METHOD FOR ESTIMATING A CARDIAC FREQUENCY AND ASSOCIATED DEVICE

Abstract

The invention is a method for estimating a cardiac frequency via the detection of radiation backscattered or transmitted by a bodily zone. The part is illuminated, simultaneously or successively, by light radiation extending over a first spectral band and a second spectral band. A photodetector detects radiation emitted by the bodily zone under the effect of its illumination, in each of the spectral bands. A first detection function and a second detection function are formed from the radiation detected in each spectral band, respectively. The method allows the cardiac frequency to be determined via the determination of characteristic instants that are identified from the first detection function and the second detection function simultaneously.

Claims

1. A Method for estimating a cardiac frequency of a living being including the following steps: a) illuminating a bodily zone of the living being with an incident light beam in a first spectral band; b) detecting light radiation transmitted or backscattered, in the first spectral band, by the bodily zone under the effect of the illumination; c) determining a first detection function, representing a variation as a function of time of an intensity of the light radiation thus detected; d) identifying characteristic instants from the first detection function, and calculating an occurrence frequency of the characteristic instants; e) estimating a cardiac frequency from the occurrence frequency calculated in step d); wherein: steps a) to b) are also implemented in a second spectral band, such that step c) includes determining a second detection function representing a variation as a function of time of an intensity of the light radiation detected in the second spectral band; step d) includes identifying characteristic instants from the second detection function and selecting characteristic instants, identified from each detection function, and appearing in temporal coincidence, the occurrence frequency being calculated from the characteristic instants thus selected.

2. The method according to claim 1, wherein the second spectral band is different from the first spectral band.

3. The method according to claim 1, wherein step d) includes calculating a first derived function, derived from the first detection function, and a second derived function, derived from the second detection function, and identifying characteristic instants from each of the derived functions.

4. The method according to claim 3, wherein each derived function is obtained via a difference between the value of a detection function at two different instants.

5. The method according to claim 1 wherein: the first spectral band includes wavelengths comprised between 600 and 700 nm; the second spectral band includes wavelengths comprised between 750 nm and 1 μm.

6. The method according to claim 1, wherein the first spectral band extends between 600 nm and 700 nm, whereas the second spectral band extends between 750 nm and 1 μm.

7. The method according to claim 1, wherein the detected radiation is radiation backscattered by the bodily zone under the effect of its illumination.

8. A device for estimating a cardiac frequency of a living being, including: a light source configured to emit an incident light beam that propagates towards a bodily zone of the living being, in a first spectral band and in a second spectral band; a photodetector, configured to detect, in the first spectral band and in the second spectral band, radiation backscattered or transmitted by the bodily zone under the effect of its illumination by the incident light beam; a processor, configured to process the radiations detected by the photodetector and to implement steps c) to e) of the method according to claim 1.

9. The device according to claim 8, wherein the photodetector is configured to detect radiation backscattered by the bodily zone under the effect of its illumination.

10. The device according to claim 8, wherein the first spectral band is different from the second spectral band.

11. The device according to claim 8, wherein the light source includes: a first elementary light source, configured to emit a first incident light beam in the first spectral band; a second elementary light source, configured to emit a second incident light beam in the second spectral band.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0037] FIGS. 1A and 1B show a first and a second example of a device configured to implement the invention, respectively.

[0038] FIG. 2A shows a detection function obtained following detection of radiation backscattered by a bodily zone under the effect of an illumination.

[0039] FIG. 2B shows what is called a derived function, obtained from a detection function such as shown in FIG. 2A.

[0040] FIG. 3A shows a detail of a detection function, and its derived function, as a function of time. It illustrates the detection of a false front.

[0041] FIG. 3B shows, as a function of time, the cardiac frequency of an individual as estimated from a signal detected in response to an illumination in a spectral band centred on 660 nm and 940 nm, respectively. These estimations were calculated using a prior-art method and were obtained by illuminating the thumb of an individual, in a back-scatter configuration.

[0042] FIG. 4A shows the main steps of a method according to the invention. FIG. 4B illustrates each illuminating sequence of the method.

[0043] FIGS. 5A and 5B show characteristic instants detected using a first detection function and a second detection function, respectively.

