Skip to main content
Skip to footer

Lung auscultation using acoustic biomarkers

Authors

Upasana Tiwari, Researcher, TCS Research
Swapnil Bhosale, Researcher, TCS Research
Rupayan Chakraborty, Senior Scientist, TCS Research
Sunil Kumar Kopparapu, Principal Scientist, TCS Research

 

Highlights

  • Deep lung auscultation using acoustic biomarkers for abnormal respiratory sound event detection.
  • Discrete wavelet transform and audio event detection model are used for automatic respiratory sound event detection.
  • The approach is evaluated against international benchmark and the results show that our system clearly outperforms the state of the art with a significant margin.

My paper in 2 minutes

Deep lung auscultation using acoustic biomarkers

 

Deep lung auscultation using acoustic biomarkers (extracted using discrete wavelet transform and deep-encoded features from a pre-trained audio event detection model) for abnormal respiratory sound event detection.

Lung auscultation is a non-invasive process of distinguishing normal respiratory sounds from abnormal ones by analyzing the airflow along the respiratory tract. With developments in deep learning and wider access to anonymized data, automatic detection of specific sounds such as crackles and wheezes have been gaining popularity. We propose to use two sets of diversified acoustic biomarkers extracted using discrete wavelet transform (DWT) and deep-encoded features from the intermediate layer of a pre-trained audio event detection (AED) model, trained using sounds from daily activities.

The first set of biomarkers highlight the time frequency localization characteristics obtained from DWT coefficients; whereas, the second set of deep encoded biomarkers captures a generalized reliable representation, and thus indemnifies the scarcity of training samples and the class imbalance in dataset. Furthermore, the ensemble of DWT based time-frequency localization and generic device agnostic deep embedding resulted into complimenting each other. We evaluated our approach on [International Conference on Biomedical Health Informatics] ICBHI-2017 Challenge dataset. The results shows that our system clearly outperforms the state of the art with a significant margin.