Sleep and Sensory Analysis Group - Sensing everything between heaven and earth
Sleep and Sensory Signal Analysis Group


Examples of our studies

Sleep analysis
Epilepsy analysis
Measurement chair
Sensing the human
Tablet interface
Other user interface studies
Engine signal processing
Health informatics standardisation
Secret company projects
Ancient web history


Doctoral theses
Publications

Epileptic brain signal analysis

 
Detecting epileptic seizures from long-term recordings

Spike and wave complex in EEG
An example of an epileptic spike-and-wave discharge

Epilepsy is a disease in which the electrical activity of the brain (EEG) is somewhat distorted. Usually the EEG looks normal for the most of the time but sometimes kind of "electrical storms" occur in the brain. These are often associated by visible symptoms in patient. In mild cases the patient is simply "absent" for a couple of seconds and in more severe cases the patient falls and has cyclic convulsions.

The suspicion of epilepsy needs to be confirmed with EEG measurements. A normal 30 minute laboratory recording might reveal epileptic activity in the EEG with proper stimulus, e.g. with flashing lights but usually a more long-term recording is needed. The visual analysis of these long-term recordings is tedious and time-consuming for the EEG experts. Computerized EEG analysis can help here to pinpoint the events of interest in the EEG to the expert thus saving him a lot of time.

Our epileptic EEG analysis methods aimed at detecting epileptic spikes and epileptic seizures form the EEG recordings. The detection of spikes is based on an adapted version of FIR-Median Hybrid filters developed in the Department. The detection of seizures is a more complicated process. The signal is first segmented to variable size segments based on the signal content. Several time and frequency domain features are calculated form these segments. The resulting feature vector is compared to a library of feature vectors representing an EEG class which has been classified by EEG experts. The closest mach is chosen as the class of the segment and the occurrence of epileptic class segments is reported to the clinicians.

The epilepsy signal analysis studies were performed in co-operation with the departments of Clinical Neurophysiology and Neurology in Tampere University Hospital.  Some sample EEG data is provided for research purposes in our public FTP site, but note that you need and expert in EEG to actually understand what is going on in these recordings.

Links
Epilepsy, Wikipedia
EEG tutorial
International League Against Epilepsy

E-mail to Alpo.Varri (at) tut.fi
Designed by Alpo Värri 28.1.2013
Department of Signal Processing Tampere University of Technology
P.O. Box 553, FI-33101 Tampere, Finland, Tel. +358-40-8490780
Sometimes epilepsy falls the patient to the ground