Electrical impedance plethysmography technique in estimatingpulmonary function status

R. AGARWAL{*, R. YADAV{, S. ANAND{, J. C. SURIx and J. GIRIJA{

{National Institute of Health & Family Welfare, Munirka, New Delhi 110 067, India{Center for Biomedical Engineering, Indian Institute of Technology, Hauz Khas,

New Delhi 110 016, IndiaxRespiratory Unit, Safdarjung Hospital, New Delhi 110 029, India

{Center for Atmospheric Science, Block VI, Indian Institute of Technology, Hauz Khas,New Delhi 110 016, India

An attempt has been made to study the appropriateness of electrical impedance

plethysmography (EIP) in detecting pulmonary health status. A feasibility study was

conducted on normal, obstructive and restrictive subjects aged between 20 and 60 years.

Quantitative assessment of pulmonary function was made by way of both EIP and

pulmonary function tests (PFT). Amongst the various EIP parameters, a statistically

significant difference was observed for the respiratory band power, between normal

(129.7) and obstructive (35.8) subjects, indicating that this EIP variable can be used to

distinguish between the two pulmonary function states. A significant positive correlation

was observed between a spirometry parameter, peak expiratory flow rate (PEFR), and an

EIP parameter, respiratory amplitude (r¼ 0.372, p 5 0.05), thereby indicating that EIP

information by way of respiratory amplitude is comparable to that provided by PEFR.

Hence, both respiratory amplitude and respiratory band power were seen to provide

useful information on pulmonary health status.

1. Introduction

Changes of body electrical impedance during the cardiac

cycle have been reported since the beginning of the

twentieth century, and there have been excellent reviews

on detection of physiological events by the impedance

technique [1]. Nyboer [2] carried out pioneer work in this

field and advocated the use of a tetra-polar electrode system

to overcome the problems associated with contact im-

pedance. He passed a high frequency current across the

human thorax and recorded the associated impedance

changes with each heart beat. He called the recording a

‘radiocardiogram’. In the early 1940s, Nyboer and associ-

ates investigated resistive bioimpedance changes of such

physiologic phenomena as peripheral circulation, body

water content, and respiratory and cardiac activity. In

1964, Nyboer first obtained pulmonary function data from

right/left lungs with strip electrodes by EIP [3]. Weng et al.

in 1979 demonstrated the feasibility of monitoring regional

lung function in children [4]. He concluded that tetra-polar

electrical impedance plethysmography (EIP), being non-

invasive, is technically the most suitable method for use in

children and can provide safe, reliable regional information

about ventilation/perfusion of normal and diseased lungs.

Kira et al. [5] studied the electrical impedance variations

associated with respiration as voltage changes between

electrodes. They found that transthoracic impedance in an

open chest condition was greater in the expiratory state than

in the inspiratory state. Researchers measured tidal volume

(VT) using transthoracic impedance variations in animals

and found a linear relationship between change in im-

pedance and VT [6,7]. Guha and Anand [8] investigated the

EIP technique in comparison with electrical field plethys-

mography for evaluating cardiac and respiratory EIP pulses.

Studies have also been carried out to quantify and

compare the instantaneous heart rate dynamics and

cardio-pulmonary interactions while performing medi-

tation with different breathing patterns. Observations

recorded by way of ECG and two respiration signals

(abdominal and chest elastic transducer bands) recorded

*Corresponding author. E-mail: rachna_india@yahoo.com

Journal of Medical Engineering & Technology, Vol. 31, No. 1, January/February 2007, 1 – 9

Journal of Medical Engineering & TechnologyISSN 0309-1902 print/ISSN 1464-522X online ª 2007 Informa UK Ltd.

http://www.tandf.co.uk/journalsDOI: 10.1080/03091900500199919

at 200Hz showed high amplitude, low frequency (0.05 –

0.1Hz) oscillations due to respiratory sinus arrhythmia

(RSA) and an increased coherence (p5 0.05) between heart

rate and breathing during both relaxation and segmented

breathing maneuvers [9].

Hyndman et al. [10] first used power spectral analysis to

assess heart rate variability. By 1971 it was clearly established

that when the respiratory rate is above 0.15Hz, the peak at

the respiratory frequency is strictly related to respiration and

is mediated entirely by modulation of vagal efferent activity

[11 – 14]. The amplitude or area of the respiratory peak in the

heart rate power spectrum is said to depend not only on

mean vagal activity, but on both the frequency and tidal

volume of respiration [15]. Recently, Berger et al. [16]

devised a method for RSA over a broad range of physio-

logical respiratory frequencies from 0 to 0.5Hz.

