MICROCOMPUTERIZED RESPIRATORY SOUND RECORDER

A Low Cost Device

Daniel F. Ponte, Raimes Moraes

Electrical Engineering Department, Federal University of Santa Catarina, 88040-900, Florian´opolis, SC, Brazil

Federal Institute of Education in Science and Technology, 64000-040, Teresina, PI, Brazil

Leila J. M. Steidle, Renata Cristina T. P. Viana

Medical School Hospital Department, Federal University of Santa Catarina, 88080-350, Florian´opolis, SC, Brazil

Deborah C. Hizume, Adriano M. Alencar

Physiotherapy Department, State University of Santa Catarina, 88080-350, Florian´opolis, SC, Brazil

Pathology Department, Medical School of University of S˜ao Paulo, 01246-903, S˜ao Paulo, SP, Brazil

Keywords:

Respiratory sounds, CORSA, Flow waveform, Respiratory diseases.

Abstract:

Auscultation of breathing sounds is a common practice since the antiquity. In 1819, La¨ennec invented the

stethoscope and published the ﬁrst work on pulmonary disorders and their associated sounds. Since then, the

auscultation was incorporated into medicine. The ﬁrst electronic device to record and analyze physiological

sounds was built in 1955, being followed by many other developments. In 2000, a task force of the European

Respiratory Society established guidelines for computerized respiratory sound analysis (CORSA). Our work

describes a low cost microcomputerized system, based on the CORSA guidelines, developed to acquire and

record breathing sounds as well as respiratory ﬂow waveforms. It consists of a four channel micro-controlled

device that can simultaneously record sounds from three different sources and ﬂow waveform. These signals

are transmitted to a microcomputer running dedicated software that shows the waveforms on the screen and

stores them into the hard disk. The developed device was tested in patients with heart failure, idiopathic

pulmonary ﬁbrosis, pneumonia and asthma. Examples of the registered signals and results of a qualitative

assessment of the developed system are presented.

1 INTRODUCTION

In 1819, La¨ennec developed a noninvasive diagnostic

tool for the assessment of pulmonary diseases named

stethoscope (La¨ennec, 1819). Based on necropsies,

he associated respiratory sounds auscultated on the

thorax to pathologies, such as: edema, pneumo-

nia, tuberculosis, bronchitis and emphysemas. How-

ever, lung sound auscultation carried out with the

stethoscope is a subjective procedure, since it de-

pends on the experience and hearing acuity of the

examiner (Garcia, 2002). Besides, the human au-

ditory system is not very sensitive to the frequency

response of stethoscopes that attenuate components

above 120Hz (Sovij¨arvi et al., 2000).

In 1955, the ﬁrst electronic device to record and

analyze biological sounds was developed (McKusic

et al., 1955), being followed by other projects in the

decades of 60 and 70 (Forgacs, 1969; Weiss and Carl-

son, 1972). In 1987, the International Lung Sounds

Association proposed a common nomenclature that

has been used internationally (Mikami et al., 1987).

Computerized methods to record and analyze respi-

ratory sounds may overcome many limitations of the

auscultation. Nevertheless, the conclusive character-

ization of the sounds belonging to different respira-

tory disorders was being hampered by the fact that

researchers were employing systems with different

technical speciﬁcations for the sound acquisition. To

circumvent that, the European community promoted

the CORSA (Computerized Respiratory Sound Anal-

ysis) project that established guidelines for the inves-

tigation of lung sounds (Sovij¨arvi et al., 2000).

The adventitious respiratory sounds are classi-

ﬁed in discontinuous or continuous (Sovij¨arvi et al.,

2000). Pulmonary crackles are discontinuous, be-

F. Ponte D., Moraes R., J. M. Steidle L., Cristina T. P. Viana R., C. Hizume D. and M. Alencar A. (2010).

MICROCOMPUTERIZED RESPIRATORY SOUND RECORDER - A Low Cost Device.

