Stereoencephalography (SEEG) is an increasingly popular, minimally invasive method used to identify and localize seizure onset zones of patients with drug resistant epilepsy. As a part of this procedure, electrodes are inserted deep within the brain and can capture both white matter and grey matter structure activity. Upon visual inspection, white matter and grey matter structures of close distances produce signals with different characteristics. Proper analysis of these differences can lead to better understanding of SEEG interpretation best practices. This study presents a preliminary signal processing analysis of the white matter and grey matter SEEG electrodes for 6 temporal lobe epilepsy patients. Power spectral density, Signal to Noise Ratio (SNR), and Mutual Information measures were extracted and averaged over the interictal, preictal, ictal and postictal seizure periods for information localized to white matter and grey matter structures. Power spectral density measures revealed that white matter power is on average lower than grey matter power for all frequency bands. This difference is especially marked in the ictal seizure time period. Power spectral density of white matter localized electrodes was also compared as a function of distance from the nearest grey matter structure. These measures revealed that power decreases as distance from the nearest grey matter structure increases. This difference was marked in the interictal and ictal seizure time periods. Signal to Noise Ratio comparisons of white matter localized electrodes revealed that electrodes 0 to 3 mm away from grey matter structures show the highest SNR in the ictal period. Mutual information measures are consistent with the power spectral density results as they reveal that mutual information between white matter localized SEEG signals and nearest grey matter localized SEEG signals decrease as a function of distance between the two structures. Overall, preliminary results suggest that grey matter and white matter electrodes within close distances may contain similar information. As this distance increases, the white matter information may be more clinically relevant. This finding may hold future implications for the SEEG procedure methodology. Future steps of this study include increasing the sample size and comparing pre-processing methodologies. 

This study would not be possible without the support of Dr. Kathryn Davis, Andy Revell, Alex Silva, and the Lila R. Gleitman mindCORE Undergraduate Fellowship