Table 5 Comparison of seizure detection methods using the benchmark Bonn EEG dataset
Article | Year | Selected features | Classifier | Case | Accuracy (%) |
|---|---|---|---|---|---|
Riaz et al. [6] | 2016 | Time matrix + spectral features | SVM | A-E D-E | 97.00 92.00 |
Raghu et al. [14] | 2017 | Wavelet Packet norm Entropy | REN | C-E | 99.60 |
Jiang et al. [37] | 2017 | WPD | TSK | A-E | 91.40 |
Jaiswal and Banka [39] | 2017 | EEG | 1D-Local Gradient Patterns (LGP) + SVM | C-E D-E | 99.10 99.07 |
Jaiswal et al. [24] | 2018 | PCA | SVM | D-E ABCD-E | 95.50 97.40 |
Tripathi and Agrawal [13] | 2018 | FuzzyEn | SVM | C-E D-E | 98.62 97.00 |
Lu et al. [22] | 2018 | Kraskov entropy + instantaneous area | LS-SVM | C-E D-E | 99.00 97.00 |
Wang et al. [21] | 2019 | STFT + average energy + PCA | RF + GSO | C-E D-E | 98.50 98.10 |
Zhao and Wang [31] | 2020 | EEG | CNN | D-E | 98.50 |
Shoeibi et al. [40] | 2021 | Timedomain + Power spectrum + Nonlinear features + Lyapunov index | Fisher + CNN | C-E | 96.67 |
Banupriya and Devi [20] | 2021 | EEG | VSPO-SVM | D-E | 98.13 |
Al-Hadeethi et al. [38] | 2021 | Max + Min + Mode + range + var + standard deviation | KST + AdaBoost | C-E AB-E CD-E | 98.50 98.00 98.20 |
Aayesha et al. [29] | 2022 | Time domain + spectrum + nonlinear features + Local Binary Pattern | Feedforward Neural Network | A-E B-E C-E D-E AB-E CD-E ABCD-E | 96.67 91.67 91.67 85.00 90.00 91.11 90.67 |
Xin et al. [36] | 2022 | DWT decompose EEG | AMWCNN | C-E D-E | 99.39 99.11 |
Hemachandira and Viswanathan [7] | 2022 | DWT Haar + db4 + Sym8 | Particle Swarm Optimization (PSO) + SVM | A-E | 98.00 |
Proposed study | 2022 | Time–frequency + nonlinear features | RF + CNN | A-E B-E C-E D-E AB-E AC-E AD-E BC-E BD-E CD-E ABC-E ABD-E ACD-E BCD-E ABCD-E | 99.30 98.10 99.90 99.20 99.00 99.40 99.28 98.40 97.46 99.07 98.95 97.30 98.65 97.65 98.47 |