FIGURE 1 The algorithm design of the proposed gaWD method. (a)
The 6 levels of wavelet decomposition. (b) The process to find threshold
T. (c) The whole process of de-noising method.
3 | SIGNAL DE-NOISING RESULTS
A typical acquired PA signal, from cancerous colorectal tissues,
corrupted by noise and limited by resolution is shown in FIGURE
2 . The final de-noising result using our proposed gaWD method, low pass
filter, sqtwolog threshold, minimaxi threshold, hersure threshold, and
rigrsure threshold method are shown in FIGURE 3 , respectively.
From the signal de-noising results of 6 different methods shown inFIGURE 3 , we can find that the classic wavelet de-noising
methods, like sqtwolog, minimaxi, hersure, and rigrsure methods, all
induce impulse noise distortion. They also fail to filter out the large
coupled signal at the transducer surface. In comparison, sqtwolog
threshold method performs better than the other 3 classic wavelet
threshold methods, although it induces some signal distortion. Low pass
filter and our proposed method both perform well in removing the coupled
signal. However, the low pass filter preserves more noise in the
waveform. On the contrast, our proposed