Empirical Mode Decomposition and a Bidirectional LSTM Architecture Used to Decode Individual Finger MI-EEG Signals

Abstract

Brain-Computer Interface (BCI) paradigms based on Motor Imagery Electroencephalogram (MI-EEG) signals have been developed because the related signals can be generated voluntarily to control further applications. Researches using strong and stout limbs MI-EEG signals reported performing significant classification rates for BCI applied systems. However, MI-EEG signals produced by imagined movements of small limbs present a real classification challenge to be effectively used in BCI systems. It is due to a reduced signal level and increased noisy distorted effects. This study aims to decode individual right-hand fingers’ imagined movements for BCI applications, using MI-EEG signals from C3, Cz, P3, and Pz channels. For this purpose, the Empirical Mode Decomposition (EMD) preprocesses the non-stationary and non-linear EEG signals to finally use a Bidirectional Long Short-Term Memory (BiLSTM) to classify corresponding feature sequences. An average accuracy of 98.8 % was achieved for ring-finger movements decoding using k-fold cross-validation on a public dataset (Scientific-Data). The obtained results support that the proposed framework can be used for BCI control applications.

Deep temporal networks for EEG-based motor imagery recognition

“…In a conventional pipeline, raw EEG data is pre-processed using dimensionality reduction by independent component analysis (ICA) 5 , multiscale principle component analysis (MSPCA) 6 and denoising techniques such as regression or blind source separation (BSS) 7 may be applied. After pre-processing, data may be decomposed using short-time Fourier transform (STFT) 8 , wavelet decomposition (WD) 9 , empirical mode decomposition (EMD) 6 , 10 or Fourier decomposition method (FDM) 11 , 12 . The statistical and temporal features 13 are extracted from the decomposed signal and classified using several machine learning classifiers, such as support vector machine(SVM), k-nearest neighbour (kNN), Naïve Bayes (NB), and decision tree (DT) 14 , 15 .…”

“…With the advent of deep learning techniques, researchers have explored their usage in MI-EEG signal classification. Various deep learning approaches based on convolutional neural network (CNN) 16 , recurrent neural network (RNN) 17 , and long-short-term memory (LSTM) 10 have been investigated. Taheri et al 18 selected the most discriminant features using a CNN by making a triple frame matrix, combining EMD, Fourier transform, and common spatial patterns.…”