(19)
(11)EP 3 375 368 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
17.06.2020 Bulletin 2020/25

(21)Application number: 16864128.0

(22)Date of filing:  04.11.2016
(51)International Patent Classification (IPC): 
A61B 5/08(2006.01)
A61B 5/00(2006.01)
A61B 5/0456(2006.01)
A61B 5/113(2006.01)
(86)International application number:
PCT/JP2016/082799
(87)International publication number:
WO 2017/082165 (18.05.2017 Gazette  2017/20)

(54)

RESPIRATION FREQUENCY ESTIMATING METHOD AND DEVICE

VERFAHREN UND VORRICHTUNG ZUR SCHÄTZUNG DER ATEMFREQUENZ

PROCÉDÉ ET DISPOSITIF D'ESTIMATION DE LA FREQUENCE RESPIRATOIRE


(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

(30)Priority: 10.11.2015 JP 2015220157

(43)Date of publication of application:
19.09.2018 Bulletin 2018/38

(73)Proprietor: Nippon Telegraph and Telephone Corporation
Tokyo 100-8116 (JP)

(72)Inventors:
  • OGASAWARA,Takayuki
    Musashino-shi Tokyo 180-8585 (JP)
  • TAJIMA,Takuro
    Musashino-shi Tokyo 180-8585 (JP)
  • KUWABARA,Kei
    Musashino-shi Tokyo 180-8585 (JP)
  • MATSUURA,Nobuaki
    Musashino-shi Tokyo 180-8585 (JP)
  • KASAHARA,Ryoichi
    Musashino-shi Tokyo 180-8585 (JP)

(74)Representative: Samson & Partner Patentanwälte mbB 
Widenmayerstraße 6
80538 München
80538 München (DE)


(56)References cited: : 
WO-A1-2009/043087
US-A1- 2010 004 552
WO-A1-2013/033393
US-A1- 2014 228 692
  
  • CHAN, ALEXANDER M. ET AL.: 'Ambulatory Respiratory Rate Detection using ECG and a Triaxial Accelerometer' 35TH ANNUAL INTERNATIONAL CONFERENCE OF THE IEEE EMBS 2013, pages 4058 - 4061, XP032489033
  • BATES, A. ET AL.: 'Respiratory rate and flow waveform estimation from tri-axial accelerometer data' INTERNATIONAL CONFERENCE ON BODY SENSOR NETWORKS 2010, pages 144 - 150, XP031705051
  
Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


Description

Technical Field



[0001] The present invention relates to a respiration estimation method and apparatus for estimating the respiration condition of a subject using a biological information sensor.

Background Art



[0002] Continuous respiratory monitoring of continuously observing breathing of a person is expected to be applied to diagnosis of respiratory diseases, the rehabilitation of a patient with a respiratory disease, remote monitoring of a fetus or patient, stress diagnosis, and the like. As a general respiration measurement method, a respiratory flow meter by a mask, heat measurement by a thermistor arranged in the nasal cavity, a chest deflection measurement band, an electrical impedance meter for a chest, or the like is used. These methods pose a problem that a subject has an unnatural feeling when wearing the device.

[0003] On the other hand, in recent years, along with the progress of wearable devices, there have been developed devices that give an improved wearing feeling to a human body. For example, there is proposed a wearable sensor in which electrodes 101 and 102 are embedded in clothing 100 such as a shirt, as shown in Fig. 8 (patent literature 1). This wearable sensor functions as a wearable electrocardiograph by arranging, near a heart 103, the electrodes 101 and 102 made of conductive fiber, and connecting the electrodes 101 and 102 to terminals (not shown) via wiring lines.

[0004] Attempts have been made to estimate the respiration condition of a person from a cardiac potential. However, when an attempt is made to estimate the respiration condition of a subject using an R-R interval as the interval between the R wave and the immediately preceding R wave of a cardiac potential, the influence of the autonomic nervous system of the subject becomes significantly apparent, and an estimation result changes depending on the age of the subject or the mental condition of the subject at the time of measurement or whether it is before or after the measurement. When an attempt is made to estimate the respiration condition of a subject using the R-wave amplitude of a cardiac potential, a measurement error occurs in accordance with a change in contact impedance between the electrode and the skin, that is caused by the body motion or skin condition of the subject, or the individual difference of the body structure or skin condition of the subject.

[0005] Thus, for example, non-patent literature 1 discloses a method of estimating the respiration condition (breathing rate) of a subject based on acceleration data of the subject in addition to conventional electrocardiograph data. The method disclosed in non-patent literature 1 improves estimation accuracy by obtaining the weighted mean of three indices of the R-R interval, R-wave amplitude, and acceleration of a cardiac potential. In the method disclosed in non-patent literature 1, when BRRSA represents a respiratory rate estimated from the R-R interval, BRQRSa represents a respiratory rate estimated from the R-wave amplitude, and BRaccel represents a respiratory rate estimated from the acceleration data, a respiratory rate BRcombined obtained by the weighted mean of the above respiratory rates is given by:



[0006] The weighting constants QRSA, QQRSa, and Qaccel of equation (1) change depending on the respective indices of the R-R interval, R-wave amplitude, and acceleration, and are adjusted within the range of 0 to 1 depending on the signal quality. By peak detection of time-series data, an error constant E is calculated as the linear sum of four parameters, that is, the standard deviation of peak amplitudes, the average value of the peak amplitudes, a time-varying constant between minimum values, and the ratio between the number of maximum and minimum values and the total number of maximum and minimum values, and signal quality Q is given by:
where τ represents an empirically determined constant. Note that if the signal amplitude is set as a noise level, Q is set to zero.

[0007] In the method disclosed in non-patent literature 1, since a respiratory sampling interval is restricted by the cardiac cycle, it is difficult to measure a pneumogram with a short cycle. Furthermore, in the method disclosed in non-patent literature 1, variations of the R-wave amplitude of the cardiac potential and variations of the signal waveform amplitude of the accelerator also occur due to an artifact caused by the body motion of the subject, and a constant to be used to calculate the respiratory rate needs to be changed in accordance with the individual difference. Hence, the SN (Signal-to-Noise) ratio does not always improve, and respiration estimation accuracy cannot be improved.

