-
Thorax 1995;50:1285-1291
Significant differences in flow standardisedbreath sound spectra
in patients with chronicobstructive pulmonary disease, stable
asthma,and healthy lungs
L Pekka Malmberg, Leena Pesu, Anssi R A Sovijarvi
AbstractBackground - Spectral characteristics ofbreath sounds in
asthma and chronic ob-structive pulmonary disease (COPD) havenot
previously been compared, althoughthe structural differences in
these dis-orders mightbe reflected in breath sounds.Methods - Flow
standardised inspiratorybreath sounds in patients with COPD (n =17)
and stable asthma (n= 10) with sig-nificant airways obstruction and
in controlpatients without any respiratory disorders(n =11) were
compared in terms of es-timates of the power spectrum. Breathsounds
were recorded simultaneously atthe chest and at the trachea.Results
- The median frequency (F50) ofthe mean (SD) breath sound spectra
re-corded at the chest was higher in asth-matics (239 (19) Hz) than
in both thecontrol patients (206 (14) Hz) and thepatients with COPD
(201 (21) Hz). Thetotal spectral power of breath sounds re-corded
at the chest in terms of root meansquare (RMS) was higher in
asthmaticsthan in patients with COPD. In patientswith COPD the
spectral parameters werenot statistically different from those
ofcontrol patients. The Fso recorded at thetrachea in the
asthmatics was significantlyrelated to forced expiratory volume in
onesecond (FEVy) (r= -0.77), but this was notseen in the other
groups.Conclusions - The observed differencesin frequency content
of breath sounds inpatients with asthma and COPD may re-flect
altered sound generation or trans-mission due to structural changes
of thebronchi and the surrounding lung tissue inthese diseases.
Spectral analysis ofbreathsounds may provide a new
non-invasivemethod for differential diagnosis of ob-structive
pulmonary diseases.(Thorax 1995;50:1285-1291)
Keywords: respiratory sounds, asthma, chronic ob-structive
pulmonary disease.
Computer assisted techniques for lung soundanalysis1`6 have
provided new data on the re-lationship between respiratory sounds
anddifferent pathophysiological conditions of thelung. By using
these methods, changes in pul-monary and tracheal breath sounds as
well
as adventitious breath sounds - for example,wheezes and crackles
- can be objectively andaccurately recorded. In early studies
Forgacs7showed that breath sound intensity at themouth was
associated with forced expiratoryvolume in one second (FEVI) in
patients withasthma and chronic obstructive bronchitis, butnot in
those with emphysema. Later it wasshown that lung sound intensity
is related toregional distribution of ventilation8 and is re-duced
primarily by airflow limitation in patientswith emphysema.9
Spectral changes dis-tinguishing the asthma patients from
healthysubjects have been described.3`510 In addition,the frequency
content of breath sounds wasfound to be dependent on the degree of
bron-chial obstruction during challenge tests, evenin the absence
of wheezing sounds."'-"
Standards for the diagnosis of obstructivepulmonary diseases are
available.'4 In clinicaland functional patterns there is a
considerableoverlap between asthma and chronic ob-structive
pulmonary disease (COPD), althoughthe morphological and structural
changes inthe bronchi and lung parenchyma, as well asthe site of
these changes in the bronchial tree,are partly different.516 The
different structuraland morphological changes in COPD andasthma
might be reflected in breath sounds.At present there are few data
available on
the spectral characteristics of breath sounds inCOPD, and no
comparative data on the breathsounds of patients with asthma and
COPD.The aim of the present study was to investigatethe spectral
characteristics of patients withCOPD and stable asthma with
significant air-ways obstruction under standardised con-ditions,
and to compare them with those of sexand age matched control
patients with healthylungs.
MethodsPATIENTSThirty eight male outpatients referred to theLung
Function Laboratory at Helsinki Uni-versity Central Hospital were
selected for thestudy. Their anthropometric and lung functiondata
are presented in table 1.
