用信号的自相关函数，并基于该自相关函数进行傅里叶变化得到有用信号的功率谱。基于得到的功率谱获取其对应的偏度和峰度。该过程可以定义为对功率谱相关指标可进行评估的第三层级评估过程。可以理解，第一层级和第二层层级的评估过程为从整体指标的角度去评价信号，当信号中局部小区域出现间断点和尖脉冲时，这些突变会导致功率谱系数将分布整个频率轴上，无法得到很好的稀疏表示。如图6和图7所示，此时通过功率谱指标能很好区分信号中是否出现突变区域造成的异常。
在一具体实施例中，其可以对有用信号x s[n]做单边归一化的自相关函数，其具体函数R[k]表达式如下：
Figure PCTCN2022115490-appb-000008
可以理解，由于N为有用信号的点数，对信号进行单边归一化的自相关函数的长度也是1000点。为了区分原始信号x[n]，把自相关函数称为R[k]，其中k为所述自相关函数的索引值，其取值范围是0到N-1，n代表着有用信号的索引值，其取值的范围也是0到N-1。
由维纳-辛钦定理可知，自相关函数的傅里叶变换对应为信号的功率谱。对自相关函 数R|k|进行傅里叶变换得到有用信号的功率谱P xx(f)满足如下表达式：
Figure PCTCN2022115490-appb-000009
功率谱的峰度P SK和偏度P KT分别如下：
Figure PCTCN2022115490-appb-000010
Figure PCTCN2022115490-appb-000011
其中，μ为P xx(f)的均值，σ为P xx(f)的方差。
对功率谱的偏度和峰度同时进行判断，在其均满足对应的门限要求时，即均大于其对应的第二门限值要求时，则判定功率谱的信号为正常，可以继续执行后面步骤S6的动作。当功率谱的偏度和峰度中任意一个不满足其对应的第二门限值，则可以判断该原始PPG信号的存在突变区域，此时直接执行后面的判断动作。即可以直接判定原始PPG信号为第三等级信号，并结束该原始PPG信号的质量评估过程。
在一实施例中，在步骤S5中，功率谱的偏度对应的第二门限值大于或等于12，功率谱的峰度对应的第二门限值大于或等于150。即，可以分别设置功率谱的偏度对应的第二门限值和功率谱的峰度对应的第二门限值。
S6、获取自相关函数的平均峰值周期，确认平均峰值周期是否在第二预设范围内，若是，则执行步骤S7，否则执行步骤S11；具体的，由于自相关函数计算的信号本身与其时间延迟之间的相关性，当信号存在周期性成分时，自相关函数会在周期的整数倍出现极大值。因此可以通过检测自相关函数的峰值检验其周期性。即对自相关函数的平均峰值周期进行判定。当其在第二预设范围内，则判定原始PPG信号在被采样信号例如心率信号的正常范围内，其可以继续步骤S7的动作。当自相关函数的平均峰值周期不在第二预设范围时，则判定原始PPG信号不在心率的正常范围内，此时可以直接执行后面的判断动作，即可以直接判定原始PPG信号为第三等级信号，并结束该原始PPG信号的质量评估过程。
在一实施例中，第二预设范围为大于0.3s且小于2s。即可以在一具体实施例中，设置当平均峰值周期大于0.3s小于2s时，判定其在第二预设范围内。
S7、获取自相关函数的最大峰值和峰值周期标准差，确认是否出现最大峰值大于第三门限值或峰值周期标准差小于第四门限值，若是，则执行步骤S8，否则执行步骤S11；具体 的，在自相关函数的平均峰值周期满足要求即判定原始PPG信号在检测信号的正常范围内时，对自相关函数的最大峰值和峰值周期标准差进行判断，即对原始PPG信号周期性进行判断。当判定原始PPG信号存在周期性时，其可以继续步骤S8的动作。当判定原始PPG信号不存在周期性时，可以直接执行后面的判断动作，即可以直接判定原始PPG信号为第三等级信号，并结束该原始PPG信号的质量评估过程。其具体判定过程为在同时出现自相关函数的最大峰值大于第三门限值和峰值周期标准差小于第四门限值时，判定原始PPG信号不存在周期性，否则可以判定原始PPG信号存在周期性。
在一实施例中，第三门限值大于或等于0.6，第四门限值小于或等于5；具体的，其可以在相关函数的最大峰值大于0.6时判定原始PPG信号存在周期性，其也可以在自相关函数的峰值周期标准差小于5时判定原始PPG信号存在周期性。