[0008] A respiratory rate measurement using a combination of respiration signals is known from US 2014/228692 A1 and WO 2013/033393 A1. Both documents describe an R-wave amplitude detection and an R-R interval detection from a cardiac potential waveform and an acceleration displacement detection from an acceleration signal by a respiration motion of the subject. In US 2014/228692 A1 the method and system allow for measuring a respiratory rate using a combination of respiration signals. A quality metric is computed for each of the respiration signals and the resulting metrics are combined to compute a weighted respiratory rate.

[0009] Finally, US 2010/004552 A1 discloses a method and device for the determination of breath frequency. The respiratory rate is determined on the basis of a combination of signals made either in the time domain or in the frequency domain.

Related Art Literature


Patent Literature



[0010] Patent Literature 1: Japanese Patent Laid-Open No. 2015-83045

Non-Patent Literature



[0011] Non-Patent Literature 1: A.M. Chan, N. Ferdosi, and R. Narasimhan, "Ambulatory Respiratory Rate Detection using ECG and a Triaxial Accelerometer", 35th Annual International Conference of the IEEE EMBS, pp. 4058-4061, July, 2013

Disclosure of Invention


Problem to be Solved by the Invention



[0012] The present invention has been made in consideration of the above problems, and has as its object to provide a respiration estimation method and apparatus capable of improving respiration estimation accuracy under various conditions.

Means of Solution to the Problem



[0013] According to the present invention, there is provided a respiration estimation method as defined in the independent claim 1.

[0014] According to the present invention, there is also provided a respiration estimation apparatus as defined in the independent claim 5.

Effect of the Invention



[0015] According to the present invention, by extracting a frequency as a candidate of the respiration frequency of a subject from each of the frequency spectrum of an R-wave amplitude, that of an R-R interval, and that of an angular displacement, and selecting the best data from the frequencies as the respiration frequency of the subject, it is possible to estimate the respiration frequency of the subject with respect to the individual difference or a wide variety of measurement conditions, and reduce the influence of the artifact of the body motion of the subject, thereby improving respiration estimation accuracy.

Brief Description of Drawings



[0016] 

Fig. 1 is a block diagram showing the arrangement of a respiration estimation apparatus according to an embodiment of the present invention;

Fig. 2 is a flowchart for explaining the operation of the respiration estimation apparatus according to the embodiment of the present invention;

Fig. 3 is a block diagram showing the arrangement of a signal selection unit of the respiration estimation apparatus according to the embodiment of the present invention;

Fig. 4 is a flowchart for explaining the operation of the signal selection unit of the respiration estimation apparatus according to the embodiment of the present invention;

Figs. 5A and 5B are graphs each showing an example of a frequency spectrum as a result of Fourier-transforming data according to the embodiment of the present invention;

Fig. 6 shows timing charts of respiration frequency signal strings each obtained when each of the peak frequencies of an angular displacement, an R-R interval, and an R-wave amplitude is selected as a respiration frequency, and a respiration frequency signal string obtained by the respiration estimation apparatus according to the embodiment of the present invention;

Fig. 7 shows timing charts of a respiration frequency signal string obtained by the respiration estimation apparatus according to the embodiment of the present invention and a respiration frequency signal string obtained by a related respiration sensor; and

Fig. 8 is a schematic view showing a state in which a wearable sensor is mounted on a human body.


Best Mode for Carrying Out the Invention



[0017] An embodiment of the present invention will be described below with reference to the accompanying drawings. Fig. 1 is a block diagram showing the arrangement of a respiration estimation apparatus according to the embodiment of the present invention. The respiration estimation apparatus includes an electrocardiograph 1, a triaxial accelerometer 2, storage units 3 and 4, an R-wave amplitude detection unit 5, an R-R interval detection unit 6, an acceleration displacement detection unit 7, a sampling unit 8, a bandpass filter 9, a Fourier transform unit 10, and a signal selection unit 11.

[0018] Fig. 2 is a flowchart for explaining the operation of the respiration estimation apparatus. The electrocardiograph 1 measures the cardiac potential waveform of a subject, and outputs the time-series signal string of the cardiac potential waveform (step S100 of Fig. 2). The storage unit 3 stores the time-series signal string of the cardiac potential waveform output from the electrocardiograph 1.

[0019] As is well known, the cardiac potential waveform is formed from continuous heartbeat waveforms, and one heartbeat waveform is formed from components such as P, Q, R, S, and T waves reflecting the activities of atriums and ventricles. The R-wave amplitude detection unit 5 detects the amplitude of the R wave from the signals of the cardiac potential waveform stored in the storage unit 3 (step S101 of Fig. 2). As a method of recognizing the R wave of the cardiac potential waveform, there is provided, for example, a technique disclosed in Japanese Patent Laid-Open No. 2003-561. This technique recognizes an R wave using a threshold based on an amplitude between the peak and valley of the cardiac potential waveform. The R-wave amplitude detection unit 5 detects an amplitude for each R wave of the cardiac potential waveform.

[0020] The R-R interval detection unit 6 detects an R-R interval from the signals of the cardiac potential waveform stored in the storage unit 3 (step S102 of Fig. 2). As a technique of detecting the R-R interval of the cardiac potential waveform, for example, there is provided a technique disclosed in "ECG Implementation on the TMS320C5515 DSP Medical Development Kit (MDK) with the ADS1298 ECG-FE", Texas Instruments Incorporated, <http://www.ti.com/lit/an/sprabj1/sprabj1.pdf>, 2011. This technique obtains an R-R interval based on a change in value of the time difference of the cardiac potential waveform. The R-R interval detection unit 6 detects an R-R interval for each R wave of the cardiac potential waveform.

[0021] On the other hand, the triaxial accelerometer 2 is mounted on the chest of the subject, and detects a triaxial acceleration by the respiratory motion of the subject and outputs the time-series signal string of the triaxial acceleration (step S103 of Fig. 2). The storage unit 4 stores the time-series signal string of the triaxial acceleration output from the triaxial accelerometer 2.

