Mobile Brain-Body Imaging of ASD Participants During Natural Movement

Background: Motor dysfunction (e.g., abnormalities of gait, balance, muscle tone, head and eye movement and coordination) is a prominent feature in autism that may be a contributing factor in cognitive and social impairments. Isolating specific underlying mechanisms that lead to a variety of motor impairments (e.g., timing, anticipation) would inform effective intervention that may in turn improve not only motor competence but also behavioral problems that are affected by motor dysfunction.  We are currently conducting a first quantitative study of motor function in ASD that relates motor dysfunction to both underlying brain structure and function as well as to behavioral. Central to this study is the ground-breaking mobile brain/body imaging (MoBI) system (Makeig et al., 2009).  This novel system uses a combination of cameras and LED emitters for motion capture with simultaneous collection of high-density scalp EEG to quantify the accuracy, coordination, and timing of motor functions and to allow modeling of cortical network function during specific phases of motor operations. Development of analytic methods is critical to the success of this work.

Objectives: The goals of the experiment presented here were 1) to test the feasibility of integrating EEG with motor activity collected during a task in which the participant moved freely around a large room, and 2) to determine the most appropriate analytic methods for these novel multi-modal datasets.

Methods: We recorded 128-channel EEG and 66-sensor motion-capture data from boys between the ages of 13-17 (ASD and typically developing controls).  The task, embedded in a simple video game, required participants to walk across a large room in order to reach a cartoon ‘alien’ projected on a wall. On STRAIGHT trials, the participant could reach the alien by walking straight across the room; on TURN trials the alien would move to one of the adjacent walls requiring the participant to change direction.

Motion-capture data was used to determine the point at which the participant began a turn to follow the alien. EEG data were decomposed using Independent Components Analysis, and activity of individual Independent Components (ICs) was time-locked to the turning time point identified from the motion-capture data.  For each IC of interest, we performed time-frequency analyses, and estimated the cortical solution; we also modeled causal relationships between pairs of ICs.

Results: We successfully separated cortical EEG activity from movement-generated artifact.  After integrating motion and EEG data, we found the markers best related to task behavior, and used these markers to identify EEG networks associated with that behavior.  We identified ICs from two categories: those whose activity was associated with leg/foot movement, and those whose activity was associated with, but more importantly preceded the act of turning.  Cortical source modeling indicated that these components were located in brain areas consistent with the motor network.

Conclusions: This pilot work established the feasibility of recording and analyzing EEG activity from freely-moving participants.  Modeling causal relationships between the various EEG networks revealed interactions that will allow us to differentiate between motor planning and execution.