Probabilistic mapping of language networks from high frequency activity induced by direct electrical stimulation

Direct electrical stimulation (DES) at 50 Hz is used as a gold standard to map cognitive functions but little is known about its ability to map large-scale networks and specific subnetwork. In the present study, we aim to propose a new methodological approach to evaluate the specific hypothesis suggesting that language errors/dysfunction induced by DES are the result of large-scale network modification rather than of a single cortical region, which explains that similar language symptoms may be observed after stimulation of different cortical regions belonging to this network. We retrospectively examined 29 patients suffering from focal drug-resistant epilepsy who benefitted from stereo-electroencephalographic (SEEG) exploration and exhibited language symptoms during a naming task following 50 Hz DES. We assessed the large-scale language network correlated with behavioral DES-induced responses (naming errors) by quantifying DES-induced changes in high frequency activity (HFA, 70-150 Hz) outside the stimulated cortical region. We developed a probabilistic approach to report the spatial pattern of HFA modulations during DES-induced language errors. Similarly, we mapped the pattern of after-discharges (3-35 Hz) occurring after DES. HFA modulations concurrent to language symptoms revealed a brain network similar to our current knowledge of language gathered from standard brain mapping. In addition, specific subnetworks could be identified within the global language network, related to different language processes, generally described in relation to the classical language regions. Spatial patterns of after-discharges were similar to HFA induced during DES. Our results suggest that this new methodological DES-HFA mapping is a relevant approach to map functional networks during SEEG explorations, which would allow to shift from local to network perspectives.