[0022] The acceleration displacement detection unit 7 detects the angular displacement of an acceleration vector from the signals of the triaxial acceleration stored in the storage unit 4 (step S104 of Fig. 2). To detect the angular displacement, after defining an acceleration variation surface from the average of the acceleration displacements in three axial directions, that is, the X, Y, and Z directions, the angle of a projection vector obtained when an acceleration vector formed from a combination of the acceleration data of the three axes in the X, Y, and Z directions is projected on the variation surface is calculated as the angular displacement. This method is disclosed in, for example, A. Bates, M.J. Ling, J. Mann and D.K. Arvind, "Respiratory rate and flow waveform estimation from triaxial accelerometer data", International Conference on Body Sensor Network, pp. 144-150, June 2010. The acceleration displacement detection unit 7 detects an angular displacement for each sampling period of the acceleration.

[0023] Subsequently, the sampling unit 8 samples each of the time-series signal of the R-wave amplitude output from the R-wave amplitude detection unit 5, the time-series signal of the R-R interval output from the R-R interval detection unit 6, and the time-series signal of the angular displacement output from the acceleration displacement detection unit 7 at a sampling frequency (for example, a 1-Hz interval) lower than the sampling frequency of the electrocardiograph 1 and that of the triaxial accelerometer 2 (step S105 of Fig. 2).

[0024] The bandpass filter 9 limits the band of each of the time-series signal of the R-wave amplitude, that of the R-R interval, and that of the angular displacement, all of which have been acquired by the sampling unit 8 (step S106 of Fig. 2). The reason why the bandpass filter 9 is used is that the respiration frequency of a person is limited to only a low frequency. The pass band of the bandpass filter 9 is, for example, 0.15 to 0.4 Hz.

[0025] After multiplying, by a Hamming window function, each of the time-series signal of the R-wave amplitude, that of the R-R interval, and that of the angular displacement, whose band has been limited by the bandpass filter 9, the Fourier transform unit 10 performs fast Fourier transform for each of the time-series signal of the R-wave amplitude, that of the R-R interval, and that of the angular displacement, thereby obtaining the frequency spectrum of each of the signals of the R-wave amplitude, the R-R interval, and the angular displacement (step S107 of Fig. 2). As is well known, the Hamming window function is used to cut desired input data. The Fourier transform unit 10 performs such Fourier transform for each sampling period of the sampling unit 8. The number of data used for Fourier transform is, for example, 128 points (128 sec).

[0026] The signal selection unit 11 extracts a frequency as a candidate of the respiration frequency of the subject from each of the frequency spectrum of the R-wave amplitude, that of the R-R interval, and that of the angular displacement, all of which have been obtained by the Fourier transform unit 10, and selects the best data from the frequencies as the respiration frequency of the subject, thereby outputting a respiration frequency signal (step S108 of Fig. 2).

[0027] Fig. 3 is a block diagram showing the arrangement of the signal selection unit 11. Fig. 4 is a flowchart for explaining the operation of the signal selection unit 11. The signal selection unit 11 includes a peak frequency detection unit 110, a detection unit 111 for a peak full width at half maximum, a frequency selection unit 112, and a respiration frequency signal reconstruction unit 113.

[0028] The peak frequency detection unit 110 detects a peak frequency of each of the frequency spectrum of the R-wave amplitude, that of the R-R interval, and that of the angular displacement, all of which have been obtained by the Fourier transform unit 10 (step S200 of Fig. 4). When a plurality of peaks are detected for one frequency spectrum, the peak frequency detection unit 110 selects the frequency of the peak whose intensity is the maximum.

[0029] Subsequently, the detection unit 111 detects the full width at half maximum of the peak detected by the peak frequency detection unit 110 for each of the frequency spectrum of the R-wave amplitude, that of the R-R interval, and that of the angular displacement (step S201 of Fig. 4). The full width at half maximum indicates the width of the frequency at which the intensity is 50% of the peak intensity in the waveform of the peak detected by the peak frequency detection unit 110.

[0030] Fig. 5A shows an example of the frequency spectrum when the SN ratio is high, and Fig. 5B shows an example of the frequency spectrum when the SN ratio is low. Note that Figs. 5A and 5B are graphs obtained by normalizing the intensity on the ordinate. In this example, normalization is performed by obtaining an area by integrating the power of each frequency (accumulating discrete values) so that the area of the power spectrum is 1. The power spectrum is obtained by cutting, for example, a section of 30 sec from the measured time-series data, and performing Fourier transform. In general, the breathing rate fluctuates but stays constant while it is stable, and has a peak P at a given frequency. The peak P of the respiration frequency is normally considered to fall within the range of about 0.2 to 0.4 Hz. However, a full width W at half maximum of the peak tends to extend due to the influence of the body motion and autonomic nervous system of the subject. It is considered that as the full width W at half maximum is narrower, the respiration detection accuracy is higher.

[0031] The frequency selection unit 112 selects, as the respiration frequency of the subject, the peak frequency at which the full width at half maximum detected by the detection unit 111 is equal to or smaller than a predetermined threshold (for example, 0.0625 Hz) among the three peak frequencies (the peak frequency of the R-wave amplitude, that of the R-R interval, and that of the angular displacement) detected by the peak frequency detection unit 110 (step S202 of Fig. 4), and discards the unselected peak frequencies (step S203 of Fig. 4).

[0032] When there are a plurality of peak frequencies at which the full width at half maximum is equal to or smaller than the predetermined threshold (for example, the full width at half maximum of the R-wave amplitude and that of the angular displacement are equal to or smaller than the threshold), the frequency selection unit 112 selects, as the respiration frequency, the peak frequency at which the full width at half maximum is narrower. Note that if all the three full widths at half maximum (the full width at half maximum of the R-wave amplitude, that of the R-R interval, and that of the angular displacement) exceed the predetermined threshold, the frequency selection unit 112 performs processing based on an assumption that data of the current data point at which the respiration frequency is to be confirmed is defective.

[0033] The respiration frequency signal reconstruction unit 113 generates a respiration frequency signal string by arranging, in time-series, the data of the respiration frequencies selected by the frequency selection unit 112 (step S204 of Fig. 4), and outputs it. Note that when the data of the respiration frequency at given time is defective, the respiration frequency signal reconstruction unit 113 complements the data by using the data of the respiration frequency at immediately preceding time (one sampling period before).