Seventeen patients had stable chronic ob-structive pulmonary
disease diagnosed pre-viously according to ATS criteria'4; all
wereheavy current or ex-smokers. The degree ofobstruction assessed
by spirometric testsranged from mild to very severe. None of
the
Department ofMedicine,Division of PulmonaryDiseases andClinical
Physiology,Helsinki UniversityCentral Hospital,00290
Helsinki,FinlandL P MalmbergA R A Sovijarvi
Laboratory ofBiomedicalEngineering,HelsinkiUniversity
ofTechnology,Helsinki, FinlandL Pesu
Reprint requests to:Professor A R A Sovijarvi.Received 12
January 1995Returned to authors23 March 1995Revised version
received12 June 1995Accepted for publication30 August 1995
1285
-
Malmberg, Pesu, Sovijarvi
Table 1 Median (range) anthropometric and lung function data of
the patients with asthma, chronic obstructivepulmonary disease
(COPD) and controls
Asthma COPD Controls(n= 10) (n = 17) (n = 11)
Age (years) 50 (26-72) 58 (38-73) 50 (44-66)Height (cm) 176
(164-180) 172 (166-185) 174 (165-176)Weight (kg) 76 (58-99) 76
(57-97) 82 (67-87)FEV, (1) 2-59 (1-02-3-61)* 1-26 (0-59-3-20)* 3-55
(3-26-5.05)FEV, (% predicted'7) 67-5 (31-79)* 36 (16-79)* 95
(85-124)TLCO (% predicted'7) 57 (35-93)Kco (% predicted'7) 67
(40-105)
FEV, forced expiratory volume in one second; TLco =carbon
monoxide lung transfer factor; Kco =carbon monoxide
transfercoefficient.* p
-
Flow standardised breath sound spectra
of the recorder was digitised in a data ac-quisition and control
unit (HP 3852A) with13-bit analogue-to-digital conversion; the
sam-pling rate was 12 kHz for sound and 100 Hzfor the airflow. A
Bessel type low pass filterwith a cut off frequency of 4 kHz (24
dB/oct)was used to prevent aliasing. The data fromthe flow and
sound signals were stored on amagneto-optical disk of a Unix work
station(HP 9000/330C). The sound signal was di-gitally high pass
filtered (Kaiser-FIR) with acut off frequency of 100 Hz (24
dB/oct).The flow and sound signals were sub-
sequently analysed on a computer by usingcustom made software
based on Labview toolpackage. For the spectral analysis of
soundsignals 2048-point fast Fourier transforms wereperformed with
50% overlapping of adjacentHanning data windows; the spectra were
av-eraged over 8-10 consecutive breathing cyclesby the method of
Welch.'9 The effective fre-quency resolution of the spectral
estimates wasapproximately 5 Hz. Only sound samples ofinspiratory
sounds that occurred at flows from1-0/s to target flow (1 -25 /s)
were used forcalculation of averaged spectra. The averagepower
spectra of background noise were sim-ilarly estimated from sound
samples within aflow gate of 0 + 0-1/s. These spectra wereused to
evaluate the signal-to-noise ratio ofthe recordings. Expiratory
sounds were notanalysed because the breathing pattern couldnot be
standardised among the patient groupsdue to expiratory flow
limitation in somepatients with asthma or COPD.The upper frequency
limits for the second
quartile (F50, the median frequency) and thethird (F75) quartile
of the spectral power werecalculated within a measuring band
of100-2000 Hz on the linear amplitude scale.The frequency of
maximum intensity (Fmax)and the lung sound amplitude in terms of
theroot mean square (RMS) value of the totalspectral power were
also determined within thesame measuring band.The presence or
absence of wheezes in the
sound samples was evaluated by visual ex-amination ofthe sound
signal in time-expandedwaveform analysis (1000 mm/s) and in
fre-
quency domain (FFT). The criteria for awheeze were the presence
of a sinusoidal wave-form with a duration of more than 200 ms oran
intensity peak distinctly separated from itssurrounding intensity
in the FFT spectrum.Within the inspiratory flow gate chosen
foranalysis (1-0-1-25 l/s) none of the patients wasfound to
wheeze.
