JP2019129996A - Heartbeat detection system and heartbeat detection method - Google Patents

Abstract

To improve precision of heartbeat detection.SOLUTION: This heartbeat detection system comprises: a Doppler sensor which receives a reflection wave from a subject and acquires an I signal and a Q signal of a time region; and a sparse spectrum reconstitution unit which converts the I signal and the Q signal of the time region to an I signal and a Q signal of a frequency region. The sparse spectrum reconstitution unit reconstitutes the sparse spectrum of heartbeats by using an algorithm belonging to a stochastic gradient descent method.SELECTED DRAWING: Figure 4

Description

  The present invention relates to a heartbeat detection system and a heartbeat detection method.

  In recent years, in order to grasp the health state and mental state of a subject, for example, a heart rate detection system that detects a heart rate and a heart rate interval without contact using a Doppler sensor has been studied. In a heart rate detection system using a Doppler sensor, improving heart rate detection accuracy is an important issue, and various proposals have been made for this purpose.

  One of various proposals includes, for example, a spectrogram method. The spectrogram method determines the frequency band including the heartbeat component from the frequency that is considered to be responding to the heartbeat on the spectrogram, integrates the amplitude of the frequency due to the heartbeat in the frequency band, and detects the heartbeat from the change in the integrated value over time. It is a technique to do.

  In the spectrogram method, excellent heartbeat detection accuracy is obtained when the subject is seated and stationary, but sufficient heartbeat detection accuracy cannot be obtained during movements involving various body movements such as typing and driving.

Mogi, Otsuki, "Heartbeat detection by Doppler radar using spectrogram", IEICE ASN Study Group 2017.

  The present invention has been made in view of the above points, and an object thereof is to improve heartbeat detection accuracy.

  The heartbeat detection system receives a reflected wave from a subject and acquires a time domain I signal and a Q signal, and converts the time domain I signal and the Q signal into a frequency domain I signal and a Q signal. A sparse spectrum reconstruction unit for converting, and the sparse spectrum reconstruction unit is required to reconstruct a sparse spectrum of a heartbeat using an algorithm belonging to the stochastic gradient descent method.

  According to the disclosed technology, heartbeat detection accuracy can be improved.

It is a figure which illustrates schematic structure of the heart-rate detection system concerning a 1st embodiment. It is an example of the signal waveform obtained with the Doppler sensor. It is a figure which illustrates the hardware block of the signal processing part which concerns on 1st Embodiment. It is a figure which illustrates the functional block of the signal processing part which concerns on 1st Embodiment. It is an example of the flowchart which shows operation | movement of the heart rate detection system which concerns on 1st Embodiment. It is an example of the signal waveform obtained at the predetermined step. It is a figure explaining the adaptive filter used for SSR. It is a figure explaining the SSR problem. It is a figure explaining a ZA-SLMS algorithm. It is FIG. (1) which shows the typical measurement result of an Example. It is FIG. (2) which shows the typical measurement result of an Example. It is FIG. (The 3) which shows the typical measurement result of an Example. It is FIG. (The 4) which shows the typical measurement result of an Example. It is the figure which put together the evaluation result of the heart rate in case the 5 test subjects are sitting still. It is the figure which put together the evaluation result of the heartbeat interval in case the 5 test subjects are seated still. It is the figure which put together the evaluation result of the heart rate at the time of typing about five test subjects. It is the figure which put together the evaluation result of the heartbeat interval in the case of typing about five test subjects.

  Hereinafter, embodiments will be described with reference to the drawings. In addition, in each drawing, the same code | symbol is attached | subjected to the same component and the overlapping description may be abbreviate | omitted.

  FIG. 1 is a diagram illustrating a schematic configuration of a heartbeat detection system according to the first embodiment. As shown in FIG. 1, the heartbeat detection system 1 includes a Doppler sensor 10 and a signal processing unit 20 as main components.

  The Doppler sensor 10 is a sensor that detects the movement of the observation target (subject) by observing the frequency shift of the transmission signal and the reception signal due to the Doppler effect. In the present embodiment, as an example, a non-modulated continuous wave (CW) is used as a transmission wave.

