ORIGINAL ARTICLE

A novel autonomic activation measurement method for stressmonitoring: non-contact measurement of heart rate variabilityusing a compact microwave radar

Satoshi Suzuki Æ Takemi Matsui Æ Hayato Imuta ÆMaki Uenoyama Æ Hirofumi Yura Æ Masayuki Ishihara ÆMitsuyuki Kawakami

Received: 9 August 2007 / Accepted: 28 November 2007 / Published online: 9 January 2008

� International Federation for Medical and Biological Engineering 2007

Abstract We developed a novel method for non-contact

monitoring of stress-induced autonomic activation through

the back of a chair, using a compact 24 GHz microwave

radar (8 9 5 9 3 cm), without large-scale equipment and

placing a heavy burden on the monitored individual.

Following a silent period of 120 s, audio stimuli using a

composite tone of 2,120 and 2,130 Hz sine-waves at

95 dB were conducted for 120 s. From dorsal, LF/HF of

HRV reflecting sympatho-vagal balance was determined

by microwave radar with the maximum entropy method

using eight volunteers (mean age 23 ± 1 years). Mean

LF/HF measured by non-contact and contact (using electro-

cardiography for reference) methods during audio stimuli

increased 34 and 37%, respectively, as compared with

those of the silent period. Maximum cross-correlations

between contact and non-contact measurements aver-

aged 0.73 ± 0.10. Our method appears to be promising

for future monitoring of stress-induced autonomic

activation of operators and may reduce stress-induced

accidents.

Keywords Non-contact � Microwave radar �Heart-rate variability � Audio stimuli �Autonomic activation � Safety precaution

1 Introduction

Using a ceiling attached microwave antenna, we have

proposed a system to monitor the respiratory rates of

subjects on a bed through a thick comfortable bed covering

the subject [15]. Min et al. [6] has reported a non-contact

method to capture the respiratory motion of a subject by the

Doppler ultrasound.

To monitor the autonomic activation induced by mental

stress, without placing any burden on the monitored indi-

vidual, we developed a non-contact autonomic monitoring

method using a 24-GHz compact microwave radar. We

have previously reported non-contact methods to monitor

heart and respiratory rates in experimental animals exposed

to toxic materials or under a hypovolemic state to determine

pathophysiological condition of the subject, such as expo-

sure to toxins or shock induced by hemorrhage [2, 4, 5].

Single photon emission tomography (SPECT) with radio-

isotope (99mTc-FBPBAT) is good for mapping the

autonomic nervous system, but is impractical for autonomic

activation monitoring due to the need for large-scale

equipment [10]. Using continuous electrocardiography

(ECG) with conventional electrodes, rhythmic components

of heart-rate variability (HRV) can be assessed using power

spectral analysis and modifications in autonomic activities

induced by mental stress have been reported in HRV power

spectra [12, 14]. However, long-term electrocardiographic

S. Suzuki � T. Matsui (&) � H. Imuta � M. Uenoyama �H. Yura � M. Kawakami

Department of Management Systems Engineering,

Tokyo Metropolitan University, 6-6 Asahigaoka,

Hino, Tokyo 191-0065, Japan

e-mail: tmatsui@cc.tmit.ac.jp

H. Yura

NeTech, Inc., Sakado 3-2-1, Kanagawa 213-0012, Japan

e-mail: info@netech.jp

M. Ishihara

Division of Biomedical Engineering, National Defense Medical

College, Research Institute, Namiki 3-2, Tokorozawa,

Saitama 359-8513, Japan

123

Med Biol Eng Comput (2008) 46:709–714

DOI 10.1007/s11517-007-0298-3

monitoring using electrodes places a heavy burden on

monitored individuals.

To determine human stress while driving or operating

equipment, we monitored human autonomic activation

induced by stressful sound using non-contact measurement

of HRV with a 24-GHz compact microwave radar, which

can easily be attached to the rear surface of back of a chair

without using either radioisotope or electrodes.

2 Non-contact autonomic activation measurement

system

We designed a non-contact autonomic activation mea-

surement system for non-contact measurement of HRV.

