Vesicular lung sound amplitude mapping by automated flow-gated

Vesicular lung sound amplitude mapping by automated flow-gated phonopneumography DENNIS M. O’DONNELL AND STEVE S. KRAMAN Department of Medicine, Section of Pulmonary Medicine, Veterans Administration and Division of Pulmonary Medicine, University of Kentucky Medical Center, Lexington, Kentucky 40511

Medical

Center,

. O'DONNELL,DENNIS M., AND STEVE S.KRAMAN. Vesicular lung sound amplitude mapping by automated flow-gated phonopneumography. J. Appl. Physiol.: Respirat. Environ. Exercise Physiol. 53(3): 603-609, 1982.-A recently developed automated apparatus capable of determining vesicular lung sound amplitude rapidly and accurately was used to construct detailed inspiratory vesicular sound amplitude maps in eight healthy male subjects to determine the normal amplitude patterns on the chest wall. The sounds were recorded in 2-cm steps along the following lines bilaterally: A, vertically, clavicle to abdomen, 6 cm from the sternal border; B, vertically, from the level of T1 to the lung bases, 6 cm from the spine; and C, horizontally, from the sternal border to the spine at the level of the nipple. Sound amplitude was measured at an airflow rate of 1.3 l/s. The resulting amplitude maps revealed considerable intra- and intersubject variation with frequent amplitude heterophony. Th patterns for the subjects as a group were as follows: series A, amplitude decreasing with distance from the clavicle; series B, amplitude increasing with distance from Ti with a peak at the bases; and series C, approximately equal amplitude at all positions. The findings in series B and C are, in general, consistent with an explanation of ventilation following hydrostatic gradients. The series A pattern and the intersubject variability in amplitude are inconsistent with this explanation and suggest that the inspiratory vesicular sound amplitude is not simply a result of ventilation distribution but involves other as yet undefined factors. breath sounds; ventilation;

amplitude

heterophony

a technique of visually displaying the amplitude and/or frequency of lung sounds, has previously been used by investigators to study regional ventilation and to evaluate the amplitude and frequency characteristics of lung sounds (1, 3, 9, 11). One inadequately studied characteristic of the vesicular lung sound is the relative amplitude of this sound at various locations on the chest wall. To our knowledge, the most detailed vesicular sound amplitude maps have been constructed by Ploy-Song-Sang et al. (9, 10) and involved only four sites on the right anterior chest wall. Other investigators (7, ll), when studying comparative lung sound amplitudes, have included even fewer sites. The major reason that vesicular sound amplitude patterns have been so little studied lies with the difficulty in comparing one breath with another when the peak flow rates are not identical, as is almost invariably the case; also, no automated measurement apparatus has yet been developed, PHONOPNEUMOGRAPHY,

0161-7567/82/0000-0000$1.25

Copyright

0 1982 the American

Physiological

so that measurements must be done manually, thus limiting the number of breaths that may be conveniently measured. The simultaneous display of an airflow signal along with the lung sound signal allows measurement of the latter at comparable flow rates but is so time consuming that few breaths can be so evaluated. An attractive alternative technique, using a large array of microphones attached to many points on the chest wall and measuring each breath simultaneously, could be employed but becomes more complex and expensive as more microphones and strip-chart recorder channels are added. Detailed knowledge of lung sound amplitude patterns is important if intelligent interpretation of amplitude differences within and between subjects is to be accomplished. We have recently developed an automated technique that permits rapid, accurate lung sound amplitude determinations. We describe herein its use in mapping the amplitude patterns of the vesicular breath sound in normal subjects. METHODS

Eight healthy medical personnel between the ages of 24 and 36 yr volunteered to participate in this study after informed consent was obtained. All subjects were lifetime nonsmokers and had normal pulmonary function as determined by history, physical examination, and spirometry (8). The subjects received training in breathing only insofar as was necessary to allow them to achieve an even increase in the rate of flow on inspiration. The recordings were performed with the subjects standing while they breathed through a pneumotachograph and watched a graphic display of the airflow rate at the same instant on an oscilloscope screen. The subjects were instructed to inspire from resting lung volume and to attempt to reach a peak inspiratory flow of approximately 2 l/s. They then exhaled normally. The method of sound amplitude analysis has been previously reported (6); briefly, (Fig. 1) a condenser microphone with chest piece of 15 mm diameter and 9 mm depth was placed on the chest at each point of interest. The signal was pass-band limited to a center frequency of 400 Hz (6-dB points, 150 and 700 Hz) to attenuate extraneous sounds, passed through an oscilloscope for visual monitoring, various stages of amplification, and was then sent to a multichannel analog-to-digital converter and computer. The output signal of a Fleisch Society

603

604

D. M.

O’DONNELL

AND

S. S. KRAMAN

PNEUMOTACHOGRAPH 6 AMPLIFIER

osCILLOSCOPE COMPUTER

STEREO

AMP.

