Published online Sep 15, 1996. doi: 10.3748/wjg.v2.iSuppl1.123
Revised: April 28, 1996
Accepted: August 17, 1996
Published online: September 15, 1996
AIM: This abstract introduces our study work on De-noising of electrogastrogram signals by wavelet shrinkage, which aims at removing noise from the noisy Electrogastrogram (EGG) signals.
METHODS: The noisy EGG signals are transformed into the wavelet domain using an orthogonal periodic wavelet transform based on Nearly Symmetric wavelets with 8 vanishing moments (S8). The wavelet coefficients are subjected to soft thresholding, theta;—t(ω) = sgn(ω)(|ω|—t)—+, with threshold t—tn = (2log(n)σ)-1/2. The results is then inverse transformed. If we let W—S8 denote the wavelet transform with Symmlet 8 wavelets, the whole de noising process amounts to a nonlinear operator Ttheta;, S8(Y), where Ttheta;, S8 = W1—S8°theta;—tn°W—S8.
RESULTS: The computer simulation are performed for four typical sets of noisy EGG data (Table 1), which represents Normal, Tachygastria, Bradygastria and Arrhythmia (Figure 1), stored in files, GE33A, MOT13C, LINA and GE70A, respectively, Noisy signals and de-noised signals are illustrated in Figure 2. It is shown that the proposed method used de noising EGG signals is efficient.