Area and Power Efficient Architecture For Muscular Artifacts Removal From EEG Signal

Neurodevelopmental Disorders (NDD) are diagnosed by medical practitioners using a portable electroencephalogram (EEG) headset. However, EEG signals are often contaminated by muscular artifacts, which need to be removed by employing an efficient technique. In this paper, an efficient wavelet thresholding technique is proposed to remove muscular artifacts from EEG signals. The EEG signal contaminated by muscular artifacts has more power compared to normal EEG signals. The proposed wavelet thresholding technique identifies and removes muscular artifacts from EEG signals, which involves the computation of the sum of absolute value (SAV) of a signal instead of power. The architecture developed for this technique consumes less area and power compared to the hardware realization of the techniques available in the literature. The proposed technique is modeled using Verilog HDL and synthesized using Synopsys tools employing 32 nm cell libraries. The performance comparison of ASIC implementation of the proposed technique with the technique available in the literature indicates that the proposed architecture can be preferred in portable EEG signal applications.