Source reconstruction of clinical resting-state EEG reveals differences in power and functional connectivity in children with developmental dyslexia

Abstract

Background: Developmental dyslexia is a neurodevelopmental disorder characterized by significant difficulties in reading and spelling. Despite lacking routine neuroimaging markers for dyslexia, recent resting-state electroencephalography (EEG) studies have detected atypical functional connectivity in children with dyslexia compared to controls. These methods are based on measures of EEG data at a sensor-level, but it remains unclear if routine clinical resting-state EEG can be used to detect source-level differences in power or functional connectivity (FC) between children with dyslexia and controls. It is also unknown if differences in these EEG metrics correlate with difficulties in reading and spelling. Methods: Using retrospective data, we investigated the source-reconstructed power and FC of 70 children with recently diagnosed dyslexia and 50 typically developing controls. We analyzed 50 seconds of awake resting-state routine clinical EEG in five frequency bands (1-29 Hz) using power, imaginary part of coherency (ImCoh), and weighted phase lag index (wPLI). Additionally, we calculated correlations between power or FC and IQ, reading, and spelling performance. Results: Children with dyslexia had a decrease in theta FC in left temporo-parieto-occipital regions and an increase in alpha FC in left fronto-temporo-parietal regions. A decrease of theta FC was observed for right parieto-occipital regions and an increase of alpha FC in right inferior fronto-temporal regions. Furthermore, children with dyslexia demonstrated lower power in delta and theta within the left parieto-occipital regions. An age-stratified sub-analysis indicated that children with dyslexia in 5th-8th school grades exhibit greater alpha FC mainly in left fronto-temporo-parietal regions. Finally, lower scores in spelling showed a positive and significant association to theta power within left parieto-occipital regions in dyslexia. Conclusions: Significant group differences in power and FC in the theta-alpha range in left cortical language and visual regions, as well as in multiple resting-state networks (RSNs), suggest abnormal oscillations as a pathophysiological sign of dyslexia reading and spelling deficits. These findings demonstrate the potential of source-reconstructed clinical routine EEG data to inform clinicians about brain network alterations in neurodevelopmental disorders such as dyslexia.