A Novel Method for Automatic Identification of Breathing State

Jinglong Niu,Joseph M Reinhardt,Weiqing Xu,Shuai Ren,Yan Shi,Zujin Luo,Wei Gao,Maolin Cai

doi:10.1038/s41598-018-36454-5

Abstract

Sputum deposition blocks the airways of patients and leads to blood oxygen desaturation. Medical staff must periodically check the breathing state of intubated patients. This process increases staff workload. In this paper, we describe a system designed to acquire respiratory sounds from intubated subjects, extract the audio features, and classify these sounds to detect the presence of sputum. Our method uses 13 features extracted from the time-frequency spectrum of the respiratory sounds. To test our system, 220 respiratory sound samples were collected. Half of the samples were collected from patients with sputum present, and the remainder were collected from patients with no sputum present. Testing was performed based on ten-fold cross-validation. In the ten-fold cross-validation experiment, the logistic classifier identified breath sounds with sputum present with a sensitivity of 93.36% and a specificity of 93.36%. The feature extraction and classification methods are useful and reliable for sputum detection. This approach differs from waveform research and can provide a better visualization of sputum conditions. The proposed system can be used in the ICU to inform medical staff when sputum is present in a patient’s trachea.

Highlights

Patients supported by mechanical ventilation are intubated with an endotracheal tube to allow for gas transport to and from the lungs
We describe a new waterproof sound sensor that can be connected to an existing endotracheal tube and permits the acquisition of high-quality respiratory sound data
This paper presents a complete system for automatic sputum detection from respiratory sounds and can be part of a method to evaluate patients’ breath state

Summary

Introduction

Patients supported by mechanical ventilation are intubated with an endotracheal tube to allow for gas transport to and from the lungs. One common condition in intubated patients is the accumulation of excess sputum in the trachea and endotracheal tube. Auscultation is performed by placing a stethoscope on the chest and using it to listen to the respiratory sounds for several breaths. Yamashita et al focused on the method used to detect the sputum condition and used a sparse representation method to extract features for detection of sputum accumulation using audio samples[9]. They classified the sounds into sounds with sputum present and sounds without sputum present; the average accuracy of classification was only 87% and the data were sourced from only three patients. If the microphone is in direct contact with the skin, it can lead to patient discomfort and the patient’s movements create friction noise that affects the recording results

Methods

Results

Discussion

Conclusion