Evaluation of Preprocessing for Functional Magnetic Resonance Imaging Brain Networks

In order to explore the effect of using different preprocessing methods on fMRI image data, this study used two different preprocessing tools to process the same batch of fMRI data and analyzed the results. The trial recruited 6 subjects with Alzheimer’s disease from the Alzheimer’s Disease Neuroimaging Project Database (ADNI). In this study, MATLAB-based software packages GRETNA and DPARSF were used to preprocess the collected subjects, and brain networks were constructed to calculate the network characteristics including clustering coefficient (C
p
), characteristic path length (L
p
), and global efficiency (E
g
), local efficiency (E
1
) and small-worldness ($\sigma$). The results showed that there was no significant difference in the network characteristics of the two groups and functional connection network of the whole brain, while there existed a significant difference in the distribution of core nodes in the whole brain. It was concluded that the chosen preprocessing methods had significant effect on the network construction, analysis and visualization of image data.