Computer simulations and signal processing techniques

have been used to predict heart rate variability with the

help of respiratory signal, RSA [17,18]. In addition,

researchers have found a significant positive correlation

between the amplitude of the sympathetic skin response

(SSR) evoked by forced maximal expiratory stimulation

and forced expiratory volume in first second of maximum

expiration (FEV1) (r¼ 0.94, p5 0.01) [19]. They found that

forced expiration was more useful as a respiratory stimulus

than forced inspiration. A non-invasive method by way of

surface EGG (electrogastrogram) has been successfully

employed to identify gastric contractions using the back-

propagation neural network [20].

The present study attempts to estimate the appropriate-

ness/utility of EIP in detecting pulmonary health status.

A feasibility study was conducted on normal, obstructive

and restrictive patients to assess pulmonary function

status. The study involved quantitative assessment of the

pulmonary function in human subjects by way of both

EIP and spirometry technique. While the spirometry

technique is well established, its only drawback is that it is

effort dependent and often requires elaborate explanation to

the subject. As a result, critically ill patients find it difficult to

perform the test. In view of the above constraints in using

conventional spirometry, an alternative technique, electrical

impedance plethysmography (EIP) was investigated. The

following sections describe the selection of the subjects, data

acquisition, signal processing, analysis of digitized signals

and statistical analysis of both digital signals and pulmonary

function test (PFT) results. The outcome of the comparative

study follows the analysis.

2. Methodology

2.1. Subject selection

A total of 31 subjects—11 normal and 20 patients (nine

obstructive and 11 restrictive)—were studied at Pulmonary

Function Test Laboratory, Respiratory Unit, Safdarjung

Hospital,NewDelhi.Volunteerswere in theagegroupof20 –

60 years. Subjects suffering from any cardiac disorder were

not selected.Consentwas taken from the studied subjects and

a brief questionnaire was filled on their behalf. Subjects’

height and weight were recorded prior to the experiment.

Subjects were instructed to sit upright for two minutes

without coughing or any movement. They were instructed to

breathe normally and not to hold their breath at any time.

2.2. EIP data acquisition

The EIP data were obtained from the healthy normal,

obstructive and restrictive subjects. The study was designed

to record the EIP signals for duration of 120 seconds from

EIP (Bionics India Ltd, Delhi). Subjects were asked to

breathe normally while a signal was acquired by placing

tetra polar, Ag-AgCl band electrodes. Figure 1 shows the

placement of electrodes and signal acquisition. The

analogue signal from EIP was digitized for analysis using

an ICPDAS PCI 1002L, multifunction data acquisition

card. Simultaneous PFT data were taken from the selected

subjects to get the lung function status of the subjects.

2.3. EIP data processing

The base impedance Z was noted for each subject but was

subsequently dropped from further analysis as it is known to

be dependent on the physique (height and diameter of the

thorax) of the subject and thus varies from person to person.

The digitized signal was stored by an interface made in

LABVIEW 7.0 (National Instruments, USA) in ASCII

format. Further processing was carried out usingMATLAB

6.5. The signal was acquired at a sampling rate of 256Hz.

Appropriate filters were applied for separating cardiac

components from respiratory and other interferences were

also excluded. The spectral estimates were calculated using

Figure 1. Placement of electrodes.

2 R. Agarwal et al.

fast Fourier transform (FFT) and Welch power spectral

density (PSD). The results were plotted with raw signal, filter

signal and PSD. Dominant frequency shift was observed in

patients and was in tune with their clinical profile.

2.4. EIP data analysis

The signals were analysed using MATLAB software and

their spectral estimates were plotted for all three groups of

Figure 2. Respiratory and heart rate time series and corresponding spectra for (a) normal, (b) obstructive and (c) restrictive

subjects.

EIP for estimating pulmonary function status 3

subjects. Figure 2 gives plots for original signal, respiratory

and cardiac components as well as FFT plots for the three

subject categories—normal, obstructive and restrictive.

Short term changes in thoracic impedance reflect the filling

and emptying of the lungs. These physical influences of

respiration result in amplitude variations in the observed

respiratory signal.

Table 1 and figure 3 give a summary of the respiratory

and cardiac rate, average amplitudes for one minute signal

(i.e. 60 – 120 seconds) as well as the EIP results, peak

frequency, band power and respiratory/cardiac ratio.