[0034] The signal selection unit 11 performs the above processing for each sampling period of the sampling unit 8 (every second in the example of this embodiment).

[0035] In Fig. 6, (a) is timing chart showing a respiration frequency signal string obtained when the peak frequency of the angular displacement is selected as a respiration frequency, (b) is a timing chart showing a respiration frequency signal string obtained when the peak frequency of the R-R interval is selected as a respiration frequency, (c) is a timing chart showing a respiration frequency signal string obtained when the peak frequency of the R-wave amplitude is selected as a respiration frequency, and (d) is a timing chart showing a respiration frequency signal string obtained by the respiration estimation apparatus according to this embodiment. In Fig. 7, (a) is a timing chart showing a respiration frequency signal string obtained by the respiration estimation apparatus according to this embodiment, and (b) is a timing chart showing a respiration frequency signal string obtained by a related respiration sensor.

[0036] The actual measured value of the respiration frequency measured by the respiration sensor is at about 0.2 Hz. As is apparent from (a) of Fig. 7, according to this embodiment, it is understood that the transition of the respiration frequency which matches the result obtained by the respiration sensor is obtained.

[0037] According to this embodiment, the respiration frequency of the subject is estimated by integrating the plurality of sensor data and extracting the best data in order to cope with the individual difference of the age, autonomic nervous system, skin condition, and body structure of the subject.

[0038] In a case in which data of the triaxial acceleration is used to estimate the respiration frequency, it is difficult to exclude a measurement error caused by the individual difference of a body motion or body structure. When data of the R-R interval is used to estimate the respiration frequency, there is restriction on calculation of a respiration cycle by the cardiac cycle, and it is thus difficult to exclude a measurement error caused by the influence of the autonomic nervous system that changes depending on the mental condition and age. When data of the R-wave amplitude is used to estimate the respiration frequency, it is difficult to exclude a measurement error caused by the individual difference of the skin condition, a change in contact impedance in accordance with the body motion or skin condition, or the like. To the contrary, in this embodiment, even if the S/N ratio degrades, it is possible to estimate the respiration frequency by selecting the best data from the plurality of sensor data.

[0039] The storage units 3 and 4, R-wave amplitude detection unit 5, R-R interval detection unit 6, acceleration displacement detection unit 7, sampling unit 8, bandpass filter 9, Fourier transform unit 10, and signal selection unit 11, all of which have been described in this embodiment, can be implemented by a computer including a CPU (Central Processing Unit), a storage device, and an interface, and a program for controlling these hardware resources. The CPU executes the processing described in this embodiment in accordance with programs stored in the storage device.

[0040] Note that the electrocardiograph 1 includes an electrode attached to clothing such as a shirt, and a cardiac potential waveform signal processing unit in a monitoring apparatus attached to the clothing, and the electrode and the cardiac potential waveform signal processing unit are connected by a wiring line. Similarly, the triaxial accelerometer 2 includes a sensor unit attached to the clothing and an acceleration signal processing unit provided in the monitoring apparatus, and the sensor unit and the acceleration signal processing unit are connected by a wiring line.

[0041] In this embodiment, the electrocardiograph 1 and the triaxial accelerometer 2 may be provided together with or separately from the wearable device attached to the clothing. That is, the storage units 3 and 4, R-wave amplitude detection unit 5, R-R interval detection unit 6, acceleration displacement detection unit 7, sampling unit 8, bandpass filter 9, Fourier transform unit 10, and signal selection unit 11 may be provided in the monitoring apparatus or provided in another apparatus. When the storage units 3 and 4, R-wave amplitude detection unit 5, R-R interval detection unit 6, acceleration displacement detection unit 7, sampling unit 8, bandpass filter 9, Fourier transform unit 10, and signal selection unit 11 are provided in another apparatus, a cardiac potential waveform signal obtained by the electrocardiograph 1 and a triaxial acceleration signal obtained by the triaxial accelerometer 2 are wirelessly transmitted to the apparatus.

Industrial Applicability



[0042] The present invention is applicable to continuous respiratory monitoring of continuously observing breathing of a person.

Explanation of the Reference Numerals and Signs



[0043] 1...electrocardiograph, 2...triaxial accelerometer, 3, 4...storage unit, 5...R-wave amplitude detection unit, 6...R-R interval detection unit, 7...acceleration displacement detection unit, 8...sampling unit, 9...bandpass filter, 10...Fourier transform unit, 11...signal selection unit, 110...peak frequency detection unit, 111...detection unit for peak full width at half maximum, 112...frequency selection unit, 113...respiration frequency signal reconstruction unit


Claims

1. A respiration estimation method comprising:

an R-wave amplitude detection step (S101) of detecting an amplitude of an R wave from a cardiac potential waveform of a subject;

an R-R interval detection step (S102) of detecting an R-R interval as an interval between an R wave and an immediately preceding R wave from the cardiac potential waveform of the subject;

an acceleration displacement detection step (S104) of detecting an angular displacement of an acceleration vector from a triaxial acceleration signal by a respiratory motion of the subject;

characterized in that the method further comprises:

a Fourier transform step (S107) of Fourier-transforming each of a time-series signal of the R-wave amplitude, a time-series signal of the R-R interval, and a time-series signal of the angular displacement to obtain a frequency spectrum of each of the signals of the R-wave amplitude, R-R interval, and the angular displacement; and

a signal selection step (S108) of extracting a frequency as a candidate of a respiration frequency of the subject from each of the frequency spectrum of the R-wave amplitude, the frequency spectrum of the R-R interval, and the frequency spectrum of the angular displacement, all of which have been obtained in the Fourier transform step, wherein the signal selection step (S108) includes

a peak frequency detection step (S200) of detecting a peak frequency, as a candidate of the respiration frequency of the subject, for each of the frequency spectrum of the R-wave amplitude, the frequency spectrum of the R-R interval, and the frequency spectrum of the angular displacement, all of which have been obtained in the Fourier transform step,

a step (S201) of detecting a full width at half maximum of the peak detected in the peak frequency detection step, for each of the frequency spectrum of the R-wave amplitude, the frequency spectrum of the R-R interval, and the frequency spectrum of the angular displacement, and

a frequency selection step (S202) of selecting, as the respiration frequency of the subject, a peak frequency at which the full width at half maximum detected in the step of detecting the full width at half maximum is not larger than a predetermined threshold among the three peak frequencies detected in the peak frequency detection step.


