EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Self-Supervised Learning with Adaptive Frequency-Time Attention Transformer for Seizure Prediction and Classification

Yajin Huang, Yuncan Chen, Shimin Xu, Dongyan Wu, Xunyi Wu

Brain Sciences · 07.04.2025 · 15 (4) · 382

DOI: 10.3390/brainsci15040382 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Brain Sciences, online erschienen am 07.04.2025. Verfasst von Yajin Huang und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Neurology, Huashan Hospital, Fudan University, Shanghai 200040, China. Die Arbeit wurde bisher 18-mal zitiert (Crossref) und nennt 64 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: The National Key R&D Program of China, The National Key R&D Program of China.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

In deep learning-based epilepsy prediction and classification, enhancing the extraction of electroencephalogram (EEG) features is crucial for improving model accuracy. Traditional supervised learning methods rely on large, detailed annotated datasets, limiting the feasibility of large-scale training. Recently, self-supervised learning approaches using masking-and-reconstruction strategies have emerged, reducing dependence on labeled data. However, these methods are vulnerable to inherent noise and signal degradation in EEG data, which diminishes feature extraction robustness and overall model performance.

Methods

In this study, we proposed a self-supervised learning Transformer network enhanced with Adaptive Frequency-Time Attention (AFTA) for learning robust EEG feature representations from unlabeled data, utilizing a masking-and-reconstruction framework. Specifically, we pretrained the Transformer network using a self-supervised learning approach, and subsequently fine-tuned the pretrained model for downstream tasks like seizure prediction and classification. To mitigate the impact of inherent noise in EEG signals and enhance feature extraction capabilities, we incorporated AFTA into the Transformer architecture. AFTA incorporates an Adaptive Frequency Filtering Module (AFFM) to perform adaptive global and local filtering in the frequency domain. This module was then integrated with temporal attention mechanisms, enhancing the model’s self-supervised learning capabilities.

Result

Our method achieved exceptional performance in EEG analysis tasks. Our method consistently outperformed state-of-the-art approaches across TUSZ, TUAB, and TUEV datasets, achieving the highest AUROC (0.891), balanced accuracy (0.8002), weighted F1-score (0.8038), and Cohen’s kappa (0.6089). These results validate its robustness, generalization, and effectiveness in seizure detection and classification tasks on diverse EEG datasets.

Schlagwörter

Stichwörter der Autor:innen

attentionepilepsyself-supervisedtransformer

Autor:innen und Institutionen

  • Department of Neurology, Huashan Hospital, Fudan University, Shanghai 200040, China.

Förderung

  • The National Key R&D Program of China
  • The National Key R&D Program of China

Zitieren

APA

Huang, Y., Chen, Y., Xu, S., Wu, D., & Wu, X. (2025). Self-Supervised Learning with Adaptive Frequency-Time Attention Transformer for Seizure Prediction and Classification. Brain Sciences, 15(4), 382. https://doi.org/10.3390/brainsci15040382

Vancouver

Huang Y, Chen Y, Xu S, Wu D, Wu X. Self-Supervised Learning with Adaptive Frequency-Time Attention Transformer for Seizure Prediction and Classification. Brain Sciences. 2025;15(4):382. doi:10.3390/brainsci15040382

BibTeX
@article{Huang2025,
  author = {Huang, Yajin and Chen, Yuncan and Xu, Shimin and Wu, Dongyan and Wu, Xunyi},
  title = {Self-Supervised Learning with Adaptive Frequency-Time Attention Transformer for Seizure Prediction and Classification},
  journal = {Brain Sciences},
  year = {2025},
  volume = {15},
  number = {4},
  pages = {382},
  doi = {10.3390/brainsci15040382}
}
Kostenlos starten →