Impact of Airﬂow Rate on Amplitude and Regional Distribution of

Normal Lung Sounds

Elmar Messner

, Martin Hagm¨uller

, Paul Swatek

, Freyja-Maria Smolle-J¨uttner

and Franz Pernkopf

Signal Processing and Speech Communication Laboratory, Graz University of Technology, Graz, Austria

Division of Thoracic and Hyperbaric Surgery, Medical University of Graz, Graz, Austria

Keywords:

Lung Sounds, Multichannel Recording, Respiratory Flow, Acoustic Thoracic Images.

Abstract:

In computerized lung sound research, the usage of a pneumotachograph, deﬁning the phase of respiration

and airﬂow velocity, is essential. To obviate its need, the inﬂuence of airﬂow rate on the characteristics of

lung sounds is of great interest. Therefore, we investigate its effect on amplitude and regional distribution of

normal lung sounds. We record lung sounds on the posterior chest of four lung-healthy male subjects in supine

position with a 16-channel lung sound recording device at different airﬂow rates. We use acoustic thoracic

images to discuss the inﬂuence of airﬂow rate on the regional distribution. At each airﬂow rate, we observe

louder lung sounds over the left hemithorax and a constant regional distribution above an airﬂow rate of 0.7 l/s.

Furthermore, we observe a linear relationship between the airﬂow rate and the amplitude of lung sounds.

1 INTRODUCTION

In auscultation, beside distinct ﬁndings like adven-

titious lung sounds, also the lung sound intensity is

used as a diagnostic marker. For example, physicians

examine the differences in intensity between left- and

right-sided lung sounds at pneumothorax condition.

Therefore, a basic knowledge about the regional dis-

tribution of normal lung sound intensity, but also its

dependence on airﬂow rate is essential. Moreover,

a good understanding of this dependence could ren-

der the pneumotachographdispensable for lung sound

research, because of airﬂow estimation directly from

lung sounds.

Several research groups already investigated the

effect of airﬂow rate on the amplitude and the re-

gional distribution of lung sound. Differing relation-

ships were observed in (Kraman, 1984), (Gavriely

and Cugell, 1996), (Hossain and Moussavi, 2002)

and (Shykoff et al., 1988). Recently, the authors

in (Yosef et al., 2009) showed the effect of airﬂow

rate on Vibration Response Imaging measurement in

healthy lungs during expiration, but also discussed the

relationship between lung sound energy and airﬂow

rate. The authors in (Torres-Jimenez et al., 2008a)

used a 5x5 microphone array and generated respi-

ratory acoustic thoracic images (RATHI) to discuss

the regional distribution of lung sounds, by com-

paring its performance with clinical physicians. In

(Torres-Jimenez et al., 2008b), the authors further

show RATHIs at different airﬂow rates.

In this paper, we independently investigate the im-

pact of airﬂow rate on amplitude and regional distri-

bution of normal lung sounds. For that, we recorded

lung sounds on the posterior chest of four lung-

healthy male subjects with a 16-channel lung sound

recording device (Messner et al., 2016) at airﬂow

rates of 0.3, 0.7, 1.0, 1.3 and 1.7 l/s during inspiration.

In contrast to other research groups, we recorded lung

sounds in supine position. Another differentiation

is the usage of uncontaminated lung sound record-

ings, i.e. free of heart and other interfering sounds.

By means of acoustic thoracic images (Charleston-

Villalobos et al., 2004), we discuss the regional distri-

bution of lung sounds dependent on airﬂow rate. To

generate the surface acoustic thoracic images from the

multiple lung sound signals, we use 2D-interpolation.

For each subject, we illustrate the acoustic thoracic

images at the ﬁve airﬂow rates independently. We ob-

serve a constant regional distribution above an airﬂow

rate of 0.7 l/s. Furthermore, we observe a linear rela-

tionship between the airﬂow rate and the amplitude of

lung sounds. Our results most closely correspond to

the ﬁndings in (Yosef et al., 2009) and are indepen-

dent of the recording position.

We organized the paper as follows. Section 2 de-

Messner E., HagmÃijller M., Swatek P., Smolle-JÃijttner F. and Pernkopf F.

Impact of Airﬂow Rate on Amplitude and Regional Distribution of Normal Lung Sounds.

DOI: 10.5220/0006134400490053

In Proceedings of the 10th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2017), pages 49-53

ISBN: 978-989-758-212-7

scribes the data acquisition, the subjects, the record-

ing material, the signal pre-processing and the gen-

eration of the acoustic thoracic images. Section 3

presents our observations for the regional distribu-

tion and the lung sound amplitude for different air-

ﬂow rates. In Section 4, we discuss the results and

Section 5 concludes the paper.