 
2. The respiration estimation method according to claim 1, wherein the frequency selection step includes a step of selection, when there is a plurality of peak frequencies at which the full width at half maximum is not larger than the threshold, the peak frequency at which the full width at half maximum is narrower as the respiration frequency.
 
3. The respiration estimation method according to claim 1, wherein the frequency selection step includes a step of determining, when all the three full widths at half maximum exceed the threshold, that data of a current data point is defective.
 
4. The respiration estimation method according to any one of claims 1 to 3, further comprising:

a sampling step (S105) of sampling each of the time-series signal of the R-wave amplitude obtained in the R-wave amplitude detection step, the time-series signal of the R-R interval obtained in the R-R interval detection step, and the time-series signal of the angular displacement obtained in the acceleration displacement detection step at a sampling frequency lower than a sampling frequency of the cardiac potential waveform and a sampling frequency of the triaxial acceleration signal; and

a filtering step (S106) of limiting a band of each of the time-series signal of the R-wave amplitude, the time-series signal of the R-R interval, and the time-series signal of the angular displacement, all of which have been acquired in the sampling step,

wherein the Fourier transform step includes a step of Fourier-transforming each of the time-series signal of the R-wave amplitude, the time-series signal of the R-R interval, and the time-series signal of the angular displacement, in which each the band has been limited in the filtering step.


 
5. A respiration estimation apparatus comprising:

an R-wave amplitude detection unit (5) configured to detect an amplitude of an R wave from a cardiac potential waveform of a subject;

an R-R interval detection unit (6) configured to detect an R-R interval as an interval between an R wave and an immediately preceding R wave from the cardiac potential waveform of the subject;

an acceleration displacement detection unit (7) configured to detect an angular displacement of an acceleration vector from a triaxial acceleration signal by a respiratory motion of the subject;

characterized in that the respiration estimation apparatus further comprises:

a Fourier transform unit (10) configured to Fourier-transform each of a time-series signal of the R-wave amplitude, a time-series signal of the R-R interval, and a time-series signal of the angular displacement to obtain a frequency spectrum of each of the signals of the R-wave amplitude, the R-R interval, and the angular displacement; and

a signal selection unit (11) configured to extract a frequency as a candidate of a respiration frequency of the subject from each of the frequency spectrum of the R-wave amplitude, the frequency spectrum of the R-R interval, and the frequency spectrum of the angular displacement, all of which have been obtained by the Fourier transform unit (10), wherein the signal selection unit (11) includes

a peak frequency detection unit (110) configured to detect a peak frequency as a candidate of the respiration frequency of the subject for each of the frequency spectrum of the R-wave amplitude, the frequency spectrum of the R-R interval, and the frequency spectrum of the angular displacement, all of which have been obtained by the Fourier transform unit (10),

a detection unit (111) for a peak full width at half maximum configured to detect a full width at half maximum of the peak detected by the peak frequency detection unit (110), for each of the frequency spectrum of the R-wave amplitude, the frequency spectrum of the R-R interval, and the frequency spectrum of the angular displacement, and

a frequency selection unit (112) configured to select, as the respiration frequency of the subject, a peak frequency at which the full width at half maximum detected by the detection unit (111) for the peak full width at half maximum is not larger than a predetermined threshold among the three peak frequencies detected by the peak frequency detection unit (110).


 
6. The respiration estimation apparatus according to claim 5, wherein when there is a plurality of peak frequencies at which the full width at half maximum is not larger than the threshold, the frequency selection unit (112) selects, as the respiration frequency, the peak frequency at which the full width at half maximum is narrower.
 
7. The respiration estimation apparatus according to claim 5, wherein when all the three full widths at half maximum exceed the threshold, the frequency selection unit (112) determines that data of a current data point is defective.
 
8. The respiration estimation apparatus according to any one of claims 5 to 7, further comprising:

a sampling unit (8) configured to sample each of the time-series signal of the R-wave amplitude obtained by the R-wave amplitude detection unit (5), the time-series signal of the R-R interval obtained by the R-R interval detection unit (6), and the time-series signal of the angular displacement obtained by the acceleration displacement detection unit (7) at a sampling frequency lower than a sampling frequency of the cardiac potential waveform and a sampling frequency of the triaxial acceleration signal; and

a bandpass filter (9) configured to limit a band of each of the time-series signal of the R-wave amplitude, the time-series signal of the R-R interval, and the time-series signal of the angular displacement, all of which have been acquired by the sampling unit (8),

wherein the Fourier transform unit (10) Fourier-transforms each of the time-series signal of the R-wave amplitude, the time-series signal of the R-R interval, and the time-series signal of the angular displacement, in which each the band has been limited by the bandpass filter (9).


 


Ansprüche

1. Ein Verfahren zur Schätzung der Atmung umfassend:

einen R-Wellen-Amplituden-Erfassungsschritt (S101) zum Erfassen einer Amplitude einer R-Welle aus einer Herzpotential-Wellenform einer Person;

einen R-R-Intervall-Erfassungsschritt (S102) zum Erfassen eines R-R-Intervalls als ein Intervall zwischen einer R-Welle und einer unmittelbar vorhergehenden R-Welle aus der Herzpotential-Wellenform der Person;

einen Beschleunigungsverschiebungs-Schritt (S104) zum Erfassen einer Winkelverschiebung eines Beschleunigungsvektors aus einem dreiachsigen Beschleunigungssignal durch eine Atmungsbewegung der Person;

dadurch gekennzeichnet, dass das Verfahren ferner umfasst:

einen Fourier-Transformationsschritt (S107) zur Fourier-Transformation jedes eines Zeitreihensignals der R-Wellen-Amplitude, eines Zeitreihensignals des R-R-Intervalls und eines Zeitreihensignals der Winkelverschiebung, um ein Frequenzspektrum jedes der Signale der R-Wellen-Amplitude, des R-R-Intervalls und der Winkelverschiebung zu erhalten; und