Figure 4 gives FFT plots of each of the three cases—

normal, obstructive and restrictive. The graph shows shifts

in peak frequencies for both respiratory (0.2 – 0.75Hz) and

cardiac (1.5 – 2.5Hz) components from normal to obstruc-

tive to restrictive subjects.

2.5. PFT data collection

Pulmonary function tests (PFT) were conducted on subjects

after measuring their height and weight at Safdarjung

Hospital. Morgan’s spirometer was used. The instrument

was calibrated periodically. The test procedures were

explained to the subjects after making them rest for at least

two minutes. A mouth piece and nose clips were applied.

Spirometry was performed on all 31 subjects in sitting

position. Three basic tests, vital capacity (VC), forced vital

capacity (FVC) and maximum ventilatory volume (MVV),

were performed on each subject according to American

Thoracic Society (ATS) protocol. Each subject was asked to

repeat the test several times until able to perform to the best

of individual capacity. The three most acceptable efforts

were selected for comparison. The best of the three efforts

was stored for every subject according to ATS selection

criteria. Precautions were taken to prevent cross-contam-

ination in subjects by sterilizing the mouth pieces.

2.6. Statistical analysis

Data for the two independent tests, i.e. PFT and EIP, were

statistically analysed for all 31 subjects (11 normal, nine

obstructive and 11 restrictive). While the PFT parameters

provided information about lung capacities (both volume

and flow), the EIP parameters gave information related

to respiratory rate as well as strength of the respiratory

signal.

Table 1. EIP results: respiratory and cardiac rates, amplitude, peak frequency, band power and respiratory/cardiac ratio.

Subject category n

Respiratory (average) Cardiac (average)

Respiratory/

cardiac ratioRate Amplitude*

Peak

frequency

(Hz) Band power Rate Amplitude*

Peak

frequency

(Hz) Band power

Normal 11 19.91 0.0919 0.32 129.75 82.5 0.0645 1.52 26.04 63.84

Obstructive 9 21.33 0.0611 0.35 35.8 101.9 0.0568 1.7 45.64 0.94

Restrictive 11 25.27 0.0848 0.40 95.74 98.6 0.0631 1.56 24.67 7.06

* Average amplitude was calculated from 1-minute plots (60 – 120 seconds).

Figure 3. Graphs for average respiratory amplitude, band

power, peak frequency and respiratory/cardiac ratio.

4 R. Agarwal et al.

The six PFT parameters were:

. FVC;

. forced expiratory volume in first second of maximum

expiration (FEV1);

. average forced expiratory flow at 25 – 75% of FVC

(FEF25-75);

. peak expiratory flow rate (PEFR);

. forced inspiratory vital capacity (FIVC); and

. peak inspiratory flow rate (PIFR).

Comparison was made with regard to these PFT para-

meters and the six EIP parameters, i.e. respiratory and

cardiac amplitudes, band powers and peak frequencies,

amongst the three categories (normal, obstructive and

restrictive) using one-way analysis of variance (ANOVA).

Post hoc Tukey’s test was performed to evaluate difference

between pairs of variables for the three groups. The test was

performed at 5% level of significance. Statistical analysis

was performed using Statistical Package for Social Science

(SPSS) for Windows (version 10.1). Table 2 gives the results

of ANOVA and post hoc test.

Pearson’s coefficient of correlation was also calculated to

determine association amongst PFT and EIP variables. The

coefficient of correlation, r, along with its significance level,

is presented in table 3.

The above analysis showed significant correlation

between PEFR and respiratory amplitude, besides others.

Therefore, bi-variate linear regression analysis (direct

method) was performed for PEFR with respiratory ampli-

tude as the independent variable. The following regression

equation was obtained at an r2 value of 0.138: PEFR¼40þ 181 (respiratory amplitude).

3. Results and discussion

A comparison of the PFT results (table 2) showed a gradual

fall in the means of FVC, FIVC and PIFR from normal–

obstructive–restrictive subjects, while for FEV1, FEF25-75

and PEFR mean values, a gradual fall was observed from

normal–restrictive–obstructive subjects. The result is in

agreement with the disease physiology [21] which docu-

ments that in restrictive condition lung volumes are

affected, while in obstructive condition airflows are

affected. The lowest mean FEF25-75 %P value in obstruc-

tive subjects indicates that small airway impairment is more

pronounced in obstructive than in restrictive subjects. This

result is clinically accurate, as by far the greatest number of

patients with obstructive respiratory difficulties have lower

airway obstruction [21].

