阅读说明：本技术 动态生理信息检测系统与方法 (Dynamic physiological information detection system and method ) 是由 吴芳铭 于 2019-09-19 设计创作，主要内容包括：本发明提供一种动态生理信息检测系统与方法。该动态生理信息检测系统包含射频检测装置,产生多个检测信号；校正装置,用以校正检测信号；特征萃取装置,其根据至少一种特征,处理经校正的检测信号,用以得到多个萃取值,并过滤掉非稳定的萃取值；及生理信息决定装置,其根据过滤后的萃取值以决定生理信息。本发明可以提高测量准确度。(The invention provides a dynamic physiological information detection system and a method. The dynamic physiological information detection system comprises a radio frequency detection device, a signal processing device and a signal processing device, wherein the radio frequency detection device is used for generating a plurality of detection signals; a correction device for correcting the detection signal; a feature extraction device that processes the corrected detection signal according to at least one feature to obtain a plurality of extraction values and filters out unstable extraction values; and a physiological information determining device for determining physiological information based on the filtered extracted value. The invention can improve the measurement accuracy.)

1. A dynamic physiological information detection system, comprising:

a radio frequency detection device for generating a plurality of detection signals;

a calibration device for calibrating the plurality of detection signals;

a feature extraction device, which processes the corrected detection signals according to at least one feature to obtain a plurality of extraction values and filters unstable extraction values; and

and the physiological information determining device determines physiological information according to the plurality of extracted values after filtering.

2. The system for detecting dynamic physiological information of claim 1, wherein the calibration device comprises:

a filter for removing unwanted frequency components of the plurality of detection signals;

a nonlinear suppressing unit for suppressing nonlinear components of the detection signals; and

a regularization unit regularizes the plurality of detection signals.

3. The system for detecting dynamic physiological information of claim 1, wherein the feature extraction device comprises:

a sliding window unit, which selects a plurality of signal sections to be processed from the plurality of detection signals according to the size of a preset window and time sequence;

a physiological information estimation unit for estimating corresponding preliminary physiological information of the plurality of signal segments; and

and the characteristic unit is used for extracting corresponding extracted values of the plurality of signal sections according to the at least one characteristic and filtering unstable extracted values through a preset threshold value.

4. The dynamic physiological information detection system of claim 3, wherein the physiological information estimation unit estimates the preliminary physiological information using a zero-crossing rate method.

5. The physiological information detecting system of claim 3, wherein the detecting signals include an in-phase signal, a quadrature signal and a phase signal.

6. The physiological information monitoring system of claim 5, wherein the physiological information determining device performs the following steps:

(a) counting the preliminary physiological information corresponding to the plurality of signal sections of the phase signal, and taking the preliminary physiological information corresponding to the maximum accumulated times which is greater than a first preset value as the physiological information;

(b) if the physiological information cannot be determined in the step (a), counting the preliminary physiological information corresponding to the signal sections of the in-phase signal and the orthogonal signal, and taking the preliminary physiological information corresponding to the maximum accumulated times which is greater than a second preset value as the physiological information;

(c) if the physiological information cannot be determined in the step (b), counting the preliminary physiological information corresponding to the signal sections of the in-phase signal and the orthogonal signal, and averaging the preliminary physiological information corresponding to the accumulated times greater than a third preset value to obtain an average value as the physiological information; and

(d) if the physiological information cannot be determined in the step (c), counting preliminary physiological information corresponding to the plurality of signal sections of the in-phase signal, the orthogonal signal and the phase signal, and taking the preliminary physiological information corresponding to the maximum accumulated times that the preliminary physiological information is greater than a preset asphyxia critical value and greater than a fourth preset value as the physiological information, or taking the preliminary physiological information corresponding to the maximum accumulated times that the preliminary physiological information is not greater than the preset asphyxia critical value and greater than a fifth preset value as the physiological information;

wherein the preliminary physiological information counted in the steps (a) to (c) is the preliminary physiological information after the filtering of the feature unit, but the preliminary physiological information counted in the step (d) is the preliminary physiological information before the filtering of the feature unit.

7. The system according to claim 1, wherein the at least one characteristic comprises one or more of the following: full width at half maximum, strongest gain value, peak value, root mean square, standard deviation, and peak-to-peak difference.