In Proceedings of the Third International Conference on Biomedical Electronics and Devices, pages 21-27

DOI: 10.5220/0002721900210027

 SciTePress

Figure 1: Microcomputerized system for acquisition of breathing sounds and ﬂow waveforms. There are three channels to

record sounds and one channel to register the ﬂow signal.

(a) (b) (c)

Figure 2: (a) Microphones and acoustic coupler manufactured in nylon; (b) Connections of the pressure transducer to the

pneumotacograph (PT); (c) View of the PT resistance that provides pressure drop proportional to the ﬂow velocity.

ing described as short, explosive and transient sound

waves, characterized by a rapid initial pressure deﬂec-

tion, called a spike, followed by a short duration ring-

ing and are usually associated with recruitment of ob-

structed airways (Alencar et al., 2001; Hantos et al.,

2004). They occur in heart failure, ﬁbrosis, pneumo-

nia and others. The wheezes and rhonchi are contin-

uous adventitious sounds and are caused by narrowed

upper airways and secretion in bronchial airways, re-

spectively. Wheezes are periodic, containing a single

tone (monophonic) or many related harmonic tones

(polyphonic).

This work presents a low cost system to record

breathing sounds that was developed according to the

CORSA recommendations. It also records ﬂow wave-

form simultaneously.

To assess the developed system, sounds and ﬂow

waveforms were recorded from patients with heart

failure, ﬁbrosis alveolitis, pneumonia and asthma.

The quality of the recorded sounds was analyzed by

specialists. The recorded waveforms as well as the

results of the qualitative assessment of the developed

system are presented.

2 MATERIALS AND METHODS

Figure 1 depicts the block diagram of the developed

system. It consists of a conditioning module (micro-

phones, pressure sensor, ampliﬁers and ﬁlters), a con-

BIODEVICES 2010 - International Conference on Biomedical Electronics and Devices

Figure 3: Screen of the software showing four acquired waveforms. On the top, left to right, lung sounds of the channels 1

and 2, respectively. On the bottom left, lung sound of the channel 3. On the bottom right, the ﬂow waveform.

trol module (microcontroller and communication in-

terface) and software for Windows

 OS. The next

subsections describe each module of the developed

system.

2.1 Conditioning Module

The breathing sounds are acquired on the thorax using

3 electret microphones (MD9745APA-1 – Knowles

Acoustics) that are housed by acoustic couplers man-

ufactured in nylon (Fig. 2) according to the dimen-

sions recommended by previous studies (Kraman

et al., 1995). This microphone model has small di-

mensions (9.7 × 4.5 mm), low weight (about 1g), a

ﬂat frequency response from 100 to 3000 Hz, signal

to noise ratio of 55 dB, output impedance of 2.2 kΩ,

and sensitivity of 9 mV/Pa.

2.1.1 Flow Transducer

To sample the ﬂow waveform, the volunteer breaths

through an acrylic tube containing a pneumotaco-

graph (PT). The pressure drop across the pneumo-

tacograph resistance is proportional to the ﬂow veloc-

ity (Doeblin, 1990). The transducer (DC030NDC4

– Honeywell) measures the differential pressure

through the obstacle in a range of ±76.2 cmH

O. It

has a sensitivity of 52.36 mV/cmH

O, generating a

voltage output of 2.25±2.0 V. Figure 2 shows how the

transducer inputs are connected to the PT apertures as

well as the ﬂow resistor. The pressure drop is negative

for inspiratory ﬂow and positive for expiratory ﬂow.

Electrical signals generated by the microphones

have low amplitude, requiring ampliﬁcation and ﬁl-

tering (anti-aliasing) before being sampled. The sig-

nal generated by each microphone is applied to a

preampliﬁer with a gain of 2 that also accomplishes

impedance matching to the next circuit stage (Sedra

and Smith, 2004). The ampliﬁed signal has its band-

width limited from 60 to 2500Hz by two second or-

der Butterworth ﬁlters in cascade. The high pass ﬁl-

ter attenuates the low frequency sounds produced by

the heart that may saturate the circuit, distorting the

sampled signal. The low pass one attenuates sounds

that are above the expected frequency content of the

crackles. These respiratory sounds have the highest

frequency components that may achieve up to 2000

Hz (Sovij¨arvi et al., 2000). The ﬁltered signal is

further ampliﬁed to 600 times by the inherent ﬁlters

gains and by software programmable gain ampliﬁer

(IC AD526 – Analog Devices).