S8、获取自相关函数的峰峰差值标准差，确认峰峰差值标准差是否小于第五门限值，若是，则执行步骤S9，否则，执行步骤S10；S9、判定原始PPG信号为第一等级信号，并结束；S10、判定原始PPG信号为第二等级信号，第二等级信号质量低于第一等级信号质量，并结束；S11、判定原始PPG信号为第三等级信号，第三等级信号质量低于第二等级信号质量，并结束。具体的，在根据自相关函数的最大峰值和峰值周期标准差即判定原始PPG信号在检测信号的存在周期性时，对自相关函数的峰峰差值标准差进行判断，根据峰峰差值标准差判定原始PPG信号的各个周期的信号形态是否相似，其在自相关函数的峰峰差值标准差小于第五门限值时，可以判定其信号形态相似，此时可以判定该判定原始PPG信号为第一等级信号，否则判定该原始PPG信号为第二等级信号。其中第一等级信号的质量优于第二等级信号，第二等级信号的质量优于第三等级信号。
可选的，本发明的PPG信号质量评估方法还包括：基于二阶导数的波谷检测算法对自相关函数进行峰值检测以获取若干峰值坐标；基于若干峰值坐标分别获取自相关函数的平均峰值周期、最大峰值、峰值周期标准差和峰峰差值标准差。具体的，其可以先使用基于二阶导数的波谷检测算法对自相关函数进行峰值检测，得到w个所有峰值坐标：
Figure PCTCN2022115490-appb-000012
则自相关函数的最大峰值
Figure PCTCN2022115490-appb-000013
为：
Figure PCTCN2022115490-appb-000014
相邻的峰峰间隔
Figure PCTCN2022115490-appb-000015
为：
Figure PCTCN2022115490-appb-000016
则自相关函数的平均峰值周期R period为：
Figure PCTCN2022115490-appb-000017
则自相关函数的峰值周期标准差
Figure PCTCN2022115490-appb-000018
为：
Figure PCTCN2022115490-appb-000019
峰峰差值
Figure PCTCN2022115490-appb-000020
为：
Figure PCTCN2022115490-appb-000021
平均峰峰差值
Figure PCTCN2022115490-appb-000022
为：
Figure PCTCN2022115490-appb-000023
则自相关函数的峰峰差值的标准差
Figure PCTCN2022115490-appb-000024
为：
Figure PCTCN2022115490-appb-000025
对与自相关函数的相关指标判定的过程可以定义为第四层级评估。其通过相比直接对有用信号进行周期性检验，计算自相关函数，能够消除加性噪声对检验信号周期性的影响。自相关函数的平均峰值周期对应原始PPG信号的周期，根据正常心率30次～200次每分钟，平均峰值周期应该在0.3s和2s之间，超过正常心率范围的评估为差的信号(如图8所示)。当自相关函数的最大峰值大于0.6或者峰值周期标准差小于5时，可以判断出信号是存在周期性的，用具有周期性的信号计算心率等生理指标其准确性会更高，此时，当相邻周期PPG信号形态的一致性、相似性越高时，自相关函数应该是一个逐步衰减的正弦波信号，而个别存在周期性的信号其波形形态可能受到通带内的噪声影响，相邻周期PPG信号的波形形态各不相同，此时其自相关函数的峰值高低起伏不一，所以通过计算自相关函数的峰峰差值的 标准差，能够判断自相关函数峰值的波动情况，当自相关函数的峰峰差值的标准差小于0.07时，评价为好的信号(如图9所示)，否则评价为中等信号(如图10所示)。
根据以上过程中信号质量评估的层级关系，如果前一层的量化指标无法满足要求，就没必要进行下一层级的判断了，这样可以有效减少误检的概率，从而提升质量评估算法的准确性。
