Mining Time-Resolved Functional Brain Graphs to an EEG-Based Chronnectomic Brain Aged Index (CBAI).
Stavros I Dimitriadis1,2,3, Christos I Salis4
1Institute of Psychological Medicine and Clinical Neurosciences, Cardiff University School of MedicineCardiff, United Kingdom.
Frontiers in Human Neuroscience
|September 23, 2017
Summary
This study introduces a novel method using resting-state EEG to create a Chronnectomic Brain Aged Index (CBAI). This index accurately predicts age and distinguishes age groups by analyzing dynamic functional connectivity, offering insights into brain aging.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- The brain's resting state involves interconnected regions forming intrinsic connectivity networks (ICNs).
- Resting-state electroencephalography (rs-EEG) offers a way to study brain networks without task-related confounds.
- Dynamic functional connectivity analysis is crucial for understanding brain network evolution.
Purpose of the Study:
- To introduce a novel framework for studying dynamic functional connectivity using functional connectivity microstates (FCmicrostates) and symbolic dynamics.
- To develop an objective Chronnectomic Brain Aged Index (CBAI) from rs-EEG data.
- To assess the age-predictive and age-discriminative capabilities of EEG-based spatio-temporal features.
Main Methods:
- Construction of a single integrated dynamic functional connectivity graph (IDFCG) capturing connection strength and dominant intrinsic coupling modes (DICM).
- Definition of novel features based on symbolic dynamics, FCmicrostates, and DICM, including transition rates, Markovian Entropy, complexity index, and a Flexibility Index.
- Application of feature selection and machine learning (Extreme Learning Machine, Support Vector Regressor) to rs-EEG data from 94 subjects (eyes-open and eyes-closed conditions).
Main Results:
- The dynamic reconfiguration of dominant coupling modes was the most significant feature for age prediction and discrimination.
- High age prediction accuracy was achieved for eyes-open (R² = 0.60) and eyes-closed (R² = 0.48) conditions.
- Accurate classification of young vs. middle-aged adults was achieved (97.8% for eyes-open, 87.2% for eyes-closed).
- Results were validated on a second dataset, confirming the robustness of the methodology.
Conclusions:
- The proposed methodology effectively characterizes intrinsic dynamic functional connectivity properties using rs-EEG.
- The developed Chronnectomic Brain Aged Index (CBAI) shows significant potential for age-related brain analysis and understanding developmental differences.
- Dynamic functional connectivity features derived from EEG are valuable for non-invasively assessing brain aging.
