Table 2.
Performance comparison of wavelet-based epileptic seizure detection.
Decomposition Method | Sensitivity (%) | Specificity (%) | Accuracy (%) | Mean Parametric Value (%) | Classifiers | CV |
---|---|---|---|---|---|---|
Empirical wavelet transform [1] | 97.91 | 99.57 | 99.41 | 98.96 | RF | 10 |
CWT with Morlet [5] | 87.90 | 82.40 | 82.40 | 84.23 | 3RF | 5 |
Tunable Q wavelet transform [6] | 97.00 | 99.00 | 97.75 | 97.92 | LS_SVM | 10 |
Wavelet decomposition (5L-db4) [8] | 92.10 | 99.50 | 99.40 | 97.00 | RBF_SVM | 5 |
Stockwell transform—ictal [9] | 99.42 | 99.89 | 99.73 | 99.68 | k-NN | 5 |
Wavelet decomposition (5L-db6) [10] | 99.40 | 99.90 | 99.60 | 99.63 | SVM | 5 |
Dual-tree complex wavelet transform (DT-CWT) [13] | 98.0 | 100 | 98 | 98.6 | SVM | 10 |