  The Doppler sensor 10 is disposed in the vicinity of the subject, can receive a signal (reflected wave) reflected at or near the subject's heart, and can obtain a signal generated by the movement of the observation target. Examples of the test subject include an operator typing on a PC, a vehicle driver, and the like.

  FIG. 2 is an example of a signal waveform obtained by the Doppler sensor. The signal shown in FIG. 2 is obtained by obtaining a signal representing a frequency shift between the transmission signal and the reception signal as a function of time. The I signal which is an in-phase component of the transmission signal and the quadrature phase ( Quadrature) component Q signal.

  Movements of the body surface such as heartbeat, breathing, blinking, and body movement (movement by the body) can be observed by the Doppler sensor 10. When performing heart rate detection, it is preferable to appropriately remove noise using a high-pass filter, a band-pass filter, or the like.

  Returning to FIG. 1, the signal processing unit 20 detects the heart rate and heart rate interval of the subject based on the output signal of the Doppler sensor 10. The signal processing unit 20 appropriately uses the I signal and Q signal of the signal received by the Doppler sensor 10 as they are, or generates various signals (amplitude, phase, their integrated values, etc.) based on the I signal and the Q signal. You can do it.

  FIG. 3 is a diagram illustrating a hardware block of the signal processing unit according to the first embodiment. Referring to FIG. 3, the signal processing unit 20 includes a CPU 21, a ROM 22, a RAM 23, an I / F 24, and a bus line 25. The CPU 21, ROM 22, RAM 23, and I / F 24 are connected to each other via a bus line 25.

  The CPU 21 controls each function of the signal processing unit 20. The ROM 22 serving as a storage unit stores a program executed by the CPU 21 to control each function of the signal processing unit 20 and various types of information. The RAM 23 serving as storage means is used as a work area for the CPU 21. The RAM 23 can temporarily store predetermined information. The I / F 24 is an interface for connecting the heartbeat detection system 1 to other devices. The heartbeat detection system 1 may be connected to an external network or the like via the I / F 24.

  However, part or all of the signal processing unit 20 may be realized only by hardware. Examples of hardware include an application specific integrated circuit (ASIC), a digital signal processor (DSP), and a field programmable gate array (FPGA). The signal processing unit 20 may be physically configured by a plurality of devices.

  FIG. 4 is a diagram illustrating a functional block of the signal processing unit according to the first embodiment. Referring to FIG. 4, the signal processing unit 20 includes, as functional blocks, a signal acquisition unit 201, a singular spectrum analysis unit 202, a time difference processing unit 203, a sparse spectrum reconstruction unit 204, a peak tracking unit 205, have. Specific functions of each functional block will be described later in the description of FIG.

  FIG. 5 is an example of a flowchart showing the operation of the heartbeat detection system according to the first embodiment. The heartbeat detection method according to the first embodiment will be described with reference to FIG. 5 and other drawings as appropriate.

  First, in step S11, the signal acquisition unit 201 acquires a signal by receiving a reflected wave from the subject by the Doppler sensor 10 (signal acquisition step). The signal acquired here is composed of, for example, an I signal and a Q signal as shown in FIG. The I signal and Q signal shown in FIG. 2 are sampled signals, and the sampling rate is, for example, 1000 Hz.

  Next, in step S12, the signal acquisition unit 201 down-samples the I signal and the Q signal received in step S11. The sampling rate of downsampling may be any value as long as it is sufficiently higher than twice the normal heart rate (about 1 to 1.6 Hz) (3.2 Hz), and can be set to about 125 Hz, for example.

  For example, if the sampling rate of the Doppler sensor 10 acquired in step S11 is 1000 Hz and the time window (observation time) for acquiring the I signal and the Q signal is 8 seconds, the number of samplings is 8000. By downsampling this at a sampling rate of 125 Hz (frequency band: 0 to 125 Hz), the number of samplings becomes 1000, and the amount of calculation in signal processing can be greatly reduced.

