Combining tomographic single subject, single trial activity into time-dependent grand-summaries of activated areas and functional connectivity

Standard statistical analysis and mutual information are employed to summarize the millisecond-by-millisecond single subject, single trial tomographic estimates of brain activity extracted from biomagnetic signals. The first approach (MFT-SPM) characterizes the spatial profile of significant activations at a particular latency in peristimulus time. The second grand-summary provide a time- and lag-dependent inventory of pair-wise interactions between brain areas. The analysis of median nerve stimulation data is used as an example to demonstrate the advantages and limitations of the grand summaries. Expected but also somewhat surprising, and hence easy to overlook, patterns emerge in the left and right hemisphere activations of each subject. These patterns dominate the grand-average summaries across subjects, leading to established results but also new insights. Finally, the amazing spatiotemporal accuracy achievable in MFT tomographic estimates of activity provides a warning that much and potentially valuable information is lost in the average and grand-average summaries.