另，本发明的一种PPG信号处理方法，包括，通过上面任意一项的PPG信号质量评估方法获取原始PPG信号的质量评估结果；并在原始PPG信号为第三等级信号时，剔除原始PPG信号；原始PPG信号为第二等级信号时，将原始PPG信号用于部分预设生理指标计算；原始PPG信号为第一等级信号时，将原始PPG信号用于与所述PPG信号相关的所有生理指标计算。即，在PPG信号处理过程中，基于上面的PPG信号质量评估方法，对得到的不同等级的PPG信号进行不同的操作。对第三等级信号直接剔除，对第二等级信号用于心率等部分指标的计算过程，该部分指标为预设的部分指标。对第一等级信号则可以正常的使用，即可以用于心率和血压等所有的生理指标的计算，这里所有生理指标为与PPG相关的生理指标。
另，如图11所示，本发明的一种PPG信号质量评估装置100，包括：
原始信号获取单元110，用于获取预设时长的原始PPG信号，其中，PPG信号为光电容积脉搏波形信号；
第一判断单元121，用于预处理原始PPG信号以得到若干预设长度的波形片段，获取若干波形片段的平均幅度差，判断平均幅度差是否在第一预设范围内，若是，则输出肯定结果，否则输出否定结果；
有用信号获取单元130，用于对原始PPG信号分别进行高通滤波和低通滤波，基于高通滤波结果和低通滤波结果获取原始PPG信号的有用信号；
第二判断单元122，用于基于有用信号和高通滤波结果获取原始PPG信号的高频信噪比，基于有用信号和低通滤波结果获取原始PPG信号的低频信噪比，判断高频信噪比和低频信噪比是否分别大于其对应的第一门限值，若是，则输出肯定结果，否则输出否定结果；
第三判断单元123，用于获取有用信号的自相关函数，对自相关函数进行傅里叶变换以得到有用信号的功率谱，分别获取功率谱的峰度和偏度，判断功率谱的峰度和偏度是否分别大于其对应的第二门限值，若是，则输出肯定结果，否则输出否定结果；
第四判断单元124，用于获取自相关函数的平均峰值周期，判断平均峰值周期是否在第二预设范围内，若是，则输出肯定结果，否则输出否定结果；
第五判断单元125，用于获取自相关函数的最大峰值和峰值周期标准差，确认是否出现最大峰值大于第三门限值或峰值周期标准差小于第四门限值，若是，则输出肯定结果，否则输出否定结果；
第六判断单元126，用于获取自相关函数的峰峰差值标准差，确认峰峰差值标准差是否小于第五门限值，若是，则输出肯定结果，否则输出否定结果；
结果确认单元140，用于在第一判断单元121、第二判断单元122、第三判断单元123、第四判断单元124和第五判断单元125中任意一个输出否定结果时，判定原始PPG信号为第三等级信号，在第六判断单元126输出否定结果时，判定原始PPG信号为第二等级信号，在第六判断单元127输出肯定结果时，判定原始PPG信号为第一等级信号，其中，第三等级信号质量低于第二等级信号质量，第二等级信号质量低于第一等级信号质量。具体的，这里的PPG信号质量评估装置各单元之间具体的配合操作过程具体可以参照上述PPG信号质量评估方法，这里不再赘述。
另，如图12所示，本发明的一种PPG信号处理***，包括如上面的PPG信号质量评估装置100，以及处理单元200；处理单元用于获取原始PPG信号的质量评估结果；并在原始PPG信号为第三等级信号时，剔除原始PPG信号；原始PPG信号为第二等级信号时，将原始PPG信号用于部分预设生理指标计算；原始PPG信号为第一等级信号时，将原始PPG信号用于与PPG信号相关的所有生理指标计算。即，PPG信号处理***处理过程中，通过上面的PPG信号质量评估装置得到不同的等级的PPG信号，并通过处理单元对得到的不同等级的PPG信号进行不同的操作。对第三等级信号直接剔除，对第二等级信号用于心率等部分指标的计算过程，该部分指标为预设的部分指标。对第一等级信号则可以正常的使用，即可以用于心率和血压等所有的生理指标的计算，这里所有生理指标为与PPG相关的生理指标。
可以理解的，以上实施例仅表达了本发明的优选实施方式，其描述较为具体和详细，但并不能因此而理解为对本发明专利范围的限制；应当指出的是，对于本领域的普通技术人员来说，在不脱离本发明构思的前提下，可以对上述技术特点进行自由组合，还可以做出若干变形和改进，这些都属于本发明的保护范围；因此，凡跟本发明权利要求范围所做的等同变换与修饰，均应属于本发明权利要求的涵盖范围。

