Space time frequency (STF) code tensor for the characterization of the epileptic preictal stage

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Understanding the preictal period (the time before a seizure) is crucial for developing seizure prediction systems.
• Multiway models and tensor decomposition have shown potential in revealing hidden structures within electroencephalogram (EEG) data.

Purpose of the Study:

• To investigate the utility of multiway models, specifically Parallel Factor Analysis (PARAFAC), for characterizing the epileptic preictal period.
• To identify spatial, temporal, and spectral signatures associated with the preictal state using PARAFAC.

Main Methods:

• The study employed Parallel Factor Analysis (PARAFAC), a tensor decomposition technique, on EEG segments from patients experiencing seizures.
• Data from 4 patients, encompassing a total of 30 seizures, were analyzed.

Main Results:

• The analysis revealed common underlying structures potentially involved in epileptic seizure generation.
• A distinct spatial signature was identified, possibly correlating with the seizure onset region.
• Specific frequency sub-bands were found to be more significant during preictal stages.

Conclusions:

• Multiway modeling with PARAFAC can effectively characterize the epileptic preictal period.
• The identified signatures provide valuable information for advancing seizure prediction frameworks.
• Further research into these spatial and spectral signatures may enhance understanding of seizure pathophysiology.