The system consists of a prototype compact microwave

radar (TAU GIKEN Co., Yokohama, Japan), control unit

for the microwave radar, an A/D converter ADA 16-32/

2(B)F (CONTEC Co., Tokyo, Japan) and a personal

computer. The compact microwave radar with an output

power of 10 mW incorporates an oscillator unit and

microwave antenna unit (a quadrupole plane antenna,

1.1 9 1.2 cm) in a small rectangular box (8 9 5 9 3 cm)

and generates a stable microwave signal at 24 GHz with an

output power of 10 mW. The antenna gain is 10dBi and the

diffusion angle is approximately 40�. Microwave (24 GHz)

was adopted to achieve high spatial resolution in order to

monitor the small body surface movements induced by

heartbeats. In our previous studies, 1.215-GHz radar was

used in cardiac and respiratory monitoring; we adopted the

24-GHz radar in order to achieve higher space resolution

required to HRV determination. Microwave radar output

was transferred to the personal computer through the

control unit by the A/D converter (Fig. 1). Without con-

ducting ECG recording by use of conventional electrodes,

the non-contact autonomic activation measurement system

is designed in order to assess the rhythmic components of

HRV, the heartbeat intervals were derived from the peak

intervals of the compact-size microwave radar output sig-

nal using a general-purpose analysis software Bimutus II

(KISSEI COMTEC Co., Nagano, Japan). Power spectra of

the time series of heartbeat intervals were calculated using

a maximum entropy method (MEM) with MemCalc soft-

ware (GMS Co., Tokyo, Japan). MEM offers a higher

spectrum resolution and shorter sampling duration than

those of fast Fourier transform (FFT). The powers of low-

frequency components of HRV (LF 0.04–0.15 Hz) have

been shown to estimate mainly sympathetic activities and

the powers of high-frequency components of HRV (HF

0.15–0.4 Hz) reflect parasympathetic activities. LF/HF can

thus be used as a parameter indicating sympatho-vagal

balance [8].

3 Testing of the non-contact autonomic activation

measurement system under audio stimuli

The system was tested using eight healthy male volunteers

with a mean age of 23 ± 1 years (range 22–25 years).

Subjects sat on a chair with headphones on and the com-

pact microwave radar was attached to the rear surface of

the back of the chair with a 30-mm spacer and about

60 mm left of the spine at around the level of the fourth

intercostal space, where cardiac induced skin surface

motion is larger than that of V5 position of precordial ECG

in our pilot study using laser distance meter (unpublished).

Subjects wore a 1-mm thick cotton T-shirt and the mesh

chair back is made of 2 mm thick polyester plastic. The

length from compact microwave radar to chair back was

30 mm (Fig. 2), in pilot study, the compact microwave

Controller

Bio. Amp.

Contact Measurement System

compact microwave radar

Electrode

Non-contactautomatic activationmeasurement system

PC

A/DConverter

Headphone

SubjectRR

IntervalsMEM LF, HF,

LF/HF

Non-Stimuli Stressful

AudioStimuli

compact microwave radar

e

t

Fig. 1 Schematic diagram of

apparatus for non-contact

monitoring of autonomic

activation

710 Med Biol Eng Comput (2008) 46:709–714

123

radar functioned well up to 50 mm. We did not give sub-

jects any instructions on breathing, such as, holding the

breath. We asked subjects to sit still leaning their head back

against the back of a chair.

Following a silent period of 120 s, audio stimulus

comprising a composite tone of 2,120 and 2,130 Hz sine-

waves at 95 dB was conducted using the headphone for

120 s. The tone range used in clinical nystagmus test

induced by audio stimuli (from 250 to 3,000 Hz at 95 dB)

was adopted for safety precautions [7]. Around 2,100 Hz is

within the most sensitive frequency domain of human

audible field. The beat tone of complex sounds causes

discomfort and alpha brain wave reduction to humans [1].

Power spectra of heartbeat intervals, as LF (0.04–

0.15 Hz), HF (0.15–0.4 Hz) and LF/HF, were calculated

using MEM with the MemCalc software. As a reference,

pectoral ECG was monitored along the non-contact moni-

toring, the output signals from both of the contact and non-

contact system were sampled through A/D converter with

the same sampling rate of 100 Hz. In real time, the

microwave radar output signal was displayed on the liquid

crystal display of the controller of a microwave radar, it

was also shown on the graphic terminal of a personal

computer. The power spectra of HRV (i.e., LF, HF and

LF/HF) for RR intervals derived by ECG were also cal-

culated using MEM. Cross-correlations of non-contact-

derived LF, HF and LF/HF with contact-derived LF, HF

and LF/HF were examined using statistic add-in software

for Excel (SSRI Co., Tokyo, Japan).