TAPE RECORDER

MICROPHONES

FIG. 1. Schematic diagram of sound recording offset voltages present at output of oscilloscope.

equipment

used in this study.

pneumotachograph (HP 2107 and 47304A) was also fed to the computer. The computer measured the average amplitude of the vesicular sound within a 0.1-s window once an inspiratory airflow rate of 1.3 l/s was achieved. Three breaths were recorded at each site and the average amplitude determined. Previous studies had shown that theA consistency of amplitude measurement by this method yielded a standard deviation of t5%. In practice, two identical microphones and audio channels were used to allow the right and left lungs to be studied simultaneously. While each subject was studied without varying the amplifier gain of the test system, some subjects required more amplification than others, so that absolute amplitude comparisons between subjects cannot be made. Lung sounds were recorded along the following lines bilaterally: anteriorly, in a vertical hne starting immediately below the clavical 6 cm from the midsternum and proceeding toward the abdomen in 2-cm increments until inspiratory sounds could no longer be heard (Fig. 2); posteriorly, along a vertical line 6 cm from the spine in 2cm increments beginning at a level even with the first thoracic vertebra and proceeding to the lung base (Fig. 3); laterally, beginning at the sternal edge at the level of the nipple in 2-cm steps following h.orizontal line around to the spine (Fig. 4). This resulted in the total of six series of recordings on the chest wall of each subject. The mean amplitude of the three inspiratory sounds at each position was plotted against that position in each series. One of the eight subjects (5) due to the finding of an unusual pattern of sound amplitude when compared with the other seven subjects, was further studied to determine whether chest wall sound transmission, distribution of ventilation, or anatomic abnormalities were responsible for these differences. These additional studies involved the comparison of expiratory as well as inspiratory amplitude maps posteriorly, a study of the posterior amplitude in two dimensions bilaterally (vertical and horizontal), the performing of a radioactive xenon ventilation lung scan, a flexible fiber-optic bronchoscopic

Stereo

amplifier

B is used to isolate

the computer

input

from

DC

study, and bronchography of the left lower lobe. This subject was one of the authors (SK) so that informed consent was implied. RESULTS The results of the series recorded along the anterior vertical axis are shown in Fig. 2. There was a remarkable similarity in the trend for all subjects. In general, the amplitudes were much greater at the apex and steadily decreased as the microphones were moved towards the base. The average trend in all subjects supported this impression and also tended to dampen out many of the individual large amplitude variations that were present in most subjects. These individual variations both in this series and the subsequent series were found to be repeatable in three subjects subsequently restudied weeks to months later. Along the posterior vertical series (Fig. 3) the trend was opposite from that of the anterior series. The a.mplitude of sound in all individual subjects tended to be greater at the base than at the apex. There was more individual variation from subject to subject in this series than the anterior series. In the lateral horizontal series, there tended to be much point-to-point intrasubject variability as in the other series (Fig. 4). When averaged, however, the various peaks in individuals were dampened and a relatively uniform average amplitude pattern was obtained bilaterally. Of all the subjects, subject 5 was found to be unusual because the average amplitude of the sounds recorded over the left lung in all three series was approximately twice that recorded over the right. In addition, there were several unusually high-amplitude peaks in the three series but especially the vertical posterior on the left. It was felt that if the reason for these peaks could be determined in this subject, it might provide insight into the reason for point-to-point variation in the other subjects. A three-dimensional amplitude contour map of the

VESICULAR

LUNG

SOUND

605

MAPPING SOUNO PRESSURE

(VOLTS)

JEW

FIG.

terior

2. Vesicular lung sound amplitude vertical positions depicted. Average

patterns patterns

subsequent figures all points recorded

recorded from an(AVE) in this and SOW0 /c

0

R

DISTANCE

CM

FROM Tl

L

4 0 CM

40

0

0.5

PRESSURE 1.0

1.5

were constructed at each position

by calculating in all subjects.

mean

amplitude

of

(VOLTS) -

3.0

@ 2.5

rL 1

8

SUBJECT

AVE.

i ‘L i-. a_-nt 1'- 7 : 0' I, :: ..' ' :\ i 8 li-

FIG.

overlying

3. Vesicular scapula.

lung

sound

amplitude

patterns

recorded

from

posterior

area of maximum amplitude over the posterior left lung and its homologous area on the right was constructed by a method very similar to that used for the two-dimen-

vertical

positions

depicted.