EIP results (i.e. respiratory amplitude, band power and

respiratory/cardiac band power ratio; figure 3) also showed

a gradual decrease in the mean values from normal–

restrictive–obstructive subjects. A similar trend in results

was observed for PFT results representing flow rates (i.e.

FEV1, FEF25-75 and PEFR). This shows that application of

Figure 4. Combined FFT plot of normal, obstructive and restrictive subjects showing peak shift.

EIP for estimating pulmonary function status 5

Table

2.Statisticsresults(m

ean,standard

deviation,ANOVA,Tukey’stest

ofhomogeneity)fornorm

al,obstructiveandrestrictivesubjects.

(a)PFTresults(FVC,FEV1,FEF25-75,PEFR,FIV

CandPIF

R)

Category

n

FVC

FEV1

FEF25-75

PEFR

FIV

CPIF

R

Best

%P

Best

%P

Best

%P

Best

%P

Best

%P

Best

%P

Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.

I.Norm

al

11

3.4

0.8

91.8

9.5

2.8

0.6

90.3

8.6

3.13

0.6

80

15.3

5.6

1.25

71.7

12

3.2

0.77

88

9.2

4.31

1.53

71

23

II.

Obstructive

92.4

0.7

65.6

16

1.5

0.59

49.3

18

1.15

0.6

28.7

14.7

3.4

1.36

44.5

17

2.1

0.54

59

15

2.86

0.99

48

15

III.

Restrictive

11

1.8

0.45

53.4

12

1.5

0.45

53.5

13.5

1.66

0.98

44.7

23

3.37

1.61

46.3

20

1.61

0.45

47

11.8

2.25

0.81

39

15

ANOVA

0.000

0.000

0.000

0.001

0.000

0.001

Tukey’stest

(5%

significance

level)

ExceptforII

&IIIthe

resultsare

significantly

different

ExceptforII

&IIIthe

resultsare

significantly

different

ExceptforII

&IIIthe

resultsare

significantly

different

ExceptforII

&IIIthe

resultsare

significantly

different

ExceptforII

&IIIthe

resultsare

significantly

different

ExceptforII

&IIIthe

resultsare

significantly

different

(b)EIP

results(respiratory

andcardiacamplitude,

bandpower

andpeakfrequency)

Category

n

Age

Respiratory

amplitude

Respiratory

bandpower

Respiratory

peak

frequency

Cardiacampli-

tude

Cardiacband

power

Cardiacpeak

frequency

Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.Mean

S.D

.

I.Norm

al

11

40

11.7

0.09

0.05

129

119

0.342

0.1

0.06

0.02

26

36.3

1.516

0.31

II.

Obstructive

937

12.1

0.06

0.02

35.8

28.9

0.351

0.07

0.05

0.02

45.6

26

1.69

0.295

III.

Restrictive

11

47

15.8

0.08

0.037

95.7

57

0.399

0.15

0.063

0.02

24.67

16

1.563

0.22

ANOVA

0.259

0.047

0.352

0.729

0.195

0.343

Tukey’stest

(5%

significance

level)

Noneare

different

Only

I&

IIare

different

Noneare

different

Noneare

different

Noneare

different

Noneare

different

6 R. Agarwal et al.

Table3.Correlationmatrix

onPFT(FVC,FEV1,FEF25-75,PEFR,FIV

CandPIF

R%Pvalues)andEIP

(respiratory

andcardiac–amplitude,bandpower,peakfrequency).

Only

significantcorrelationsare

shown.

FEV1

FEF25-75

PEFR

FIV

CPIF

R

Respiratory

amplitude

Respiratory

bandpower

Respiratory

peak

frequency

Cardiac

amplitude

Cardiac

bandpower

Cardiac

peak

frequency

FVC

0.899{

0.679{

0.653{

0.943{

0.706{

–0.334x

FEV1

0.883{

0.744{

0.887{

0.595{

FEF25-75

0.697{

0.672{

0.400{

0.295x

0.304x

PEFR

0.693{

0.440{

0.372{

0.275{

–0.291{

FIV

C0.766{

–0.273{

Respiratory

amplitude

0.672{

Respiratory

bandpower

–0.303x

Respiratory

peakfrequency

–0.292{

–0.297x

Cardiacamplitude

0.381{

–0.455{

n¼31.

{p5

0.01.

{p5

0.05.

xp4

0.10.

{p4

0.15.

EIP for estimating pulmonary function status 7

EIP in determining pulmonary function gives information

that corresponds more appropriately to the respiratory

airflows rather than to lung volumes.

