Abstracts

Rationale: Intraoperative Electrocorticography (ECoG) is used to demarcate epileptic cortex. Focal attenuation (FA) on ECoG can help localize the epileptogenic zone but can be hard to recognize visually and is subject to varying interpretation. Quantitative EEG techniques provide an alternate, objective means to assess FA. We compared agreement between a fast Fourier transform (FFT) method and visual assessment of ECoG attenuation.Methods: We sampled 1 ECoG segment from each of 34 epilepsy surgery patients. Selected segments were recorded with 5 x 4 grid, had minimal artifacts, and lasted 1-5 minutes. A board-certified EEGer determined for each ECoG segment which electrodes visually were attenuated and if the attenuated electrodes spatially clustered suggesting an area of FA. FA for each ECoG segment was scored as definite FA, definite no FA and possible FA. Fast Fourier transform (FFT) was applied to the same ECoG segments to create amplitude spectra of the frequency range 14 24 Hz. The sampling rate was 128 Hz with epoch duration of 1 second and epoch step of 1 second. At every epoch, each electrode s amplitude spectral value was divided by the second highest amplitude spectral value for the 20 electrodes on the same grid. This ratio was averaged over the ECoG segment. An averaged ratio value <0.5 was scored attenuated. Overall absolute agreement between the visual and the quantitative assessment of attenuation was determined. Kappa scores were calculated overall controlling for ECoG segment and overall within each of the three groups defined by FA.Results: Of a total 680 electrodes, visual and FFT agreed in 85.7%. The overall kappa adjusted for ECoG segment was .75 (95% CI=.69-.81). In 11 segments with definite focal attenuation, the adjusted kappa was .84 (76-.91); in the 15 with possible focal attenuation, kappa was .64 (.54-.74); and in the 8 with no focal attenuation, kappa was .58 (.38-.77). Excluded from the calculation in this last group were 4 grids in which there was perfect agreement that no electrode was attenuated.Conclusions: Overall there is good to excellent agreement between visual and FFT identification of attenuation. Agreement is best in the group that showed definite focal attenuation. We speculate that when focal attenuation is clearly visible, attenuation most likely results from the epileptogenic zone. Our findings also suggest that when focal attenuation is clearly absent, the quantitative method is unlikely to produce false positives. FFT can be used to help EEGers recognize focal attenuation in the operating room in real time. In addition it can be used in training EEGers to identify attenuation, and to standardize how attenuation is characterized among various epilepsy surgery centers.