Claims (13)

  1. 一种PPG信号质量评估方法，其特征在于，包括步骤：
    S1、获取预设时长的原始PPG信号，其中，所述PPG信号为光电容积脉搏波形信号；
    S2、预处理所述原始PPG信号以得到若干预设长度的波形片段，获取若干所述波形片段的平均幅度差，判断所述平均幅度差是否在第一预设范围内，若是，则执行步骤S3，否则执行步骤S11；
    S3、对所述原始PPG信号分别进行高通滤波和低通滤波，基于高通滤波结果和低通滤波结果获取所述原始PPG信号的有用信号；
    S4、基于所述有用信号和所述高通滤波结果获取所述原始PPG信号的高频信噪比，基于所述有用信号和所述低通滤波结果获取所述原始PPG信号的低频信噪比，判断所述高频信噪比和所述低频信噪比是否分别大于其对应的第一门限值，若是，执行步骤S5，否则执行步骤S11；
    S5、获取所述有用信号的自相关函数，对所述自相关函数进行傅里叶变换以得到所述有用信号的功率谱，分别获取所述功率谱的峰度和偏度，判断所述功率谱的峰度和偏度是否分别大于其对应的第二门限值，若是，执行步骤S6，否则执行步骤S11；
    S6、获取所述自相关函数的平均峰值周期，确认所述平均峰值周期是否在第二预设范围内，若是，则执行步骤S7，否则执行步骤S11；
    S7、获取所述自相关函数的最大峰值和峰值周期标准差，确认是否出现所述最大峰值大于第三门限值或所述峰值周期标准差小于第四门限值，若是，则执行步骤S8，否则执行步骤S11；
    S8、获取所述自相关函数的峰峰差值标准差，确认所述峰峰差值标准差是否小于第五门限值，若是，则执行步骤S9，否则，执行步骤S10；
    S9、判定所述原始PPG信号为第一等级信号，并结束；
    S10、判定所述原始PPG信号为第二等级信号，所述第二等级信号质量低于所述第一等级信号质量，并结束；
    S11、判定所述原始PPG信号为第三等级信号，所述第三等级信号质量低于所述第二等级信号质量，并结束。
  2. 根据权利要求1所述的PPG信号质量评估方法，其特征在于，在所述步骤S1中，所述预设时长为大于或等于5s小于或等于15s。
  3. 根据权利要求1所述的PPG信号质量评估方法，其特征在于，在所述步骤S2中，
    所述预处理所述原始PPG信号以得到若干预设长度的波形片段，包括：
    对所述原始PPG信号进行中值滤波，对滤波后的信号以所述预设长度的滑动窗口依次进行 无间隔截取以获取若干所述波形片段，其中所述波形片段之间不重叠；和/或，
    所述获取所述若干波形片段的平均幅度差，包括：
    获取每一波形片段的幅度差，并获取所有波形片段的幅度差的平均值为所述平均幅度差。
  4. 根据权利要求1所述的PPG信号质量评估方法，其特征在于，
    所述第一预设范围大于或等于150且小于或等于90000；和/或，
    所述预设长度为1s。
  5. 根据权利要求1所述的PPG信号质量评估方法，其特征在于，在所述步骤S3中，所述对所述原始PPG信号分别进行高通滤波和低通滤波，基于高通滤波结果和低通滤波结果获取所述原始PPG信号的有用信号；包括：
    S31、以截止频率为第一频率的巴特沃斯高通滤波器对所述原始PPG信号进行高通滤波，以得到高频噪声信号；
    S32、以截止频率为第二频率的巴特沃斯低通滤波器对所述原始PPG信号进行低通滤波，以得到低频噪声信号；
    S33、对所述原始PPG信号分别剔除所述高频噪声信号和所述低频噪声信号，以剩余信号为所述有用信号。
  6. 根据权利要求5所述的PPG信号质量评估方法，其特征在于，在所述步骤S4中，
    所述基于所述有用信号和所述高通滤波结果获取所述原始PPG信号的高频信噪比，包括：获取所述高频噪声信号与所述有用信号的比值为所述高频信噪比；
    所述基于所述有用信号和所述低通滤波结果获取所述原始PPG信号的低频信噪比，包括：获取所述低频噪声信号与所述有用信号的比值为所述低频信噪比。
  7. 根据权利要求1所述的PPG信号质量评估方法，其特征在于，在所述步骤S4中，所述高频信噪比对应的第一门限值大于或等于6dB，所述低频信噪比对应的第一门限值大于或等于-15dB；和/或，
    在所述步骤S5中，所述功率谱的偏度对应的第二门限值大于或等于12，所述功率谱的峰度对应的第二门限值大于或等于150。
  8. 根据权利要求1所述的PPG信号质量评估方法，其特征在于，在所述步骤S5中，所述获取所述有用信号的自相关函数，包括：对所述有用信号做单边归一化的自相关函数，其中所述自相关函数满足以下公式：
    Figure PCTCN2022115490-appb-100001
