Relevance and redundancy analysis of PCG signals in the detection of heart murmurs by ANOVA

One of the most important operating problems of the human heart valves is the murmur, which can be detected by auscultation employing an stethoscope. In order to design an aided diagnosis system for detecting the mentioned pathology, it is required to establish the discriminat capacity of the calculated features, which are used during the decision making stage. In this work, the analysis of variance (ANOVA) is employed to select the relevant features from an original feature set. The analysis allows to reduce the feature space dimension without decrementing the classification performance of the automatic system, compared to the classification performance obtained when the full feature set is employed.