Topological Characterization of the Multi-feature based Network in Patients with Alzheimer's Disease and Mild Cognitive Impairment

Abstract

Characterization of morphological organization pattern in individual brain has long been an open question. Recently, a novel single-subject network that built upon multiple morphological features (MFN) was introduced [1], which exhibited extraordinary power in diagnosing patients with Alzheimer's disease (AD) and mild cognitive impairment (MCI). This study aims to characterize the cortico-cortical topological organization in MCI and AD cohorts via the MFN, and uncover the structural substrate that leads to the high classification accuracy. The MFNs were constructed for 165 normal controls (NCs), 221 patients with MCI, and 142 patients with AD from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Results from graph theoretical analysis showed ‘small-world’ property and modular organization of the MFN in three groups; as well as increased local clustering and characteristic path length, and altered hub regions in patients with AD and MCI. More importantly, we found the primary atrophic regions were accompanied by increased number of in-flow connections; whereas, the connections that linked to the less atrophic regions were mostly out-flow connections. This phenomenon, we speculated, might indirectly reflect the trophic supporting effects in the brain. Our results demonstrated the basic organizational principles of the MFN and its changes in patients with AD and MCI, providing important implications of what makes the MFN powerful in auto-diagnosis of mental disorders.