Ince, Nuri F.

Abstract BACKGROUND Few studies have examined concordance between intraoperative electrocorticography (ECoG) and direct electrical stimulation (DES) language mapping, especially across varied naming tasks. METHODS Twelve patients (Mdn age=59; 75% HGG) underwent awake craniotomy with ECoG and DES language mapping. Mapping utilized iPad stimulus presentation coupled to a 4x8 electrode array for both ECoG and DES. Time-frequency ECoG gamma band activity (Early and Late) was visualized for 3 naming paradigms (Object, Action, Auditory). Bipolar DES was also conducted for each task at each electrode pair. ECoG activations and DES hits were overlayed on preoperative MRI. DES hits represented ground truth for eloquence. ECoG accuracy was examined for each task with 3 different criteria: exact (identical electrodes for ECoG activation and DES hits), adjacent (ECoG <1cm of DES hits), and adjacent within a predefined peritumoral region of interest (<2 cm from boundaries). RESULTS 32 DES hits (40% Auditory, 32% Action, 28% Obj) were observed across patients. Exact ECoG-DES accuracy ranged from 78%-90% [Object: Sens=11%, Spec=91%, PPV=6%, NPV=95%; Action: Sens=40%, Spec=93%, PPV=24%, NPV=97%; Auditory: Sens=69%, Spec=78%, PPV=19%, NPV=78%]. Adjacent ECoG-DES accuracy ranged from 82%-91% [Object: Sens=33%, Spec=92%, PPV=18%, NPV=97%; Action: Sens=44%, Spec=93%, PPV=25%, NPV=97%; Auditory: Sens=91%, Spec=81%, PPV=37%, NPV=99%]. When restricted to the peritumoral region, adjacent ECoG-DES accuracy ranged from 94%-97% [Object: Sens=14%, Spec=99%, PPV=25%, NPV=97%; Action: Sens=25%, Spec=99%, PPV=25%, NPV=99%; Auditory: Sens=100%, Spec=94%, PPV=54%, NPV=100%]. For patients with DES hits (N=10), all had at least 1 task activation <1 cm of DES hits, with 7/10 having convergence of activations across at least 2 tasks. CONCLUSIONS Negative ECoG mapping is strongly predictive of non-eloquence confirmed by DES for all naming tasks. Additionally, gamma band activation elicited by auditory naming within the peritumoral region is strongly predictive of eloquence. Findings support task-based ECoG language mapping for prediction of eloquence and identification of targets for DES confirmation.

Abstract INTRODUCTION Accurately mapping sensorimotor regions is crucial for preserving motor and sensory functions in the treatment of gliomas located in the peri-Rolandic areas of the brain. OBJECTIVE We examined the use of high-density electrocorticography (ECoG) for passive mapping of sensorimotor regions in treating peri-Rolandic gliomas by analyzing the spatial-temporal and spectral features of real-time somatosensory evoked potentials (SSEPs), and their spectral patterns in differentiating states of consciousness. METHODS During surgery, SSEPs were recorded using high-density ECoG on fourteen patients (n=14), both in anesthetized and awake states. Neural data from 0.6Hz median nerve stimulation were captured at 2.4kHz and processed in real-time using MATLAB Simulink. The system displayed SSEPs’ peak activations as a 2D heat map on a screen, particularly around the 20ms time point (N20), and generated the spectral power in the gamma range using the Stockwell transform. Accuracy in delineating the central sulcus (CS) was assessed using the ROC curve, and paired t-tests compared gamma oscillations between states of consciousness. RESULTS Patients consistently showed clear discrimination between anterior and posterior channels at the 20ms time point, with 93.6±14.9% accuracy. Color contrast delineated the CS, correlating with the sulcus in the 3D reconstructed magnetic resonance imaging of each patient. Late gamma (60–250 Hz) modulation occurred 50 ms after stimulation onset, extending up to 250 ms in the primary somatosensory area. It was suppressed in the anesthetized state and significantly increased in the awake state (delineation accuracy: 81±10.3% and 91±13.4%, respectively). CONCLUSIONS These results demonstrate that SSEPs and gamma modulations can delineate sensorimotor areas and assess consciousness, significantly impacting neurosurgical planning.

Abstract Studying mammalian brain function aids our understanding of human brain evolution. We implanted a beagle with a prototype human neuromodulation platform that measures activity from the brain surface. One year later, a set of simple sensory tasks was performed, finding visual and somatosensory representation in the canine homologs of the expected areas in humans. Surprisingly, the canine anterior ectosylvian gyrus, which is anatomically homologous to human receptive speech areas, was selectively active during independent social reinforcement tasks. This suggests that human speech understanding may have evolved from more general mammalian brain structures that are specialized for social reinforcement.