Applications of signal recognition algorithms to diagnosis and monitoring in chest medicine

Electronic stethoscopes capable of recording and digitally storing lung sounds have been available for several years. The clinical advantage of these devices is that recordings can be stored and replayed for comparison, either at a later time or to another physician. The next logical step is to offer the user an automatic, objective ´diagnosis´ of the clinical condition. This paper presents wavelet-based algorithms that have been developed to detect and quantify crackles and wheezes, associated with asthma, chronic obstructive pulmonary disease and lung fibrosis. However, signal processing is not the only challenge. Practical application for real clinical uses requires a robust and appropriate user platform and the definition of suitable uses, which will complement expert judgement. The potential for applications in chronic disease monitoring, medication regime development, emergency triage and routine screening is discussed. Issues regarding user acceptability, ergonomics and clinical validation are outlined, based on clinical feedback from users of existing commercial devices.