Combining Residual U-Net and Data Augmentation for Dense Temporal Segmentation of Spike Wave Discharges in Single-Channel EEG
This work addresses automated detection of absence seizures in EEG for medical monitoring, but it is incremental as it builds on existing U-Net methods with enhancements for cross-subject generalization.
The study tackled the labor-intensive manual annotation of spike-wave discharges (SWDs) in EEG by developing AugUNet1D, a 1D U-Net with residual connections and data augmentation, which outperformed 15 other classifiers and a recent algorithmic approach on a dataset of 961 hours of EEG recordings with 22,637 labeled SWDs.
Manual annotation of spike-wave discharges (SWDs), the electrographic hallmark of absence seizures, is labor-intensive for long-term electroencephalography (EEG) monitoring studies. While machine learning approaches show promise for automated detection, they often struggle with cross-subject generalization due to high inter-individual variability in seizure morphology and signal characteristics. In this study we compare the performance of 15 machine learning classifiers on our own manually annotated dataset of 961 hours of EEG recordings from C3H/HeJ mice, including 22,637 labeled SWDs and find that a 1D U-Net performs the best. We then improve its performance by employing residual connections and data augmentation strategies combining amplitude scaling, Gaussian noise injection, and signal inversion during training to enhance cross-subject generalization. We also compare our method, named AugUNet1D, to a recently published time- and frequency-based algorithmic approach called "Twin Peaks" and show that AugUNet1D performs better on our dataset. AugUNet1D, pretrained on our manually annotated data or untrained, is made public for other users.