EVALUATION OF CONNECTIVITY MEASURES AND ANATOMICAL FEATURES FOR STATISTICAL BRAIN NETWORKS

Statistical brain connectivity is a relatively new approach for inferring large scale anatomical organization for both cortical and subcortical structures. This paper presents a a comparison of network connectivity measures and anatomical features used for extraction of such networks. In this paper, we use structural information from three cortical features, i) area, ii) gray-matter volume, and iii) cortical thickness. Based upon these features, the connectivity graph is discretized at different sparcity levels and both global and local efficiencies of the network structure are computed using i) correlation, ii) partial-correlation, and iii) mutual information. The results show that different aspects of the connectivity is captured by different structural features and connectivity measures.