Effect of Early and Late Fusion on Heart Sound Classification

Abstract

Heart sounds; together with important clues in the evaluation of heart diseases, is a guide for advanced diagnostic examination and is used for the early diagnosis of cardiovascular diseases. Auscultation of heart sounds is an important part of the physical examination and allows understanding of many pathological heart diseases such as heart valve diseases, arrhythmic heart beats, heart failure. The correct detection of Fundamental Heart Sounds (FHS) in phonocardiography (PCG) provides the presence of systolic and diastolic regions and the understanding of pathological conditions. In this study; the leading classification approaches of Computing in Cardiology (CinC) 2016 challenge have been examined with different methods and it has been found that it contributes to performance according to single use. Purpose of the study; PCGs are correctly classified as normal or abnormal by the classifier stacking algorithm. It is aimed to make access to expert diagnosis easier by the proposed method.