Abstracts

Effects of Spatial Downsampling on Source Localization of Interictal Events Recorded with 128 Channels

Abstract number : 1.251
Submission category :
Year : 2001
Submission ID : 2008
Source : www.aesnet.org
Presentation date : 12/1/2001 12:00:00 AM
Published date : Dec 1, 2001, 06:00 AM

Authors :
K.J. Eriksen, PhD, Electrical Geodesics, Inc, Eugene, OR; D.M. Tucker, PhD, Electrical Geodesics, Inc., Eugene, OR; N.K. So, MD, Neurodiagnostics, Legacy Health System, Portland, OR

RATIONALE: High density scalp EEG recording may provide additional useful localizing information for epileptic activity. Michel et al. [2001] have reported recently that, with a weighted minimum norm source localization technique, 128 channel EEG provided significantly better results than 32 channel EEG. We wished to determine if discrete, spatiotemporal dipole modeling would show a similar effect.
METHODS: We recorded 128 channel scalp EEG on four patients. The nature of the electrode application system (Geodesic Sensor Nets) precluded long recording times, so only one hour of high-density EEG was recorded with each patient. Interictal activity consisting of spikes and sharp waves was identified and marked by a clinical electroencephalographer (NKS). One second EEG epochs surrounding spikes were analyzed with the BESA2000 program using a four sphere head model and regional sources. Optimal source configurations (from one to six regional sources) were detemined with a genetic algorithm. The EEG was then spatially downsampled from 128 to 32 channels to simulate clinical recording with a typical 10-10 montage. The source modeling was then repeated with the reduced data set.
RESULTS: As the number of sources being modeled increased, the source locations found with 128 channels were more stable than those found using 32 channels. In general, with 128 channels, as each new source was added, the previous source locations stayed close to where they were originally found. Each new source was localized to a new area, or in some cases a previous source split into two sources straddling the old one. With 32 channels, however, there was a lack of consistency seen as new sources were added. Previous sources changed positions significantly, and many more were localized to positions outside the brain case.
CONCLUSIONS: While we cannot judge how successful the source modeling itself has been until after surgery and long-term follow-up, the use of high density EEG data in the modeling appears to provide more robust and consistent results using discrete dipole spatiotemporal analysis.

Michel CM, Lantz G, Spinelli L, Thut G, Perrig S, Ducommun C, Blanke O, and Seeck M, [dsquote]Localization of Interictal Epileptic Activity with 128-channel EEG[dsquote], Abstract accepted for the 7th Annual Meeting of the Organization for Human Brain Mapping, Brighton, June 10-14, 2001.
Disclosure: Salary - Electrical Geodesics, Inc. (Eriksen, Tucker); Ownership - Electrical Geodesics, Inc. (Tucker); Materials - Electrical Geodesics, Inc.; Stock - Electrical Geodesics, Inc. (Eriksen)