Abstracts

Rationale: Subdural hemorrhage (SDH) is a potential complication associated with placement of subdural electrode grids in the course of EEG monitoring. In our experience, hemorrhage may not immediately manifest clinically, and detection by visual interpretation of EEG changes may be delayed for a period of several hours. We have observed that the associated EEG parameter eventually perceived is most commonly a gradual decrease in high frequency power (HFP). We therefore set out to create an automated EEG screening tool for detection of changes associated with SDH. Such an automated algorithm would allow for early detection and intervention of potentially devastating, iatrogenic complications. Methods: In order to test our hypothesis that the development of SDH may be reliably detected by a drop in high frequency energy measured with a subdural electrode grid, we initially assessed four potential variables: compressed spectral array (CSA), spectral edge, alpha/delta ratio, and HFP using Persyst software. While all four instruments appeared equal in accuracy, eventually HFP was chosen as the screening tool of choice based on ease of interpretation. Data from three patients known to have this outcome were retrospectively analyzed. Three patients who had subdural electrode grids placed and did not experience hemorrhage were also studied for comparison. In each case, pre and post-hemorrhage periods were determined retrospectively based on brain imaging (CT or MRI) findings. Samples of continuous intracranial EEG recordings for both pre and post-hemorrhage periods were analyzed for each day that data was available, allowing for HFP trend analysis. We chose a frequency band of 8 – 30 Hz based on our initial observations. Results: For each of the cases of SDH studied, a gradual and persistent trend of decreasing HFP was noted. For each non-hemorrhagic case, transient false positive drops in HFP were noted. In contrast to the hemorrhagic cases however, these decreases were state-dependent and invariably followed by return to prior power levels. Causes of false positives included administration of benzodiazepines, slow wave sleep, and post ictal changes. In one case of SDH, multiple CT scans were available and showed evolution of a progressively larger hemorrhage. Frequency analysis in this case suggests a relationship between extent of hemorrhage and degree of high frequency loss.Conclusions: The purpose of this study was to provide data on the utility of automated monitoring of HFP loss for the early detection, and intervention of SDH occurring as a complication of placement of intracranial electrodes during planning for surgical treatment of epilepsy. Our results indicate that a persistent decrease in HFP may be used to detect SDH in this setting, allowing for earlier treatment of a potentially devastating iatrogenic complication.