[0044] FIG. 6A is identical to FIG. 3B. FIG. 6B shows the cardiac frequency of an individual as estimated, while implementing the invention, from the detected signals shown in FIG. 6A.

[0045] FIG. 6C corresponds to the cardiac frequency of an individual, estimated according to a prior-art method by illuminating the wrist of an individual. FIG. 6D corresponds to the cardiac frequency estimated, while implementing the invention, from the detected signal shown in FIG. 6C.

[0046] FIGS. 6A to 6D were obtained by implementing the method in a back-scatter configuration.

DETAILED DESCRIPTION OF THE ILLUSTRATED EMBODIMENTS

[0047] FIG. 1A shows an example of a device 1 allowing the invention to be implemented. A light source 10 emits an incident light beam 12 that propagates towards a sample 20, along a propagation axis Z. The term sample designates a bodily zone of a living being, the cardiac frequency of which it is desired to acquire. T

[0048] he photons composing the incident light beam 12 penetrate into the sample and some thereof are backscattered in a direction parallel to the propagation axis, in a direction opposite to the latter. These backscattered photons form backscattered radiation 14. The backscattered radiation 14 may be detected by a photodetector 30 placed facing the surface 21 of the sample. The photodetector may be configured so as to detect backscattered radiation emanating from the sample at a distance d, called the back-scatter distance, which is generally non-zero, smaller than a few millimetres and typically smaller than 15 mm or 10 mm.

[0049] In this example, the light source 10 includes two elementary light sources 10.sub.1 and 10.sub.2. The first elementary light source 10.sub.1 is a light-emitting diode emitting in a first spectral band Δλ.sub.1 centred on a first wavelength λ.sub.1 equal to 660 nm. It is a light-emitting diode sold by the manufacturer Kingbright under the reference APT1608SURCK. The second elementary light source 10.sub.2 is a light-emitting diode emitting in a second spectral band Δλ.sub.2 centred on a second wavelength λ.sub.2 equal to 940 nm. It is a light-emitting diode sold by the manufacturer Kingbright under the reference APT1608F3C. Thus, the first spectral band Δλ.sub.1 preferably extends between 600 and 700 nm, this covering the red visible spectral band, whereas the second spectral band Δλ.sub.2 preferably extends between 700 and 1000 nm, and more preferably between 810 nm-1000 nm, this corresponding to a spectral band in the near infrared. Preferably, the first spectral band Δλ.sub.1 and the second spectral band Δλ.sub.2 are different and do not overlap. By do not overlap, what is meant is that most of the emission spectrum, and preferably 80% or even more than 90% of the emitted intensity, is not located in the same spectral range.

[0050] A microcontroller 15 commands the sequential activation of the elementary light sources 10.sub.1 and 10.sub.2. Thus, the sample is successively illuminated by a first incident light beam 12.sub.1, in the first spectral band Δλ.sub.1, and by a second incident light beam 12.sub.2, in the second spectral band Δλ.sub.2.

[0051] A photodetector 30 detects first backscattered radiation 14.sub.1, in the first spectral band Δλ.sub.1, under the effect of the illumination by the first incident light beam 12.sub.1, and second backscattered radiation 14.sub.2, in the second spectral band Δλ.sub.2 under the effect of the illumination by the second incident light beam 12.sub.2. In the example shown, the photodetector is a photodiode sold by VISHAY under the reference BPW345, the spectral band of detection of which allows the first and second backscattered radiation to be detected. The back-scatter distance d is, in this example, 7 mm.

[0052] A processor 32 is configured to establish a detection function, corresponding to a variation as a function of time of the intensity of radiation detected by the photodetector, in each of the spectral bands. It may be connected to a memory 33 configured to store instructions allowing a method described in this description to be implemented. It may also be connected to a display unit 34.

[0053] According to another embodiment, which is shown in FIG. 1B, the device 1 is placed in what is called a transmission configuration. In such a configuration, the sample 20 extends between the light source 10 and the photodetector 30. The latter is then configured to detect radiation 14 transmitted by the sample along the propagation axis Z. However, such a device is less compact than the back-scatter device described with reference to FIG. 1A. In addition, a back-scatter device, such as shown in FIG. 1A, is more versatile than a device operating in transmission. Specifically, its operation is independent of the thickness of the analysed bodily zone. It may therefore be placed against various bodily zones, for example a wrist, a finger, or a leg.