Quantitative data are expressed as mean ± SD. Statis-

tical analysis was performed using statistic add-in software

for Excel. Sample size was determined to achieve sufficient

assurance for paired t test for relatively uniform subjects.

All study protocols were reviewed and approved (seven

votes in favor versus none against) by the institutional

committee on human studies (Faculty of System Design,

Tokyo Metropolitan University, Tokyo, Japan). Informed

consent was obtained from all subjects.

4 Results

When subjects sat on the chair, the compact microwave

radar output through the control unit showed a cyclic

oscillation, corresponding to ECG (Fig. 3). In both non-

contact (compact microwave radar) and contact (ECG as

reference) measurements, the HRV parameter, LF of a

subject, reflecting mainly sympathetic activation, showed a

peak during cessation of audio stimuli (Fig. 4a). Mean LF

of eight subjects measured by non-contact and contact

methods during audio stimuli increased by 62 and 65%,

Body

Front Back

Clothes(Cotton)

D=30mm

Transmission

Compactmicrowaveradar

Backrest of the chair(Nylon mesh)

SpatiumIntercostale IV

Body

Front Back r

Fig. 2 Setting position of a compact 24-GHz microwave radar to

measure cardiac activity from the dorsal of subjects

-0.05

0

0.05

0.1

0.15

0.2

0 1 2 3 4 5 6

0.001.00

2.003.00

4.005.00

6.00

0 1 2 3 4 5 6

Time (sec)

Time (sec)

Rad

ar: N

on-C

onta

ct (

V)

EC

G: C

onta

ct (

V)

(a)

(b)

Fig. 3 A compact microwave

radar output (lower) showing a

cyclic oscillation that

corresponds to the cardiac

oscillation measured by ECG

(upper)

Med Biol Eng Comput (2008) 46:709–714 711

123

respectively, compared with those of the silent period

before audio stimuli [Fig. 4b; non-contact measurement

995 ± 775 m s2 (silent period), 1,617 ± 1,036 m s2 (dur-

ing audio stimuli), p \ 0.0001; contact measurement

781 ± 904 m s2 (silent period), 1,287 ± 1,359 m s2 (dur-

ing audio stimuli), p \ 0.0001]. Cross-correlation of LF

between non-contact and contact measurements in a same

subject showed a maximum value of 0.89 (Fig. 4c) and

maximum cross-correlation values in LF averaged

0.73 ± 0.14 in the eight subjects.

HF of a same subject, reflecting parasympathetic acti-

vity did not show any distinctive change during audio

stimuli (Fig. 5a) and mean HF of eight subjects increased a

very little during audio stimuli in both non-contact and

contact measurements [Fig. 5b; non-contact measurement

1,249 ± 626 m s2 (silent period), 1,424 ± 474 m s2 (dur-

ing audio stimuli), p \ 0.0001; contact measurement:

854 ± 595 m s2 (silent period), 945 ± 801 m s2 (during

audio stimuli), p \ 0.01]. Cross-correlation of HF between

non-contact and contact measurements in a same subject

showed a maximum value of 0.39 (Fig. 5c) and maximum

cross-correlation values of HF averaged 0.64 ± 0.15 in

eight subjects.

LF/HF of a same subject, reflecting sympatho-vagal

balance, exhibited a peak during audio stimuli (Fig. 6a).

Mean LF/HF of eight subjects measured by non-contact

and contact methods during audio stimuli increased by 34

and 37%, respectively, as compared with the silent period

before audio stimuli [Fig. 6b; non-contact measurement

0.86 ± 0.50 (silent period), 1.16 ± 0.59 (during audio

stimuli), p \ 0.0001; contact measurement 1.23 ± 1.18

(silent period), 1.68 ± 1.72 (during audio stimuli),

p \ 0.0001]. Cross-correlation of LF/HF between contact

and non-contact measurements of a same subject showed a

maximum value of 0.81 (Fig. 6c) and maximum cross-

correlation values of LF/HF averaged 0.73 ± 0.10 in the

eight subjects. Without using radioisotope or electrodes,

stress-induced autonomic activation was monitored.

P<0.0001P<0.0001

0ECG Radar

Audio Stimuli

-1

-0.5

0

0.

