Research

Biomedical Context


The objective of my research is to address real problems posed by medicine or physiology concerning the interpretation of biomedical signals and electro-physiological modelling. The main current fields of application are electrocardiography (ECG), cardio-respiratory coupling during physical effort and rest, analysis of electromyograms (EMG) and respiration in the context of physical exercise. The most recent theme concerns the analysis of episodes of atrial fibrillation and flutter, in particular the quantification of the spatio-temporal complexity of this phenomenon. This action is carried out in collaboration with the Centre Hospitalier Princesse Grace (Monaco), the department of data science and knowledge engineering (Maastricht) and the University of Kuala Lumpur (UniKL).

This research conducted at the level of the organ is also completed with an approach developed at the level of the cell (New York Medical College). The analysis of the effect of diabetes on the variability of cardiac waves and on the action potentials of cardiac ventricular cells is a fine example of multi-scale work. The fine measurement of the beat frequency of the eyelashes of the ependymal cells of the cerebral ventricles is another example of research carried out on analysis tools at the cellular level.

The general problems encountered in signal processing are the estimation of average signals most often associated with permanent or stable phenomena over time, the characterization of natural variability and finally the measurement of significant variations in parameters or more generally in shape, related to an external agent such as a pathology, an effort or a drug.

The theoretical approaches concern detection, estimation of parametric, semi-parametric or non-parametric models and classification, in the non-stationary case, source separation and data-driven signal decomposition.

In order to improve the performance and robustness of classical signal separation and extraction methods, new research aims to optimally incorporate the a priori information provided by physiological knowledge, thus generating more semi-blind techniques. specific to the biomedical problem to be treated (eg atrial fibrillation). Finally, a recently highlighted feature is the analysis of the variability of biomedical signals. In fact, more than in any other field, this variability can either be eliminated when it hinders a reliable detection or an accurate estimation, or on the contrary be studied as it may carries valuable information.

Liste des publications sur HALĀ : https://cv.hal.science/olivier-meste