New clustering of the spikes morphology based on dynamic clouds in partial epilepsy

O. Khouma (1) , M.L. Ndiaye (1) , I. Diop (1) , A.K. Diop (1) , S.M. Farsi (1) , Babacar Diouf (1) , J.-J. Montois (2)

1 UCAD - Université Cheikh Anta Diop de Dakar [Sénégal]
2 LTSI - Laboratoire Traitement du Signal et de l'Image

The acquisition of EEG signals can be done during days. In this paper, the signals of depth Stereo-Electroencephalography (SEEG) are used. In SEEG signal, several inter-ictal paroxystic events (IPE) are found, they appear between the crises. To analyze their distribution, they should initially be detected by separating them from the basic activity. After the detection step, we implement a clustering process to distinguish IPE or spike according to their morphology. Results of detection and clustering would characterize a link between space-time distribution of IPE and the arrival of the crises. It would be thus a contribution on the diagnosis of the partial epilepsies. © 2016 IEEE.

Mots clés: Clustering algorithms; Electrophysiology; Morphology; Neurology; Signal processing; clustering; Clustering process; Epilepsy; ON dynamics; Page-Hinkley; Partial epilepsy; Space time distribution; Spike detection; Electroencephalography

Domaines: Ingénierie biomédicale

O. Khouma, M.L. Ndiaye, I. Diop, A.K. Diop, S.M. Farsi, et al.. New clustering of the spikes morphology based on dynamic clouds in partial epilepsy. 2016 SAI Computing Conference, SAI 2016, Jul 2016, London, United Kingdom. pp.354--360, ⟨10.1109/SAI.2016.7556006⟩. ⟨hal-01381262⟩