Table 4.
Summery of signal processing methods.
| References | Summary of signal processing methods |
|---|---|
| Blokland et al. (2013) | Pre-processing: fNIRS: High pass filters (HPF) (0.01 Hz), low pass filter (LPF) (0.2 Hz), EEG: down sampling to 256 Hz Features: For fNIRS HbO, HbR, For EEG power spectral features Classifier: L2−regularized linear logistic regression classifier |
| Do et al. (2013) | Pre-processing: EEG prediction model, approximate information discriminant analysis Features: Spatio-spectral features Classifier: Classwise principal component analysis (PCA), linear Bayesian classifier |
| Rea et al. (2014) | Pre-processing: Wavelet-minimum description length algorithm, Gaussian Low pass filter Features: Mean changes in HbT concentration in PMC and PPC were used as discriminatory features Classifier: Linear discriminant analysis (LDA) |
| Bulea et al. (2014) | Pre-processing: Butterworth filter, ASR filter, band-pass filter (BPF), Low pass filter (LPF) Feature extraction method (FEM): Teager-Kaiser energy (TKE) Classifier: PCA, LDA, GA, LFDA-GMM |
| Salazar-Varas et al. (2015) | Pre-processing: Common average reference (CAR) FEM: Common spatial pattern Classifier: LDA |
| Sburlea et al. (2015) | Pre-processing: Software's EEGLAB and FastICA were used for artifacts sLORETA FEM: Features were extracted using MRCP and event-related (de)synchronization Classifier: Sparse linear discriminant analysis |
| Lopez-Larraz et al. (2016) | Pre-processing: Z-score based automatized artifacts removal FEM: Sliding window method (SWM), Sparse discriminant analysis |
| Hortal et al. (2016) | Pre-processing: LPF, 8th Order Butterworth filter FEM: Fast Fourier Transform is used to obtain spectral power and then further used for feature extraction Classifier: Support Vector Machine (SVM) |
| Li et al. (2016) | Pre-processing: Regression analysis algorithm FEM: LORETA method is used to measure cortical activity, fourteen regions of interest were obtained from the segmentation of Brodmann areas |
| Gui et al. (2017) | Pre-processing: 4th order Butterworth filter Features: Four locomotion modes to be identified by SSVEP: (ST, 12.5 Hz), (NW, 8.33 Hz), (AC, 7.5 Hz), and (DE, 6.82 Hz) Classifier: LDA |
| Liu D. et al. (2017) | Pre-processing: CAR Classifier: Random Forest |
| Zhang et al. (2017) | Pre-processing: ASR filter, 2nd Butterworth filter and standardized (z-score) Features: Multiple learning algorithm (MKL) Classifier: Kernel-based learning (KBL) |
| Contreras-Vidal et al. (2018) | Pre-processing: LPF (3 Hz), ASR filter, Butterworth filter, Kalman filter Classifier: PCA |
| Tobar et al. (2018) | Pre-processing: BPF, down sampling FEM: Variational Bayesian Multimodal Features: Down sampled epochs Classifier: Sparse logistic regression |
| Liu et al. (2018) | Pre-processing: : 6th Order Butterworth filter, CAR, Weighted average (WAVG) filter FEM: TKE operator Classifier: Random Forest |
| Hedian et al. (2018) | Pre-processing: Mathematical morphology filter (MMF) FEM: Power spectrum analysis Features: HbO, HbR, and HbT Classifier: SVM |
| Khan R. A. et al. (2018) | Pre-processing: Kalman, Wiener, Gaussian, hemodynamic response filter, band-pass, finite impulse response FEM: Spatial Averaging Features: Signal Mean, signal slope, signal variance, slope kurtosis, signal peak, and signal skewness Classifier: k-Nearest neighbor (KNN), Quadratic discriminant analysis (QDA), LDA, Naïve Bayes, SVM |
| Costa-Garciacutea et al. (2019) | Pre-processing: Maximum value threshold FEM: Maximum entropy method Features: Features contain information about synchronizations and desynchronization Classifier: LDA |
| Elvira et al. (2019) | Pre-processing: BPF, signals with a standard deviation greater than 40 microV have been removed, Channels with artifact are manually removaed Classifier: LDA |
| Li et al. (2020a) | Pre-processing: 2nd order low pass Chebyshev filter, MMF FEM: Genetic algorithm (GA), average over regions of interest by entropy weight method, time-domain, and correlation analysis feature extraction Features: Mean, standard deviation, coefficient of variation, energy, range, skewness, kurtosis, peak, and Hjorth parameters Classifier: Light gradient boost decision tree, two layer-GA-SVM, PCA |
| Li et al. (2020b) | Pre-processing: Chebyshev bandpass filter, z-score FEM: TKE Features: HbO, HbR, and HbT Classifier: Gradient boosting decision tree model |