Aims: Our first aim was to evaluate the validity of high-density electroencephalography (EEG) for analysing functional brain networks in children and adolescents. Secondly, we focused on determining whether EEG-derived functional connectivity can be used to estimate brain network changes after perinatal stroke. Finally, our main goal was to analyse cognitive outcome in children after perinatal stroke, and to identify correlations between functional network properties and specific cognitive functions.
Hypotheses: 1. EEG derived functional brain networks reflect brain maturation changes through early life. 2. The functional brain networks of children after perinatal stroke will have a longer characteristic path length, lower global efficiency, a lower clustering coefficient, lower modularity, lower small world index, and lower interhemispheric strength. 3. Poorer cognitive outcome will correlate with longer characteristic path length, lower global efficiency, lower clustering coefficient, lower modularity, lower small world index and lower interhemispheric strength.
Methods: In the first part of our research, we analysed two high-density resting-state EEG data sets (rs-EEG) in healthy children and adolescents. We recorded a 64-channel EEG and calculated functional connectomes in 27 participants aged 5–18 years. To validate our results, we used publicly available data and calculated functional connectomes from a 128-channel rs-EEG in another 86 individuals aged 6–18 years. We analysed connectomes on the basis of functional connectivity in the alpha, beta and theta frequency bands.
In the second part of our study, we explored functional brain networks after perinatal stroke, using the same methodology in 12 participants, aged 5–14 years and compared them to 16 healthy controls. Each subject from the test group was compared with multiple control subjects, matched by sex and age. Functional connectomes were calculated from the alpha frequency band, and then the differences in network graph metrics between the two groups were analysed. We studied whole brain networks and intrahemispheric networks, as well as the correlation between lesion volume and brain network changes.
In the last part of our research, we performed a matched test-control study of functional brain network properties and cognitive outcome after perinatal stroke. Using the same methodology, we analysed the connectomes of 24 children with a history of perinatal stroke and compared them to the functional connectomes of 24 healthy controls. All participants underwent psychological evaluation to estimate their cognitive performance and any deficits in specific domains. After the data from the EEG recordings and psychometric tests were gathered, we compared the graph metric values and psychometric test results between the two groups and analysed the correlations between the network characteristics and specific cognitive functions.
Results: When studying age-related changes in healthy children and adolescents, we observed an age-related increase in characteristic path length, clustering coefficients and interhemispheric strength in the alpha and beta frequency bands. We also observed an increase in individual peak alpha frequency with age. When studying functional brain networks after perinatal stroke, we found evidence of functional network disruption many years after the insult, and the scale of the brain network changes appeared to be influenced by the lesion volume. The networks were more segregated after perinatal stroke, with a higher clustering coefficient and small world index, but they had a less modular structure. Higher values of characteristic path length and interhemispheric strength indicated higher synchronization/functional connectivity at whole brain, intra- and interhemispheric levels. When it comes to cognitive outcome, after perinatal stroke children had a lower intelligence quotient (IQ), slower processing speed, poorer working memory, and poorer visual-motor integration, and scored higher rates of omissions, commissions, and preservations in a continuous performance test. Network modularity correlated positively with IQ and processing speed. The graph metrics, indicating network segregation (clustering coefficient) correlated negatively with IQ, and positively with impulsive behaviour. Impulsive behaviour also correlated positively with graph metrics, reflecting increased network connectivity (characteristic path length and interhemispheric strength).
Conclusions: Our research confirmed that high-density EEG can be used to study functional brain networks. We showed that after perinatal stroke functional brain networks remain disrupted years after the injury, and that changes exist both on a whole-brain network scale and intrahemispherically. Functional connectomes had a less modular structure, and increased interhemispheric, intrahemispheric and whole-brain connectivity after perinatal stroke. Various graph metrics correlated with specific cognitive functions. Individuals with a less modular brain network organization had lower IQ and slower processing speed. Individuals with higher network segregation and increased functional connectivity showed indices of impulsive behaviour on a continuous performance test.
|
