Seizure Detection in EEG Signals

This is the authors' abstract. We don't add key points for this paper.

Neurological disorders, particularly epilepsy, affect millions of people worldwide and can significantly impact quality of life. Electroencephalography (EEG) provides a non-invasive method for monitoring brain activity and detecting seizures, but manual analysis is time-consuming and prone to error. Automated seizure detection systems, leveraging signal processing and machine learning techniques, have the potential to improve accuracy, efficiency, and real-time monitoring. This thesis investigates methods for EEG-based seizure detection, including signal and embedding based feature extraction, supervised and patient-specific learning, and time series anomaly detection approaches. We evaluate different detection strategies, explore patient-specific adaptations, and assess methods for localizing seizure activity in the brain. Experimental results demonstrate that leveraging both feature extraction and patient-specific learning improves detection performance, highlighting the importance of personalized approaches. The findings contribute to the development of more reliable and efficient automated seizure detection systems, which can support clinicians, enhance patient monitoring, and reduce the burden of epilepsy management.