The output voltage signal of the pressure trans-

MICROCOMPUTERIZED RESPIRATORY SOUND RECORDER - A Low Cost Device

ducer is ﬁltered by a second order Butterworth low

pass ﬁlter with a cut-off frequency of 40Hz. The sig-

nal amplitude is adjusted to the maximum value of the

acquisition module (0 to 2.5V), considering the max-

imum expected ﬂow range (-70 to +70ℓ/min). For

that, it is ampliﬁed and summed to a dc offset to

achieve positive values. The obtained resolution is

17.9mV(ℓ/min)

−1

2.2 Control Module

The main IC of the control module is the ADuC841

microcontroller (Analog Devices), an optimized

single-cycle 20 MHz 8052 core. It has a 12-bit analog

to digital converter (ADC) fed by an 8-channel ana-

log input multiplexer, four different memory blocks

(62 kiB of ﬂash for code, 4 kiB of ﬂash for data, 256

bytes of general-purpose RAM and 2 kiB of internal

XRAM), 3 timers, serial communication interfaces

(UART, SPI, I2C) and 2 digital to analog converter

(DAC) of 12 bits.

After being ﬁltered and ampliﬁed, the three respi-

ratory signals and the ﬂow waveform are simultane-

ously sampled at 10 kSPS by a sample - and - hold (IC

SMP04 - Analog Devices). The ADuC841 gets, one

by one, the sampled voltage values and converts them

to digital. It carries out the conversion in 8 µs with a

voltage resolution of 0.61mV (1LSB=2.5 V/4096 ).

The digital samples are sent to the data-transfer

device (IC FT245BM – Future Technology Devices

Intl.) that establishes the USB communication with

a notebook, transferring data at the rate of 1MiB/sec

(Axelson, 2005).

2.3 PC Software

Software developed in C++ Builder establishes the

communication between the computer and the control

module. For that, USB driver made available by the

FT245BM manufacturer is employed.

The received data contain multiplexed samples of

each channel. The samples have a header with the

number of the channel to which they belong.

The software demultiplexes the received data and

shows them on the screen in real time. To achieve

that, a scientiﬁc chart library for plotting multiple

curves (Scope) is used (Scientiﬁc Plotting Library)

since the native C++ Builder library is quite slow for

real time applications. Each sampled waveform is

stored into the hard disk in individual wave ﬁles. Fig-

ure 3 shows the screen of the developed software.

3 RESULTS

To assess the qualitative performance of the devel-

oped system, adventitious sounds were recorded from

patients of a Medical School Hospital (Federal Uni-

versity of Santa Catarina) after the approval by its Re-

search Ethics Committee (Process number:181/2007).

The patients were in a room without noise level

control (inﬁrmary). Based on the medical records,

clinical signs, chest x-rays and lung function studies,

the patients were diagnosed with the following pul-

monary conditions: heart failure (2 men), idiopathic

pulmonary ﬁbrosis (3 men), pneumonia (2 men) and

asthma (5 women).

Figures 3 and 4 show examples of sound curves

as well as ﬂow waveforms that were simultaneously

recorded using the developed system.

The sounds collected from twelve patients (with-

out any post-processing) were reproduced to seven

respiratory sound specialists of the Therapeutic Labo-

ratory in the Medical School of the University of S˜ao

Paulo (LTFMUSP). The specialists ﬁlled up a ques-

tionnaire on the quality of the recorded sounds. 42

opinions were obtained (Table 1). Besides the sound

quality, the questionnaire aimed to evaluate the need

for further processing of the sounds to improve the

diagnosis.