  Next, in step S13, noise components of the I signal and Q signal that have been downsampled in step S12 are removed. The removal of the noise component can be performed by hardware using a band pass filter, for example. Alternatively, the signal processing unit 20 may perform digital signal processing (digital filter or the like). In general, since the heartbeat in a normal time changes in the range of about 1 to 1.6 Hz, the pass band of the bandpass filter can be set to about 0.4 Hz to 5 Hz, for example.

  FIG. 6A shows an example of a signal waveform. In the signal waveform shown in FIG. 6A, when the time window is set to 8 seconds, the I signal and Q signal with a sampling rate of 1000 Hz are downsampled at 125 Hz in step S12, and noise components are removed in step S13. It is a Doppler signal (that is, FIG. 6A is data for 8 seconds).

  Next, in step S14, the singular spectrum analysis unit 202 performs singular spectrum analysis (SSA: Singular Spectrum Analysis). The SSA decomposes the time-domain I and Q signals into vibrational components and noise components, and includes noise principal components (remaining noise components for heartbeats) due to breathing and body movement contained in the time-domain I and Q signals. The heartbeat component is extracted. In step S14, the previous processing value can be used as necessary. For example, the accuracy may be improved by limiting the range from the previous processing value and obtaining the current processing value.

Next, in step S15, the time difference processing unit 203 performs time difference processing that suppresses non-periodic noise with respect to heartbeats included in the I signal and Q signal in the time domain. Specifically, a singular spectrum analysis is performed in step S14 to obtain a sampling signal y = [y ₁ , y ₂ ,..., Y _M ] from which noise main components due to respiration and body movement are removed (for example, M = 1000), and obtain a first order difference y ′ = [y ₂ −y ₁ , y ₃ −y ₂ ,..., Y _M −y _M−1 ] of the sampling signal y. By performing the time difference process, the peak of the spectrum reconstructed in step S16 can be brought close to the actual periodic heartbeat peak.

  FIG. 6B shows an example of a signal waveform. The signal waveform shown in FIG. 6B is obtained by performing a specific spectrum analysis on the I signal and the Q signal in FIG. 6A in step S14 to obtain a sampling signal y from which noise main components due to breathing and body movement are removed ( M = 1000) and the primary difference y ′ of the sampling signal y is obtained.

  Next, in step S16, the sparse spectrum reconstruction unit 204 performs sparse spectrum reconstruction (SSR) that converts the time domain I and Q signals into frequency domain I and Q signals (sparse spectrum reconstruction). Spectrum reconstruction step). The sparse spectrum reconstruction unit 204 reconstructs a sparse spectrum of heartbeats using an algorithm belonging to a stochastic gradient descent (SGD). In the present embodiment, as an example, a ZA-LMS (Zero Attracting Least Mean Square) algorithm belonging to the stochastic gradient descent method is applied to SSR, and an example of removing noise due to breathing or body movement is shown.

  The sparse spectrum is simply a heartbeat spectrum in which there is a heartbeat peak and energy corresponding to other frequencies is almost zero.

  FIG. 6C shows an example of a signal waveform. The signal waveform shown in FIG. 6C is an I signal after the sparse spectrum reconstruction is performed on the I signal and the Q signal in FIG. 6B by the ZA-LMS algorithm. In FIG. 6C, GT (Ground Truth: true value) is a heart rate (corresponding to FIG. 6D) measured by an electrocardiograph, and EST (Estimation: estimated value) is a spectrum obtained by performing sparse spectrum reconstruction. The peak frequency. In FIG. 6 (c), the EST is about 1.2 Hz, that is, the heart beats 72 times per minute (72 BPM). In the example of FIG. 6C, the deviation (absolute value error) between this estimated value (72 BPM) and GT is 0.78 BPM.

  FIG. 6D shows an actual heartbeat signal measured by an electrocardiogram by the same subject for reference. GT can be obtained from FIG.

Here, the ZA-LMS algorithm will be described in detail. FIG. 7 is a diagram illustrating an adaptive filter used for SSR. As shown in FIG. 7, an impulse response hori is applied to the input signal x (n) that has entered the unknown system 501, and an inner product x ^T (n) hori is obtained. x ^T (n) hori is interfered by additional noise z (n) to be a target signal d (n) = x ^T (n) hori + z (n), and is input to the adaptive filter 502.

