Normalization enhances brain network features that predict individual intelligence in children with epilepsy.

Area of Science:

• Neuroscience
• Medical Imaging
• Machine Learning

Background:

• Cerebral network architecture correlates with IQ in children with epilepsy.
• Subject-level prediction of cognitive function using network analysis remains a challenge.

Purpose of the Study:

• To compare two network normalization strategies for optimizing subject-level predictions.
• To assess the relationship between brain network architecture and brain function in pediatric epilepsy.

Main Methods:

• Resting-state fMRI data from 27 children with epilepsy (ages 8-18).
• Brain network construction using anatomic parcellation in patient and template spaces.
• Calculation of network metrics (clustering coefficient, path length, etc.).
• Machine learning for IQ prediction using normalized and non-normalized network metrics.

Main Results:

• Brain networks exhibited small-world properties.
• Normalization strategies, including transformation to standard space and random network comparison, significantly improved prediction accuracy.
• Accounting for population heterogeneity enhanced predictive power.

Conclusions:

• Network normalization is crucial for accurate subject-level prediction of cognitive function in children with epilepsy.
• Quantitative network approaches hold promise for clinical applications in pediatric neurological disorders.