2 MATERIALS AND METHODS

2.1 Data Acquisition

We recorded the lung sounds with a multichannel

recording device, which enables the simultaneous

recording of airﬂow (Messner et al., 2016). The

device is equipped with a multichannel lung sound

recording front-end and a pneumotachograph.

The recording front-end is a foam pad covered

with artiﬁcial leather, a similar construction as the

Stethographics STG 16 (Murphy, 2007). On the sur-

face of the pad, we arranged 16 lung sound transduc-

ers (LSTs) with a ﬁxed pattern. The pattern is compa-

rable with the one proposed in (Sen and Kahya, 2006).

Based on the approach with air-coupled electret-

6cm

6cm 6cm

7cm

1 2

5 6

11 12

15 16

Figure 1: Multichannel recording front-end of the lung

sound recording device (Messner et al., 2016). 16 lung

sound transducers are distributed on the foam pad. The cen-

ter line represents the spine.

condenser microphones (Pasterkamp et al., 1993), we

modiﬁed Littmann Classic II chest pieces for the LST

design. By placing the foam pad under the back of the

subject, we perform the recording of the lung sounds

in supine position on an examination table.

We use standard audio recording equipment for

the analogue pre-ﬁltering, pre-ampliﬁcation, and dig-

itization of the LST signals. The sampling frequency

is f

= 16 kHz and the resolution is 24 bit. Before

the analog-to-digital conversion of the LST signals,

we apply a Bessel high-pass ﬁlter with a cut-off fre-

quency of f

= 80 Hz and a slope of 24 dB/oct.

We measure the airﬂow with a pneumotachograph

Schiller SP 260. The airﬂow signal is sampled with a

frequency f

= 400 Hz.

We calibrate the recording device with a Br¨uel &

Kjær sound calibrator Type 4231, a sound source with

a sinusoidal waveform at a frequency of f = 1 kHz

and with a sound pressure level of 94 dB. We adjusted

the microphone preampliﬁers of the LSTs to reach the

same signal level for the sound calibrator.

2.2 Subjects and Material

At airﬂow rates of 0.3, 0.7, 1.0, 1.3 and 1.7 l/s, we

recorded lung sounds over the posterior chest of four

lung-healthy subjects. The subjects held the pneu-

motachograph with both hands and wore a nose clip.

The subjects breathe steadily during inspiration at the

given airﬂow rates and with natural breathing dur-

ing expiration. The recording setup provided a real-

time feedback for the airﬂow rate. The subjects were

placed on the pad with a deﬁned distance d ≈ 7 cm

between the 7th cervical vertebra (C7) and the cen-

ter line of the topmost row of sensors. The recording

material of one subject consists of 16-channel lung

sound recordings at ﬁve different airﬂow rates, with

4-8 breathing cycles within 30 seconds, respectively.

The subjects were four male volunteers, with no diag-

nosed lung diseases and with the following metadata:

age (27, 27, 26, 27 years), weight (78, 75, 75, 75 kg)

and height (1.8, 1.78, 1.89, 1.72 m).

The used multichannel recording front-end is ro-

bust against ambient noise. However, in lung sound

recordings, interfering signals are caused by the heart,

bowels and body movement. These can distort the

signal energy values from lung sound signals. To en-

sure uncontaminated lung sound recordings, we man-

ually labeled the sections containing heart, bowel and

other interfering sounds.

2.3 Signal Pre-processing

We applied a bandpass ﬁlter, with a lower cut-off fre-

quency f

= 150 Hz and an upper cut-off frequency

= 250 Hz, to the 16 lung sound signals. To cal-

culate the energy, we used a sliding window with a

length of 50 ms and an overlap of 75 %.

BIOSIGNALS 2017 - 10th International Conference on Bio-inspired Systems and Signal Processing

2.4 Acoustic Thoracic Images

To illustrate the regional distribution of the lung

sound energy, we use acoustic thoracic images. We

generate the images for the left and the right hemitho-

rax independently. In particular, we use the energy

signal of the left-sided (Sensors 3, 4, 7, 8, 11, 12, 15

and 16) and right-sided sensors (Sensors 1, 2, 5, 6, 9,

10, 13 and 14), respectively. To generate an acoustic

thoracic image at a certain airﬂow rate, we used the

appropriate segments of the recording. We average

the energy values of all the uncontaminated segments,

i.e. labeled as free of interfering sounds (cf. Section

2.2), where the subjects reached the proper airﬂow

rate. For the interpolation between the energy values,

obtained from the eight sensor signals, we use the bi-

harmonic spline interpolation. The resulting acoustic

thoracic images are grayscale images (see Figure 4).

The white color indicates the minimum value and the

black color the maximum value.

3 RESULTS

3.1 Amplitude

Figure 2 shows the square root of the sound energy

as a function of airﬂow rate for all of the four subjects

independently. We performed linear regression for the

values from each subject independently (The results

are not shown in Figure 2).