By subtracting the inner product x ^T (n) h (n) of the input signal x (n) and the filter coefficient h (n) from the target signal d (n), a regression error e (n) is obtained. Using an adaptive algorithm, the regression update of the filter coefficient h (n) can be performed based on the stochastic gradient descent method so as to minimize the regression error e (n), and hori can be estimated.

  The SSR problem is to restore a high-resolution heart rate spectrum using a Fourier observation matrix Φ with a small number of sampling signals y with respect to the Doppler signal. In the present embodiment, based on the correspondence relationship between the SSR problem and the parameters of the adaptive filter 502, the adaptive filter 502 shown in FIG. 7 is used to apply the SSR problem shown in FIG.

  Specifically, the Fourier observation vector φj shown in FIG. 8 corresponds to the input signal x (n), the spectrum unknown solution s (n) corresponds to the filter coefficient h (n), and the sampling signal element yj corresponds to the target signal d (n). Let In FIG. 8, 503 schematically shows a Doppler signal in the time domain, and 504 schematically shows a Doppler signal in the frequency domain obtained by Fourier transforming 503. HR (Heart Rate) schematically shows a peak corresponding to the heart rate.

  In the case of the ZA-LMS algorithm, it is assumed that the impulse response hori includes two variables. One input signal x (n) is stochastically used for each regression, the gradient ∇f of the objective function due to the secondary regression error e (n) is reversed, the coefficient is corrected by multiplying the step size μ. By the coefficient correction, the filter coefficient h (n) gradually approaches hori.

  The above is the explanation of the ZA-LMS algorithm. In a non-Gaussian noise environment, there is a high possibility that the regression error is unstablely updated, and it may be difficult to achieve an ideal decoding result with the ZA-LMS algorithm. is there. Therefore, in order to effectively cope with the impulse noise, as shown in FIG. 9, the regression error is limited to a certain range, and the ZA-SLMS (Zero It is more preferable to apply the Attracting Sign Least Mean Square) algorithm.

  In FIG. 9, a broken-line circle and a solid-line arrow indicate a gradient drop that limits the regression error, and a dashed-dotted arrow indicates a gradient drop that is interfered by impulse noise. S (n) represents a spectrum unknown solution, and Sori represents a heartbeat spectrum. Thus, by introducing a sparse constraint to the ZA-LMS algorithm, the spectrum unknown solution S (n) can be closer to the actual sparse heartbeat spectrum Sori.

  The objective function of ZA-LMS can be expressed by equation (1).

式（１）において、ｅ（ｎ）＝ｄ（ｎ）−ｘ^Ｔ（ｎ）ｈ（ｎ）である。目的関数の導関数である式（２）により、ＺＡ−ＬＭＳの更新方程式である式（３）が得られる。

In the formula (1), e (n) = d (n) −x ^T (n) h (n). Equation (2), which is a derivative of the objective function, yields Equation (3), which is an update equation for ZA-LMS.

式（３）において、γ＝μλである。ここで、γはゼロ吸引ファクタ、μはステップサイズ、λは制御パラメータである。又、ｓｇｎ関数が式（４）のように定義される。

In equation (3), γ = μλ. Here, γ is a zero suction factor, μ is a step size, and λ is a control parameter. Also, the sgn function is defined as in equation (4).

ＺＡ−ＬＭＳの更新方程式である式（３）において、ｅ（ｎ）に式（４）に示すｓｇｎ関数を導入することで、ＺＡ−ＳＬＭＳの更新方程式である式（５）が得られる。

In Formula (3) which is an update equation of ZA-LMS, Formula (5) which is an update equation of ZA-SLMS is obtained by introducing the sgn function shown in Formula (4) into e (n).

