Skip to Main content Skip to Navigation
Theses

Analyse de distributions spatio-temporelles de transitoires dans des signaux vectoriels. Application à la détection-classification d'activités paroxystiques intercritiques dans des observations EEG

Abstract : Electroencephalographic signals (EEG) recorded in epileptic patients reflect, apart from periods corresponding to seizures, transient signals referred to as "interictal epileptiform events" (IEEs). The analysis of IEEs can contribute to the study of intractable partial epilepsies. A method of characterization of the spatio-temporal distribution of IEEs in depth-EEG recordings is presented. The method proceeds in four steps:

1. Detection of monochannel IEEs. The detection method, based on a heuristic approach, uses a wavelet filter banks to enhance the sharp part of IEEs. The mean value of the obtained statistics is analyzed by a Page-Hinkley algorithm aimed at detecting abrupt changes corresponding to spikes.

2. Fusion of monochannels IEEs. The procedure looks for co-occurents monochannels IEEs using a sliding temporal window and builts multichannels IEEs.

3. Extraction of subsets of channels frequently and significantly active during multichannels IEE (called "activation sets").

4. Evaluation of the existence of a reproducible temporal order (possibly partial) in extracted activation sets.

Methods proposed in each step have been evaluated with simulated EEG signals (step 1) and Markovian models (steps 2-4). The results show that the whole method is robust to false alarms. This method is applied to long duration signals recorded in 8 patients (each one containing several thousands of IEEs). The results show a high reproducibility of the spatio-temporal distributions of IEEs and allow identification of particular anatomo-functional networks.
Complete list of metadata

https://tel.archives-ouvertes.fr/tel-00007178
Contributor : Jérôme Bourien <>
Submitted on : Friday, October 22, 2004 - 11:46:46 AM
Last modification on : Thursday, January 14, 2021 - 11:25:00 AM
Long-term archiving on: : Friday, April 2, 2010 - 8:53:01 PM

Identifiers

  • HAL Id : tel-00007178, version 1

Collections

Citation

Jérôme Bourien. Analyse de distributions spatio-temporelles de transitoires dans des signaux vectoriels. Application à la détection-classification d'activités paroxystiques intercritiques dans des observations EEG. Traitement du signal et de l'image [eess.SP]. Université Rennes 1, 2003. Français. ⟨tel-00007178⟩

Share

Metrics

Record views

1050

Files downloads

1941