8. The physiological information detecting system of claim 7, wherein the feature extracting device uses a plurality of fitting methods to fit the DC levels of the detecting signals to obtain a plurality of fitting curves.

9. The system of claim 1, wherein the physiological information comprises a respiration rate.

10. A dynamic physiological information detection method is characterized by comprising the following steps:

(I) generating a plurality of detection signals;

(II) correcting the plurality of detection signals;

(III) processing the corrected plurality of detection signals according to at least one characteristic to obtain a plurality of extracted values and filtering out unstable extracted values; and

(IV) determining physiological information according to the plurality of extracted values after filtration.

11. The method for detecting dynamic physiological information of claim 10, wherein the step (II) comprises:

(IIa) removing unnecessary frequency components of the plurality of detection signals;

(IIb) suppressing nonlinear components of the plurality of detection signals; and

(IIc) regularizing the plurality of detection signals.

12. The method for detecting dynamic physiological information of claim 11, wherein the step (IIa) comprises:

passing the detection signals below a cut-off frequency, wherein the cut-off frequency is greater than a breathing frequency, but attenuating other frequency components.

13. The method for detecting dynamic physiological information of claim 10, wherein the step (III) comprises:

(IIIa) framing a plurality of signal sections to be processed in the plurality of detection signals in a time sequence according to the size of a preset window;

(IIIb) estimating respective preliminary physiological information for the plurality of signal segments; and

(IIIc) extracting to obtain corresponding extracted values of the plurality of signal sections according to the at least one characteristic, and filtering unstable extracted values through a preset threshold value.

14. The method according to claim 13, wherein the step (IIIb) comprises:

the preliminary physiological information is estimated using a zero-crossing rate method.

15. The method as claimed in claim 13, wherein the detection signals include an in-phase signal, a quadrature signal and a phase signal.

16. The method according to claim 15, wherein the step (IV) comprises:

(IVa) counting preliminary physiological information corresponding to the plurality of signal segments of the phase signal, wherein the preliminary physiological information corresponding to the maximum accumulated number of times greater than a first preset value is taken as the physiological information;

(IVb) if the physiological information is not determined in step (IVa), counting the preliminary physiological information corresponding to the signal segments of the in-phase signal and the orthogonal signal, and using the preliminary physiological information corresponding to the maximum accumulated number of times greater than a second predetermined value as the physiological information;

(IVc) if the physiological information is not determined in the step (IVb), counting the preliminary physiological information corresponding to the plurality of signal sections of the in-phase signal and the orthogonal signal, and averaging the preliminary physiological information corresponding to the accumulated times greater than a third preset value to obtain an average value as the physiological information; and

(IVd) if the physiological information is not determined in the step (IVc), counting preliminary physiological information corresponding to the plurality of signal segments of the in-phase signal, the orthogonal signal and the phase signal, and using the preliminary physiological information corresponding to the maximum accumulated number of times that the preliminary physiological information is greater than a preset asphyxia threshold value and greater than a fourth preset value as the physiological information, or using the preliminary physiological information corresponding to the maximum accumulated number of times that the preliminary physiological information is not greater than the preset asphyxia threshold value and greater than a fifth preset value as the physiological information;

wherein the preliminary physiological information counted in the steps (IVa) to (IVc) is the preliminary physiological information after the filtering in the step (III), but the preliminary physiological information counted in the step (IVd) is the preliminary physiological information before the filtering in the step (III).

17. The method according to claim 10, wherein the at least one characteristic comprises one or more of the following: full width at half maximum, strongest gain value, peak value, root mean square, standard deviation, and peak-to-peak difference.

18. The method of claim 17, wherein the step (III) uses a plurality of fitting methods to fit the DC levels of the detected signals to obtain a plurality of fitting curves.

19. The method as claimed in claim 10, wherein the physiological information includes respiration rate.

Technical Field

The present invention relates to physiological information detection, and more particularly, to a system and method for non-contact dynamic physiological information detection.

Background

Body Temperature (BT), Blood Pressure (BP), Heart Rate (HR) and Respiration Rate (RR) are four major physiological information (visual signals). The detection of physiological information can be used to assess the health of the body and can provide clues to the disease.

