Performance comparison of wavelet based denoising methods on discontinuous adventitious lung sounds

Area of Science:

• Medical Signal Processing
• Respiratory Medicine
• Biomedical Engineering

Background:

• Lung crackles are vital indicators of various respiratory diseases.
• Traditional de-noising methods like wavelet transforms and total variation struggle with crackle signals, introducing artifacts.
• Preserving information-bearing parts of crackles is crucial for accurate parameter estimation.

Purpose of the Study:

• To develop an advanced de-noising technique for lung crackles corrupted by noise.
• To overcome limitations of existing wavelet and total variation de-noising methods.
• To preserve the essential waveform characteristics of crackles for improved analysis.

Main Methods:

• A novel wavelet total variation de-noising algorithm was proposed.
• The method was tested on synthetically generated crackles with varying levels of white Gaussian noise.
• Performance was evaluated against classical wavelet de-noising methods using root mean square error (RMSE).

Main Results:

• The proposed wavelet total variation method successfully removed artifacts from both classical wavelet and total variation de-noising.
• The new method demonstrated superior performance in de-noising crackles across different noise levels (0-20 dB SNR).
• Visual validation confirmed the preservation of time and frequency domain representations without waveform deformation.

Conclusions:

• The proposed wavelet total variation de-noising method offers a significant improvement for analyzing lung crackles.
• This technique effectively removes noise while preserving critical signal features essential for diagnosing lung diseases.
• The method holds promise for enhancing the accuracy of respiratory sound analysis in clinical settings.