DATA ANALYSISThe breath sound variables between the studygroups
were compared by the non-paired t testwith the p values being
adjusted for multiplecomparison according to the Bonferroni
equa-tion (p*=pK; K=k(k-1)/2 where p* is theadjusted value of p and
k is the number ofgroups).20 For correlative analysis within
thegroups, Pearson's correlation coefficient wascalculated. A p
value of less than 0-05 wasconsidered significant.
ResultsExamples ofphonopneumograms and averagedinspiratory
spectra recorded at the chest inpatients with asthma, COPD, and
normal lungsare illustrated in fig 1. The spectra are cal-culated
from sound samples at an airflow gateof 1 -0-1-25 1/s. In each
spectrum plotted ona log-log scale the F50 is indicated and
thebackground spectra (sound at 0 +0-1 /s flow)presented. The
expiratory flow profiles differedbetween the patients with
different diagnosesas shown in fig 1, but inspiratory flow
profilesappeared quite similar. The spectra in fig 1 aresmooth in
shape without irregularities or peaksthat could indicate wheezing
sounds. The spec-trum of the asthma patient is broader in
shape,indicating increased sound energy at a fre-quency range of
200-1000 Hz; consequently,the F50 is increased compared with that
of thecontrol patient. In the patient with COPD theattenuation of
the sound power is more gradualover the whole frequency range, but
the globalfrequency distribution in terms of F50 is
notsignificantly different from that of the controlpatient. In the
whole group of patients withCOPD we did not find any characteristic
spec-
Control patienti1s .0
.$ 1.001 .b. [N. iiif N. &a flOa.040.00~ ~ ~ ~ VIFPY
0.00
-1.00 I -10
0 1 205 Hz
-25-310-35--40--45-
100 1000 2000Frequency (Hz)
Asthmatic patient
& 0.00 1;Ibo -I-Eu. ___
rwqm
m'a
a)
0
0-
1 s s1.00 -.t 00
* 1 00 E ,, -1 0E0- 267 Hz
-5 --10 -.__ _ _ _-15-20--25 --30 --35-40 --45.-.
100
a
L-
0)
0~
1000 2000Frequency (Hz)
u - _1 190 Hz-10-15 ..-20' .. s-25-10.-35
-40.-45-
Patient with COPD1 s
i"-.AQib1.00 a.0.00 E-1.00 C*E
1000 2000Frequency (Hz)
Figure 1 Examples ofphonopneumograms and flow standardised
inspiratory spectra recorded at the chest in a control patient and
patients with asthmaand COPD. Only sound samples at a flow gate of
1 -0-1 -25 1/s (indicated as lines under the flow curve) were used
for cakulation of the averagedspectra. In each spectrum the median
frequency (solid vertical line) is indicated and the background
(sound at 0+ 0-1 lls flow) spectra presented (dottedline).
0
lm'a0a-
;Al
1 287
100
-
Malmberg, Pesu, Sovijarvi
A
p
-
Flow standardised breath sound spectra
not explained by differences in the severity ofairways
obstruction.High quality equipment was used for re-
cording of breath sounds.6 Different micro-phones were used at
the trachea and at thechest for anatomical reasons so the sound
spec-tra cannot be critically compared between thetwo locations. It
should be noted that the fre-quency characteristics of the
microphones,27and specifically the dimensions of the air coup-ling
cavity," have been shown to modify theshape of lung sound spectra.
