Spatio-Temporal Clustering of Epileptic ECOG

Area of Science:

• Neuroscience
• Computational Neuroscience
• Medical Technology

Background:

• Epileptic seizure generation mechanisms remain poorly understood.
• Quantifying spatio-temporal brain dynamics is crucial for epilepsy research.
• Existing methods may not fully capture seizure progression complexity.

Purpose of the Study:

• To quantify spatio-temporal interactions in an epileptic brain.
• To explore the utility of a SOM-based Similarity Index (SI) for seizure analysis.
• To investigate spatial mappings during different seizure stages.

Main Methods:

• Utilized a Self-Organizing Map (SOM)-based Similarity Index (SI).
• Applied spectral clustering to SOM-SI values interpreted as affinity matrices.
• Analyzed data from two pairs of seizures from an epileptic patient.

Main Results:

• Demonstrated the effectiveness of spectral clustering for mapping brain activity.
• Interpreted SOM-SI values to reveal spatial dynamics.
• Observed no consistent spatio-temporal patterns during inter-ictal to post-ictal transitions.

Conclusions:

• The SOM-SI measure and spectral clustering offer novel insights into seizure dynamics.
• Epileptic seizure progression may exhibit irregular spatio-temporal channel interactions.
• Further research is needed to fully elucidate seizure generation mechanisms.