4 DISCUSSION

Figure 4a shows that crackles occurred during the in-

spiration and expiration for a ﬁbrosis patient. Patient

with heart failure had crackles only at the end of the

inspiration (Figure 4b). It should be noted that this

condition may generate crackles during expiration as

well (Piiril¨a et al., 1991; Vyshedskiy et al., 2009).

Figure 4c and d shows an expanded crackle wave-

form. The American Thoracic Society (ATS) uses

time intervals of the expanded crackle waveforms

(initial deﬂection width (IDW) and two-cycle dura-

tion (2CD)) to classify the crackles in two classes:

ﬁne, or high pitched crackles, and coarse, or low

pitched crackles (American Thoracic Society, 1977).

Figure 5a shows crackles and wheezes acquired from

a patient with pneumonia. Figure 5b shows the ex-

panded wheezes also known as squawks (Paciej et al.,

2004).

Figure 6 shows a short acoustic interval contain-

ing wheezes (and its sonogram) acquired from a pa-

tient with acute asthma. It is possible to see that the

recorded sound has more than one tone, being named,

therefore, polyphonic wheezes.

BIODEVICES 2010 - International Conference on Biomedical Electronics and Devices

IDW

2CD

(b)(a)

(d)(c)

Figure 4: Crackles and ﬂow waveforms recorded from patients with idiopathic pulmonary ﬁbrosis (a) and heart failure (b).

The sound and ﬂow waveforms are shown by the black and blue lines, respectively. (c) Zooming of the crackles shown in (b).

(d) The waveforms, initial deﬂection width (IDW) and two-cycle duration (2CD) waveforms of the strongest crackle shown

in (b).

Table 1 summarizes the answers of the 42 ques-

tions posed to specialists on the quality of the sounds

recorded with the developed system. The specialists

heard background noise interference in nearly 15% of

the recordings. They reported heart sounds superim-

posed to the adventitious sounds in about 27% of the

recordings. The majority of the specialists considered

that the noise level contained in the recordings was

low enough to allow the sound classiﬁcation.

In the inﬁrmary, background noise is always

present and it will be also heard when using a stetho-

scope. When the sound is recorded, digital signal pro-

cessing techniques may be applied to reduce the inter-

ferences due to the background noise and due to the

heart sounds.

5 CONCLUSIONS

Auscultation of breathing sounds using stethoscopes

is a relevant medical practice. Nevertheless, it does

not allow the information to be quantiﬁed, stored, re-

produced, visualized or processed. Therefore, it is

difﬁcult for the specialists to exchange information

and educate new professionals.

Breathing sounds acquired with microphones and

recorded with eletronic devices contain information

of a given patient lung condition that can be stored

and does not depend on the specialist subjectivity.

The application of digital signal processing tech-

niques to these signals may allow the development of

quantitative methods to assist the diagnosis.

This work presented a low cost system that is able

to acquire and store respiratory sounds and ﬂow wave-

MICROCOMPUTERIZED RESPIRATORY SOUND RECORDER - A Low Cost Device

Table 1: Specialists answers about the respiratory sounds recorded with the developed system.

No Yes

Is there environmental noise

in the respiratory sounds? 85.7% 14.3%

Is there heart sounds mixed

with the respiratory sounds? 73.8% 26.2%

Low Acceptable High Unacceptable

Classify the level of the

intrinsic system noise added 31% 47.5% 21.5% 0%

to the respiratory sounds.

(b)

(a)

Figure 5: Crackles recorded from patient with pneumonia

followed by wheezes (squawks - encircled region) (a) and

the zoomed view of the squawks (b).

forms. The system presented a good performance.

It was possible to record different respiratory sounds

with a good quality (crackles, wheezes and others).

The development of such low cost systems may allow

the dissemination of computerized respiratory sound

analyzes, contributing to a more objective diagnosis

of pulmonary disorders in the clinical practice.

(a)

(b)

Figure 6: Wheezes from patient with acute asthma (a) and

its sonogram using Hamming window (b).