With the presentcondenser microphone of 5 mm cavity depthsome high
frequency components may there-fore have been attenuated in the
chest wallrecordings. However, the same microphoneswere used for
all patients and their char-acteristics cannot therefore explain
the differ-ences among the patient groups.The frequency spectra
were averaged to
minimise the effect of random noise. A cutofffrequency level of
100 Hz was chosen to elim-inate the muscular29 and
cardiovascular30sounds from the respiratory sound
analysis.Calculation of background noise enabled us toevaluate the
signal-to-noise ratio of the sound.The background spectra were not
subtractedfrom the analysed breath sound spectra becauserelevant
information may have been lost fortwo reasons. Firstly, the
background spectrawere derived under dynamic conditions
(duringbreathing with flows up to + 0.01 l/s) and in-evitably
contained a small contribution of lungsounds. Secondly, in some
individuals therewere insufficient data points to create an
ac-curate estimate from the background spectrum(even if the level
of signal-to-noise ratio couldbe estimated). We believe that, if
subtractionofbackground noise is preferred, sound duringbreath
holding should be used.3' Our pref-erence to use background noise
at zero flow asquality control is the same as that ofPasterkampet
al.27 The same recording room and conditionswere used for every
patient so the ambientnoise during recordings did not vary
sig-nificantly from patient to patient.The flow at the mouth was
carefully con-
trolled during lung sound recordings so thatinspiratory tidal
peak flow and pattern weresimilar in the patient groups. The
spectral shapehas been shown to be quite stable at high flowrates
above 1'0 I/s,3233 but to be dependent onflow at low flow rates.310
Flow gated samplingof respiratory sounds was originally describedby
O'Donnel and Kraman.34 In this study weused only sound samples of a
defined in-spiratory flow gate (1I0-1 -25 U/s) for the
spectralanalysis, ensuring that breath sound spectraamong the
patient groups were derived fromexactly the same flows. Differences
in airflowor breathing pattern cannot therefore explainthe
differences in frequency distribution oflungsounds among the
patient groups. Since onlyinspiratory breath sounds were analysed,
ex-piratory flow limitation occurring in somepatients with COPD did
not influence the res-ults. In contrast, clear differences in
expiratoryflow patterns did not allow valid comparisonof expiratory
breath sounds so the latter werenot included in the study.
Compared with control and COPD patients,the different frequency
content of breathsounds in the asthmatics was reflected inchanges
in F50, F75, and the total spectral power.The presence of
adventitious lung sounds suchas wheezes and crackles could change
thesespectral indices, but there were no wheezingsounds during the
inspiratory sound samplesselected for analysis. Crackling sounds
were notcounted, but on visual examination appeared tobe rare in
our patients with COPD. Fur-thermore, the averaging method used is
rel-atively insensitive to non-stationary changes ofbreath sounds
such as crackles. It is thereforeunlikely that adventitious breath
sounds wouldhave markedly contributed to the changes ob-served in
averaged frequency spectra.
It may be argued that the present parameterschosen to
characterise the breath sound spectra(RMS, Fmax, and the quartiles)
are not suffi-cient as they do not provide specific
informationabout the pattern of the spectral curve. How-ever, they
have been widely used in clinicalstudies of breath sounds in
asthma'0 11 13 and,in particular, the quartiles give an estimateof
the global frequency distribution of breathsounds, the primary
interest of our study.Normal tracheal sounds are characterised
by
a broad spectrum of frequencies with a sharpdecrease ofpower
above 800-1600 Hz. Normalvesicular lung sounds in the chest contain
aspectrum of frequencies up to approximately500 Hz; the frequency
components of lungsounds above this are effectively filtered by
thelung tissue and the chest wall.35 In normalmen inspiratory
sounds containing substantialpower relative to background noise up
to2000 Hz have been reported when high airflowrates are used (up to
2-5 l/s).27 In the presentstudy the usual frequency range at which
thebackground noise was reached in chest re-cordings was 1 0-1 2
kHz. However, the targetflow was set at a relatively low level (1I
25 l/s) dueto the reduced ventilatory function in patientswith
asthma and COPD.
In previous studies the frequency spectra ofasthmatics were
found to differ from those ofhealthy controls during episodes
ofwheezing.3In patients with acute asthma the proportionof the
respiratory cycle occupied by wheezingwas found to correlate with
FEVI.36 In previousstudies on asthmatics during challenge tests
thefrequency distribution of breath sounds wasshown to be related
to the degree of bronchialobstruction, with higher F50 values being
foundduring bronchial obstruction11-3 and also inthe absence
ofwheezing sounds. Schreur et all'recently found similar changes in
frequencydistribution of lung sounds in asthma patientseven with
normal lung function. The presentfinding that sound energy is
shifted towardshigher frequencies in asthmatics compared
withcontrols is in accordance with previously re-ported
studies.
In the previous literature there are only fewdata available
about the spectral characteristicsof breath sounds in COPD.