Conventional medical detection devices can be classified into contact (contact) and non-contact (non-contact) types. The contact detection device can be worn on the body and is used for collecting physiological information through a sensor (sensor). The non-contact detection means, for example, sense radar, by which radio frequency signals are emitted and the reflected radio frequency signals are analyzed for physiological information.

The touch sensing device needs to be worn on the body, which causes inconvenience in use or causes erroneous estimation due to an erroneous use. The non-contact detection device is susceptible to interference from ambient noise, thereby causing erroneous estimates.

Therefore, it is desirable to provide a novel mechanism for improving the shortcomings of the conventional non-contact medical detection device.

Disclosure of Invention

In view of the above, an objective of the embodiments of the invention is to provide a method for detecting dynamic physiological information, which dynamically determines physiological information by feature extraction of signals, thereby improving measurement accuracy.

According to an embodiment of the present invention, the dynamic physiological information detecting system includes a radio frequency detecting device, a calibrating device, a feature extracting device and a physiological information determining device. The radio frequency detection device generates a plurality of detection signals. The correction means corrects the detection signal. The characteristic extraction device processes the corrected detection signal according to at least one characteristic to obtain a plurality of extraction values and filters unstable extraction values. The physiological information determining device determines the physiological information according to the filtered extraction value.

According to the embodiment of the invention, the dynamic physiological information detection method comprises the following steps: (I) generating a plurality of detection signals; (II) correcting the plurality of detection signals; (III) processing the corrected plurality of detection signals according to at least one characteristic to obtain a plurality of extracted values and filtering out unstable extracted values; and (IV) determining physiological information according to the plurality of extracted values after filtration.

The invention can improve the measurement accuracy.

Drawings

FIG. 1A is a block diagram of a dynamic physiological information detection system according to an embodiment of the present invention.

FIG. 1B is a detailed block diagram of the calibration apparatus shown in FIG. 1A.

FIG. 1C shows a detailed block diagram of the feature extraction apparatus of FIG. 1A.

Fig. 2 shows a flow of a dynamic physiological information detection method according to an embodiment of the invention.

Fig. 3A illustrates a normal in-phase signal I and a quadrature signal Q.

Fig. 3B illustrates a normal phase signal P.

Fig. 3C illustrates the in-phase signal I and the quadrature signal Q after the waveforms and dc levels are distorted.

Fig. 3D illustrates the phase signal P subjected to the distortion change.

Fig. 4A illustrates a spectrum of a low-pass filter.

Fig. 4B illustrates a constellation diagram of an in-phase signal I and a quadrature signal Q.

Fig. 5 illustrates an in-phase signal I, a quadrature signal Q, and a window.

Fig. 6A illustrates the full width at half maximum of the signal and the strongest gain value.

Fig. 6B to 6E show the signals and autocorrelation signals in the active state, the mobile state, the leaving state and the inactive state, respectively.

Fig. 7A illustrates three different states among the signals.

FIG. 7B shows the regularized autocorrelation signals for each state.

FIG. 7C shows the autocorrelation signals for each state prior to regularization.

Fig. 8A illustrates a steady state (time domain) signal.

Fig. 8B illustrates a non-steady state (time domain) signal.

Fig. 9A illustrates a steady-state in-phase signal I, quadrature signal Q, and a fitted curve.

Fig. 9B illustrates the non-steady-state in-phase signal I, the quadrature signal Q, and the fitted curve.

Reference numerals

100 dynamic physiological information detection system

11 radar

12 correcting device

121 low pass filter

122 nonlinear suppression unit

123 regularization unit

13 characteristic extraction device

131 sliding window unit

132 physiological information estimation unit

133 feature cell

14 physiological information determining device

200 dynamic physiological information detection method

21 obtain a signal I, Q, P

22A Low pass filtered Signal I, Q, P

22B suppresses the non-linearity of signal I, Q and filters out DC values

22C regularization signal I, Q, P

23A sliding Window to frame Signal sections

23B estimation of physiological information and feature extraction of signal segments

23C extracting to obtain the extraction value of the signal section

24A counts the respiration rate of the signal P and determines the one with the highest cumulative count

24B counts the respiration rate of signal I, Q and determines the highest accumulated count

