Respiratory sounds - Advances beyond the stethoscope

Lung sounds have been valuable indicators of respiratory health and disease since ancient times. Laennec's stethoscope raised their diagnostic significance but other methods, more sensitive and specific for respiratory assessment, have largely replaced auscultation in clinical pulmonary diagnosis. We are now witnessing the next phase in the evolution of pulmonary assessment by acoustical means. Although the complexity of the respiratory system has slowed the formulation of a comprehensive model of chest and lung acoustics, there have been major advances in understanding lung sounds during the last decade. More powerful yet smaller computers have made digital respiratory sound analysis possible in ambulatory care and at the bedside. The next years will likely bring about an integration of respiratory sound analyzers with more established computer-based spirometry. The most promising areas for respiratory acoustical measurements are in upper airway diagnosis and monitoring, e.g., in patients with obstructive sleep apnea, in the assessment of lower airway dynamics, e.g., in patients with asthma or bronchiolitis, and in the assessment of regional ventilation. It may soon be possible to determine the site of upper airway obstruction by the analysis of respiratory sounds and to follow the effect of therapy, e.g., the application of continuous positive airway pressure, by acoustical means. Spirometry will remain the standard for assessing lower airway flow obstruction, but lung- sound analysis can extend the assessment to younger patients. Objective characterization of wheezing should improve the epidemiologic understanding of acute and chronic lower airway obstruction, especially in children. Monitoring of regional ventilation by chest surface acoustical topography may now be possible with faster computers that allow the simultaneous processing of sounds from multiple recording sites. The multisite recording of respiratory sounds and of passively transmitted sounds could be particularly useful in critical care, e.g., to monitor regional ventilation and lung water content in intubated patients. Certain technical challenges need to be resolved before lung sound analysis can enter into routine clinical practice. In particular, a robust and inexpensive sensor for lung sound recording that is relatively immune to ambient noise has yet to be developed. Furthermore, the automated recognition and rejection of artifacts as well as the separate processing of adventitious and basic respiratory sounds need to be refined. However, advances in digital sound processing have already translated into enhancements of the traditional stethoscope, e.g., the use of active noise cancellation in high noise environments (195, 196). Also, the teaching of chest auscultation to medical students has been enhanced by computer-aided instruction (197-199). Lung sounds of high fidelity can be transferred via telecommunication, as interested readers with access to the Internet can verify. The sound repertoire of the lung may indeed be limited when heard through a stethoscope, but it clearly exhibits a much wider range of information content when digitally analyzed. Computer analysis is now reaching beyond the capabilities of the human ear, e.g., to resolve changes in respiratory sounds during narrowing of the intrathoracic- or extrathoracic airways. With the disappearance of auscultation as the standard to judge the clinical significance of acoustical findings, it becomes even more important to integrate lung sound analysis and traditional measurements of respiratory mechanics. One should keep in mind, however, that voice recognition of continuous speech, an easy task for the human listener, is still not possible on standard computers after decades of research and substantial investment from industry. Thus, one should not expect that computer-based lung sound analyzers will replace the stethoscope-bearing clinician anytime soon, but they will expand the noninvasive diagnostic capabilities in respiratory medicine.