ANOVA results showed significant (p4 0.001) differ-

ence for all PFT parameters between the three categories,

while for EIP variables, only respiratory band power

showed significant difference (p5 0.05), and in addition

that was between normal and obstructive categories only

(table 2). A right shift in the average respiratory peak

frequency was observed from normal (0.32Hz) to ob-

structive (0.35Hz) to restrictive (0.40Hz) subjects. A right

shift in the average respiratory peak frequency for

obstructive (0.35Hz) and restrictive (0.40Hz) subjects

(figure 4) indicates increased respiratory rate for obstruc-

tive (21 breaths min71) and restrictive (24 breaths min71)

subjects from the normal (19 breaths min71). While in

obstructive subjects there is a slight right shift, in

restrictive subjects there is a greater right shift. The result

is in tune with the physiology of the restrictive and obs-

tructive cases. In restrictive cases there is a decrease in

lung volume which leads to an increase in respiratory rate

to meet the physiological requirements. With an increased

respiratory rate a decrease in respiratory amplitude is also

seen in restrictive subjects. In obstructive cases there is

narrowing of airways, thereby affecting the airflows. This

also leads to decreased respiratory amplitude but an

increase in respiratory rates to meet the body’s oxygen

requirement.

Thus, in both cases (restrictive and obstructive) there was

decrease in the respiratory amplitude and an increase in the

respiratory rate. A right shift in the respiratory frequency

for restrictive and obstructive subjects clearly shows an

increased respiratory rate. The obtained results were in tune

with the actual recorded mean respiratory rates (table 1),

i.e. normal¼ 19.9 breaths min71, obstructive¼ 21.3

breaths min71 and restrictive¼ 25.27 breaths min71. Re-

duced respiratory amplitude amongst restrictive (0.085) and

obstructive (0.06) subjects as compared to the normal (0.092)

was also in agreement with the physiology of the two

respiratory states.

Results showed positive significant (p5 0.01) correlation

between all six PFT parameters (table 3). Significant

positive correlation was also seen between respiratory

amplitude and band power (r¼ 0.67, p5 0.01) as well as

between cardiac amplitude and band power (r¼ 0.381,

p5 0.05). Significant negative correlation was observed

between cardiac amplitude and peak frequency (r¼ –0.455,

p5 0.01). Significant positive correlation was observed

between PEFR and respiratory amplitude (r¼ 0.372,

p5 0.05), thereby indicating that EIP information by way

of respiratory amplitude can be useful in estimating peak

expiratory flow rate. An estimate of the PEFR can therefore

be given from the regression equation: PEFR¼ 40þ181(respiratory amplitude).

4. Conclusion

Experimental findings revealed that amongst various EIP

variables, a considerable difference existed for the mean

values of respiratory band power amongst normal (129.75),

restrictive (95.74) and obstructive (35.8) subjects. A

statistically significant difference was however observed

between normal and obstructive categories only, thereby

indicating that respiratory band power can be used to

distinguish between these two pulmonary function states.

Moreover, the respiratory amplitude for obstructive and

restrictive subjects was 11% and 33% lower than the

normal, respectively. On the other hand, the respiratory

rate was seen to increase from normal (19.9) to obstructive

(21.3) to restrictive (25.2) subjects. A comparison between

the two techniques showed significant positive correlation

between PEFR and respiratory amplitude (r¼ 0.372,

p5 0.05), thereby indicating that EIP information by way

of respiratory amplitude can be useful in estimating peak

expiratory flow rate. An estimate of PEFR can therefore be

given from the regression equation: PEFR¼ 40þ 181 AR

(respiratory amplitude). Since the objective of the study

was quantitatively to determine the feasibility of using EIP

technique in determining pulmonary function status, it has

been established that both respiratory amplitude and

respiratory band power can provide useful information

on pulmonary health status. EIP information corresponds

more appropriately to the respiratory airflows rather than

the lung volumes.

Acknowledgements

The authors are thankful to the laboratory technicians at

Safdarjung Hospital for their kind cooperation while

conducting experiments.

References

[1] Geddes, L.A. and Baker, L.E., 1972, Thoracic impedance change

following saline injection into right and left ventricles. Journal of

Applied Physiology, 33, 278 – 286.

[2] Nyboer, J., Bango, S., Barnett, A. and Hasley, R.H., 1940, Radio-

cardiograms. Journal of Clinical Investigation, 19, 773 – 781.

[3] Nyboer, J., Murray, P. and Sedensky, J.A., 1974, Blood-flow indices in

amputee and control limbs by mutual electrical impedance plethysmo-

graphy. American Heart Journal, 87, 704 – 711.