24C statistics of the respiration rate of signal I, Q, averaging all values for which the cumulative number of times is greater than a predetermined value

24D counts (unfiltered) the respiration rate of signal I, Q, P and determines the highest accumulated count

51 window

61 full width at half maximum

62 strongest gain value

71 state

State 72

73 state

81 peak value

82 peak value

91 fitting curve

92 fitting curve

93 fitting curve

94 curve of fit

I in-phase signal

Q quadrature signal

P phase signal

Detailed Description

Fig. 1A is a block diagram of a dynamic physiological information detecting system 100 according to an embodiment of the invention, and fig. 2 is a flowchart of a dynamic physiological information detecting method 200 according to an embodiment of the invention. The blocks of FIG. 1A and the steps of FIG. 2 may be implemented using hardware, software, or a combination thereof. Although the following embodiments illustrate detecting respiration rate, the embodiments can also be used to detect other physiological information.

In this embodiment, the dynamic physiological information detecting system (hereinafter referred to as a detecting system) 100 may include a Radio Frequency (RF) detecting device, such as a radar 11, for transmitting a radio frequency signal to a subject, receiving a reflected radio frequency signal, and converting the radio frequency signal to obtain detecting signals, such as an in-phase (polarized) signal I, a quadrature (polarized) signal Q, and a phase (phase) signal P (step 21). The phase signal P is a relative phase between the in-phase signal I and the quadrature signal Q. The radar 11 of the present embodiment may be a continuous-wave (CW) radar or an ultra-wideband (UWB) radar (e.g., a Frequency Modulated Continuous Wave (FMCW) radar).

Radio frequency signals are susceptible to interference from ambient noise, causing non-linear or time-varying variations that result in distorted changes in the amplitude, phase or Direct Current (DC) level of the signal. Fig. 3A illustrates a normal in-phase signal I and a quadrature signal Q, and fig. 3B illustrates a normal phase signal P. Fig. 3C illustrates the in-phase signal I and the quadrature signal Q after the waveforms and dc levels are distorted. Fig. 3D illustrates the phase signal P subjected to the distortion change. In this example, the signal duration is about 10 seconds, and the estimated number of breaths is 3 (within 10 seconds) according to FIG. 3B, but after the distortion change, the estimated number of breaths is 12 (within 10 seconds) according to FIG. 3D. In view of this, the present embodiment proposes the following mechanism to improve this problem.

In the present embodiment, the detection system 100 may include a calibration device 12 for calibrating the in-phase signal I, the quadrature signal Q and the phase signal P to eliminate or reduce distortion (distortion) of the signals, thereby improving the signal accuracy. FIG. 1B is a detailed block diagram of the calibration device 12 of FIG. 1A. In the present embodiment, the calibration device 12 may comprise a (digital) filter for removing unwanted (unwanted) frequency components. The filter of this embodiment may comprise a low pass filter 121 that passes the in-phase signal I, the quadrature signal Q, and the phase signal P below a cutoff frequency (e.g., 6Hz), but attenuates other frequency ranges (step 22A). Fig. 4A illustrates a spectrum of the low-pass filter 121. Generally, the target range for the respiration rate is about 0 Hz to about 1 Hz. However, considering that the subsequent processing of the detection system 100 (e.g. the feature extraction means 13) requires additional frequency components, the selected cut-off frequency needs to be greater than the breathing frequency. In the embodiment, the cut-off frequency is selected to be 6Hz, but the disclosure is not limited thereto. For example, the cut-off frequency is selected appropriately for different subjects to be detected (e.g., elderly people, children, or middle aged people who breathe slower than babies). In another embodiment, if the detection system 100 is to detect the heartbeat frequency, the cut-off frequency should be selected to be greater than the heartbeat frequency, so as to determine the degree of the signal affected by the environmental noise by using the additional frequency components.

