DINO-EEG: Event-level, Montage-robust Seizure Detection in Clinical EEG (Model Weights)
Description
This record contains pretrained model weights for the paper:
"DINO-EEG: Event-level, Montage-robust Seizure Detection in Clinical EEG"
Overview
DINO-EEG is an end-to-end framework for seizure detection that formulates seizures as temporal objects in EEG signals and performs event-level detection using an object detection paradigm.
The model operates on single-channel EEG signals and produces channel-wise detections, which are aggregated into global seizure events. This design enables montage robustness and clinically interpretable outputs.
Dataset and Training
The models were trained on the Temple University Hospital EEG Seizure Corpus (TUSZ v2.0.3), and evaluated using the SzCORE event-based evaluation framework.
Notes
-
Pretrained model weights are provided.
-
External datasets used for evaluation include CHB-MIT and Siena EEG datasets.
Please refer to the paper for full methodological details.
Files
readme.md
Additional details
Software
- Repository URL
- https://github.com/XploreAI-Lab/DINO-EEG
- Programming language
- Python
References
- Zhang, H. et al. DINO: DETR with improved denoising anchor boxes for end-to-end object detection. ICLR, 2023.
- Shah, V. et al. The Temple University Hospital Seizure Detection Corpus. Frontiers in Neuroinformatics 12, 83 (2018).
- Dan, J. et al. SzCORE: A seizure community open-source research evaluation framework for the validation of EEG-based automated seizure detection algorithms. Epilepsia (2024).