Automatic Classification of Phonocardiogram
This paper presents a novel method of automatic classification of phonocardiogram. The noisy phonocardiogram is segmented to form noisy datasets of one heart cycle. The dataset is created for murmur and normal sounds. The datasets are LMS filtered to produce two new datasets to form the filtered datasets. The test sound which is a noisy phonocardiogram is taken, segmented and each segment of one cycle duration is checked for normal and murmur segments. The test sound is LMS filtered and re-segmented. The segments are checked for normal and murmur sounds and the results are compared with the previously obtained results.