For Subject 1, the coefﬁcient of determination is

= 0.98, for Subject 2 it is R

= 0.96, for Subject 3

it is R

= 0.99 and for Subject 4 it is R

= 0.99. Figure

3 shows the spectral characteristics (i.e. power spec-

tral density (PSD)) of the lung sounds at different air-

ﬂow rates, generated from the lung sound recording of

the sensor at position 6 (see Figure 1) from Subject 1.

0 0.5 1 1.5 2

Airﬂow Rate [l/s]

√

E [Arb. Units]

Subject 1

Subject 2

Subject 3

Subject 4

Figure 2: Square root of the sound energy

√

E as a function

of airﬂow rate for all four subjects.

Frequency [Hz]

PSD [dB]

-120

-110

-100

-90

-80

-70

-60

-50

-40

0.3 l/s

0.7 l/s

1.0 l/s

1.3 l/s

1.7 l/s

Figure 3: Spectral characteristics of the lung sounds at dif-

ferent airﬂow rates, generated from the lung sound record-

ing of the sensor at position 6 (see Figure 1) from Subject 1.

3.2 Regional Distribution

Figure 4 shows the acoustic thoracic images of four

lung-healthy subjects, evaluated at ﬁve different air-

ﬂow rates. The grayscale images are normalized for

the respective airﬂow rate. For an airﬂow rate of

0.3 l/s, we observe that most of the energy is in the

middle right area. Already for an airﬂow rate of

0.7 l/s, the lung sound energy is higher towards the

base of the lungs. Above an airﬂow rate of 0.7 l/s, the

regional distribution remains almost constant.

Table 1 shows the mean and standard deviation of

the percentage of the summed up energy from the sen-

sors over the left, right, upper (sensor 1 to 8) and

lower hemithorax (sensor 9 to 16). The signal en-

ergy over the left lung is distinctly higher than over

the right lung. This is further reﬂected in the acous-

tic thoracic images, especially above an airﬂow rate

of 0.3 l/s. The values in Table 1 show that with in-

creasing airﬂow rate the percentage for the left lung

increases. Regarding the ratio of the upper to lower

hemithorax, for an airﬂow value of 0.3 l/s the energy

in the upper lungs is higher. With increasing airﬂow,

the percentage for the lower lung increases, but for

1.7 l/s it decreases again.

Table 1: Mean and standard deviation of the percentage of

the summed up energy from the sensors over the left, right,

upper and lower hemithorax for different airﬂow values.

0.3 l/s 0.7 l/s 1.0 l/s 1.3 l/s 1.7 l/s

Left 53±8 62±9 59±3 62±3 65±7

Right 47±8 38±9 41±3 38±3 35±7

Upper 56±13 43±5 34±3 33±4 40±1

Lower 44±13 57±5 66±3 67±4 60±1

Impact of Airﬂow Rate on Amplitude and Regional Distribution of Normal Lung Sounds

Subject 1

Subject 2

Subject 3

Subject 4

R L R L R L R L R L

0.3 l/s 0.7 l/s 1.0 l/s 1.3 l/s 1.7 l/s

Figure 4: Acoustic thoracic images from four lung-healthy subjects, evaluated at ﬁve different airﬂow rates. The orientation

is indicated by the capital letters R (right hemithorax) and L (left hemithorax).

4 DISCUSSION

To compare our ﬁndings with those in (Yosef et al.,

2009), we used a similar bandpass ﬁlter, with a lower

cut-off frequency f

= 150 Hz and an upper cut-off

frequency f

= 250 Hz (see Section 2.3). Although

we lose important information from the signal in the

higher frequency range, due to the dominance of the

signal energy in the low frequency range, a higher up-

per cut-off frequency f

would not have a huge im-

pact on the acoustic thoracic images. According to

Figure 3 a bandpass ﬁlter with an upper cut-off fre-

quency of f

≈ 600 Hz could be considered.

Our ﬁndings regarding amplitude and regional

distribution of lung sounds correspond most closely

with those in (Yosef et al., 2009), although we

recorded the lung sounds in supine position. The au-

thors in (Fiz et al., 2008) already observed that, com-

pared with sitting, the supine position does not cause

a substantial change in lung sound intensity. The au-

thors in (Torres-Jimenez et al., 2008b) also observed

a constant regional distribution for RATHIs at airﬂow

rates of 1.0, 1.5 and 2.0 l/s. The authors in (Yosef

et al., 2009) showed the same for Vibration Response

Images at airﬂow rates of 1.0, 1.3 and 1.7 l/s. Re-

garding the relationship between airﬂow rate and the

square root of lung sound energy (see Section 3.1),

the authors in (Yosef et al., 2009) achieved for linear

regression a coefﬁcient of determination of R

= 0.95.

A limitation of our experiment is the small number

of subjects n = 4 and the lack of female subjects.

