Alteration of the resting-state brain networks: a multi-aspect study of functional and effective connectivity in Parkinson's disease

Purpose: Parkinson disease (PD) is known as the second most common and growing neurodegenerative disorder in the world. It is widely known that the appearance of motor and non-motor symptoms causes disorders in the brain network of such patients. Methods This study evaluates the brain connectivity of PD patients' (n = 15) in comparison with healthy individuals (n = 15) using functional magnetic resonance imaging (fMRI). First, independent component analysis (ICA) was implemented on the preprocessed data to extract resting-state networks (RSNs) as functional connectivity (FC) for evaluating intra-network connectivity values. Granger causality analysis (GCA) and transfer entropy (TE) are extracted as effective connectivity (EC) methods for assessing the network and regional time courses for assessing inter and intra-network connectivity measures. Finally, F-test was used to find the discriminative regions between the groups. Results Thirty ICA maps were identified as independent components, and seven were selected as RSNs. In terms of spatial maps, we found that patients had altered connectivity between Auditory and dorsal Default Mode Network. Several networks and regions were also discriminative between the groups in inter and intra network connectivity analysis, respectively. Conclusions The results show that EC methods such as GCA and TE are promising in extracting local information of PD. The reason can be considered in terms of being directional and causal in this type of connectivity, which is in relation to the concept of neuronal substrates. Also, TE might be more accurate than GCA, since TE is nonlinear which is consistent with the nature of the data. To the best of our knowledge, there was not any research that employed EC and ICA on Parkinson's resting-state fMRI data, and analyzed it using ICA time courses as well as regional time series.