Cortical surface shape assessment via sulcal/gyral curve-based gyrification index

Abstract

We propose novel gyrification index (GI) based on sulcal and gyral curves on the human cortical surface. Instead of using the widely employed methods based on Euclidean or geodesic kernels of uniform size, we biologically determine local regions based on a curve-wise correspondence between sulcal and gyral curves. Specifically, we initially extract sulcal and gyral curves from the cortical surface. For each sulcal point, we then find two corresponding gyral points associated with the closest gyral curves. This process requires a categorical optimization that generally possesses an intractable parameter space, which is addressed via a novel evolutionary algorithm. Finally, we propagate sparse measurements of the proposed sulcal GI at each sulcal point to the entire cortical surface in order to yield a complete GI map. The experimental results show that our measurement achieves reasonable reliability across a scan-rescan dataset and provides a complementary information of cortical folding, compared with a recent kernel-based metric in a longitudinal study. © 2016 IEEE.