    其中，R[k]为所述自相关函数，k为所述自相关函数的索引值，x s[n]为所述有用信号，n为所述有用信号的索引值，N为所述有用信号的点数。
  9. 根据权利要求1所述的PPG信号质量评估方法，其特征在于，所述方法还包括：
    基于二阶导数的波谷检测算法对自相关函数进行峰值检测以获取若干峰值坐标；
    基于所述若干峰值坐标分别获取所述自相关函数的平均峰值周期、最大峰值、峰值周期标准差和峰峰差值标准差。
  10. 根据权利要求1所述的PPG信号质量评估方法，其特征在于，所述方法包括以下参数设置中的一个或多个：
    所述第二预设范围为大于0.3s且小于2s；
    所述第三门限值大于或等于0.6；
    所述第四门限值小于或等于5；
    所述第五门限值小于或等于0.07。
  11. 一种PPG信号处理方法，其特征在于，包括，通过权利要求1至10任意一项所述的PPG信号质量评估方法获取原始PPG信号的质量评估结果；并在所述原始PPG信号为第三等级信号时，剔除所述原始PPG信号；
    所述原始PPG信号为第二等级信号时，将所述原始PPG信号用于部分预设生理指标计算；
    所述原始PPG信号为第一等级信号时，将原始PPG信号用于与所述PPG信号相关的所有生理指标计算。
  12. 一种PPG信号质量评估装置，其特征在于，包括：
    原始信号获取单元，用于获取预设时长的原始PPG信号，其中，所述PPG信号为光电容积脉搏波形信号；
    第一判断单元，用于预处理所述原始PPG信号以得到若干预设长度的波形片段，获取若干所述波形片段的平均幅度差，判断所述平均幅度差是否在第一预设范围内，若是，则输出肯定结果，否则输出否定结果；
    有用信号获取单元，用于对所述原始PPG信号分别进行高通滤波和低通滤波，基于高通滤波结果和低通滤波结果获取所述原始PPG信号的有用信号；
    第二判断单元，用于基于所述有用信号和所述高通滤波结果获取所述原始PPG信号的高频信噪比，基于所述有用信号和所述低通滤波结果获取所述原始PPG信号的低频信噪比，判断所述高频信噪比和所述低频信噪比是否分别大于其对应的第一门限值，若是，则输出肯定结果，否则输出否定结果；
    第三判断单元，用于获取所述有用信号的自相关函数，对所述自相关函数进行傅里叶变换以得到所述有用信号的功率谱，分别获取所述功率谱的峰度和偏度，判断所述功率谱的峰度和偏度是否分别大于其对应的第二门限值，若是，则输出肯定结果，否则输出否定结果；
    第四判断单元，用于获取所述自相关函数的平均峰值周期，判断所述平均峰值周期是否在第二预设范围内，若是，则输出肯定结果，否则输出否定结果；
    第五判断单元，用于获取所述自相关函数的最大峰值和峰值周期标准差，确认是否出现所述最大峰值大于第三门限值或所述峰值周期标准差小于第四门限值，若是，则输出肯定结果，否则输出否定结果；
    第六判断单元，用于获取所述自相关函数的峰峰差值标准差，确认所述峰峰差值标准差是否小于第五门限值，若是，则输出肯定结果，否则输出否定结果；
    结果确认单元，用于在所述第一判断单元、所述第二判断单元、所述第三判断单元、所述第四判断单元和所述第六判断单元中任意一个输出否定结果时，判定所述原始PPG信号为第三等级信号，在所述第六判断单元输出否定结果时，判定所述原始PPG信号为第二等级信号，在所述第六判断单元输出肯定结果时，判定所述原始PPG信号为第一等级信号，其中，所述第三等级信号质量低于所述第二等级信号质量，所述第二等级信号质量低于所述第一等级信号质量。
  13. 一种PPG信号处理***，其特征在于，包括如权利要求12所述的PPG信号质量评估装置，以及处理单元；
    所述处理单元用于获取原始PPG信号的质量评估结果；并在
    所述原始PPG信号为第三等级信号时，剔除所述原始PPG信号；
    所述原始PPG信号为第二等级信号时，将所述原始PPG信号用于部分预设生理指标计算；
    所述原始PPG信号为第一等级信号时，将原始PPG信号用于与所述PPG信号相关的所有生理指标计算。
PCT/CN2022/115490 2022-05-27 2022-08-29 Ppg信号质量评估方法及装置以及ppg信号处理方法及*** WO2023226223A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210585033.1 2022-05-27
CN202210585033.1A CN114676738B (zh) 2022-05-27 2022-05-27 Ppg信号质量评估方法及装置以及ppg信号处理方法及***

