Abstracts

Circadian Variation in an Electrocorticographic Feature Detected by a Responsive Neurostimulator System

Abstract number : 1.031
Submission category : Clinical Neurophysiology-Computer Analysis of EEG
Year : 2006
Submission ID : 6165
Source : www.aesnet.org
Presentation date : 12/1/2006 12:00:00 AM
Published date : Nov 30, 2006, 06:00 AM

Authors :
1Robert B. Duckrow, and 2Tom Tcheng

An implanted responsive neurostimulator system (RNS[trade], NeuroPace, Inc.) is being developed to treat localization-related epilepsy. The recording and analysis functions of this device detect localized epileptiform activity and trigger responsive electrical stimulation to abort a developing seizure. Algorithmic feature detectors in this device provide measures of ongoing electrocorticographic (ECoG) activity. Based on the observation that epileptic seizures can occur at specific times of the day, we hypothesized that one such feature of the baseline ECoG, the line length, would show circadian variation and that this variation could be interpreted in light of what is known about cortical activity or seizure susceptibility., Data were collected during intracranial monitoring of patients with medically intractable epilepsy as part of a safety and feasibility trial of an external responsive neurostimulator system (eRNS). Using bipolar recordings, the eRNS sampled a total of 87 brain regions from 24 consecutive patients. The ECoG was characterized by calculating line length, a feature sensitive to high frequency content or signal complexity and defined for a uniformly sampled time series as the normalized sum of the absolute value of the difference between consecutive elements. The feature was derived from 32.7 second-long baseline segments that preceded interictal activity during 2 to 54 day monitoring sessions. When collapsed to a 24-hour cycle, a median of 494 values per recording site were available to be fit with a cosine function. These functions were analyzed with respect to time of peak and cerebral location., Line length values were represented by a cosine function ([italic]F[/italic][lt]0.05) in 41 of 54 valid data sets (76%). The average daily variation was [plusmn]7% of the mean. The time of the peak had a bimodal distribution with local maxima at 0530 and 1500 hours. There was no segregation of the peak to day or night based on recordings from cerebral lobes. However, all data sets obtained from lateral or inferior temporal lobe had a peak during the day while none occurred at night (Fisher[apos]s exact test: [italic]p[/italic]=0.03). Peak times from mesial temporal sites were evenly distributed over day and night., The line length feature of the ECoG shows circadian (nychthemeral) variation. The segregation of peak amplitude to either 0530 or 1500 hours suggests a complex modulation of the ECoG that is independent of the sleep cycle but possibly relates to known fluctuations in spectral energy measured from the scalp during wakefulness. Speculative correlations can be made with the propensity of certain seizures types to occur at specific times of the day. This study shows how implanted devices could be used to provide chronic out-patient recording of similar features to monitor seizure susceptibility., (Supported by NeuroPace, Inc., Mountain View, CA.)
Neurophysiology