[0054] FIG. 2A shows what is called a detection function S(t) representing the intensity, S, as a function of time, of backscattered (or transmitted) radiation detected by the photodetector 30, in the first spectral band. Under the effect of cardiac activity, the intensity of the backscattered radiation varies periodically as a function of time, the period of this variation depending on the cardiac frequency hr. FIG. 2B shows the variation, as a function of time, of what is called a derived function representing a time derivative of the detection function. It will be understood that it is easy to estimate the frequency of the function S′(t), the latter also corresponding to the frequency of the function S(t), which also corresponds to the sought cardiac frequency hr. A known prior-art method consists in using a low threshold Th.sub.low and a high threshold Th.sub.high, so as to detect a descending front F.sub.low and a rising front F.sub.high, respectively, in the derived function S′(t). The instants corresponding to a descending front, or to a rising front, are called characteristic instants. They are obtained by determining, to the nearest sampling period, the instant at which the derived function S′(t) crosses the low threshold Th.sub.low and the high threshold Th.sub.high, respectively. The frequency f of the derived function is obtained by calculating a period between two successive characteristic instants, or an average of the time interval between a plurality of consecutive characteristic instants.

[0055] Such an estimation is generally calculated in a single spectral band, whether it be a red spectral band or an infrared spectral band. However, this type of estimation lacks robustness. More particularly, movements of the illuminated bodily zone, exposure to parasitic light sources or simple electronic noise may corrupt the estimation of cardiac frequency. FIG. 3A shows a detail of a detection function and of a derived function in a time interval of 1500 ms. This figure illustrates an example of detection of what is called a false descending front F.sub.false and what is called a true descending front F.sub.true. The inventors have estimated that it is necessary to decrease the detection of false fronts, because their presence corrupts the estimation of cardiac frequency.

[0056] FIG. 3B shows estimations of cardiac frequency hr.sub.1, hr.sub.2, these estimations being calculated from a first detection function S.sub.1(t) obtained by illuminating a sample in the first spectral band λΔ.sub.1, and a second detection function S.sub.2(t) obtained by illuminating the sample in the second spectral band Δλ.sub.2, respectively. In each of these estimations, false variations in the estimations, due to the detection of false fronts such as the front F.sub.false described with reference to FIG. 3A, are observed.

[0057] The inventors have defined a method allowing the detection of false characteristic instants to be avoided. This method, which combines the detection of signals backscattered or transmitted in two spectral bands, will be described below with reference to FIG. 4A, the main steps of which are presented below.

[0058] Step 100: arranging the device 1 in such a way that the light source 10 is configured to illuminate a sample, i.e. a bodily zone 20 of a living being, and that the photodetector is configured to detect radiation backscattered or transmitted by the bodily zone consecutively to this illumination.

[0059] Step 110: illuminating the sample 20 in the first spectral band Δλ.sub.1 (substep 110.sub.1) and in the second spectral band Δλ.sub.2 (substep 110.sub.2). Depending on the photodetector used, this illumination may be simultaneous or successive. In this example, a single non-spectrally resolved photodetector is used. The sample is illuminated successively by each elementary source 10.sub.1 and 10.sub.2, the duration of each illumination being 1.66 ms. The successive activation of each elementary light source, which successive activation is designated by the term “illuminating sequence”, is controlled by the microcontroller 15. Alternatively, the light sources may be continuously activated, the backscattered (or transmitted) radiation being detected by two different photodetectors, each being configured to detect the radiation in the first spectral band Δλ.sub.1 or the second spectral band Δλ.sub.2, respectively. According to another variant, the photodetector may be spectrally resolved, thereby also allowing the bodily zone 20 to be illuminated simultaneously in the two spectral bands. Preferably, but optionally, after the activation of the second light source, no light source is activated for 1.66 ms (substep 110.sub.3). The signal S.sub.B detected by the photodetector 30 is thus representative of a dark current of the latter.

