Abstracts

Rationale: Direct cortical stimulation is crucial for identifying and preserving the language function for patients with seizure onset zone or lesions adjacent to the language networks. After-discharges, byproducts of cortical stimulation, are found to be more frequently triggered in the superior and middle temporal cortex with a lower current threshold, impeding reliable assessments of the language function. Additionally, it is practically challenging to test all implanted electrodes and have patients repeatedly participated in language tasks. High-gamma activities recorded passively during a language task without any electrical stimulation could potentially localize the areas involved in language processes, which may help narrow down the areas that need to be tested with cortical stimulation. In this study, we will explore the feasibility of using high-gamma activities triggered during a timed picture naming task to map the areas involved in language processes and determine their roles in speech generation.

Methods: The picture naming task utilized in this pilot study consists of 25 simple objects. Each picture was presented for 3 seconds following by a 3-second fixation. StimTracker Quad (Cedrus) was controlled by a Python script to present the picture naming task and deliver event markers to the intracranial EEG recording system (Natus Quantum). The intracranial EEG was recorded at 2048 Hz. Patients (n=4, left dominant for language) were instructed to vocally name the pictures and remained still during the task. Time-frequency analysis was performed on the recorded data. The averaged power of the high-gamma frequency band (80-110 Hz) was compared between task and control segments using the Wilcoxon Signed-Rank test.

Results: Among all 551 implanted stereo-EEG electrodes, 118 electrodes (~21%) showed a significant power increase in the high-gamma frequency band during the picture naming task (p< 0.05), in comparison to the fixation. Vision-related areas in the occipital lobe showed an early increase around 0.5-1 second after the picture presentation. Subsequently, anterior frontal regions, excluding Broca’s area, had some activations around 1 second. Areas in the left-side temporal lobe, consistent with language dominance, usually showed power increases in high-gamma band later, ranging from 1 to 2.5 seconds after the picture onset.