Gavriely et al37used lung sound analysis for respiratory
healthscreening and found that 30 out of 62 subjectswith clinical
COPD had abnormal breath
1289
-
Malmberg, Pesu, Sovijarvi
sound spectra. They used regression analysisof the spectral
curve to define abnormality,which is a particularly useful method
to detectcurve irregularities caused by wheezing sounds.The
quartile frequencies were not used in thisstudy, and the patterns
of abnormalities inCOPD were not compared with those inasthma. In
our study flow standardised in-spiratory breath sounds that did not
containwheezes did not differ between COPD andcontrol patients, but
in asthmatic patients sig-nificantly higher RMS, F50 and F75 values
werefound than in patients with COPD. The spec-tral shapes did not
differ visually betweenpatients with COPD and control patients,
andthe frequency range that contained breathsounds with substantial
power relative to back-ground spectra was about the same in thetwo
groups. However, the frequency contentof inspiratory vesicular lung
sounds in patientswith COPD seemed to be affected by the pres-ence
ofemphysema, patients with reduced Kcohaving lower F50 values.
It has been shown that the auscultatory find-ing of damped
breath sounds in patients withemphysema is predominantly due to
airflowlimitation, and that at standardised airflows thebreath
sound intensity is similar in patientswith emphysema and healthy
controls.9 TheRMS values in the present study were cal-culated from
the power spectrum and are nottherefore equivalent to breath sound
intensity.However, our finding that the total spectralpower did not
differ between the COPDpatients and controls was consistent with
theprevious literature. During episodes of normallung function in
asthmatic patients Schreur etallo found that, at a given airflow,
expiratorysounds were lower in intensity than those ofthe healthy
controls. In contrast, they did notfind any difference in
inspiratory sound in-tensity between their study groups as we
didbetween the control patients and asthmaticpatients with airways
obstruction.The generally accepted view is that vesicular
lung sounds are generated predominantly byturbulence in the
larger airways (main, lobar,and segmental bronchi) and transmitted
fromthese locations to the chest wall.2" The cal-culations from gas
density, viscosity, andvelocity predict laminar and,
presumably,soundless flow in more peripheral airways.However, there
is experimental evidence toshow that at least the inspiratory
portion ofvesicular lung sounds is produced locally insmall airways
by non-turbulent mechanisms.38In asthmatic patients the increase in
frequencycontent of breath sounds during broncho-constriction can
be explained by the local in-crease of flow velocity through the
narrowedbronchi which results in increased kinetic en-ergy and
turbulence, characterised by a higherpitch sound.39 According to
the model of un-stable vortices by Hardin and Patterson40 thesound
frequency produced is inversely relatedto the diameter of the
airways at a given massflow; higher sound frequencies are thus
ex-pected to be found in asthmatics with narrowedairways. The
present finding that the F5,, issignificantly correlated with FEV,
in asthmatic
patients also indicates the close relationshipbetween frequency
distribution of breathsounds and narrowing of the airways.The
present results indicate that, in patients
with COPD, the vesicular lung sounds areeither unchanged in
spite of airway narrowingor some other mechanisms are involved
thatcounteract the effect of increased turbulencein the bronchi.
The sound generation in asthmaand COPD may be different since the
structuralchanges in the bronchi in patients with COPDare situated
more peripherally in small airwayswhere turbulence does not occur
in normalconditions. The trend between Kco and F50 ofinspiratory
lung sounds in COPD suggests thatthe sound is affected by the
degree of em-physema. The lung parenchyma and the chestwall act
like a band pass filter. The averagedensity of lung tissue is
reduced in emphysemadue to tissue destruction; this should affect
thetransmission of lung sounds from their sourceto the chest wall22
and accentuate the low passfiltration effect of the thorax.
Consequently,lower frequency lung sounds would be detectedat the
chest, a possible explanation for thedifference between asthma and
COPD.