einen Signalauswahlschritt (S108) zum Extrahieren einer Frequenz als ein Kandidat einer Atemfrequenz der Person aus jedem des Frequenzspektrums der R-Wellen-Amplitude, des Frequenzspektrums des R-R-Intervalls und des Frequenzspektrums der Winkelverschiebung, die alle in dem Fourier-Transformationsschritt erhalten wurden, wobei der Signalauswahlschritt (S108) aufweist:

einen Spitzenfrequenzerfassungsschritt (S200) zum Erfassen einer Spitzenfrequenz als ein Kandidat für die Atemfrequenz der Person für jedes des Frequenzspektrums der R-Wellen-Amplitude, des Frequenzspektrums des R-R-Intervalls und des Frequenzspektrums der Winkelverschiebung, die alle im Fourier-Transformationsschritt erhalten wurden,

einen Schritt (S201) des Erfassens einer vollen Breite bei der Hälfte des Maximums der Spitze, die in dem Spitzenfrequenzerfassungs-Schritt erfasst wurde, für jedes des Frequenzspektrums der R-Wellen-Amplitude, des Frequenzspektrums des R-R-Intervalls und des Frequenzspektrums der Winkelverschiebung, und

einen Frequenzauswahlschritt (S202) zum Auswählen als die Atemfrequenz der Person eine Spitzenfrequenz, bei der die volle Breite bei halbem Maximum, die in dem Schritt des Erfassens der vollen Breite bei halbem Maximum erfasst wird, nicht größer als ein vorbestimmter Schwellenwert unter den drei Spitzenfrequenzen ist, die in dem Spitzenfrequenzerfassungs-Schritt erfasst wurden.


 
2. Verfahren zur Schätzung der Atmung nach Anspruch 1, wobei der Frequenzauswahlschritt einen Auswahlschritt aufweist, wenn es eine Mehrzahl von Spitzenfrequenzen gibt, bei denen die volle Breite bei halbem Maximum nicht größer als der Schwellenwert ist, die Spitzenfrequenz, bei der die volle Breite bei halbem Maximum schmaler als die Atemfrequenz ist.
 
3. Verfahren zur Schätzung der Atmung nach Anspruch 1, wobei der Frequenzauswahlschritt einen Schritt zur Bestimmung, dass Daten eines aktuellen Datenpunktes fehlerhaft sind, umfasst, wenn alle drei vollen Breiten bei der Hälfte des Maximums den Schwellenwert überschreiten.
 
4. Verfahren zur Schätzung der Atmung nach einem der Ansprüche 1 bis 3, ferner umfassend:

einen Abtastschritt (S105) zum Abtasten jedes des Zeitreihensignals der R-Wellen-Amplitude, das im R-Wellen-Amplituden-Erfassungsschritt erhalten wurden, des Zeitreihensignals des R-R-Intervalls, das im R-R-Intervall-Erfassungsschritt erhalten wurde, und des Zeitreihensignals der Winkelverschiebung, das im Beschleunigungsverschiebungs-Erfassungsschritt bei einer Abtastfrequenz erhalten wurde, die niedriger ist als eine Abtastfrequenz der Herzpotenzialwellenform und eine Abtastfrequenz des dreiachsigen Beschleunigungssignals; und

einen Filterschritt (S106) zum Begrenzen eines Bandes von jedem des Zeitreihensignals der R-Wellen-Amplitude, des Zeitreihensignals des R-R-Intervalls und des Zeitreihensignals der Winkelverschiebung, die alle im Abtastschritt aufgenommen worden sind,

wobei der Fourier-Transformations-Schritt einen Schritt der Fourier-Transformation jedes des Zeitreihensignals der R-Wellen-Amplitude, des Zeitreihensignals des R-R-Intervalls und des Zeitreihensignals der Winkelverschiebung aufweist, wobei jedes der Bänder in dem Filterschritt begrenzt worden ist.


 
5. Eine Vorrichtung zur Schätzung der Atmung umfassend:

eine R-Wellen-Amplituden-Erfassungseinheit (5), die so konfiguriert ist, dass sie die Amplitude einer R-Welle aus einer Herzpotenzials-Wellenform einer Person erfasst;

eine R-R-Intervall-Erfassungseinheit (6), die so konfiguriert ist, dass sie ein R-R-Intervall als ein Intervall zwischen einer R-Welle und einer unmittelbar vorhergehenden R-Welle aus der Herzpotential-Wellenform der Person erfasst;

eine Beschleunigungsverschiebungs-Erfassungseinheit (7), die so konfiguriert ist, dass sie eine Winkelverschiebung eines Beschleunigungsvektors aus einem dreiachsigen Beschleunigungssignal durch eine Atmungsbewegung der Person erfasst;

dadurch gekennzeichnet, dass die Atmungsschätzvorrichtung ferner umfasst:

eine Fourier-Transformationseinheit (10), die so konfiguriert ist, dass sie jedes eines Zeitreihensignals der R-Wellen-Amplitude, eines Zeitreihensignals des R-R-Intervalls und eines Zeitreihensignals der Winkelverschiebung fouriertransformiert, um ein Frequenzspektrum jedes der Signale der R-Wellen-Amplitude, des R-R-Intervalls und der Winkelverschiebung zu erhalten; und

eine Signalauswahleinheit (11), die so konfiguriert ist, dass sie eine Frequenz als ein Kandidat für eine Atemfrequenz der Person aus jedem des Frequenzspektrums der R-Wellen-Amplitude, des Frequenzspektrums des R-R-Intervalls und des Frequenzspektrums der Winkelverschiebung extrahiert, die alle von der Fourier-Transformationseinheit (10) erhalten wurden, wobei die Signalauswahleinheit (11) umfasst:

eine Spitzenfrequenz-Erfassungseinheit (110), die so konfiguriert ist, dass sie eine Spitzenfrequenz als ein Kandidat der Atemfrequenz der Person für jedes des Frequenzspektrums der R-Wellen-Amplitude, des Frequenzspektrums des R-R-Intervalls und des Frequenzspektrums der Winkelverschiebung erfasst, die alle von der Fourier-Transformationseinheit (10) erhalten wurden,

eine Erfassungseinheit (111) für eine Spitze voller Breite bei halbem Maximum, die so konfiguriert ist, dass sie eine volle Breite bei halbem Maximum der Spitze erfasst, die von der Spitzenfrequenz-Erfassungseinheit (110) erfasst wird, für jedes des Frequenzspektrums der R-Wellen-Amplitude, des Frequenzspektrums des R-R-Intervalls und des Frequenzspektrums der Winkelverschiebung, und

eine Frequenzauswahleinheit (112), die so konfiguriert ist, dass sie als Atemfrequenz der Person eine Spitzenfrequenz auswählt, bei der die volle Breite bei halbem Maximum, die von der Erfassungseinheit (111) erfasst wird, nicht größer als ein vorbestimmter Schwellenwert unter den drei Spitzenfrequenzen ist, die von der Spitzenfrequenz-Erfassungseinheit (110) erfasst werden.