No recording

was made

from

an area

sional maps previously described. This is shown in Fig. 5, A and B. These maps confirmed that the left lung was louder than the right and also showed the extent of the

606

D. M. SOUND

PRESSURE

O’DONNELL

AND

S. S. KRAMAN

(VOLTS)

SUBJECT

FIG.

4, Vesicular

16 CM

lung sound

amplitude

)

patterns

recorded

horizontal

positions

depicted

(sternum

to spine).

16 CM

1

FIG. 5. Three-dimensional amplitude contour maps of depicted. Each of the 80 points on each side represents mean of 3 inspirations recorded at that point. Distance between cm vertically and 2 cm horizontally. Clearly seen in these

from

the areas amplitude points is 3 maps are

major amplitude peak that had been seen in the onedimensional map. Since previous studies (7,9) had linked the amplitude of the inspiratory vesicular sound with

relative loudness of left loud area in left midlung present over scapulae.

side (A) compared with right (B), unusually zone described in text, and sound attenuation

distribution of ventilation, especially in the upright posture, a radioactive xenon ventilation scan of the posterior aspect of the chest was performed. This revealed gener-

VESICULAR

LUNG

SOUND

607

MAPPING

ally even ventilation with scintillation counts indicating that 55% of the ventilation was going to the right lung and 45% to the left. However, in approximately the same area as the large left posterior amplitude peak, there was a small area of decreased ventilation. Scintillation counts of this area compared with the homologous area on the right showed 70% ventilation on the right, 30% on the left. Thus there was decreased ventilation in that small area of greatest sound intensity. Washin and washout rates of the xenon were equal. To determine whether chest wall thickness was responsible for these amplitude differences, a two-dimensional vertical posterior map of the expiratory component of the vesicular lung sound was constructed by a method identical to the inspiratory map. This showed equal amplitude patterns with amplitude increasing as the base was approached and the total absence of the large amplitude peak in the middle of the posterior chest that had appeared in the inspiratory sound map. A sound intensity vs. airflow phonopneumogram of inspiration and expiration recorded from the left and right midposterior lung zones is shown in Fig. 6 and reveals the great disparity in inspiratory amplitudes and similarity in expiratory amplitudes between the left and right sides. Because chest wall thickness would be expected to affect inspiratory and expiratory sounds equally, this finding suggested that chest wall transmission characteristics were not responsible for the observed difference in amplitude. The subject then underwent flexible fiber-optic bronchoscopy, which revealed normal endobronchial anat-

omy. A catheter was placed within the trachea and a left lower lobe bronchogram was performed using oily dionosil. This bronchogram was interpreted by two chest radiologists and two pulmonary physicians and determined to be within normal limits. DISCUSSION

The present study was undertaken to document the relative intensity patterns of lung sound on the chest wall. To our knowledge, the most detailed amplitude map in the past (9, 10) included only four positions on the chest wall. Having further refined the technique of phonopneumography by using computerized analysis, we were able to record more than 50 positions on each side of the chest in eight subjects within a relatively short period of time. Formerly, the technique of phonopneumography has been tedious and the analysis of data very time consuming. These previous techniques also suffered in that it was difficult to accurately control airflow, thus the cumbersome indices of sound intensity (see below) were developed in an attempt to compensate for this. Our technique of computerized flow-gated phonopneumography overcomes many of these problems. In 1970, Leblanc et al. (7) used a phonocardiograph and microphone to record lung sounds from two positions on the chest wall: a) apex of left lung anteriorly and b) 3-4 cm below the scapula on the left posteriorly. They found that the intensity of lung sounds recorded on the chest wall varied with the following. 1) Lung volume-at low lung volumes the sound intensity was highest at the apex; at high lung volumes, the sound intensity was

EX

FIG. 6. Phonopneumogram of inspiratory (IN) and expiratory (EX) breath sounds recorded over left (L) and right (R) posterior midlung zones. Point 0 on abscissa represents zero flow and is starting point of recording. Inspiration and expiration cause defection to left and right, respectively.