5 CONCLUSIONS

In this paper, we investigate the impact of airﬂow rate

on amplitude and regional distribution of normal lung

sounds. Therefore, we record lung sounds with a mul-

tichannel recording device at different airﬂow rates.

We illustrate the regional distribution with acoustic

thoracic images.

We observe a linear dependence between airﬂow

rate and lung sound amplitude. In our recordings, the

signal energy from lung sounds over the left lung is

distinctly higher than from those over the right lung.

Above an airﬂow rate of 0.7 l/s, we observe a con-

stant regional distribution for the lung sound energy.

Although we recorded lung sounds on the posterior

chest in supine position instead of sitting, our ﬁnd-

ings match most closely with those in (Yosef et al.,

2009).

The observed relationship can be used for the ex-

traction of the airﬂow from lung sounds. Further-

more, the ﬁndings are helpful for future work, re-

garding the standardized recording of lung sounds and

subsequent classiﬁcation by means of machine learn-

ing techniques.

BIOSIGNALS 2017 - 10th International Conference on Bio-inspired Systems and Signal Processing

ACKNOWLEDGEMENTS

This project was supported by the govern-

ment of Styria, Austria, under the project call

HTI:Tech

for Med. The authors acknowledge 3M

for providing Littmann

 stethoscope chest pieces

and Schiller AG for the support with a spirometry

solution.

REFERENCES

Charleston-Villalobos, S., Cort´es-Rubiano, S., Gonz´alez-

Camerena, R., Chi-Lem, G., and Aljama-Corrales,

T. (2004). Respiratory acoustic thoracic imaging

(RATHI): assessing deterministic interpolation tech-

niques. Medical and Biological Engineering and

Computing, 42(5):618–626.

Fiz, J. A., Gnitecki, J., Kraman, S. S., Wodicka, G. R., and

Pasterkamp, H. (2008). Effect of body position on

lung sounds in healthy young men. CHEST Journal,

133(3):729–736.

Gavriely, N. and Cugell, D. W. (1996). Airﬂow effects

on amplitude and spectral content of normal breath

sounds. Journal of applied physiology, 80(1):5–13.

Hossain, I. and Moussavi, Z. (2002). Relationship be-

tween airﬂow and normal lung sounds. In Proceed-

ings of the 15th IEEE Canadian Conference on Elec-

trical and Computer Engineering (CCECE’02), pages

1120–1122.

Kraman, S. (1984). The relationship between airﬂow and

lung sound amplitude in normal subjects. CHEST

Journal, 86(2):225–229.

Messner, E., Hagm¨uller, M., Swatek, P., and Pernkopf, F.

(2016). A robust multichannel lung sound recording

device. In Proceedings of the 9th Annual Interna-

tional Conference on Biomedical Electronics and De-

vices (BIODEVICES’16), pages 34–39.

Murphy, R. (2007). Development of acoustic instruments

for diagnosis and management of medical condi-

tions. Engineering in Medicine and Biology Maga-

zine, 26(1):16–19.

Pasterkamp, H., Kraman, S., DeFrain, P., and Wodicka,

G. (1993). Measurement of respiratory acoustical

signals. Comparison of sensors. CHEST Journal,

104(5):1518–1525.

Sen, I. and Kahya, Y. (2006). A multi-channel device for

respiratory sound data acquisition and transient detec-

tion. In Proceedings of the 27th Annual International

Conference of the IEEE Engineering in Medicine and

Biology Society (EMBS’06), pages 6658–6661.

Shykoff, B. E., Ploysongsang, Y., and Chang, H. (1988).

Airﬂow and normal lung sounds1. 2. Am Rev Respir

Dis, 137:872–876.

Torres-Jimenez, A., Charleston-Villalobos, S., Gonzalez-

Camarena, R., Chi-Lem, G., and Aljama-Corrales,

T. (2008a). Asymmetry in lung sound intensities

detected by respiratory acoustic thoracic imaging

(RATHI) and clinical pulmonary auscultation. In Pro-

ceedings of the 30th Annual International Conference

of the IEEE Engineering in Medicine and Biology So-

ciety (EMBS’08), pages 4797–4800.

Torres-Jimenez, A., Charleston-Villalobos, S., Gonzalez-

Camarena, R., Chi-Lem, G., and Aljama-Corrales,

T. (2008b). Respiratory acoustic thoracic imaging

(RATHI): Assessing intrasubject variability. In Pro-

ceedings of the 30th Annual International Conference

of the IEEE Engineering in Medicine and Biology So-

ciety (EMBS’08), pages 4793–4796.

Yosef, M., Langer, R., Lev, S., and Glickman, Y. A. (2009).

Effect of airﬂow rate on vibration response imaging in

normal lungs. The open respiratory medicine journal,

3(1).

Impact of Airﬂow Rate on Amplitude and Regional Distribution of Normal Lung Sounds