Design and development of mobile application to characterize and classify lung sound based on Frequential Analysis and Wavelet Transform

Resumen

Throughout the last decades, the lung auscultation has been used as one of themost popular procedures to evaluate the state of the respiratory airways with relative confidence based on the interpretation of the audio signals given by the stethoscope, a medical tool that has not changed significantly over the last years. The recent development of digital stethoscopes provides them with several capabilities that enhance auscultation interpretability with novel features like audio amplification, noise rejection and filtering. Even better is a digital stethoscope,which gives the chance of computerized signal analysis on lung sounds, which is the motivation of this work. Based on the lung sound characterization performed by Laenec [1], we consider that it is possible for a computer-based system to detect the elemental features ofa lung sound by its frequency contents and presence of discontinuities in order toclassify them into its basic types: ronchi, wheezes, and stridor. Using both traditional signal analysis tools, such as Fourier Transform, as well asnovel ones like Wavelet Transform, this work proposes to implement a mobile application for the Android OS along with a digital stethoscope that, beyond just classifying the content of a lung sound signal, provides a graphical representationof its characteristics like the frequency on discontinuous sounds found in the sound that might be helpful to make the diagnosis about the respiratory state of a patient easier.