Denoising of the impedance cardiographie signal (ICG) for a best detection of the characteristic points

This paper is part of the electrophysiological signal processing framework. The impedance cardiography signal (ICG) is considered as one of the most interesting electrophysiological signals. This signal provides information about the physio-pathological statue of the cardiovascular system of the patients. A suitable preprocessing is always required before any type of treatment applied to this signal. Our study is interested, in a first step, on the denoising of ICG signals: a comparison is carried out between several denoising techniques in order to specify the most effective method without distorting the shape of the studied signal. In a second step, this paper presents our detection algorithm of the characteristic points B, C and X of the denoised ICG signal. An evaluation test of this technique is performed and allowed to choose the Savitzky-Golay filter which gave an excellent positive predictivity P⁺ varying from 93.1% to 100% according to the characteristic points and a low detection error rate DE varying from 0 to 6.9%.