Automatic detection of generalized paroxysmal fast activity in interictal EEG using time-frequency analysis
Amir Omidvarnia1, Aaron E L Warren2, Linda J Dalic3
1Institute of Bioengineering, Center for Neuroprosthetics, Center for Biomedical Imaging, EPFL, Geneva, Switzerland; Department of Radiology and Medical Informatics, University of Geneva, Geneva, Switzerland.
Computers in Biology and Medicine
|May 22, 2021
Summary
Automated detection of generalized paroxysmal fast activity (GPFA) in Lennox-Gastaut syndrome (LGS) is now possible using a novel EEG framework. This method offers a faster, objective approach to identifying epilepsy events, aiding clinical decisions.
Area of Science:
- Neuroscience
- Medical Imaging
- Signal Processing
Background:
- Manual marking of interictal epileptiform discharges (IEDs) on EEG is crucial for epilepsy diagnosis but is time-consuming and subjective.
- Generalized paroxysmal fast activity (GPFA) is a specific type of generalized IED observed in Lennox-Gastaut syndrome (LGS).
Purpose of the Study:
- To develop an automated framework for detecting GPFA in scalp EEG recordings of children with LGS.
- To validate the automated detection method using EEG-functional MRI (EEG-fMRI).
Main Methods:
- Studied 13 children with LGS exhibiting GPFA events.
- Utilized time-frequency information from manually marked IEDs to train an automated detection algorithm.
- Validated detection accuracy using standalone scalp EEG and simultaneous EEG-fMRI.
Main Results:
- GPFA events exhibit a distinct low-high frequency 'bimodal' pattern in the time-frequency domain, primarily on frontal EEG channels.
- The automated detection approach successfully identified events with similar spectral features to manually marked GPFAs.
- EEG-fMRI maps generated from automatically detected IEDs closely matched those from manually marked events.
Conclusions:
- GPFA events in LGS patients possess a characteristic bimodal time-frequency signature detectable via automated scalp EEG analysis.
- EEG-fMRI analysis confirms the validity of the automated detection by recapitulating known brain activity patterns associated with generalized IEDs in LGS.
- This automated methodology provides a rapid, objective tool for inspecting generalized IEDs in LGS, with potential applications for other epilepsy syndromes.