Publications (1)

Publication Number Publication Date
WO2023226223A1 true WO2023226223A1 (zh) 2023-11-30

Family

ID=82079397

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/115490 WO2023226223A1 (zh) 2022-05-27 2022-08-29 Ppg信号质量评估方法及装置以及ppg信号处理方法及***

Country Status (2)

Country Link
CN (1) CN114676738B (zh)
WO (1) WO2023226223A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114676738B (zh) * 2022-05-27 2022-08-02 广东玖智科技有限公司 Ppg信号质量评估方法及装置以及ppg信号处理方法及***
CN115670397B (zh) * 2022-11-17 2023-06-02 北京中科心研科技有限公司 一种ppg伪迹识别方法、装置、存储介质及电子设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108523877A (zh) * 2018-03-23 2018-09-14 南京中医药大学 一种心电信号质量辨识方法及其心电分析方法
US20190133533A1 (en) * 2017-11-06 2019-05-09 Tata Consultancy Services Limited System and method for photoplethysmogram (ppg) signal quality assessment
CN111493821A (zh) * 2020-03-16 2020-08-07 浙江好络维医疗技术有限公司 一种基于modwt及中值滤波的ppg信号实时去噪方法
CN114676738A (zh) * 2022-05-27 2022-06-28 广东玖智科技有限公司 Ppg信号质量评估方法及装置以及ppg信号处理方法及***