Breath sounds heard at the trachea are pre-sumably generated in
the very central airwaysby turbulent mechanisms, and the larynx
mayalso contribute to the sound production byacting as a nozzle,
creating a jet of air.2' Thecorrelation between FEV, and F50 in
asthmaticpatients may relate to increased turbulencein central
airways, probably caused by airwaynarrowing. The frequency
distribution oftrach-eal sounds has high intersubject variation
whichrelates to individual resonant frequencies withaccentuated
intensities. Thus, significantdifferences between the patient
groups couldnot be found for tracheal sounds.The finding that
breath sound spectral pat-
terns in patients with COPD and asthma differsignificantly may
have also clinical implications,since it provides new tools for
differential diag-nosis of obstructive lung disorders. In the
pres-ent study, however, the F50 values in asthmaand in COPD partly
overlapped, and the speci-ficity of a given F50 value is probably
not veryhigh. Further studies may reveal other spectralestimators
that could be used to separate thegroups more effectively.
In conclusion, the present findings indicatethat the frequency
spectra of breath sounds inpatients with stable asthma, but not in
patientswith COPD, differ significantly from those inpatients with
healthy lungs. It is postulated thatthe structural changes in the
bronchi and in thesurrounding lung tissue in COPD and asthmaresult
in different breath sound generation ortransmission. Breath sound
spectral analysismay provide a new non-invasive method
fordifferential diagnosis of bronchial obstruction.
This study was supported by the Ida Montin Foundation,Finland.
The authors are grateful to Erkki Paajanen and KariKallio for their
technical collaboration.
1 Murphy RLH, Sorensen K. Chest auscultation in thediagnosis of
pulmonary asbestosis. Jf Occup Med 1973;15:272-6.
2 Murphy RLH, Holford SK, Knowler WC. Visual lung-sound
characterization by time-expanded wave-form ana-lysis. N Engl _J
Med 1977;296:968-71.
1290
-
Flow standardised breath sound spectra
3 Charbonneau G, Racineux JL, Sudraud M, Tuchais E. Anaccurate
recording system and its use in breath soundsspectral analysis.
Appl Physiol 1983;55: 1120-7.
4 Gavriely N, Palti Y, Alroy G, Grotberg JB. Measurementand
theory of wheezing breath sounds. Jf Appl
Physiol1984;57:481-92.
5 Pasterkamp H, Carson C, Daien D, Oh Y. Digital
re-spirosonography. New images of lung sounds. Chest
1989;96:1405-12.
6 Rosqvist T, Paajanen E, Kallio K, Rajala H-M, Katila
T,Malmberg LP. A toolkit for lung sound analysis. Med BiolEng
Comput 1995;33:190-5.
7 Forgacs P, Nathoo AR, Richardson HD. Breath sounds.Thorax
1971;26:288-95.
8 Ploysongsang Y, Pare JAP, Macklem PT. Correlation ofregional
breath sounds with regional ventilation in em-physema. Am Rev
Respir Dis 1982;126:526-9.
9 Schreur HJW, Sterk PJ, Vandershoot J, van Klink HCJ,
vanVollenhoven E, Dijkman JH. Lung sound intensity inpatients with
emphysema and in normal subjects at stand-ardized airflows. Thorax
1992;47:674-9.
10 Schreur HJW, Vandershoot J, Zwinderman AH, DijkmanJH, Sterk
PJ. Abnormal lung sounds in patients withasthma during episodes
with normal lung function. Chest1994;106:91-9.
11 Andersson K, Aitken S, Carter R, MacLeod JES, Moran
F.Variation of breath sound and airway caliber inducedby histamine
challenge. Am Rev Respir Dis 1990;141:1147-50.
12 Spence DPS, Bentley S, Evans DH, Morgan MDL. Effect
ofmethacholine induced bronchoconstriction on the
spectralcharacteristics of breath sounds in asthma. Thorax
1992;47:680-3.
13 Malmberg LP, Sovijarvi ARA, Paajanen E, Piirila P, HaahtelaT,
Katila T. Changes in frequency spectra ofbreath soundsduring
histamine challenge test in adult asthmatics andhealthy controls.
Chest 1994;105:122-31.
14 American Thoracic Society. Standards for the diagnosis
andcare ofpatients with chronic obstructive pulmonary disease(COPD)
and asthma.Am Rev RespirDis 1987;136:225-44.
15 McDowell EM, Beals TF. Biopsy pathology of the
bronchi.Philadelphia: W B Saunders, 1987.
