In the context of drug resistant partial epilepsy, intra-cerebral electrical stimulation (Deep Brain Stimulation) constitutes one of the means of investigation to locate epileptic volume. This exogenous source can then activate the underlying epileptic networks and generate an electrophysiological reaction. The purpose of this work is to estimate and eliminate the overlapping electrical stimulation signal in order to subsequently explore the provoked underlying electrical activity. We propose here several methods to tackle this problem, using two different approaches based on different assumptions: BSS approach based on Independent Component Analysis (ICA) and non parametric decomposition - empirical modes decomposition (EMD) algorithms.