Frequency Change-Point Detection in physiological Signals : an Algebraic Approach

Sonia Rezk (1, 2) , Cédric Join (3, 4) , Sadok El Asmi (1) , Mohamed Dogui (5) , Mohamed Hédi Bedoui (2)

1 MASC - Département Mathématiques Appliquées, Signaux et Communications
2 Laboratoire de Biophysique [Monastir]
3 CRAN - Centre de Recherche en Automatique de Nancy
4 ALIEN - Algebra for Digital Identification and Estimation
5 Service d'Exploration Fonctionnelle du Système Nerveux

This paper considers the problem of change-point detection for noisy data. Estimation of signal frequency content relies on differential algebra and non-commutative algebra together with operational calculus. We adapt this approach to the study of changes that may be observed in EEG signal dynamics during epileptic seizure and in ECG signal during the occurrence of a QRS complex. The correlation with frequency change is what this idea is based on. The interest of our estimator is firstly il lustrated according to several academic examples. Then, the method is applied on real physiological signals to detect abrupt frequency changes.

Domaines: Automatique

Sonia Rezk, Cédric Join, Sadok El Asmi, Mohamed Dogui, Mohamed Hédi Bedoui. Frequency Change-Point Detection in physiological Signals : an Algebraic Approach. International Journal on Sciences and Techniques of Automatic control & computer engineering, 2008, 2 (1), pp.456-468. ⟨inria-00308474⟩