Automatic Absence Seizures Detection in EEG signals: An Unsupervised Module.

Area of Science:

• Neurology
• Biomedical Engineering
• Signal Processing

Background:

• Absence seizures exhibit unique spike-and-wave complexes on electroencephalogram (EEG).
• Automated detection of these seizures is crucial for accurate diagnosis and treatment.
• The chaotic nature of EEG signals presents challenges for pattern recognition.

Purpose of the Study:

• To develop and validate an unsupervised methodology for the automatic detection of absence seizures from EEG data.
• To assess the performance of the proposed method in terms of sensitivity and false detection rate.

Main Methods:

• Analysis of spectral activity within the 2.5-4.5 Hz frequency range.
• Assessment of synchronous, repetitive patterns across multiple EEG channels.
• Unsupervised learning approach applied to the TUSZ open dataset.

Main Results:

• Achieved a high sensitivity of 93.94% for absence seizure detection.
• Reported a low false detection rate of 0.168 false detections per hour.
• Demonstrated the effectiveness of the unsupervised method on a public dataset.

Conclusions:

• The proposed unsupervised method effectively detects absence seizures using EEG.
• The technique shows promise for reliable, automated seizure detection in clinical settings.
• Further validation on diverse datasets is warranted.