Delayed brain development of Rolandic epilepsy profiled by deep learning-based neuroanatomic imaging.

Area of Science:

• Neuroimaging
• Developmental Neuroscience
• Pediatric Neurology

Background:

• Rolandic epilepsy (RE) is often considered a brain developmental disorder.
• Previous neuroimaging studies have not definitively confirmed developmental delay in RE.
• Assessing brain age offers a novel approach to evaluate developmental trajectories.

Purpose of the Study:

• To investigate brain development in Rolandic epilepsy using a deep learning-based neuroanatomic biomarker.
• To quantify brain age differences in children with RE compared to typically developing children.
• To explore the relationship between brain age, neuroanatomy, and cognitive function in RE.

Main Methods:

• A 3D-CNN brain age prediction model was trained on 1155 typically developing children's MRI data.
• The model was applied to a dataset including 167 RE patients and 107 controls.
• Brain-predicted age difference was calculated and correlated with clinical and cognitive variables.

Main Results:

• The 3D-CNN model demonstrated robust brain age prediction in typically developing children.
• Children with RE showed an average brain age delay of 0.45 years.
• Delayed brain age correlated with Rolandic region neuroanatomic changes and attention deficits.

Conclusions:

• This study provides neuroimaging evidence supporting delayed brain development in Rolandic epilepsy.
• Imaging phenotypes in RE include increased gray matter and decreased white matter volume in Rolandic regions.
• Brain age delay in RE is associated with specific morphometric alterations and attentional deficits.