Empirical mode decomposition: a method to reduce low frequency interferences from surface electroenterogram

The surface electroenterogram (EEnG) is a non-invasive method of studying myoelectrical bowel activity. However, surface EEnG recordings are contaminated by cardiac activity, respiratory and motion artifacts, and other sources of interference. The aim of this work is to remove the respiration artifact and the very low frequency components from surface EEnG by means of empirical mode decomposition (EMD). Eleven recording sessions were carried out on canine model. Several parameters were calculated before and after the application of the method: signal-to-interference ratio (S/I ratio) and the attenuation level of the signal and of interference. The results show that the S/I ratio was significantly higher after the application of the method (3.68+/-5.54 dB vs. 10.45+/-3.65 dB), the attenuation level of signal and of interference is -0.49+/-0.80 dB versus -7.26+/-5.42 dB, respectively. Therefore, EMD could be a useful aid in identifying the intestinal slow wave and in removing interferences from EEnG recordings.