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103020472B (zh) * 2012-12-27 2015-12-09 中国科学院深圳先进技术研究院 基于约束估计的生理信号质量评估方法和***
WO2017089921A1 (en) * 2015-11-24 2017-06-01 Koninklijke Philips N.V. Method to quantify photoplethysmogram (ppg) signal quality
CN109598180A (zh) * 2017-09-30 2019-04-09 深圳市岩尚科技有限公司 光电容积脉搏波的质量评估方法
CN112494001B (zh) * 2020-11-25 2023-05-26 青岛歌尔智能传感器有限公司 Ppg信号质量评估方法、装置、穿戴式设备及存储介质
CN112545472B (zh) * 2020-12-02 2022-11-01 成都心吉康科技有限公司 一种ppg信号质量评估的方法、装置、设备及储存介质
CN114528888B (zh) * 2022-04-25 2022-07-12 广东玖智科技有限公司 一种ppg信号聚类中心获取方法及装置和ppg信号处理方法及装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190133533A1 (en) * 2017-11-06 2019-05-09 Tata Consultancy Services Limited System and method for photoplethysmogram (ppg) signal quality assessment
CN108523877A (zh) * 2018-03-23 2018-09-14 南京中医药大学 一种心电信号质量辨识方法及其心电分析方法
CN111493821A (zh) * 2020-03-16 2020-08-07 浙江好络维医疗技术有限公司 一种基于modwt及中值滤波的ppg信号实时去噪方法
CN114676738A (zh) * 2022-05-27 2022-06-28 广东玖智科技有限公司 Ppg信号质量评估方法及装置以及ppg信号处理方法及***

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LI SU-YI; XU ZHUANG; XIOMG WEN-JI; JIANG SHAN-QING; WU JIANG: "An Automatic Peak Identification Method for Photoplethysmography Signals", SPECTROSCOPY AND SPECTRAL ANALYSIS, BEIJING DAXUE CHUBANSHE, CN, vol. 37, no. 10, 15 October 2017 (2017-10-15), CN , pages 3145 - 3149, XP009550679, ISSN: 1000-0593 *

Also Published As

Publication number Publication date
CN114676738A (zh) 2022-06-28
CN114676738B (zh) 2022-08-02

Similar Documents

Publication Publication Date Title
WO2023226223A1 (zh) Ppg信号质量评估方法及装置以及ppg信号处理方法及***
Papadaniil et al. Efficient heart sound segmentation and extraction using ensemble empirical mode decomposition and kurtosis features
US10194821B2 (en) Medical device having automated ECG feature extraction
CN108388912A (zh) 基于多传感器特征优化算法的睡眠分期方法
CN111407315B (zh) 一种针对超声多普勒胎心信号的质量评估方法
CN110037668B (zh) 脉搏信号时空域结合模型判断年龄、健康状态及恶性心律失常识别的***
US20210267551A1 (en) Noise detection method and apparatus
CN105997043B (zh) 一种基于腕式可穿戴设备的脉率提取方法
CN112545472B (zh) 一种ppg信号质量评估的方法、装置、设备及储存介质
CN111839488A (zh) 基于脉搏波的无创连续血压测量装置和方法
KR101510522B1 (ko) 심전도 신호 분류 시스템
CN111419219A (zh) Ppg心搏信号的预处理方法、装置及房颤检测设备
WO2023165482A1 (zh) 一种心率检测方法和装置
CN113558584A (zh) 一种基于信号质量评估的脉搏波预处理方法
Lin et al. A characteristic filtering method for pulse wave signal quality assessment
US20150182140A1 (en) Arterial pulse analysis method and system thereof
CN112656392A (zh) 一种低功耗ppg心率计算方法及电子设备
Cherif et al. Detection of artifacts on photoplethysmography signals using random distortion testing
CN110368020B (zh) 一种心音信号预处理方法及装置
CN116502124A (zh) 脉搏波信号的质量评估方法及装置、计算机可读存储介质
CN110432885A (zh) 一种光电容积脉搏波噪声去除方法
Thakker et al. Outlier pulse detection and feature extraction for wrist pulse analysis
CN110368019B (zh) 一种心音信号特征提取、检测模型构建及检测装置
Sivaranjini et al. Computationally-efficient pulse rate estimation from compressed ppg measurements for continuous vital signs monitoring
CN113907770A (zh) 基于特征融合的棘慢复合波检测识别方法及***