University of Minnesota
University Relations
http://www.umn.edu/urelate
612-624-6868

Minnesota Supercomputing Institute


Log out of MyMSI

Research Abstracts Online
January 2010 - March 2011

Main TOC ...... Next Abstract

University of Minnesota Twin Cities
College of Science and Engineering
Department of Biomedical Engineering

PI: Theoden Netoff

Seizure Detection and Prediction Using EEG Signal Processing and Classification

These researchers are developing patient-specific algorithms for seizure detection and prediction based on intracranial encephalogram (iEEG) recordings and ultimately intend to establish low-powered implantable devices for reliably alarming seizures in real time. To be reliable and implantable those algorithms and devices should have excellent sensitivity and low false positive rates and should consume low power. The researchers have built a seizure prediction algorithm that works based on EEG signal processing and a machine learning classification technique of support vector machines (SVMs). They have tested their seizure prediction algorithm by extracting linear features and SVM-classification with a relatively small size of datasets in the Freiburg database (~15GB) through cross-validation and have achieved high sensitivity of 87% and a low false positive rate of 0.13 per hour for a total of 18 patients’ 80 seizures and 437 interictal EEG recordings. They are currently developing reliable seizure detection algorithms based on linear feature extraction and simple classification methods. Also, they are expanding their analysis for seizure detection and prediction into much a larger database at Mayo (~100GB) to investigate the algorithm’s reliability, which has 30 patients’ recordings and was recorded in a higher sampling rate.

Group Members

Bryce C. Beverlin, Graduate Student
Michael J. Brown, Graduate Student
Oscar Miranda Dominguez, Graduate Student
Brendan Murphy, Supercomputing Institute Undergraduate Intern
Yun Sang Park, Graduate Student
Tyler Stigen, Graduate Student