Enhancement of bowel sounds by wavelet-based filtering

This paper evaluates the performance of an automatic method for structural decomposition, noise removal and enhancement of bowel sounds (BS), based on the wavelet transform, The proposed method combines multiresolution analys is with hard thresholding to compose a wavelet transform-based stationary-n onstationary (WTST-NST) filter, for enhanced separation of bowel sounds (BS ) from superimposed noise. Quantitative and qualitative analysis of the exp erimental results, when applying the WTST-NST filter to BS recorded from co ntrols and patients with gastrointestinal dysfunction, prove that the abili ty of the WTST-NST filter to remove noise and reveal the authentic structur e of BS is excellent, By eliminating the need to record a noise reference s ignal, this method reduces hardware overhead when analysis of BS is the pri mary aim. The method is independent of subjective human judgement for selec tion of noise reference templates, is robust to different levels of signal interference, and, due to its simplicity, can easily be used in clinical me dicine.