Effects of TFD thresholding on EEG signal analysis based on the local Rényi entropy

Analysis of electroencephalographic (EEG) signals using both non-iterative and iterative modifications of the Rényi entropy, called local or short-term Rényi entropy, often requires thresholding in the time-frequency domain. The purpose of this thresholding is to reduce noise and low-energy cross-terms prior to detecting the number of components present in multicomponent EEG signals. However, preset time-frequency threshold value is often chosen empirically. In this paper, a performance analysis of the short-term Rényi entropy based method with regards to the chosen time-frequency threshold is rendered. The method was applied to both noise-free and noisy real-life EEG signals for left and right leg movements verifying the method's sensitivity to time-frequency distribution (TFD) thresholding.