1

-150 -100 -50 0 150

Non Stimuli Stressful Audio Stimuli

00 60 120 180 240 300 360

1000

2000

3000

4000

5000

6000

LF

(mse

c2 )

LF

(mse

c2 )

500

1000

1500

2000

2500

3000

Contact

Non-Contact

-1

-5 50 100

Time lag (sec)

Time (sec)

Cro

ss-C

orre

lati

on5

(a) (b)

(c)

Fig. 4 a In both non-contact

and contact (ECG)

measurement, LF of a subject

(reflecting sympathetic

activation) shows a peak during

audio stimuli. b Mean LF of

eight subjects measured by non-

contact and contact methods

during audio stimuli increased

62 and 65%, respectively,

compared with the silent period

before audio stimuli. c Cross-

correlation of LF between non-

contact and contact

measurements of the same

subject

-1

-0.5

0

0.5

1

-150 -100 -50 0 50 100 150- -5

0

1000

2000

3000

4000

5000

6000

HF

(mse

c2 )

0 60 120 180 240 300 360

Time (sec)

Audio Stimuli Contact

Non-Contact

0

HF

(mse

c2 )

500

1000

1500

2000

2500

3000

ECG Radar

Non Stimuli Stressful Audio Stimuli

P<0.001P<0.01

Time lag (sec)

Cro

ss-C

orre

lati

on

(a) (b)

(c)

Fig. 5 a In both non-contact

and contact (ECG)

measurement, HF of the same

subject (reflecting

parasympathetic activity)

changes a very little with the

start of audio stimuli. b Mean

HF of eight subjects increases a

very little during audio stimuli

in both non-contact and contact

measurement, compared with

the silent period before audio

stimuli. c Cross-correlation of

HF between non-contact and

contact measurements of the

same subject

712 Med Biol Eng Comput (2008) 46:709–714

123

5 Discussion

We monitored human autonomic activation induced by

stressful sounds through non-contact measurement of the

heart rate variability using a 24-GHz compact microwave

radar. The radar can be easily attached to the rear surface of

back of a chair. The mean LF, which mainly reflects sym-

pathetic activity during audio stimuli is significantly higher

than the mean LF of the silent period before audio stimuli.

The HF indicating parasympathetic activity changed a very

little during the audio stimuli. This can be attributed to the

activation of the sympathetic nerve system induced by

stressful sounds. Without using radio isotope or electrodes,

our system monitors the sympathetic activation through the

back of a chair. Mental stress affects the autonomic nervous

system [12, 14]. Moreover, there is a relationship between

mental stress and traffic accidents [13]. By issuing auto-

nomic activation early warnings, it may be possible to

prevent traffic accidents and industrial accident.

The long-term monitoring of LF/HF in HRV can be used

as a diagnostic test for sepsis [3]. Moreover, it has been

reported that a reduction of HRV is useful in identifying

septic patients at a risk of the development of multi organ

dysfunction syndrome (MODS) [9]. Our method of non-

contact monitoring for HRV can thus be used not only for

monitoring autonomic activation, but also as a future

diagnostic method for sepsis or as a predictor of MODS,

without touching the patient. Zheng et al. [16] has proposed

a wearable health care system for long-term continuous

vital sign monitoring of high-risk cardiovascular patient.

Our system does not require a special wear and may be

suitable for screening of high-risk patients in health check.

The electromagnetic wave damage has been discussed,

especially in the case of human application. The power

density spectrum (PDS) of our system is in the allowable

range of 0.50 mW/cm2. In frequency range over 3 GHz,

the PDS limit is 1 mW/cm2 according to the guideline for

radio waves by Telecommunication Bureau of the Ministry

of Internal Affairs and Communication in Japan.

Our method appears to be promising for future non-

contact autonomic activation monitoring of people who are

operating equipment or elderly personals with cardiac

disorders, sympathetic activation sometimes triggers fatal

cardiac events [11]. The method allows autonomic acti-

vation monitoring, without placing any burden on the

monitored individuals.

