Abstracts

Rationale: For patients with drug-resistant focal epilepsy in the dominant hemisphere, language mapping is essential for identifying eloquent cortex to be spared during surgery. The current gold standard uses electrical cortical stimulation (ECS) mapping in word confrontation paradigms; however, such mapping is time intensive, depends on patient cooperation, and may fail to reveal the full language network supporting speech processing in more naturalistic conditions. Recent studies using functional magnetic resonance imaging (fMRI) during naturalistic movie-watching conditions suggest putative language networks may be mapped without the need of an overt task1. However, the temporal and spectral dynamics of these networks, and how they may be modulated by speech input, is unknown. Here we report the results of a novel approach for passive language mapping. We used correlations between phase coherence in the brain signal with the auditory language input to identify electrodes associated with movie watching, language processing and neural communication.Methods: Four patients with drug-resistant epilepsy were monitored with a left lateral 64 channel subdural grid as part of their pre-surgical monitoring for epilepsy surgery. Electrophysiological signals were acquired while patients watched a short (approx. 3 min) movie clip, which contained periods of speech and silence. Using Fieldtrip and custom software in MATLAB, we determined the extent to which fluctuations of high-frequency broadband power tracked the audio envelope of the videos. Connectivity of electrodes that showed high correlation with the audio signal was assessed with phase coherence within theta and high gamma bands to identify other nodes within the language network.Results: For 3 of 4 patients, the amplitude envelope of the video’s acoustic signal was strongly correlated with high frequency power of electrodes located in auditory cortex (p < .001). For each patient, correlations between the brain’s phase coherence and the video’s auditory envelope revealed a network of sites located within inferior frontal, inferior and superior temporal, and temporal-parietal areas, whose connectivity fluctuated as a function of speech content. Comparison with sites identified by ECS as important for expressive language will confirm whether we have identified additional nodes in the network that support naturalistic language processing.Conclusions: Measuring connectivity during passive movie watching may provide a novel method for mapping language networks that support naturalistic speech processing, without the need for an overt task. This could be particularly useful in paediatric epilepsy populations, for whom language mapping with cortical stimulation is more challenging.