[0060] Step 120: detecting radiation backscattered (or transmitted) by the sample following the illumination in each spectral band. The photodetector generates a first detection signal S.sub.1 depending on the intensity of the radiation backscattered (or transmitted) 14.sub.1 under the effect of the illumination of the sample in the first spectral band Δλ.sub.1 (substep 120.sub.1) and a second detection signal S.sub.2 dependent on the intensity of the radiation backscattered (or transmitted) 14.sub.2 under the effect of the illumination of the sample in the second spectral band Δλ.sub.2 (substep 120.sub.2). In this example, the first detection signal S.sub.1 and the second detection signal S.sub.2 are detected during the illumination by the first elementary source and during the illumination by the second elementary source, respectively. When no light source is activated, the photodetector acquires a background-noise signal or dark-current signal S.sub.B (substep 120.sub.3). This dark current may be subtracted from the detection signals S.sub.1 and S.sub.2.

[0061] Thus, as shown in FIG. 4B, an illuminating sequence Δt lasts 5 ms, and is subdivided into three time periods of 1.66 ms, respectively corresponding to: [0062] the activation of the first elementary light source 10.sub.1, and the detection, by the photodetector, of a first signal S.sub.1 representing the intensity of first radiation 14.sub.1 backscattered (or transmitted) by the sample; [0063] the activation of the second elementary light source 10.sub.2, and the detection, by the photodetector, of a second signal S.sub.2 representing the intensity of second radiation 14.sub.2 backscattered (or transmitted) by the sample; [0064] the detection, by the photodetector, of a background-noise signal S.sub.B when no light source is activated.

[0065] Step 130: establishing a first detection function S.sub.1(t) and a second detection function S.sub.2(t) representing the variation as a function of time of the first detection signal S.sub.1 and of the second detection signal S.sub.2, respectively. Each of these functions is obtained by sampling over time the first signal S.sub.1 and the second signal S.sub.2, respectively, the sampling frequency for example being 200 Hz, this corresponding to an acquisition of a first signal S.sub.1 and of a second signal S.sub.2 every 5 ms. The establishment of each detection function may comprise a preprocessing step in which the signal is smoothed, allowing a high-frequency component of the detected signal to be removed. This preprocessing may take the form of application of a low-pass filter or of a moving average. In this example, a moving average is calculated for a time interval of 25 ms, i.e. 5 samples.

[0066] Step 140: determining a first derived function S′.sub.1(t) and a second derived function S′.sub.2(t). Each derived function is obtained via a difference of a detection function at two different times t and t+St. The time difference δt is preferably smaller than 500 ms, or even than 100 ms. In this example, t and t+δt are successive instants, i.e. instants spaced apart by the sampling period, i.e. 5 ms. The derived function may be obtained by normalizing the difference described above by the time difference, this corresponding to the conventional definition of a rate of variation.

[0067] In other words,

[00001]$\begin{array}{cc}{S}_1^{\prime }\left(t\right)=S_1\left(t+\delta .Math..Math.t\right)-S_1\left(t\right).Math..Math.\left(1\right).Math..Math.\mathrm{or}.Math..Math.S^{\prime }\left(t\right)=\frac{S_1\left(t+\partial t\right)-S_1\left(t\right).Math.}{\partial t}.& \left(1^{\prime }\right)\end{array}$

The second derived function S′.sub.2(t) is obtained identically to the first derived function S′.sub.1(t), from the second detection function S.sub.2(t).

[0068] Step 150: identifying characteristic instants. By characteristic instant, what is meant is an instant at which the detection function or its derived function reaches a particular value, crosses a threshold or reaches a local extremum, for example a local minimum or local maximum. In this example, as described with reference to FIGS. 2B and 3A, a characteristic instant corresponds to the instant at which a derived function crosses a threshold Th. Step 150 then includes a comparison of each value of each derived function to a threshold Th, so as to detect a descending front or a rising front. When the value of a derived function, for example the first derived function S′.sub.1(t), crosses such a threshold, a time window W is opened. The duration of this time window is short and for example comprised between 5 and 100 times the sampling period. It is preferably smaller than 500 ms and typically comprised between 20 ms and 200 ms. If, in this time window W, the other derived function, in the present case S′.sub.2(t), also crosses the threshold Th, the crossing of the threshold Th is validated for the two derived functions. The instant corresponding to the crossing of the threshold Th, by one of the two derived functions, is selected as being a characteristic instant t.sub.i. The duration of the time window W may be preset or adjustable.

