Adaptive Metadata-Guided Supervised Contrastive Learning for Epilepsy Detection
Published in IEEE Journal of Biomedical and Health Informatics, 2025
This work presents a novel approach for epilepsy detection using adaptive metadata-guided supervised contrastive learning. The method leverages advanced MRI analysis techniques combined with foundational models to improve presurgical decision-making accuracy. The approach demonstrates significant improvements in detecting epileptic foci using arterial spin labeling MRI data.
Key Contributions
- Novel metadata-guided contrastive learning framework
- Advanced MRI analysis for epilepsy detection
- Improved presurgical decision-making support
- Integration with foundational models for neurological disease prediction
Methods
The work introduces adaptive contrastive learning techniques that incorporate clinical metadata to guide the learning process, resulting in more robust and clinically relevant feature representations for epilepsy detection tasks.
Recommended citation: Jalali, A. et al. (2025). "Adaptive Metadata-Guided Supervised Contrastive Learning for Epilepsy Detection." IEEE Journal of Biomedical and Health Informatics.
Download Paper