The calibration apparatus 12 of the present embodiment may include a non-linear suppressing unit 122 for suppressing non-linear components of the frequency doubled (or higher) of the in-phase signal I and the quadrature signal Q and filtering out a Direct Current (DC) value (step 22B). Fig. 4B illustrates a constellation diagram (constellation diagram) of an in-phase signal I and a quadrature signal Q. For ideal in-phase signal I and quadrature signal Q, the constellation diagram is circular as shown, with the center (0,0) as the center. When the in-phase signal I and the quadrature signal Q are distorted, the constellation diagram is an ellipse as shown. In one embodiment, the non-linear suppressing unit 122 uses a matrix mirror technique to reduce the constellation diagram to a circle with the center (0,0) as the center. Meanwhile, the nonlinear suppression unit 122 removes a Direct Current (DC) value with the center (0,0) as a center.

The calibration apparatus 12 of the present embodiment may include a regularization unit 123 for regularizing the in-phase signal I, the quadrature signal Q and the phase signal P (step 22C) to improve the improper scaling of the signals by the aforementioned devices (i.e., the low pass filter 121 and the nonlinear suppression unit 122) or steps (i.e., steps 22A and 22B).

In this embodiment, the detection system 100 may include a feature extraction device 13, which processes the corrected in-phase signal I, the quadrature signal Q and the phase signal P according to at least one feature to obtain a plurality of extracted values respectively, and filters (or screens) the non-stable extracted values. FIG. 1C shows a detailed block diagram of the feature extraction apparatus 13 of FIG. 1A. The feature extraction device 13 of the present embodiment may include a sliding window unit 131, which selects signal segments to be processed according to a predetermined window size (e.g., 10 seconds) in a time-sequential frame (step 23A). Fig. 5 illustrates the in-phase signal I and the quadrature signal Q sliding right (as shown by the arrow) every 2.5 seconds apart with a window 51 of 10 seconds. Thus, a total of 21 signal segments to be processed can be framed out during one minute.

The feature extraction device 13 of the present embodiment may include a physiological information estimation unit 132 for estimating corresponding (preliminary) physiological information of the signal segments and extracting features (step 23B). In the present embodiment, the physiological information estimation unit 132 estimates the respiration rate by using a zero-crossing rate (zero-crossing rate) method, and estimates the respiration rate by interleaving the signal with a dc level of zero. Since the two interleaved times represent one breath, the total interleaved times are divided by two to obtain the breath rate.

The feature extraction device 13 of the present embodiment may include a feature unit 133 for extracting corresponding extracted values of the signal segments according to at least one feature (step 23C), and filtering out unstable extracted values (and corresponding physiological information thereof) by setting a threshold. The feature extraction device 13 of the present embodiment may perform feature extraction according to one or more of the following features: half-width (half width), peak-gain (peak-gain), peak (kurtosis), Root Mean Square (RMS), standard deviation (STD), and peak-to-peak difference (Vpp).

Fig. 6A illustrates the full width at half maximum 61 and the strongest gain value 62 of the signal. Fig. 6B to 6E respectively show signals (e.g., in-phase signal I and quadrature signal Q) and autocorrelation (autocorrelation) signals in a liveness (video) state (e.g., rest or sleep), a movement (motion) state, a leaving (leaving) state, and a non-liveness (no-video) state (e.g., left). From these signals, it can be known that the stable signal (e.g., fig. 6B) has a larger full width at half maximum, the unstable signal in the moving state (e.g., fig. 6C) has a maximum strongest gain value, and the unstable signal in the non-active state (e.g., fig. 6E) has a minimum strongest gain value. Thus, the feature unit 133 can filter out the unstable extracted values (and the corresponding physiological information thereof) by setting a threshold.

Fig. 7A illustrates three different states 71, 72 and 73 among the signals. FIG. 7B shows the normalized autocorrelation signals for each of states 71, 72, and 73. Where the full width at half maximum (0.165) of the (stable) state 71 is greater than the full width at half maximum (0.106) of the (unstable) state 72, but less than the full width at half maximum (0.282) of the other (unstable) state 73. FIG. 7C shows the autocorrelation signals for each state 71, 72, and 73 prior to regularization. Wherein the strongest gain value (153) of the (stable) state 71 is smaller than the strongest gain value (170) of the (unstable) state 72 and smaller than the strongest gain value (178) of the other (unstable) state 73. Thus, in this embodiment, the feature unit 133 can distinguish the states 71, 72, and 73 as the stable state or the non-stable state by setting the threshold, and further filter the extracted value (and the corresponding physiological information) of the non-stable state.

