Abstracts

Rationale: The clinical standard for language mapping in drug resistant epilepsy (DRE) patients undergoing invasive pre-surgical evaluation consists of electrical cortical stimulation (ECS) with stimulus response-inhibition. Such testing is crucially dependent on patient cooperation, precluding it in patients unable to participate in formal language tasks. To investigate an alternative novel method for language mapping this prospective study compared spontaneous conversation related high-γ synchronization (HGS) mapping with traditional task-inhibition based ECS. Methods: All DRE patients admitted for invasive evaluation and able to converse with the investigator were eligible. Electrocorticographic signals and synchronized audio were acquired during a 5 min. resting baseline followed by a 5-10 min. natural conversation between investigator and patient. Using Signal Modeling for Real-time Identification and Event Detection (SIGFRIED) procedure, a statistical model for baseline data was generated, for frequencies between 70-116 Hz. Then, a single score was calculated for each channel representing the probability that the power features in the experimental signal window belonged to the baseline model. Significant electrodes were determined using a Bonferroni corrected Student's t-test (α ≤0.01/channel) which compared the scores during 500ms period preceding speech (listening) onset vs. 500ms period of actual speech (or listening). For plotting activations on the 3D cortical model (-log₁₀p) was used. During clinical ECS, electrodes with language or oral motor functions were regarded as true positives, whereas other electrodes where stimulation was carried out but no such response was observed were regarded as true negatives. Sensitivity, specificity, positive and negative predictive values (PPV, NPV), and classification accuracy were calculated. Results: Seven patients were included (mean age 10.28 ± 4.07 years). In left hemisphere, significant high-γ modulation was seen in posterior superior temporal gyrus (Wernicke's area), inferior peri-rolandic sensorimotor cortex, inferior frontal gyrus (Broca's area) and antero-inferior parietal lobe. Some activation was also seen in basal temporal area. Similar high-γ modulation was also observed in some homologous areas of right hemisphere (Figure). When electrodes with picture naming interruption or oral motor phenomena were taken as true+, the sensitivity and specificity of HGS mapping was 88.89% and 63.64% respectively. The PPV and NPV for HGS mapping were 35.29% and 96.25%, with an overall accuracy of 68.24%. HGS mapping was able to correctly determine all ECS+ sites in 6/7 patients and all false− sites (ECS+, HGS− for visual naming, n=3) were due to only 1 patient (Table). Conclusions: This study supports the validity of language mapping using spontaneous conversation related HGS as compared to traditional response-inhibition ECS in childhood-onset DRE. It also demonstrates high-γ modulation associated with spontaneous conversation to be a functionally specific anatomical localizer of language network.