Electroencephalogram Data-Based Analysis of Paroxysmal Slow Wave Events Patterns in Brain Pathologies

Area of Science:

• Neuroscience
• Medical Imaging
• Computational Biology

Background:

• Brain activity slowing in electroencephalography (EEG) is typical during rest.
• A novel pattern of transient cortical slowing, termed paroxysmal slow wave events (PSWEs), has been identified in epilepsy and Alzheimer's disease patients.
• PSWEs are characterized by median power frequency (MPF) below 6 Hz and durations exceeding 5 seconds.

Purpose of the Study:

• To analyze the temporal and spatial features of PSWEs in epilepsy patients.
• To identify PSWE characteristics aiding in the diagnosis of epilepsy, particularly drug-resistant epilepsy.
• To assess the accuracy of PSWE features in distinguishing epilepsy from other brain disorders such as Alzheimer's disease and mood disorders.

Main Methods:

• Utilized clinical EEG recordings from Temple University and Bonn University databases.
• Trained machine learning models on the Temple University dataset to classify epilepsy versus non-epilepsy patients.
• Applied machine learning models to the Bonn University database for classifying drug-resistant epilepsy against seizure-free groups.

Main Results:

• Achieved 78.26% classification accuracy distinguishing epilepsy from non-epilepsy patients using the Temple University dataset.
• Attained 91.67% accuracy in classifying drug-resistant epilepsy versus seizure-free groups with the Bonn University dataset.
• Demonstrated the potential of PSWEs as diagnostic biomarkers.

Conclusions:

• PSWEs can serve as a potential biomarker for early epilepsy diagnosis and risk assessment.
• PSWE analysis aids in distinguishing between isolated seizures and chronic epilepsy.
• The association of PSWEs with neurodegenerative and cognitive disorders underscores their importance in monitoring neurological diseases.