JAMF-based representation for computational lung sound analysis

Obstructive lung disease is a category of respiratory disease characterized by an obstruction of the airflow, inflamed and/or easily collapsible airways, and mucus retention. Asthma, bronchiectasis, and chronic obstructive pulmonary disease (COPD) are the three main types of obstructive lung diseases. This type of diseases causes adventitious lung sounds which can be heard through lung auscultation. Physiotherapists can detect these sounds and use them to adapt their treatment. Physicians the world over are starting to rely more on computers than ever before. For lung auscultation data, this is mostly limited to representing the sound as a time-pressure graph or a spectrogram, while still doing the actual analysis themselves. Researchers however, have already shown the strength of computational lung sound analysis. In this paper, we propose the Joint Acoustic- and Modulation Frequency (JAMF) representation as a signal-processing technique for a lung-sound sensor which create a visually clean, simple and yet powerful representation which allows physicians to determine possible problems at first glance, with future possibilities for easy automatic analysis of the lung sounds.