Improved Detection of Lung Fluid With Standardized Acoustic Stimulation of the Chest.

Adam Rao,Samuel Zetumer,Simon Chu,Shuvo Roy,Neil Batlivala

doi:10.1109/jtehm.2018.2863366

Abstract

Accumulation of excess air and water in the lungs leads to breakdown of respiratory function and is a common cause of patient hospitalization. Compact and non-invasive methods to detect the changes in lung fluid accumulation can allow physicians to assess patients’ respiratory conditions. In this paper, an acoustic transducer and a digital stethoscope system are proposed as a targeted solution for this clinical need. Alterations in the structure of the lungs lead to measurable changes which can be used to assess lung pathology. We standardize this procedure by sending a controlled signal through the lungs of six healthy subjects and six patients with lung disease. We extract mel-frequency cepstral coefficients and spectroid audio features, commonly used in classification for music retrieval, to characterize subjects as healthy or diseased. Using the n}{}Kn-nearest neighbors algorithm, we demonstrate 91.7% accuracy in distinguishing between healthy subjects and patients with lung pathology.

Highlights

Respiratory diseases are a leading cause of death worldwide [1]
In this analysis we demonstrated that the presence of structural lung disease leads to detectable frequency differences using an principal component analysis (PCA) analysis of the frequencies from an FFT of the recorded signal
K-nearest neighbors (KNN) clustering of standard melfrequency cepstral coefficients (MFCC) and spectroid audio features extracted from the recordings resulted in successful segmentation of healthy and pathological cases

Summary

Introduction

Respiratory diseases are a leading cause of death worldwide [1]. Shortness of breath, or dyspnea, is a common chief complaint among patients presenting to the hospital, accounting for 3.7 million visits to emergency departments in the US in 2011 [2]. Exam results often suffer from a high degree of variability and a low interobserver agreement [6] This qualitative physical exam has led to the introduction of adjunctive imaging modalities to assist in the diagnosis of respiratory disease. Lung sounds are heard over the chest during inspiration and expiration They are non-stationary and non-linear signals, requiring a combined time and frequency approach for accurate analysis [15]. Previous work utilizing KNN and ANN classification to distinguish between healthy and pathological lung sounds reported values ranging from 69.7–92.4% for accurate classification [12], [14], [17], [18]. The usage of breath sound analysis shows potential for accurate classification, the large range in accuracy reported in prior work motivates the need for a standardized approach. Changes observed in recorded breath sounds could be a result of both the differences in structure of the system or the result of intersubject and intrasubject variability between breath cycles

Objectives

Methods

Findings

Conclusion