Detection of High-Frequency Oscillations from Intracranial EEG Data with Switching State Space Model
Summary
This study introduces a novel Switching State Space Model (SSSM) for automatically detecting High Frequency Oscillations (HFOs) in epilepsy. The SSSM accurately identifies HFO events and their duration, aiding in pinpointing epileptogenic zones.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- High Frequency Oscillations (HFOs) are critical biomarkers for identifying epileptogenic zones (EZs).
- Manual annotation of HFOs in long-term intracranial EEG recordings is challenging due to their short duration and the large number of electrodes.
- HFOs can be difficult to detect when obscured by background low-frequency activity.
Purpose of the Study:
- To develop and validate an automated method for detecting High Frequency Oscillations (HFOs).
- To improve the accuracy and efficiency of identifying epileptogenic zones (EZs) using HFOs.
Main Methods:
- A Switching State Space Model (SSSM) was employed for automatic HFO event detection.
- The SSSM processes intracranial EEG data without requiring feature extraction from sliding windows.
- The model's performance was evaluated on intracranial EEG recordings from human subjects.
Main Results:
- The SSSM demonstrated effective automatic and instantaneous detection of HFO events.
- The model showed improved accuracy in capturing both the occurrence and duration of HFOs.
- Validation was performed using intracranial EEG data from patients undergoing presurgical evaluation.
Conclusions:
- The Switching State Space Model (SSSM) offers a promising automated solution for HFO detection in epilepsy.
- This approach enhances the precision and efficiency of identifying epileptogenic zones (EZs).
- The SSSM's ability to capture HFO duration is crucial for accurate localization.
