Classification of mild cognitive impairment EEG using combined recurrence and cross recurrence quantification...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Mild cognitive impairment (MCI) is an early stage of cognitive decline.
• Electroencephalography (EEG) is a non-invasive tool for brain activity monitoring.
• Recurrence-based analysis offers novel features for EEG signal characterization.

Purpose of the Study:

• To classify mild cognitive impairment (MCI) EEG signals.
• To combine complexity and synchronization features for improved classification.
• To evaluate EEG patterns during resting and memory task states.

Main Methods:

• Utilized recurrence quantification analysis (RQA) for complexity and cross-recurrence quantification analysis (CRQA) for synchronization.
• Analyzed EEG data from eyes-closed (EC) resting state and short-term memory (STM) task.
• Employed geometrical signal separation in a feature space of RQA and CRQA RR values for classification.

Main Results:

• MCI EEG exhibited significantly lower complexity (higher RQA RR) and higher inter/intra-hemispheric synchronization (higher CRQA RR) compared to normal controls (NC).
• The STM task effectively highlighted MCI-specific EEG features in temporal, parietal, and frontal lobes.
• Linear classification analysis showed enhanced efficiency for MCI/NC discrimination under STM conditions versus EC.

Conclusions:

• Combining complexity and synchronization features provides an effective approach for EEG-based MCI classification.
• Task-based EEG analysis, particularly memory activation, improves the detection of MCI characteristics.
• This method holds promise for early and accurate diagnosis of mild cognitive impairment.