Automated EEG analysis: characterizing the posterior dominant rhythm

Area of Science:

• Neuroscience
• Medical Imaging
• Signal Processing

Background:

• Clinical electroencephalogram (EEG) interpretation is subjective and time-consuming.
• Automated analysis can improve objectivity and efficiency in EEG interpretation.
• Characterizing the posterior dominant rhythm (PDR) is crucial for EEG analysis.

Purpose of the Study:

• To develop and evaluate an automated method for characterizing the main properties of the PDR.
• To quantify PDR frequency, symmetry, and reactivity from clinical EEG recordings.
• To establish a feasible first step towards a fully automated EEG interpretation system.

Main Methods:

• A three-component curve-fitting technique was used to identify dominant peaks in EEG spectra during eyes-closed states.
• PDR frequency and amplitude were estimated from the fitted spectral curve.
• PDR symmetry and reactivity were assessed using spectral power at estimated PDR frequencies, with a certainty value incorporated.

Main Results:

• The automated method achieved PDR frequency estimates within 1.2Hz of visual analysis in 92.5% of 1215 clinical EEG recordings.
• Accuracy improved further when estimates with low certainty values were excluded.
• The method demonstrated matched accuracy to visual inspection for quantifying essential PDR features.

Conclusions:

• The presented automated method accurately quantifies key PDR features, comparable to visual inspection.
• This technique offers a reliable and objective approach to EEG analysis.
• It represents a significant contribution towards the development of fully automated clinical EEG interpretation systems.