Fig 1.
Block diagram of the proposed abnormal EEG signal detection framework.
Table 1.
Detailed description of the TUH Abnormal EEG dataset utilized in this study.
Fig 2.
Flowchart of EEG preprocessing.
Fig 3.
Distribution of the 21 EEG electrodes according to 10–20 system.
Fig 4.
Time-frequency feature extraction on raw EEG signals using discrete wavelet transform.
Fig 5.
Process of feature aggregation.
Fig 6.
Performance of three classifiers.
Table 2.
Confusion matrix of EEG detection.
Fig 7.
Confusion matrices of three classifiers.
Table 3.
Comparison of the classification results obtained by different EEG pathology diagnosis approaches on the real-world EEG abnormal dataset.
Table 4.
Ablation study of multi-view feature aggregation.
Fig 8.
Heatmap of p-values for time-frequency features aggregated across activity, mobility, and complexity dimensions.
The color intensity in the heatmap corresponds to the p-value size, representing feature statistical significance for classification.
Fig 9.
Performance of three classifiers (without spatial features).