Rule-Based EEG Classifier Utilizing Local Entropy of Time–Frequency Distributions

Electroencephalogram (EEG) signals are known to contain signatures of stimuli that induce brain activities. However, detecting these classify captured EEG waveforms is one the most challenging tasks analysis. This paper proposes a novel time–frequency-based method for analysis and characterization implemented in computer-aided decision-support system can be used assist medical experts interpreting patterns. The computerized utilizes spectral non-stationarity, which clearly revealed time–frequency distributions (TFDs) multicomponent signals. proposed algorithm, based on modification Rényi entropy, called local or short-term entropy (STRE), was upgraded with blind component separation procedure instantaneous frequency (IF) estimation. applied EEGs both forward backward movements left right hands, as well imagined hand movements, were by 19-channel recording system. obtained results show given virtual instrument, methods efficiently distinguish between real limb considering their terms dominant component’s IFs at specified subset channels (namely, F3, F4, F7, F8, T3, T4). Furthermore, computing number signal components, extraction, IF estimation provide important information shows potential enhance existing clinical diagnostic techniques intensity, location, type function abnormalities patients neurological motor control disorders.