Etay Ziv1, Olga Tymofiyeva1, Christopher P. Hess1, Donna M. Ferriero2, A James Barkovich1, Duan Xu1
1Department of Radiology & Biomedical Imaging, UCSF, San Francisco, CA, United States; 2Department of Pediatrics, UCSF, San Francisco, CA, United States
Structural connectivity networks derived from diffusion MRI vary with choice of brain parcellation. In adults, parcellation and subsequent assembly of the large-scale human connectome relies heavily on brain atlases. However, in the context of the rapidly developing pediatric brain, such approaches introduce problematic biases. Here we present a network-driven brain parcellation that does not rely on brain atlases and propose a method to define the optimal number and size of nodes.