図５の説明に戻り、次に、ステップＳ１７では、ピーク追跡部２０５は、スパーススペクトルのピークを追跡し、心拍数を推定する。例えば、図６（ｃ）の信号波形を例にとると、ピーク追跡部２０５は、図６（ｃ）の信号波形のピークを追跡し、約１．２Ｈｚ（７２ＢＰＭ）にピークが存在することを認識する。まず、前回の推定値を参照し最大ピークを選び、そして、閾値により今回の推定値を確認する。

Returning to the description of FIG. 5, next, in step S <b> 17, the peak tracking unit 205 tracks the peak of the sparse spectrum and estimates the heart rate. For example, taking the signal waveform of FIG. 6 (c) as an example, the peak tracking unit 205 tracks the peak of the signal waveform of FIG. 6 (c) and determines that a peak exists at about 1.2 Hz (72 BPM). recognize. First, the maximum peak is selected with reference to the previous estimated value, and the current estimated value is confirmed by a threshold value.

  The peak tracking unit 205 performs a selection operation and a confirmation operation. Specifically, the peak tracking unit 205 selects, as a current estimated value, a frequency at which the sparse spectrum peaks within a predetermined search range centered on the previous estimated value (for example, within the range of 10 BPM before and after). (Selection operation). Next, since the adjacent estimated value does not change significantly, the peak tracking unit 205 outputs the previous estimated value when the change of the estimated value exceeds a predetermined threshold (for example, threshold θ = 6 BPM) (confirmation operation) ).

  For example, taking the signal waveform of FIG. 6C as an example, if the previous estimated value is 72 BPM, the peak tracking unit 205 selects the current estimated value within a range of 62 BPM to 82 BPM, for example. For example, if the estimated value change is 6 BPM or less, the current estimated value is output. If the estimated value change exceeds 6 BPM, the previous estimated value is output.

  Through the above steps, noise can be removed and heartbeat detection accuracy can be improved. Of the spectrum peak of the I signal and the spectrum peak of the Q signal, data having a higher peak value can be adopted.

  Thus, in the heartbeat detection system 1, sparse spectrum reconstruction is performed on the I signal and the Q signal acquired by the Doppler sensor that is a non-contact sensor. At this time, by using an algorithm belonging to the stochastic gradient descent method for sparse spectrum reconstruction, it is possible to remove noise due to respiration and body movement with high accuracy. As a result, heartbeat detection accuracy can be improved not only when the user is seated and stationary, but also during movements involving various body movements such as typing and driving. Here, the heartbeat detection includes detection of a heart rate and detection of a heartbeat interval (RRI: R-R Interval).

  Note that, for the algorithm belonging to the stochastic gradient descent method, it is preferable to use a ZA-LMS algorithm having high robustness against noise, and it is more preferable to use a ZA-SLMS algorithm having higher robustness against noise. However, the present invention is not limited thereto, and a FOCUSS (Focal Underdetermined System Solver) algorithm or the like may be used as shown in the following modification of the first embodiment.

  Moreover, before performing sparse spectrum reconstruction, by performing a specific spectrum analysis to remove noise main components due to breathing and body movement and extracting a heartbeat component, heartbeat detection accuracy can be further improved. Further, after the specific spectrum analysis, the time difference process is performed to bring the spectrum peak to be reconstructed closer to the actual periodic heartbeat peak, whereby the heartbeat detection accuracy can be further improved.

<Modification of First Embodiment>
In the modification of the first embodiment, an example in which an algorithm different from that of the first embodiment is applied to the sparse spectrum reconstruction in step S16 in FIG. In the modification of the first embodiment, the description of the same components as those of the already described embodiments may be omitted.

  In the modification of the first embodiment, the FOCUSS algorithm belonging to the stochastic gradient descent method is applied to the sparse spectrum reconstruction in step S16 in FIG. 5 to remove noise due to breathing and body movement with high accuracy.

The FOCUSS algorithm is a non-convex norm algorithm that belongs to approximate convex optimization. Here, based on the compressed sensing (CS) technique, an SSR algorithm using an M × N Fourier observation matrix Φ∈CM ^{× N} is applied. Φ can be expressed by equation (6).

