Assortativity changes in Alzheimer's diesease: A resting-state FMRI study

There is a growing trend toward using resting-state functional magnetic resonance imaging (rs-fMRI) data in studying brain network, and finding altered brain regions in neurological and psychiatric disorders. In this paper, we investigated the brain network of 15 normal and 15 Alzheimer subjects, using rs-fMRI data. To overcome the shortcomings of anatomical atlases in functional connectivity studies, we defined the regions based on functional atlases. We produced two functional parcellations: an individual parcellation for each subject separately, and a group-wise parcellation based on the whole dataset. For each subject, two functional graphs were constructed through these atlases. Finally, common network measures such as clustering coefficient and also assortativity coefficient were extracted from the resulted graphs. Comparison between corresponding network measures in patients and normal groups indicate that assortativity coefficient is significantly lower in both group-wise atlas-driven graphs (p-value = 0.0429), and individual atlas-driven graphs (p-value = 0.0334). Reduced assortativity coefficient, which might reveal disturbed primary order in vertices' degrees, can help in better distinguishing Alzheimer's subjects from normal ones.