Semi-automatic Extraction of Functional Dynamic Networks Describing Patient's Epileptic Seizures

Area of Science:

• Neuroscience
• Computational Neuroscience
• Medical Imaging and Signal Processing

Background:

• Epileptic activity spreads across brain regions during seizures, necessitating accurate network characterization for surgical intervention.
• Stereotactic electroencephalography (SEEG) provides intracranial recordings, but analyzing seizure dynamics from large SEEG datasets is complex and time-consuming.
• Functional connectivity (FC) analysis of SEEG signals is used to understand statistical relationships between brain regions during seizures.

Purpose of the Study:

• To introduce a novel method, Brain-wide Time-varying Network Decomposition (BTND), for characterizing dynamic epileptogenic networks in individual patients using SEEG data.
• To enable objective, patient-specific, and time-varying network analysis of seizure spread.
• To compare the BTND method's findings with traditional visual interpretation of SEEG signals.

Main Methods:

• Development and application of the Brain-wide Time-varying Network Decomposition (BTND) method to SEEG data.
• Analysis of pathological functional connectivity (FC) subgraphs and their temporal activation patterns during seizures.
• Validation of the BTND method by comparing its results with visual interpretation of SEEG signals from 27 seizures across nine patients.

Main Results:

• The BTND method successfully identified pathological FC subgraphs representing brain regions involved during seizures.
• The temporal activation dynamics of these subgraphs were highly consistent with classical visual interpretation of SEEG data.
• The method demonstrated reproducibility of subgraph extraction across multiple seizures from the same patient.

Conclusions:

• The proposed BTND method offers an objective and efficient approach to characterizing dynamic epileptogenic networks from SEEG recordings.
• BTND can complement visual analysis by highlighting significant components of epileptic networks and their activation dynamics.
• This technique has the potential to improve surgical treatment planning for epilepsy by providing a detailed understanding of seizure spread.