 
6. Atmungsschätzvorrichtung nach Anspruch 5, wobei, wenn es eine Mehrzahl von Spitzenfrequenzen gibt, bei denen die volle Breite bei halbem Maximum nicht größer als der Schwellenwert ist, die Frequenzauswahl einheit (112) als die Atemfrequenz die Spitzenfrequenz auswählt, bei der die volle Breite bei halbem Maximum schmaler ist.
 
7. Atmungsschätzvorrichtung nach Anspruch 5, wobei die Frequenzauswahleinheit (112), wenn alle drei vollen Breiten bei halbem Maximum den Schwellenwert überschreiten, bestimmt, dass Daten eines aktuellen Datenpunktes fehlerhaft sind.
 
8. Atmungsschätzvorrichtung nach einem der Ansprüche 5 bis 7, ferner umfassend:

eine Abtasteinheit (8), die so konfiguriert ist, dass sie jedes der Zeitreihensignale der R-Wellen-Amplitude, die von der R-Wellen-Amplituden-Erfassungseinheit (5) erhalten werden, des Zeitreihensignals des R-R-Intervalls, das von der R-R-Intervall-Erfassungseinheit (6) erhalten wird, und des Zeitreihensignals der Winkelverschiebung, das von der Beschleunigungsverschiebungs-Erfassungseinheit (7) erhalten wird, mit einer Abtastfrequenz abtastet, die niedriger ist als eine Abtastfrequenz der Herzpotenzialwellenform und eine Abtastfrequenz des dreiachsigen Beschleunigungssignals; und

einen Bandpassfilter (9), der so konfiguriert ist, dass er ein Band von jedem des Zeitreihensignals der R-Wellen-Amplitude, des Zeitreihensignals des R-R-Intervalls und des Zeitreihensignals der Winkelverschiebung begrenzt, die alle von der Abtasteinheit (8) erfasst worden sind,

wobei die Fourier-Transformationseinheit (10) jedes des Zeitreihensignals der R-Wellen-Amplitude, des Zeitreihensignals des R-R-Intervalls und des Zeitreihensignals der Winkelverschiebung fouriertransformiert, wobei das Band jeweils durch das Bandpassfilter (9) begrenzt wurde.


 


Revendications

1. Procédé d'estimation de la respiration comprenant:

une étape de détection d'amplitude d'onde R (S101) consistant à détecter une amplitude d'une onde R à partir d'une forme d'onde de potentiel cardiaque d'un sujet;

une étape de détection d'intervalle R-R (S102) consistant à détecter un intervalle R-R en tant qu'intervalle entre une onde R et une onde R immédiatement précédente à partir de la forme d'onde de potentiel cardiaque du sujet;

une étape de détection de déplacement d'accélération (S104) consistant à détecter un déplacement angulaire d'un vecteur d'accélération à partir d'un signal d'accélération triaxiale par un mouvement respiratoire du sujet;

caractérisé en ce que le procédé comprend en outre:

une étape de transformée de Fourier (S107) consistant à effectuer une transformée de Fourier sur chacun d'un signal chronologique de l'amplitude d'onde R, d'un signal chronologique de l'intervalle R-R et d'un signal chronologique du déplacement angulaire pour obtenir un spectre de fréquence de chacun des signaux de l'amplitude d'onde R, de l'intervalle R-R et du déplacement angulaire; et

une étape de sélection de signal (S108) consistant à extraire une fréquence en tant que candidat d'une fréquence de respiration du sujet parmi chacun du spectre de fréquence de l'amplitude d'onde R, du spectre de fréquence de l'intervalle R-R et du spectre de fréquence du déplacement angulaire, tous ayant été obtenus à l'étape de transformée de Fourier, dans lequel l'étape de sélection de signal (S108) inclut

une étape de détection de fréquence de crête (S200) consistant à détecter une fréquence de crête, en tant que candidat de la fréquence de respiration du sujet, pour chacun du spectre de fréquence de l'amplitude d'onde R, du spectre de fréquence de l'intervalle R-R et du spectre de fréquence du déplacement angulaire, tous ayant été obtenus à l'étape de transformée de Fourier,

une étape (S201) consistant à détecter une largeur totale à mi-hauteur de la crête détectée à l'étape de détection de fréquence de crête, pour chacun du spectre de fréquence de l'amplitude d'onde R, du spectre de fréquence de l'intervalle R-R et du spectre de fréquence du déplacement angulaire, et

une étape de sélection de fréquence (S202) consistant à sélectionner, en tant que fréquence de respiration du sujet, une fréquence de crête à laquelle la largeur totale à mi-hauteur détectée à l'étape de détection de la largeur totale à mi-hauteur n'est pas supérieure à un seuil prédéterminé parmi les trois fréquences de crête détectées à l'étape de détection de fréquence de crête.


 
2. Procédé d'estimation de respiration selon la revendication 1, dans lequel l'étape de sélection de fréquence inclut une étape consistant à sélectionner, quand il y a une pluralité de fréquences de crête auxquelles la largeur totale à mi-hauteur n'est pas supérieure au seuil, la fréquence de crête à laquelle la largeur totale à mi-hauteur est plus étroite que la fréquence de respiration.
 
