Practice of Cardiac Auscultation: Clinical perspectives and its implications on computer aided diagnosis

Abstract

Study of the disease demographics in human population indicates that cardiac ailments are the primary cause of premature death, and a need for emergent technologies is felt to address the rising trend. However, development of automated heart sound analysis system and its usage at the grassroot levels for cardiac pre-screening have been hindered by the lack proper understanding of the intrinsic characteristics of cardiac auscultation. In this article we present an investigatory report based on a nationwide survey conducted on the practice of cardiac auscultation for determination of its effectiveness in diagnosis. The aims are to achieve better validation of heart sound acquisition methods and use the clinical feedback from cardiologists for improvements in the classification of the cardiac abnormalities. Results obtained from six different classifiers used in the study are illustrated, which show a remarkable specificity using an improvised classification hierarchy, derived based on clinical recommendations. The study addresses the needs for better understanding of the relevancy of heart sound signal parameters, recording transducers, recording location and the inherent complexity associated in interpretation of heart sounds, specially in noisy environments of out-patient departments and primary healthcare centers. Further, the inter-relationship between heart sound and other advance medical imaging modalities, and the need for more focused training in cardiac auscultation among the medical and paramedical staff is investigated.