例えば、図８の５０３中の破線で示す時間領域でのサンプリング数がＭ＝１０００の場合、Ｍの半分（５００）の何倍かの周波数ビン（例えば、Ｎ＝４０９６／２＝２０４８：図８の５０４中の座標軸の正半分）に広げ、スパーススペクトル再構成を行う。同時に、スペクトルに雑音成分を含むゼロに近い値を減少し、大きな値の復元を強化する。

For example, when the number of samplings in the time domain indicated by the broken line in 503 in FIG. 8 is M = 1000, frequency bins that are several times half of M (500) (for example, N = 4096/2 = 2048: FIG. And sparse spectrum reconstruction is performed. At the same time, the value close to zero including noise components in the spectrum is reduced, and the restoration of large values is enhanced.

  The SSR problem can be expressed by the equation (7).

式（７）において、ｘはＮ次元の未知ベクトルである。この問題では、ｙ'_recによりΦを用いてｘを求める。||ｘ||を最小にすることによって、式（８）に示すようにアフィン尺度変換（ＡＳＴ）法を通して方程式の解ｘ（最小ノルム解）を求める。

In Expression (7), x is an N-dimensional unknown vector. In this problem, x is obtained using Φ by y ′ _rec . By minimizing || x ||, the solution x (minimum norm solution) of the equation is obtained through the affine scale transformation (AST) method as shown in equation (8).

式（８）において、擬似逆行列がΦ^＋＝Φ^Ｔ（ΦΦ^Ｔ）^−１と定義され、ｘ（ａ）が復元結果を表す。

In Equation (8), the pseudo inverse matrix is defined as Φ ⁺ = Φ ^T (ΦΦ ^T ) ^−1, and x (a) represents the restoration result.

Next, a weighted minimum norm solution is obtained. Based on the same principle as the minimum norm solution, || q || (q = W ⁺ x, W is an N × N square matrix) is minimized as an optimization target, and Equation (9) is redefined.

そして、式（１０）、式（１１）により方程式の解ｘ（ａ）を求める。ｑ（ａ）は復元した最適化目標である。

And the solution x (a) of an equation is calculated | required by Formula (10) and Formula (11). q (a) is the restored optimization target.

ここで、基本的なＦＯＣＵＳＳでは、式（１２）〜式（１４）で表されるが、式（１５）〜式（１７）で表される拡張版のＭ−ＦＯＣＵＳＳや、式（１８）〜式（２０）で表される改良型のＲＭ−ＦＯＣＵＳＳを用いても良い。

Here, in the basic FOCUSS, it is expressed by Expression (12) to Expression (14), but the extended version of M-FOCUSS expressed by Expression (15) to Expression (17) or Expression (18) to An improved RM-FOCUSS represented by the formula (20) may be used.

式（１２）〜式（１４）において、ｎは回帰数である。

In Formula (12)-Formula (14), n is a regression number.

式（１５）において、ｐはｘ_ｎのスパース性を高めるパラメータである。ｐは、例えば、０．８程度とすることができる。

In Expression (15), p is a parameter that increases the sparsity of _xn . For example, p can be set to about 0.8.

式（１９）において、Ｉ＝ｄｉａｇ（１，１，・・・，１）は単位行列である。又、λ_ＲＭＦはスパース性の比重を調整するパラメータである。λ_ＲＭＦは、例えば、０．１程度とすることができる。なお、λ_ＲＭＦ＝０の場合がＭ−ＦＯＣＵＳＳである。

In equation (19), I = diag (1, 1,..., 1) is a unit matrix. Λ _RMF is a parameter for adjusting the specific gravity of the sparsity. λ _RMF can be set to about 0.1, for example. The case of λ _RMF = 0 is M-FOCUSS.

  As described above, even if the FOCUSS algorithm is used instead of the ZA-LMS algorithm or the ZA-SLMS algorithm as an algorithm belonging to the stochastic gradient descent method used for the sparse spectrum reconstruction, noise due to breathing or body motion can be accurately detected. Can be removed.

  However, since the FOCUSS algorithm is susceptible to noise, the ZA-LMS algorithm and ZA-SLMS algorithm, which have high robustness against noise, remove noise caused by breathing and body movement with higher accuracy. can do.

[Example]
In the example, an experiment of heart rate detection was performed using the heart rate detection system 1. Table 1 shows the specifications of the experiment.

