Abstracts

Rationale: We aimed to test the utility of a method by which we measure the presence, and extent, of cognitive responses in epileptic tissue and study the effect of pathological high frequency oscillations within the epileptic tissue on local physiological responses to relevant cognitive stimuli and behavior in a memory task.  Methods: We recruitedsix patients with intracranial electrode coverage in either visual association cortex (N=3) or in the medial temporal lobes (MTL, N=3) who participated in site-relevant cognitive experiments. We identified the epileptic focus in each patient and measured their spontaneous high frequency oscillations (HFOs, 80 - 500 Hz) as well as stimulus-locked physiological responses in high frequency broadband (HFB, 70 - 180 Hz) range and explored their interaction and behavioral correlates. We compared the temporal and spectral profiles of HFOs and HFB signals, and calculated the probability of temporal discordance during the same behavioral tasks. In addition, we introduce a computerized method on the basis of which pathological and physiological high frequency activities can be automatically differentiated, and performed the prediction with a supervised SVM classifier. Results: We successfully developed a method that automatically separates HFOs from physiological high frequency activities using a set of novel features, achieving an AUC value of 0.98 using SVM classifier. In all subjects, we found abundant spontaneous pathological HFOs as well as normal physiological responses to relevant cognitive stimuli in the epileptic sites, but these responses were likely to be missed or delayed when spontaneous HFOs occurred approximately 850-1050ms before, till about 150-250ms after, the onset of relevant cognitive stimuli. Ongoing spontaneous HFOs in the MTL also significantly impacted the subjects' reaction time, hit rate, and confidence scoring during a memory task.  Conclusions: Our findings suggest that non-lesional epileptic brain structures show normal physiological responses to cognitive stimuli, but these responses are transiently impaired by preceding spontaneous high frequency oscillations for about 800-1100ms with subsequent behavioral effects. We offer clinicians a quantitative tool for differentiating pathological and physiological activities in epileptic sites and indirectly assessing their cognitive reserve function and approximating the risk of resective surgery.  Funding: No funding