608

greatest at the base posteriorly. 2) Flow-the sound intensity increased as the airflow at the mouth increased. 3) Body position- the sounds were loudest over the portion of the lung that was dependent. These investigators felt that the regional intensity of breath sounds correlated well with the regional distribution of ventilation as determined by radioactive xenon studies. Thus they surmised that the intensity of the breath sounds heard on the chest wall were a good indicator of underlying ventilation. Ploy-Song-Sang et al. (9) further refined the technique of phonopneumography by developing indices that they felt would compensate for variability in airflow and transmission characteristics of the lung. Lung sounds were recorded from four positions on the right anterior chest wall along a vertical line at 5, 10, 15, and 20 cm from the apex while the subject inspired from residual volume to total lung capacity. Because they were unable to precisely control the airflow rate, an index of breath sound pressure was developed. This index was simply the quotient of the sound intensity recorded at any site divided by the sound intensity at the 5-cm reference position. It was felt the changes in airflow would produce similar degrees of changes in lung sound amplitude at all positions on the chest wall. It was reasoned that this sound index would compensate for variations in airflow at the mouth. Transmission characteristics of the lung were also studied by introducing white noise at the mouth and recording the amplitude of this at the same four positions on the chest wall and at the same lung volumes (lo-90% of vital capacity). A second index was developed to compensate for differences in transmission of sound to the chest wall. This was termed the compensated breath sound, whereas the sound actually recorded which was equivalent to the sound heard with a stethoscope was termed the uncompensated breath sound. The authors correlated the sound patterns that they found with the distribution of ventilation as determined by radioactive xenon ventilation scans. They found that the uncompensated breath sound intensities varied from region to region but were fair indicators of regional ventilation only in the upright position. The compensated breaths sound was found to be a more accurate indicator of regional ventilation both in the upright and supine position. In a subsequent investigation Ploy-Song-Sang and colleagues (10) applied their phonopneumographic techniques to patients with emphysema and compared the results with those of their studies in normal subjects. Much more variability in sound amplitude was found in the emphysematous patients than in the normal subjects. There were unexplained areas of increased and decreased sound intensity, and sound transmission was erratic. These findings were felt to be secondary to deranged regional ventilation. The present study was undertaken to document the relative intensity patterns of lung sounds on the chest wall. The amplitude of the sound within a 0.1-s window beginning at a mouth airflow rate of 1.3 l/s was chosen after many trials at various other window widths, and initial flows provided less consistent results. The short recording period enabled us to minimize the influence of

D. M. O’DONNELL

AND S. S. KRAMAN

artifactual sounds during the span of time in which the recordings were performed. When times much in excess of 0.1 s were used, there was an increase in breath-tobreath variation due to uneven patterns of airflow. When a measurement window of much less than 0.1 s was used, increases in variation again occurred and were due to measurements of insufficient points to obtain a representative sample. The trigger airflow of 1.3 l/s was chosen because most subjects reached or surpassed this when asked to “breathe deeply.” The patterns of amplitude obtained in this study were somewhat surprising. In the posterior vertical and sternum to spine horizontal series, the relative amplitudes generally followed those that would be expected from the presumed regional ventilation of the underlying areas. Along the anterior series, the amplitude tended to be higher at the apex and lower at the base, which is contrary to what one would expect if the amplitude varied solely with the regional ventilation. Indeed, it would seem that the loudest point on the chest wall is the area immediately below the clavical anteriorly. This pattern was observed in all subjects tested as seen in Fig. 2. While it may be reasoned that this sound is loud due to transmission from the larynx, one recent study (5) suggests that the laryngeal noise is not radiated to any part of the chest wall including this area. Subject 5 was found to have several areas on the chest wall with very high lung sound amplitudes that were reproducible when recordings were repeated at several later dates, months apart. The ventilation scan surprisingly showed decreased ventilation in the left posterior midlung zone, which was the loudest of all posterior sites. In addition, the expiratory map showed no such amplitude peak and thus strongly suggests that sound transmission through the chest wall was not involved, since were this the case, expiratory amplitude should have been similarly affected. In a further attempt to determine whether chest wall thickness had any effect at all on the sound amplitude patterns that we observed, the average patterns found in our subjects were compared with the approximate chest wall thickness of a representative thorax determined by measurements derived from an atlas of computerized axial tomography of the chest (2). The results of this comparison can only be suggestive, since the subject studied in this atlas was obviously not one of the subjects studied by us. However, it can be assumed that this subject is fairly representative of normal anatomy. This comparison is displayed in Fig. 7. From the two vertical series, it would appear that chest wall thickness may be related to the general patterns found in our study. Indeed, it would seem that the thinnest part of the chest wall is the apical anterior portion. However, at the level of the horizontal series, there seems to be considerable variation in chest wall thickness as one proceeds from the sternum to the spine. No corresponding consistent amplitude changes were found in our study that could be attributed to these differences. While still inadequately studied, it would seem that chest wall thickness is not a major determinant of the amplitude of the vesicular sound recorded on the surface of the chest. Due to the large amount of radiation attendent to a computerized tomographic examination of