具体的には、心拍検出システム１において、図５のステップＳ１６のスパーススペクトル再構成に、ＲＭ−ＦＯＣＵＳＳアルゴリズムを用いた場合、ＺＡ−ＬＭＳアルゴリズムを用いた場合、ＺＡ−ＳＬＭＳアルゴリズムを用いた場合について、心拍数及び心拍間隔（ＲＲＩ）の評価を行った。又、基準値として、心電計で心拍数及び心拍間隔の測定を行った。又、比較例として、スペクトログラム法を用いた場合について、実施例と同一条件で評価した。なお、スペクトログラム法とは、前述のように、スペクトログラム上の心拍に反応していると考えられる周波数から心拍成分を含む周波数帯域を決定し、周波数帯域内の心拍による周波数の振幅を積分し、積分値の時間変化から心拍を検出する手法である。

Specifically, in the heart rate detection system 1, when the RM-FOCUSS algorithm is used for the sparse spectrum reconstruction in step S <b> 16 in FIG. 5, the ZA-LMS algorithm is used, or the ZA-SLMS algorithm is used. Evaluation of heart rate and heart rate interval (RRI) was performed. As a reference value, the heart rate and heart rate interval were measured with an electrocardiograph. Further, as a comparative example, the case where the spectrogram method was used was evaluated under the same conditions as in the examples. The spectrogram method, as described above, determines the frequency band including the heartbeat component from the frequency that is considered to respond to the heartbeat on the spectrogram, integrates the amplitude of the frequency due to the heartbeat within the frequency band, and integrates This is a technique for detecting the heartbeat from the time change of the value.

  10 to 13 show typical measurement results. FIG. 10 shows the evaluation results of the heart rate when the subject 3 is seated and stationary among the five subjects. FIG. 11 shows an evaluation result of the heart rate when the subject 2 is typing among the five subjects. FIG. 12 shows the evaluation result of the heartbeat interval when the subject 3 is seated and stationary among the five subjects. FIG. 13 shows the evaluation result of the heartbeat interval when the subject 2 is typing among the five subjects.

  10 and 11, each data measured using the RM-FOCUSS algorithm, the ZA-LMS algorithm, and the ZA-SLMS algorithm has a time window of 8 seconds and a shift time of 2 seconds. The heart rate estimated from the peak frequency (EST) of the spectrum subjected to sparse spectrum reconstruction by the method described with reference to FIG. 6C is plotted for each time window. Further, the heart rate is based on the data shown in FIG. 10 (at the time of sitting) and FIG. 11 (at the time of typing), and the heart rate interval is calculated based on the data shown in FIG. 12 (at the time of sitting still) and FIG. It is a plot.

  FIG. 14 is a table summarizing the evaluation results of heart rate when five subjects are seated and stationary, and compares absolute value errors with reference values (unit: BPM). As shown in FIG. 14, for each algorithm, when comparing the average value of the absolute value error with respect to the reference value of the heart rate in the case of sitting still, compared to the spectrogram method that is a comparative example, It was confirmed that the average value of the absolute value error of the algorithm is greatly reduced, and the heartbeat detection accuracy can be improved. In particular, when the ZA-SLMS algorithm was used, the average value of absolute value errors was the smallest, and it was confirmed that the ZA-SLMS algorithm is a suitable algorithm for improving heartbeat detection accuracy.

  FIG. 15 is a table summarizing the evaluation results of heartbeat intervals when five subjects are seated and stationary, and compares RMSE (Root Mean Square Error) (unit: msec). As shown in FIG. 15, when comparing the average value of the RMSE of the heartbeat interval when sitting still for each algorithm, compared to the spectrogram method that is a comparative example, the three algorithms that are examples are It was confirmed that the average value was greatly reduced and the heartbeat detection accuracy could be improved. Above all, when the ZA-LMS and ZA-SLMS algorithms are used, the average value of RMSE is the smallest, and the ZA-LMS and ZA-SLMS algorithms are suitable algorithms for improving heartbeat detection accuracy. Was confirmed.

