Spatial alignment of human cortex by matching hierarchical patterns of functional connectivity

Abstract

A novel cortical surface registration method is proposed to spatially align inter-subject cortical surfaces by maximizing the similarity of their hierarchical patterns of local functional connectivity extracted from fMRI data. The cortical surface with fMRI data is characterized by functional connectivity information for each vertex of the surface to its spatial neighbors on the cortex sheet at multiple spatial scales with a hierarchical structure. Each vertex's functional connectivity information at a given scale is represented as a probability distribution of functional connectivity measures between functional signals of the vertex and its neighbors so that the functional connectivity information is independent on the vertices' spatial locations. The cortical surface registration is implemented under the spherical demons framework by matching different cortical surfaces' functional connectivity information. The experimental results for the registration of both task and resting-state fMRI data across different subjects have demonstrated that the proposed algorithm could improve the functional consistency of cortical surfaces of different subjects, and compared favorably with state-of-the-art cortical surface registration techniques.