ACKNOWLEDGEMENTS

The authors are thankful to the seven specialists

that answered the questions about the lung sounds

recorded. The authors acknowledge Dr. Henrique T.

Moriya, and Jo˜ao R. Baggio, MSc for many valuable

discussions related to technical aspects of the acqui-

BIODEVICES 2010 - International Conference on Biomedical Electronics and Devices

sition system as well as the patients of the HU/UFSC

that allowed the recordings. A. M. Alencar acknowl-

edge FAPESP (Fundac¸˜ao de Amparo a Pesquisa do

Estado de S˜ao Paulo).

REFERENCES

Alencar, A. M., Buldyrev, S. V., Majumdar, A., Stanley,

H. E., and Suki, B. (2001). Avalanche dynamics of

crackle sound. Phys. Rev. Lett., 87:088101.

American Thoracic Society (1977). Uptated nomenclature

for membership reaction. reports of the ats – accp ad

hoc committee. Am. Thorac. Soc., 3:5–6.

Axelson, J. (2005). USB Complete. Lakeview Research

LLC, 5310 Chinook Ln., Madison, USA.

Doeblin, E. O. (1990). Measurement Systems: application

and design, Chapter 7. McGraw-Hill, Singapore.

Forgacs, P. (1969). Lung sounds. Br. J. Dis. Chest, 63:1–2.

Garcia, E. A. C. (2002). Biof´ısica. Sarvier, S˜ao Paulo,

Brazil, 2nd edition.

Hantos, Z., Tolnai, J., Alencar, A. M., Majumdar, A., and

Suki, B. (2004). Acoustic evidence of airway open-

ing during recruitment in excised dog lungs. J. Appl.

Physiol., 97(2):592–598.

Kraman, S. S., Wodicka, G. R., Oh, Y., and Pasterkamp,

H. (1995). Measurement of respiratory acoustic sig-

nals: Effect of microphone air cavity width, shape and

venting. Chest, 108:1004–1008.

La¨ennec, R. T. H. (1819). De l’auscultation m´ediate ou

trait´e du diagnostic de maladies des poumons et du

coeur, fond´e principalement sur ce nouveau moyen

d’exploration. Brosson et Chaud´e, Paris.

McKusic, V. A., Jenkins, J. T., and Webb, G. N. (1955).

The acoustic basis of the chest examination: Studies

by means of sound spectrography. Am. Rev. Tuberc.,

72:12–34.

Mikami, R., Murao, M., Cugell, D. W., Chretien, J., Cole,

P., Meier-Sydow, J., Murphy, R. L., and Loudon, R. G.

(1987). International symposium on lung sounds. syn-

opsis of proceedings. Chest, 92(2):342–345.

Paciej, R., Vyshedskiy, A., Bana, D., and Murphy, R.

(2004). Squawks in pnemonia. Thorax, 59:177–179.

Piiril¨a, P., Sovij¨arvi, A. R. A., Kaisla, T., Rajala, H. M.,

and Katila, T. (1991). Crackle in patients with ﬁbros-

ing alveolitis, bronchiectasis, copd, and heart failure.

Chest, 99:1076–1083.

Sedra, A. S. and Smith, K. C. (2004). Microeletronics Cir-

cuits. Oxford University Press, Ney York.

Sovij¨arvi, A. R. A., Vanderschoot, J., and Earis, J. E.

(2000). Standardization of computerized respiratory

sound analysis – corsa. Eur. Respir. Rev., 10(77):585.

Vyshedskiy, A., Alhashem, R. M., Paciej, R., Ebril, M.,

Rudman, I., Fredberg, J. J., and Murphy, R. (2009).

Mechanism of inspiratory and expiratory crackles.

Chest, 135(1):156–164.

Weiss, E. B. and Carlson, C. J. (1972). Recording of breath

sounds. Am Rev Respir Dis., 105:835–939.

MICROCOMPUTERIZED RESPIRATORY SOUND RECORDER - A Low Cost Device