  FIG. 16 is a diagram summarizing the evaluation results of the heart rate when typing is performed for five subjects, and the absolute value error with respect to the reference value is compared (unit: BPM). As shown in FIG. 16, when comparing the average value of the absolute value error with respect to the reference value of the heart rate when typing for each algorithm, compared to the spectrogram method as a comparative example, three algorithms as an example It was confirmed that the average value of the absolute value error was significantly reduced and the heartbeat detection accuracy could be improved. In particular, when the ZA-SLMS algorithm was used, the average value of absolute value errors was the smallest, and it was confirmed that the ZA-SLMS algorithm is a suitable algorithm for improving heartbeat detection accuracy.

  FIG. 17 is a table summarizing the evaluation results of heartbeat intervals when typing is performed on five subjects, and RMSE is compared (unit: msec). As shown in FIG. 17, when comparing the average value of the RMSE of the heartbeat interval when typing for each algorithm, the three algorithms of the examples are compared with the spectrogram method of the comparative example. The value was greatly reduced, and it was confirmed that the heartbeat detection accuracy could be improved. In particular, when the ZA-SLMS algorithm was used, the average value of RMSE was the smallest, and it was confirmed that the ZA-SLMS algorithm is a suitable algorithm for improving heartbeat detection accuracy.

  The preferred embodiments and the like have been described in detail above, but the present invention is not limited to the above-described embodiments and the like, and various modifications can be made to the above-described embodiments and the like without departing from the scope described in the claims. Variations and substitutions can be added.

1 Heart Rate Detection System 10 Doppler Sensor 20 Signal Processing Unit 21 CPU
22 ROM
23 RAM
24 I / F
25 Bus line 201 Signal acquisition unit 202 Singular spectrum analysis unit 203 Time difference processing unit 204 Sparse spectrum reconstruction unit 205 Peak tracking unit

Claims (8)

A Doppler sensor that receives a reflected wave from a subject and obtains an I signal and a Q signal in a time domain;
A sparse spectrum reconstruction unit that converts the I signal and the Q signal in the time domain into an I signal and a Q signal in the frequency domain, and
The sparse spectrum reconstruction unit is a heartbeat detection system for reconstructing a sparse spectrum of a heartbeat using an algorithm belonging to a stochastic gradient descent method. A peak tracking unit for tracking a peak of the sparse spectrum and estimating a heart rate;
When estimating the two adjacent heart rates, the peak tracking unit selects the current estimated value within a predetermined search range centered on the previous estimated value, and the change in the estimated value exceeds a predetermined threshold value. The heartbeat detection system according to claim 1, wherein the previous estimated value is output. A singular spectrum analysis unit for removing a main component of noise for heartbeats included in the I signal and the Q signal in the time domain and extracting a heartbeat component;
The sparse spectrum reconstruction unit converts the time domain I signal and the Q signal from which the noise main component is removed by the singular spectrum analysis unit into a frequency domain I signal and the Q signal. The described heart rate detection system. A time difference processing unit that suppresses non-periodic noise for heartbeats included in the I signal and the Q signal in the time domain;
The sparse spectrum reconstruction unit removes the noise main component by the singular spectrum analysis unit, and converts the I signal and the Q signal in the time domain from which non-periodic noise is suppressed by the time difference processing unit to the I signal in the frequency domain The heartbeat detection system according to claim 3, wherein the heartbeat detection system converts the signal into the Q signal.   The heartbeat detection system according to any one of claims 1 to 4, wherein the algorithm belonging to the stochastic gradient descent method is a ZA-SLMS algorithm.   The heartbeat detection system according to any one of claims 1 to 4, wherein the algorithm belonging to the stochastic gradient descent method is a ZA-LMS algorithm.   The heartbeat detection system according to any one of claims 1 to 4, wherein the algorithm belonging to the stochastic gradient descent method is a FOCUSS algorithm. A signal acquisition step of receiving a reflected wave from a subject by a Doppler sensor and acquiring an I signal and a Q signal in a time domain;
A sparse spectrum reconstruction step of converting the time domain I signal and the Q signal into a frequency domain I signal and the Q signal, and
In the sparse spectrum reconstruction step, a heartbeat detection method for reconstructing a sparse spectrum of a heartbeat using an algorithm belonging to the stochastic gradient descent method.

Images