Table 1.

Summary of epileptic seizure detection approaches in the time domain.

Feature Extraction Method	Subjects	Window Size	Features	Classifier	Performance (%)
1 s non-overlapping window [21]	4 patient, 21 h recording	1 s	Time: skewness, kurtosis, No. of maxima and minima, mean, variation, standard deviation, and Shannon, entropy, ApEn, energy, standard variation, variance, and energy of auto-covariance and COV, RMS.	SVM	Sen: 90.62 Spe: 99.32
1 s non-overlapping frames [23]	21 patients (excluding patients 6, 12, 16)	1 s	Time: No. of maxima and minima, skewness, kurtosis, standard deviation, COV, RMS, Shannon entropy, ApEn, energy, standard variation, mean, variation variance, the energy of auto-covariance. Frequency: mean of the power spectrum, spectral entropy, median frequency. maximum, minimum, and Time-frequency: relative scale energy, COV, frequency regularity index, maximum, minimum, Shannon entropy, variance, mean, std-deviation, No. of extrema, and energy Nonlinear: Lyapunov exponent	SVM, multi-dimensional PSO	Sen: 89 Spe: 93
Time domain approach [28]	23 patient		Mean, std-deviation, median, skewness, kurtosis, PA value, NA_value, mean of 1st and 2nd derivative and a maximum of 1st and 2nd derivative, RMS amplitude, line length, COV	K-means clustering
PCA [34]	23 patients excluding 15	1 s	Range, quantile, IQR, Shannon entropy, RMS amplitude, COV, and energy	LDA, NB	Sen: 88.26 Spe: 93.21
SVD [36]		1 s	Classical features such as mean, variance, kurtosis, skewness, power	SVM	Acc: 94.82
PARAFAC decomposition [38]	1 patient		Spatio-spectral features	LDA, SVM, K-means
PCA and LDA [40]	171 seizures 171 non-seizures	60 s	Peak frequency, median frequency, variance, RMS, sample entropy, skewness, and kurtosis	k-NN classifier	Sen: 88 Spe: 88 Acc: 93
2 s non-overlapping window [43]	24 patient 198 seizures	600 s	Spectral energy features	Linear SVM, $D^2$ A	Sen: 95.1 Spe: 96.2
SVD [51]	23 patient	4 s	2D eigenvalues, cross bi-spectrum in the spatial and spectral direction
PCA [62]	23 patient	1 s	Quantile, Inter quantile, range, Shannon entropy, RMS, COV, and energy	SVM NB	Sen: 95.01 Selectivity: 97.97 Acc: 96.77