[0069] FIGS. 5A and 5B illustrate step 150. FIG. 5A shows a first detection function S.sub.1(t) and its derived function S′.sub.1(t). FIG. 5B shows a second detection function S.sub.2(t) and its derived function S′.sub.2(t). FIG. 5B is similar to FIG. 3A, which was described above. A first crossing of a threshold Th, by the second derived function S′.sub.2(t), is observed at t=t.sub.1. When the threshold is crossed, a time window W of 100 ms duration is opened. During this time window, the first derived function S′.sub.1(t) is not detected to cross the threshold Th. The characteristic instant t.sub.1 is therefore invalidated. A second crossing of the threshold Th by the first derived function S′.sub.1(t) is also observed at t=t.sub.2. When the threshold is crossed, a new time window W is opened, during which a second crossing of the threshold is detected for the second derived function S′.sub.2(t). The two crossings of the threshold Th are then considered to have occurred, at the instant t.sub.2, in temporal coincidence, or in other words simultaneously, to within a few sampling periods. The instant t.sub.2 is therefore selected as being a characteristic instant t.sub.i to be considered for evaluation of cardiac frequency.

[0070] Step 160: determining cardiac frequency hr. From the characteristic instants t.sub.i selected in step 150, an occurrence frequency f of the successive characteristic instants is established, this frequency corresponding to the cardiac frequency hr. For example, the occurrence frequency is obtained by averaging the occurrence frequency of a number N of successive characteristic instants t.sub.i. The frequency f.sub.i attributed to a characteristic instant t.sub.i may then be established depending on the average time difference between N successive instants preceding the characteristic instant, such that:

[00002]$\begin{array}{cc}{f}_i=\frac{1}{\frac{1}{N-1}.Math.{.Math.}_{j=i-N+2}^{j=i}.Math.\left(t_j-t_{j-1}\right).Math.}& \left(2\right)\end{array}$

The cardiac frequency hr.sub.i at the instant t.sub.i is equal to f.sub.i. The units may then be changed to obtain a cardiac frequency in min.sup.−1.

[0071] FIGS. 6A and 6B illustrate the advantage of the invention. These figures show an evaluation of the cardiac frequency of a man, the device being arranged in a back-scatter configuration, such as shown in FIG. 1A, on his thumb. FIG. 6A shows an estimation of the cardiac frequency according to the prior art, the measurements in the first spectral band and in the second spectral band being considered independently. FIG. 6B shows an estimation of the cardiac frequency from characteristic instants appearing in temporal coincidence, i.e. simultaneously, to within the width of the time window, in each spectral band. The detection functions used to establish FIGS. 6A and 6B were analogous. The cardiac frequency determined by implementing the invention is more stable. In particular, the parasitic fluctuations appearing, in FIG. 6A, at t=80 s, t=130 s and t=180 s, do not appear in FIG. 6B.

[0072] FIGS. 6C and 6D are analogous to FIGS. 6A and 6B, respectively, the device being placed level with the wrist of a man. An improvement in the precision of the estimation is also observed when the invention is applied.

[0073] The invention will possibly be implemented in devices to be worn by individuals and operating in a transmission or back-scatter mode. The back-scatter configuration is particularly suitable for integration into a compact watch-type device, a portable device for monitoring actigraphy or a dermal patch. In order to improve the reliability of the estimation, the device will preferably maintain contact with the skin of the person, or be kept a fixed distance from the latter, by means of a strap or another rigid or elastic mount structure.

[0074] The invention will possibly be used to monitor living beings, such as new-borns, elderly people, athletes or people at risk. The use of the red and infrared spectral bands are suitable for integration into pulsed oximetry devices based on the same spectral bands, so as to determine other physiological parameters, such as blood saturation, according to known methods.