3. Procédé d'estimation de respiration selon la revendication 1, dans lequel l'étape de sélection de fréquence inclut une étape consistant à déterminer, quand les trois largeurs totales à mi-hauteur dépassent toutes le seuil, que des données d'un point de données actuel sont défectueuses.
 
4. Procédé d'estimation de respiration selon l'une quelconque des revendications 1 à 3, comprenant en outre:

une étape d'échantillonnage (S105) consistant à échantillonner chacun du signal chronologique de l'amplitude d'onde R obtenue à l'étape de détection d'amplitude d'onde R, du signal chronologique de l'intervalle R-R obtenu à l'étape de détection d'intervalle R-R et du signal chronologique du déplacement angulaire obtenu à l'étape de détection de déplacement d'accélération à une fréquence d'échantillonnage inférieure à une fréquence d'échantillonnage de la forme d'onde de potentiel cardiaque et une fréquence d'échantillonnage du signal d'accélération triaxiale; et

une étape de filtrage (S106) consistant à limiter une bande de chacun du signal chronologique de l'amplitude d'onde R, du signal chronologique de l'intervalle R-R et du signal chronologique du déplacement angulaire, tous ayant été acquis à l'étape d'échantillonnage,

dans lequel l'étape de transformée de Fourier inclut une étape consistant à effectuer une transformée de Fourier sur chacun du signal chronologique de l'amplitude d'onde R, du signal chronologique de l'intervalle R-R et du signal chronologique du déplacement angulaire, dans lesquels chaque bande a été limitée à l'étape de filtrage.


 
5. Appareil d'estimation de respiration comprenant:

une unité de détection d'amplitude d'onde R (5) configurée pour détecter une amplitude d'une onde R à partir d'une forme d'onde de potentiel cardiaque d'un sujet;

une unité de détection d'intervalle R-R (6) configurée pour détecter un intervalle R-R en tant qu'intervalle entre une onde R et une onde R immédiatement précédente à partir de la forme d'onde de potentiel cardiaque du sujet;

une unité de détection de déplacement d'accélération (7) configurée pour détecter un déplacement angulaire d'un vecteur d'accélération à partir d'un signal d'accélération triaxiale par un mouvement respiratoire du sujet;

caractérisé en ce que l'appareil d'estimation de respiration comprend en outre:

une unité de transformée de Fourier (10) configurée pour effectuer une transformée de Fourier sur chacun d'un signal chronologique de l'amplitude d'onde R, d'un signal chronologique de l'intervalle R-R et d'un signal chronologique du déplacement angulaire pour obtenir un spectre de fréquence de chacun des signaux de l'amplitude d'onde R, de l'intervalle R-R et du déplacement angulaire; et

une unité de sélection de signal (11) configurée pour extraire une fréquence en tant que candidat d'une fréquence de respiration du sujet parmi chacun du spectre de fréquence de l'amplitude d'onde R, du spectre de fréquence de l'intervalle R-R et du spectre de fréquence du déplacement angulaire, tous ayant été obtenus par l'unité de transformée de Fourier (10), dans lequel l'unité de sélection de signal (11) inclut

une unité de détection de fréquence de crête (110) configurée pour détecter une fréquence de crête en tant que candidat de la fréquence de respiration du sujet pour chacun du spectre de fréquence de l'amplitude d'onde R, du spectre de fréquence de l'intervalle R-R et du spectre de fréquence du déplacement angulaire, tous ayant été obtenus par l'unité de transformée de Fourier (10),

une unité de détection (111) pour une largeur totale de crête à mi-hauteur configurée pour détecter une largeur totale à mi-hauteur de la crête détectée par l'unité de détection de fréquence de crête (110), pour chacun du spectre de fréquence de l'amplitude d'onde R, du spectre de fréquence de l'intervalle R-R et du spectre de fréquence du déplacement angulaire, et

une unité de sélection de fréquence (112) configurée pour sélectionner, en tant que fréquence de respiration du sujet, une fréquence de crête à laquelle la largeur totale à mi-hauteur détectée par l'unité de détection (111) pour la largeur totale de crête à mi-hauteur n'est pas supérieure à un seuil prédéterminé parmi les trois fréquences de crête détectées par l'unité de détection de fréquence de crête (110).


 
6. Appareil d'estimation de respiration selon la revendication 5, dans lequel, quand il y a une pluralité de fréquences de crête auxquelles la largeur totale à mi-hauteur n'est pas supérieure au seuil, l'unité de sélection de fréquence (112) détecte, en tant que fréquence de respiration, la fréquence de crête à laquelle la largeur totale à mi-hauteur est plus étroite.
 
7. Appareil d'estimation de respiration selon la revendication 5, dans lequel, quand les trois largeurs totales à mi-hauteur dépassent toutes le seuil, l'unité de sélection de fréquence (112) détermine que des données d'un point de données actuel sont défectueuses.
 
8. Appareil d'estimation de respiration selon l'une quelconque des revendications 5 à 7, comprenant en outre:

une unité d'échantillonnage (8) configurée pour échantillonner chacun du signal chronologique de l'amplitude d'onde R obtenue par l'unité de détection d'amplitude d'onde R (5), du signal chronologique de l'intervalle R-R obtenu par l'unité de détection d'intervalle R-R (6) et du signal chronologique du déplacement angulaire obtenu par l'unité de détection de déplacement d'accélération (7) à une fréquence d'échantillonnage inférieure à une fréquence d'échantillonnage de la forme d'onde de potentiel cardiaque et une fréquence d'échantillonnage du signal d'accélération triaxiale; et

un filtre passe-bande (9) configuré pour limiter une bande de chacun du signal chronologique de l'amplitude d'onde R, du signal chronologique de l'intervalle R-R et du signal chronologique du déplacement angulaire, tous ayant été acquis par l'unité d'échantillonnage (8),

dans lequel l'unité de transformée de Fourier (10) effectue une transformée de Fourier sur chacun du signal chronologique de l'amplitude d'onde R, du signal chronologique de l'intervalle R-R et du signal chronologique du déplacement angulaire, dans lesquels chaque bande a été limitée par le filtre passe-bande (9).


 




Drawing





























Cited references

REFERENCES CITED IN THE DESCRIPTION



This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

Patent documents cited in the description




Non-patent literature cited in the description