International Journal of Soft Computing and Engineering (IJSCE) ISSN: 2231-2307, Volume-2, Issue-2, May 2012 41 Abstract—In recent days, fractal compression has gained a wide popularity due to its inherent features and efficiency in compressing data. In the present communication, fractal compression technique has been applied on heart sound signals for effective compression. Fractal heart sound coding based on the representation of a heart sound signal (1D or vector) by a contractive transform, on the sound data, for which the fixed point (reconstructed heart sound) is close to the original heart sound. The work is intended to provide an approach on this process by introducing the idea of multi-scale Domain pool classification using Variance Fractal Dimension (VFD) based on complexity of the heart sound data. A pre-processing analysis of the heart sound data by VFD to identify the complexity of each sound data samples block for classification has been undertaken. The performance result of the present work has focused in terms of good fidelity signal reconstruction versus encoding time and amount of compression. Index Terms— Phonocardiogram, Fractal Compression, Variance Fractal Dimension, Domain Classification I. INTRODUCTION The heart is considered to be an analogous of electromechanical system, and as such its health can be characterized by both its electrical behavior (reflected in ECG) and its mechanical functioning. The mechanical functioning of heart generates sounds. Such sounds can be detected by using stethoscope, echocardiography, catheterization, or computerized analysis of the sounds emitted by the heart. Normal hearts generate at least two heart sounds – the first heart sound (S1) and the second heart sound (S2). S 1 and S 2 are produced due to the closure of the mitral and aortic valves at the beginning and end of the systolic process [1] respectively [2]. Depending upon an individual’s health, additional S 3 and S 4 sounds may occur (Fig. 1). S 3 is produced due to the filling of the ventricle in the early stage of diastole. If blood enters in a relatively “non -compliant” ventricle late in diastole, it generates S 4 . The four heart sounds are illustrated in Figure 1. The presence of an S 3 , for example, is a strong indication of Congestive Heart Failure (CHF) [3]. Historically, heart sounds have been observed through auscultation of the heart wherein the cardiologist listens to the heart through a stethoscope. . Manuscript received on April 14, 2012. Prof. M. K. Kowar, Director, BIT Durg, India, (e-mail:mkkowar@gmail.com), Amit Biswas, Associate Professor, BIT, Durg (e-mail: biswas_amit@rediffmail.com), Ravindra Manohar Potdar , Sr. Associate Professor BIT Durg, India (e-mail: ravi_potdar@rediffmail.com), Mayur Amtey, BIT Raipur, India (e-mail: amtey.mayur@gmail.com). Forming a diagnosis based on sounds heard through the stethoscope is subjective, varies from person to person and also depends on a cardiologist’s past experience. In recent years, there has been an effort automate the heart sound diagnosis operation in order to remove subjectivity and to make heart sound auscultation more quantitative Fig. 1 Heart Sound Signals with S 1 , S 2 , S 3 and S 4 Heart sounds analysis can provide lots of information about heart condition whether it is normal or abnormal. Heart sounds signals are time-varying signals where they exhibit some degree of non-stationary. Compression of the heart sounds (or phonocardiogram) is very convenient to reduce bandwidth in tele-diagnosis systems that aid the physician in the evaluation of the cardiovascular state. Data compression plays an important role in computational practices especially in signal analysis and transmission due to heavy memory requirement for storing and retrieving such complex data. The goal of signal compression is to reduce storage space and to save transmission complexities. The availability of the sound file and the variance fractal dimension technique helped in the development of the software and thus allows the biomedical environment to switch from old system to new system. The fractal compression is mostly based on fractal system’s ability to approximate discontinuous functions, where audio signals exhibits greater smoothness. The mechanical function of heart, blood flow and valve movements produce the heart sound during contraction and relaxation phases of the heart. The heart sound signal is important clinical information in the diagnostic process of heart malfunctions [4]. The objective of this work is to compress the sound file (heart sound) using the technique of variance fractal dimension [5, 6] and the major motive of this work is to develop a software through which heart sound data can be compressed efficiently to make it suitable for transmission over the networks. The techniques for recording and analysis have been changing as new electronic devices and signal processing techniques have become available. Fractal compression can be considered as one of this new form [5]. A more detailed description of fractal encoding Multi-Scale Domain Classification Based Heart Sound Compression Ravi M. Potdar, Manoj K. Kowar, Amit Biswas, Mayur Amtey