Proof of principle for a novel method of detecting brain network patterns during action-observation

Abstract

Objective

To investigate the feasibility of analyzing brain network patterns that correlate with movement observation using an action-observation-based technique.

Background

Understanding brain network dysfunction in Parkinson’s Disease (PD) is challenging. Despite recent advances, temporal and spatial resolution remains limited as does understanding of the pathophysiologic processes these networks reflect. Furthermore, techniques that minimize motoric and cognitive confounds, yet still provide relevant whole-brain network physiology measurement, are urgently needed to improve diagnosis, track disease progression, and understand response to therapy in PD.

Methods

Healthy control participants watched a 12min continuous video of an actor performing real-world arm-hand (AH) and leg-foot (LF) actions while undergoing high-density EEG (256 channels). EEG data were segmented corresponding to AH or LF movement observation. Resulting data were pre-processed, source localized, concatenated across participants and condition, and decomposed by independent components analysis. Dual-regression was used to identify spatial maps for each participant from the full component set to test for statistically significant differences between AH and LF conditions.

Results

AH and LF observation resulted in significantly different network patterns involving regions known to be important in action observation/production including pre- and primary motor, (pre-) supplementary motor, parietal, and insular cortices and basal ganglia and also reflected known motor cortex and putamen somatotopy (p<.05 corrected for multiple comparisons per component; n=12 participants analyzed). AH and LF networks overlapped in putamen, hippocampus, visual cortices, and temporal pole.

Conclusions

This proof of principle pilot study in healthy controls yielded three key observations: 1) whole-brain networks can be measured using this novel technique while avoiding a priori assumptions about network distribution; 2) the novel action-observation-based technique can engage networks in a physiologically relevant manner while avoiding overt-performance confounds; and 3) observation of AH and LF movements involves both spatially distinct and overlapping brain networks. Based on this proof of principle study, expanded analysis including 45 more healthy control participants’ data is ongoing as is pursuit of this paradigm with fMRI and with PD participants.