Table 1.
Feature-based approaches for abnormality detection in EGG using TUAB dataset.
| Studies | Year | Input | Architecture | ACC (%) |
|---|---|---|---|---|
| Lopez et al. [37] * | 2017 | band power-based features using Cepstral coefficients | CNN + Multilayer Perception (MLP) | 78.8 |
| Alhussien et al. [27] * | 2019 | FFT band-limited signals | AlexNet + MLP | 89.13 |
| Gemein et al. [12] | 2020 | DWT + CWT + DFT + Statistical features | RG | 85.90 |
| Cisotto et al. [40] | 2020 | Statistical features + spectral power in specific frequency bands | LSTM+attention | 79.18 |
| Sharma et al. [41] | 2020 | Wavelet-based statistical features | SVM | 79.34 |
| Albaqami et al. [21] | 2021 | WPD + Statistical features | CatBoost | 87.68 |
| Singh et al. [17] | 2021 | Spectrogram image based on STFT | VGG-19 + RF | 88.04 |
| Bajpai et al. [14] | 2021 | Spectrogram image based on STFT | SeizNet + SVM | 96.56 |
| Mohsenvand et al. [42] | 2021 | EEG contrastive learning | Simple Contrastive Learning of Visual Representations(SimCLR) | 87.45 |
| Wu et al. [16] | 2022 | Statistical features from DWT coefficients | CatBoost | 89.13 |
| Wu et al. [23] | 2022 | Statistical features from WPD coefficient | Catboost | 89.76 |
| Tasci et al. [25] | 2023 | Multilevel Discrete Wavelet Transform (MDWT) + Statistical features | KNN | 87.78 |
| Zhong et al. [38] | 2023 | Statistical features from WPD coefficients | CatBoost | 89.13 |
| Kohad et al. [15] | 2022 | EMD and EWT based features | Linear SVM | 88.48 |
* Used extra training data not included in TUAB.