VESICULAR

LUNG

SOUND

Ape#Ix.CHESTUAlL 11

SOUND

7HfclaEss 4 A

0 1

I

8

4

2

\I0

6

\

o\ \

\

\

1

I

1

1.0

1.5

2.0

T

0 CM 40 H

/2.5 /

/ 0 / /

I

/

\I

R CM

L

1 0.5

i \

\

I 10

I \

\

(VOLTS)

I \

\

PRHSURE

I

\

\

609

MAPPING

/

/

/

b

ANTERIOR

VERTICAL FIG. 7. Average inspiratory vesicular lung sound amplitude maps of left and right series in our 8 subjects compared with approximate chest wall thickness derived from measurements of a representative computerized axial tomographic study.

0

R

CM 40

POSTERIOR

0

L

VERTICAL

CM CM 40

/;i

0

R

CM CM

60 1

HORIZONTAL

the chest, it was not deemed ethical to subject our subjects to this procedure. In conclusion, we find that there is considerable interand intrasubject variability in amplitude of the inspiratory vesicular sound heard on the chest wall, and that this variability is due in large part to factors other than distribution of ventilation and chest wall thickness. We believe that these other factors involve the site of production of these sounds and their transmission through the airways and lung tissue. These factors will have to be studied in much greater detail in the future to determine

their real contribution to the distribution of amplitude of vesicular sounds. The authors acknowledge the helpful comments of Dr. Fred Zechman and the technical assistance of Laurence Kojan Ong. This study was supported by the Veterans Administration Medical Center. S. S. Kraman is supported by National Heart, Lung, and Blood Institute Grant HL-26334-OlSB. Address for reprint requests: S. S. Kraman, V. A. Medical Center, CDD, lllH, Lexington, KY 40511. Received

31 August

1981; accepted

in final

form

20 April

1982.

REFERENCES E. F., R. C. KORY, AND G. L. SNIDER. PhonbpneumogAm. Rev. Respir. Dis. 107: 449-455, 1973. GAMBARELLI, J., G. GUERINEL, L. CHEVROT, AND M. MATTER. Computerized AxiaZ Tomography. New York: Springer-Verlag, 1977, p. 116-161. GAVRIELY, N., Y. PALTI, AND G. ALROY. Spectral characteristics of normal breath sounds. J. Appl. Physiol.: Respirat. Environ. Exercise Physiol. 50: 307-314, 1981. KANEKO, K., J. MILIC-EMILI, M. B. DOLOVICH, A. DAWSON, AND D. V. BATES. Regional distribution of ventilation and perfusion as a function of body position. J. Appl. PhysioZ. 21: 767-777, 1966. KRAMAN, S. Does laryngeal noise contribute to the vesicular lung sound? Am. Rev. Respir. Dis. 124: 292-294, 1981. KRAMAN, S. Rapid lung sound amplitude measurement by automated flow-gated phonopneumography. Proceedings of the Third International Symposium on Computers in Critical Care and

1. BANASZAK,

raphy.

2.

3.

4.

5.

6.

Pulmonary Medicine. In press. 7. LEBLANC, P., P. T. MACKLEM, AND W. R. D. Ross. Breath sounds and distribution of pulmonary ventilation. Am. Rev. Respir. Dis. 102: 10-16, 1970. 8. MORRIS, J. A., A. KOSKI, AND L. C. JOHNSON. Spirometric standards for healthy non-smoking adults. Am. Rev. Respir. Dis. 103: 57-67, 1971. 9. PLOY-SONG-SANG, Y., R. R. MARTIN, W. R. D. Ross, R. G. LOUDON, AND P. T. MACKLEM. Breath sounds and regional ventilation. Am. Rev. Respir. Dis. 116: 187-199, 1977. 10. PLOY-SONG-SANG, Y., J. A. P. PARE, AND P. T. MACKLEM. Lung sounds in patients with emphysema. Am. Rev. Respir. Dis. 124: 4549, 1981. 11. WOOTEN, F. T., W. W. WARING, M. J. WAGMANN, W. E. ANDERSON, AND J. D. CONLEY. Method for respiratory sound analysis. Med. Instrum. Baltimore 12: 254-257, 1978.