References

1. Horii A, Yamamura C, Katsumata T, Uchiyama A (2004)

Physiological response to unpleasant sounds. J Int Soc Life Info

Sci 22(2):536–541

2. Kikuchi M, Ishihara M, Matsui T et al (2004) Biomedical engi-

neering’s contribution to defending the homeland. IEEE Eng Med

Biol Mag 23:175–186

3. Korach M, Sharshar T, Jarrin I et al (2001) Cardiac variability in

critically ill adults: influence of sepsis. Crit Care Med 29:1380–

1385

4. Matsui T, Hagisawa K, Ishizuka T et al (2004) A novel method to

prevent secondary exposure of medical and rescue personnel to

toxic materials under biochemical hazard conditions using

microwave radar and infrared thermography. IEEE Trans Biomed

Eng 51:2184–2188

5. Matsui T, Ishizuka T, Takase B et al (2004) Non-contact deter-

mination of vital sign alterations in hypovolemic states induced

by massive hemorrhage: an experimental attempt to monitor the

condition of injured persons behind barriers or under disaster

rubble. Med Biol Eng Comput 42:807–811

6. Min SD, Yoon DJ, Yoon SW, Yun YH, Lee M (2007) A study on

a non-contacting respiration signal monitoring system using

Doppler ultrasound. Med Biol Eng Comput (in press)

7. Ostrowski VB, Byskosh A, Hain TC (2002) Tullio phenomenon

with dehiscence of the superior semicircular canal. Otol Neurotol

22(1):61–65

00.5

11.5

22.5

33.5

4

0

1

2

3

4

5

6

7

0 60 120 180 240 300 360

LF/

HF

LF/

HF

P<0.0001P<0.0001Audio Stimuli

-1

-0.5

0

0.5

1

-150 -100 -50 0 50 100 150- - 50 15

Time (sec)

ECG Radar

Time lag (sec)

Cro

ss-C

orre

latio

n

ContactNon-Contact

Non Stimuli Stressful Audio Stimuli

(a) (b)

(c)

Fig. 6 a In both non-contact

and contact (ECG)

measurement, LF/HF of the

same subject (reflecting

sympathetic activation) exhibits

a peak during audio stimuli.

b Mean LF/HF of eight subjects

as measured by non-contact and

contact methods during audio

stimuli increase 34 and 37%,

respectively, compared with the

silent period before audio

stimuli. c Cross-correlation of

LF/HF between non-contact and

contact measurements of the

same subject

Med Biol Eng Comput (2008) 46:709–714 713

123

8. Perini R, Veicsteinas A (2003) Heart rate variability and auto-

nomic activity at rest and during exercise in various physiological

conditions. Eur J Appl Physiol 90(3–4):317–325

9. Pontet J, Contreras P, Curbelo A et al (2003) Heart rate vari-

ability as early marker of multiple organ dysfunction syndrome in

septic patients. J Crit Care 18:156–163

10. Richter S, Schaefer A, Menger MD et al (2005) Mapping of the

cardiac sympathetic nervous system by single photon emission

tomography with technetium-99 m-labeled fluorobenzylpiperi-

dine derivative(99mTc-FBPBAT): result of a feasibility study in a

porcine model and an initial dosimetric estimation in humans.

Nucl Med Commun 26(4):361–368

11. Rozanski A, Blumenthal JA, Kaplan J (1999) Impact of psy-

chological factors on the pathogenesis of cardiovascular disease

and implications for therapy. Circulation 99(16):2192–2217

12. Ruediger H, Seibt R, Scheuch K et al (2004) Sympathetic and

parasympathetic activation in heart rate variability in male

hypertensive patients under mental stress. J Hum Hypertens

18(5):307–315

13. Taylor AH, Dorn L (2006) Stress, fatigue, health, and risk of road

traffic accidents among professional drivers: the contribution of

physical inactivity. Annu Rev Public Health 27:371–391

14. Tuininga YS, Crijns HJ, Brouwer J et al (1995) Evaluation of

importance of central effects of atenolol and metoprolol mea-

sured by heart rate variability during mental performance tasks,

physical exercise, and daily life in stable postinfarct patients.

Circulation 15 92(12):3415–3423

15. Uenoyama M, Matsui T, Yamada K, Suzuki S, Takase B, Suzuki

S, Ishihara M, Kawakami M (2006) Non-contact respiratory

monitoring system using a ceiling-attached microwave antenna.

Med Biol Eng Comput 44(9):835–840

16. Zheng JW, Zhang ZB, Wu TH, Zhang Y (2007) A wearable

mobihealth care system supporting real-time diagnosis and alarm.

Med Biol Eng Comput 45(9):877–885

714 Med Biol Eng Comput (2008) 46:709–714

123