[4] Weng, T.R., Spence, J.A., Polgar, G. and Nyboer, J., 1979,

Measurement of regional lung function by tetrapolar electrical

impedance plethysmography. Chest, 76, 64 – 69.

[5] Kira, S., Hukushima, Y., Kitamura, S. and Ito, A., 1971, Transthor-

acic electrical impedance variations associated with respiration.

Journal of Applied Physiology, 30, 820 – 826.

[6] Davidson, K.G., Bersten, A.D., Nicholas, T.E., Ravenscroft, P.R. and

Doyle, T.E., 1999, Measurement of tidal volume by using trans-

thoracic impedance variations in rats. Journal of Applied Physiology,

86, 759 – 766.

8 R. Agarwal et al.

[7] Sandberg, K.L., Lindstrom, D.P., Krueger, E.D., Sundell, H. and

Cotton, R.B., 1988, Measurement of tidal volume during high

frequency ventilation by impedance plethysmography. Pediatric

Research, 23, 253 – 256.

[8] Guha, S.K. and Anand, S., 1982, Comparison of electrical field

plethysmography with electrical impedance plethysmography. Annals

of Biomedical Engineering, 10, 231 – 239.

[9] Peng, C.K., Issac, C.H., Joseph, E.M., Jeffery, M.H., Gurucharan, K.,

Herbert, B. and Ary, L.G., 2004, Heart rate dynamics during three

forms of meditation. International Journal of Cardiology, 95, 19 – 27.

[10] Hyndman, B.W., Kitney, R.I. and Sayers, B.M., 1971, Spontaneous

rhythms in physiological control systems. Nature, 233, 339 – 341.

[11] Pomeranz, B., MaCaulay, R.J.B., Caudill, M.A., Kutz, I., Adam, D.

and Gordon, D., 1985, Assessment of autonomic function in humans

by heart rate spectral analysis. American Journal of Physiology, 248,

H151 –H161.

[12] Saul, J.P., Berger, R.D., Albrecht, P., Stein, S.P., Chen, M.H. and

Cohen, R.J., 1991, Transfer function analysis of the circulation:

unique insights into cardiovascular regulation. American Journal of

Physiology, 261, H1231 – 1245.

[13] Akselrod, S.D., Gordon, F.A. and Ubel, D.C., 1981, Power spectrum

analysis of heart rate fluctuations: A quantitative probe of beat-to-

beat cardiovascular control. Science, 213, 220 – 222.

[14] Berger, R.D., Saul, J.P. and Cohen R.J., 1989, Transfer function

analysis of autonomic regulation. I. The canine atrial rate response.

American Journal of Physiology, 25, H142 –H152.

[15] Saul, J.P. and Cohen, R.J., 1994, Respiratory Sinus Arrhythmia.

In Vagal control of the Heart: Experimental Basis and Clinical

Implications, M.N. Levy and P.J. Schwartz (Eds), pp. 511 – 536

(New York, USA: Futura Publishing).

[16] Berger, R.D., Saul, J.P. and Cohen, R.J., 1989, Assessment of

autonomic response by broad-band respiration. IEEE Transactions on

Biomedical Engineering, 36, 1061 – 1065.

[17] Clynes, M., 1960, Respiratory sinus arrhythmia: laws derived from

computer simulation. Journal of Applied Physiology, 15, 863 – 874.

[18] Balocchi, R., Menicucci, D., Santarcangelo, E., Sebastiani, L.,

Gemignani, A., Ghelarducci, B. and Varanini, M., 2004, Deriving

the respiratory sinus arrhythmia from the heartbeat time series using

empirical mode decomposition. Chaos, Solitons and Fractals, 20, 171 –

177.

[19] Kira, Y., Ogura, T., Aramaki, S., Kubo, T., Hayasida, T. and

Hirasawa, Y., 2001, Sympathetic skin response evoked by respiratory

stimulation as a measure of sympathetic function. Clinical Neurophy-

siology, 112, 861 – 865.

[20] Chen, J.D.Z., Lin, Z., Wu, Q. and McCallum, R.W., 1995, Non-

invasive identification of gastric contractions from surface electro-

gastrogram using back-propagation neural networks. Medical

Engineering & Physics, 17, 219 – 225.

[21] Miller, A., 1986, Pulmonary Function Tests in Clinical and Occupa-

tional Lung Disease (New York, USA: Grune & Stratton).

EIP for estimating pulmonary function status 9