Extracting event-related field components through space-time ICA: a study of MEG recordings from children with ADHD and controls

Event-related studies have provided indirect evidence that Attention Deficit Hyperactivity Disorder (ADHD) children have abnormalities in signal detection and discrimination, and in information processing. Moreover, studies suggest that there exist very low frequency fluctuations modulating underlying neuronal events. This paper presents an event-related fields (ERF) study involving the analysis of magnetoencephalographic (MEG) recordings of children with ADHD and controls during selective attention and perceptual, stimuli-based tasks. A specific form of blind-source separation — space-time independent component analysis (ST-ICA) — is used to isolate the early M100 responses within the data, which are indicative of selective attention deficits. The properties of the extracted responses, namely the amplitude and latency, as well as the power spectral densities of their inter-trial variations are then analyzed. Preliminary results demonstrate the ability of ST-ICA to extract relevant components from multi-dimensional, noisy, ERF data, and reveal differences in the amplitude and latency variations of the M100 responses of the two groups