16 Laitinen LA, Laitinen A. Pathology and cytology of asthma.In:
Clark TJH, Godfrey S, Lee TH, eds. Asthma. London:Chapman &
Hall, 1992:232-53.
17 Vlljanen AA, Halttunen PK, Kreu K-E, Viljanen BC. Ref-erence
values for spirometric, pulmonary diffusing capacityand body
plethysmographic studies. Scand Clin Invest1982;42(Suppl
159):1-50.
18 Cotes JE, Chinn DJ, Quanjer PH, Roca J, Yemault
JC.Standardization of the measurement of transfer factor(diffusing
capacity). Eur Respir3' 1993;6(Suppl):41-52.
19 Welch PD. The use of fast Fourier transform for the
es-timation of power spectra. IEEE Trans Audio
Electroacoust1967;AU-15:70-3.
20 Snedecor GW, Cochran WG. Statistical methods. 7th edn.Ames:
The Iowa State University Press, 1982.
21 Olson DE, Hammersley JR. Mechanisms of lung soundgeneration.
Semin Respir Med 1985;6:171 -9.
22 Rice DA. Transmission of lung sounds. Semin Respir
Med1985;6:166-70.
23 Laennec RTH. A treatise on diseases of the chest.
Translatedfrom the French by John Forbes. London: T. and
C.Underwood, 1821.
24 Richardson PS, Sterling GM. Effects of
beta-adrenergicreceptor blockade on airway conductance and lung
volumein normal and asthmatic subjects. BMJ 1969;3:580-4.
25 Hidalgo HA, Wegmann MJ, Waring WW. Frequency spectraof normal
breath sounds in childhood. Chest 1991;100:999-1002.
26 Sanchez I, Pasterkamp H. Tracheal sound spectra dependon body
height. Am Rev Respir Dis 1993;148:1083-7.
27 Pasterkamp H, Kraman SS, DeFrain PD, Wodicka GR.Measurement
of respiratory signals. Comparison of sens-ors. Chest
1993;104:1518-25.
28 Wodicka GR, Kraman SS, Zenk GM, Pasterkamp H.Measurement of
respiratory acoustic signals. Effect ofmicrophone air cavity depth.
Chest 1994;106: 1140-4.
29 Kraman SS. Does the vesicular breath sound come onlyfrom the
lungs? Am Rev Respir Dis 1983;128:622-6.
30 Pasterkamp H, Fenton R, Tal A, Chernick V. Interferenceof
cardiovascular sounds with phonopneumography inchildren. Am Rev
Respir Dis 1985;131:61-4.
31 Gavriely N, Cugell D. Breath sounds methodology. Boca Ra-ton,
Florida: CRC Press Inc., 1995.
32 Kraman SS. Effects of lung volume and airflow on thefrequency
spectrum ofvesicular lung sounds. RespirPhysiol1986;66:1-9.
33 Mussel MJ, Nakazano Y, Miyamoto Y. Effect of air flowand flow
transducer on tracheal breath sounds. Med BiolEng Comput
1990;28:550-4.
34 O'Donnel DM, Kraman S. Vesicular lung sound amplitudemapping
by automated flow-gated phonopneumography.I Appl Physiol
1982;53:603-9.
35 Gavriely N, Palti Y, Alroy G. Spectral characteristics
ofnormal breath sounds. J Appl Physiol 198 1;50:307-14.
36 Baughman RP, Loudon RG. Quantitation of wheezing inacute
asthma. Chest 1984;86:718-22.
37 Gavriely N, Nissan M, Cugell DW, Rubin A-HE. Res-piratory
health screening using pulmonary function testsand lung sound
analysis. Eur Respir J 1994;7:35-42.
38 Kraman SS. Determination of the site of production
ofrespiratory sounds by subtraction phonopneumography.Am Rev Respir
Dis 1980;122:303-9.
39 Olson DE, Snyder B. Origin ofcentral airway breath sounds.Am
Rev Respir Dis 1983;127:A268.
40 Hardin JC, Patterson JL. Monitoring the state of the
humanairways by analysis of respiratory sound. Acta
Astronaut1979;6:1137-51.
1291