Speaker
Description
Introduction: Developmental Dyslexia (DD) is a prevalent learning that impairs reading acquisition and fluency. Due to its moderate-to-high heritability, the risk of DD rises by 40% with a first-degree relative's diagnosis. The present study uses a recently developed battery of non-verbal cognitive assessment tasks, the ReadFree Screening Tool (RFST), and multimodal MRI measures, to investigate whether multivariate machine learning can facilitate early DD identification.
Method: We designed a longitudinal study, including two sessions: (1) Second-grade children complete the RFST and 3T MRI scanning. As a fMRI task, we used an adapted version of the RFST’s irregular visual Go/No-Go task. (2) One year later, children are re-assessed with standardized DD diagnosis tests.
17 children (mean age = 7.5, SD = 0.18) have been tested so far. Based on parental reports of familial reading difficulties and developmental cognitive disorders, a familial risk index was calculated.
Results: Behavioral results revealed significant correlations between reading skills, phonological processing, and performance on two RFST tasks, namely the Go/No-Go, tackling inhibitory control mechanisms, and the Rhythmic Tapping task, tackling predictive anticipation and synchronization. Functional MRI results showed No-Go activation in inhibitory-control regions and associations between these activated regions, reading and phonological skills. The familial risk index was negatively associated with activation of the orthographic-phonological brain network.
Discussion: Our preliminary findings suggest that the RFST tasks could help identifying children at risk of DD. The combination of cognitive and multimodal neuroimaging measures is a promising approach for an early detection of DD.