Fig. 8A illustrates a steady state (time domain) signal with a lower peak 81, and fig. 8B illustrates a non-steady state (time domain) signal with a significantly higher peak 82. Therefore, the characterization unit 133 can determine whether the signal is in a stable state or not by the signal peak. The signal peak K can be expressed as follows:

wherein x_iRepresents the ith measurement, s represents the standard deviation, n represents the number of samples,represents the arithmetic mean.

FIG. 9A illustrates a steady state in-phase signal IThe dc levels of the in-phase signal I and the quadrature signal Q are fitted by a polynomial fitting method (fitting) to obtain fitting curves 91 and 92, respectively. Fig. 9B illustrates the non-steady-state in-phase signal I and quadrature signal Q, and the dc levels of the in-phase signal I and quadrature signal Q are fitted by a polynomial fitting method to obtain fitting curves 93 and 94, respectively. From these fitted curves, it can be known that the fitted curves 91, 92 in the steady state are approximate to straight lines and are close to the positions where the dc level is zero; the non-steady state fit curves 93, 94 are perturbed and away from the location where the dc level is zero. According to the characteristics, the Root Mean Square (RMS), the standard deviation (STD) and the peak-to-peak difference value (Vpp) can be respectively obtained for the fitting curve, so as to filter unstable extraction values (and corresponding physiological information thereof). Root mean square M, standard deviation SD and peak-to-peak difference V_ppCan be expressed as follows:

V_pp＝max(x_i)-min(x_i)

wherein x_iRepresenting the ith measurement, n represents the number of samples,represents the arithmetic mean, max () represents the maximum function, and min () represents the minimum function.

The detection system 100 of the present embodiment may comprise a physiological information determining device 14, which determines (final) physiological information (e.g. respiration rate) according to the corresponding extracted values of the signal segments (extracted by the feature unit 133) and the corresponding (preliminary) physiological information (estimated by the physiological information estimating unit 132). The calibration device 12, the feature extraction device 13 and the physiological information determination device 14 may be different signal processing devices. Alternatively, two or all of the correction device 12, the feature extraction device 13 and the physiological information determination device 14 may be integrated into the same signal processing device.

In step 24A, the respiration rates (e.g., 21 data) of the phase signals P (of the signal segments) are counted, and the accumulated number is the maximum respiration rate, but the accumulated number needs to be greater than a first preset value (e.g., 2). The reason why the physiological information determining device 14 initially performs statistics on the phase signal P is that the phase signal P has the best effect of suppressing nonlinearity.

If the respiration rate cannot be determined in step 24A, step 24B is performed to count the respiration rates (e.g., 42 data in total) of the in-phase signal I and the quadrature signal Q, and the accumulated number is output as the respiration rate at the maximum, but the accumulated number is greater than a second predetermined value (e.g., 3).

If the respiration rate is not determined in step 24B, step 24C is performed, and the respiration rates (for example, 42 data in total) of the in-phase signal I and the quadrature signal Q (for the signal segments) are counted, and all values with the accumulated number greater than a third preset value (for example, 3) are averaged to obtain an average value as the respiration rate output.

It should be noted that the respiration rate counted in the steps 24A to 24C is the respiration rate corresponding to the filtered unstable extracted value. If step 24C fails to determine the respiration rate, step 24D is entered, in which the counted respiration rate is the respiration rate for which the non-stable extracted value has not been filtered out. In step 24D, the respiration rates (e.g., 63 data in total) of the in-phase signal I, the quadrature signal Q, and the phase signal P (of the signal segments) are counted, and the maximum cumulative number of times of respiration which is greater than a (preset) apnea (apnea) threshold (e.g., 9 times) is output as the respiration rate, but the cumulative number of times is greater than a fourth preset value (e.g., 24); alternatively, the maximum cumulative number of breaths not greater than the apnea threshold is output as the breath rate, but the cumulative number is greater than a fifth preset value (e.g., 12). Typically, the fifth preset value is less than the fourth preset value. If the breathing rate cannot be determined at step 24D, the breathing rate output is set to zero.

The above description is only for the preferred embodiment of the present invention, and is not intended to limit the claims of the present invention; it is intended that all such equivalent changes and modifications be included within the scope of the present disclosure as defined by the appended claims.