Abstracts

Rationale: Surgical resections for epilepsy or neoplasms are a common cause of naming deficits. While various brain regions have been hypothesized to support separable language processes, the complete cortical language network and its functional mapping to naming remain unclear. However, such knowledge could improve the localization of critical brain regions in order to preserve language function. We used electrocorticography (ECoG) to identify cortical brain networks involved in lexical cognition in the context of heteromodal sensory input modalities with convergent design.

Methods: Data were obtained from patients with epilepsy who underwent intracranial electrode implantation. Recordings were acquired during three cued naming tasks where patients were asked to name objects in response to pictures (n=120; 16,284 electrodes), auditory descriptions (n=82; 11,215 electrodes), and orthographic descriptions (n=68; 9,282 electrodes). High gamma power (65-115 Hz) was used to measure cortical engagement throughout the tasks, and electrode recording zones were defined on the cortical surface. Surface-based mixed-effects multilevel analysis (SB-MEMA) was used to estimate group-level high gamma activity for each task and compared to identify regions with significant activation across tasks.

Results: Analyses showed heteromodal activity in the left inferior frontal gyrus (IFG), the left middle fusiform gyrus (mFus), and the left intraparietal sulcus (IPS) following initial sensory processing in response to pictures, auditory descriptions, and orthographic descriptions. Peak activation of mFus occurred 150 ms prior to stimulus offset for auditory (15.2%, p < 10^-6) and orthographic (21.4%, p < 10^-6) naming and at 350 ms following picture onset (39.0%, p < 10^-6) for visual naming. Peak activation of IPS occurred just prior to stimulus offset for auditory (38.1%, p = 7.7*10^-5) and orthographic (27.5%, p = 0.008) naming and at 450 ms following picture onset (43.0%, p < 10^-6) for visual naming. Peak activation of IFG also occurred just prior to stimulus offset for auditory (30.1%, p < 10^-6) and orthographic (40.1%, p = 10^-6) naming and at 500 ms following picture onset (36.4%, p < 10^-6) for visual naming. There was also heteromodal activity in the left posterior middle temporal gyrus (pMTG) for auditory and orthographic descriptions but not for pictures with peak activation occurring prior to the convergence of all three tasks at 250 ms before stimulus offset for auditory (34.1%, p < 10^-6) and orthographic (25.5%, p = 0.0035) naming.

Conclusions: These results reveal that a shared, heteromodal brain network consisting of IFG, mFus, and IPS supports lexical access. Furthermore, our findings also implicate the role of pMTG in phonological access as opposed to lexical retrieval.  Altogether, this work further characterizes the functional roles of key brain regions within language networks and provides important insights that could influence surgical approaches to minimize postoperative deficits.

Funding: NIH NIDCD R01 DC014589