EEG signal analysis for epileptic seizure genesis study
Résumé
Epilepsy is a neurological disorder that manifests itself as episodes od epileptic seizure characterized by an unusually sporadic neural activity observable by EEG. A model of transgenic mouse affected by epilepsy has been developed in order to better understand, predict and preventively treat these seizures. Between the seizures, we observe some interictal spikes that are expected to be used as predictive tools for the seizures, We were interested in developing algorithms to automate their detection, counting and characterization. A major obstacle was the contamination of the studied signals by artifacts. The first approach was to decompose the signal with a Convolutional Dictionary Learning framework, but it was inconclusive. The successful approach was to, first, detect and cut the artifactual zones using a threshold on the signal norm, then a rough spike detection and finally, a classification of the spikes between desirable and unwanted events using a classifier trained on a small dataset of hand-picked events.
Origine | Fichiers produits par l'(les) auteur(s) |
---|