Classification of Respiratory Diseases Using Respiratory Sound Analysis R. K. Sawant PG Department of Computer Science, Sant Gadge Baba Amravati University, Amravati, India Email: ranjitsawant@yahoo.com A. A. Ghatol K. J. Education Institutes, Pune, India Abstract—Respiratory or lung sounds recorded on the chest can be used to identify different types of diseases. These sounds are attenuated by the thorax and thorax-microphone interface. In order to proper classification of respiratory diseases waveforms similar to the ones generated within the lungs must be recovered from the attenuated sounds. The equalization of crackle sounds recorded on the chest can be done for accurate classification of respiratory sounds. From an experiment an estimation of the channel attenuation was obtained according to which the equalization is applied. For that, multiple tones between 100 and 1200 Hz were applied to each subjects’ mouth where they were acquired. These tones were also recorded on the chest. The power ratio between the one measured on the chest and that measured at the mouth is used to calculate the attenuation of each tone. After obtaining the average attenuation curve a discrete- time equalizer was applied to crackles acquired from patients with congestive heart failure, fibrosis, and pneumonia. The equalization is used to modify the maximum frequency and two cycle duration indices measured from these crackles. The equalizer improves the extraction of features from the crackles sounds. Equalization of crackles can be used to better classify the different diseases. Index Terms—equalization of crackles, lung diseases, respiratory disease classification, respiratory diseases, respiratory sound analysis I. INTRODUCTION Respiratory sounds can be recorded with the help of devices having different technical specifications. The European Respiratory Society proposed the computerized respiratory sound analysis (CORSA) guidelines for research and clinical practice [1]. Still the characterization of respiratory sound is not accurate. The attenuation of the sounds traveling from the lungs to the thorax surface provides the crackles that can be best heard [2]. The lung sounds referred as crackles are useful for classifying cardiopulmonary diseases such as fibrosis, congestive heart failure, and pneumonia [3], [4]. The crackles are usually heard on the chest with a stethoscope during patient checkups; their identification depends on the experience and hearing perception of the physician Manuscript received August 1, 2014; revised November 21, 2014. [3]. The visual inspection of crackles recorded waveform reveals an initial fast-rising deflection followed by a short ringing duration [3], [4]. The crackle can be described as short, explosive, and transient. The quantitative characterization of crackles can be done for identification of various diseases. Various electronic systems are used to record these respiratory sounds. The diseases can be identified by the two cycle duration (2CD) index (time from the beginning of the initial deflection of a crackle to the point where the waveform of the crackle has completed two cycles) and the maximum frequency of crackles [3], [4]-[7]. The parameters that are measured from the crackles can get affected as attenuation path may modify the crackle waveform. Further information for assisting the diagnosis of the different diseases can be acquired from the crackles with characteristics closer to those generated within the lungs. The acquired sounds can be equalized by knowing the transmission channel to recover characteristics that were changed during their propagation through the path. This paper presents discrete-time equalization method for compensating sound attenuation measurements of the channel consisting of the thorax and the thorax interfaces. The importance of listening to and understanding respiratory sounds is evident from the iconic and symbolic usage of the stethoscope in modern medicine. The stethoscope was invented in 1821 by the French Physician, Laennec, upon the discovery that respiratory sound analysis aids in the diagnosis of pulmonary infections and diseases, such as acute bronchitis and pneumonia [8], [9]. Since 1821, stethoscopes have become the most common diagnostic tool by doctors in the twenty-first century [9], [10]. Despite its widespread use, however, analysis of respiratory sounds using stethoscopes is rudimentary at best and requires a degree of subjectivity from the physician [9]-[11]. Analysis of respiratory sounds using stethoscopes depends on the variable factors of the diagnosing physician’s experiences, hearing, and ability to recognize and differentiate patterns [10]. In addition, stethoscope data is not typically recordable, making long-term correlation of data difficult [9], [11]. All of these factors reduce the value stethoscopes bring to a world that increasingly demands quantitative measures of disease. 62 doi: 10.12720/ijsps.4.1.62-66 International Journal of Signal Processing Systems Vol. 4, No. 1, February 2016 ©2016 Int. J. Sig. Process. Syst.