Offre de Post-doctorat

PhD proposal: Early detec2on of risk events based on cardiac protheses massive data processing

The aging of the popula/on and the western lifestyle are having an increasing impact on the prevalence of heart diseases. These chronic heart diseases represent a major challenge for healthcare systems worldwide. In addi/on to drug therapy, treatments based on cardiac implantable electronic devices (CEDs) are now highly validated. The new genera/on of CEDs open the possibility of remote monitoring of the pa/ent’s physiological variables. In this context, the SEPIA team of the LTSI has a long experience, with the proposal of a set of methods for processing data acquired by the implants and mathema/cal models allowing the early detec/on of risk events. The objec/ve of the SEPIA team is to implement effec/ve tools for the processing and analysis of large longitudinal data extracted from cardiac prostheses. These devices include an ECG module, an accelerometer, a gyroscope, and a 3D magnetometer. The data they provide have been collected from 1000 implanted pa/ents who have undergone longitudinal monitoring.

Following our work in the field, the main objec/ve of this thesis is to propose a complete data processing chain for the early detec/on of risk events, exploi/ng the massive data extracted from cardiac prostheses and integra/ng an interpretable model. These mul/dimensional and heterogenous data recordings contain markers that characterize serious cardiovascular events. The objec2ve of this thesis is the development and applica2on of Deep Learning (DL)-based methods on the available dataset, enabling robust and reliable extrac2on of these mul2variate markers. A par/cular focus will be paid to implement an op/mized DL-based models, especially in the framework of natural language processing. In fact, DL methods have achieved striking success in dealing with massive data among them ECG processing and analysis. Several performant deep learning models such as CNN, LSTM were proposed for example for ECG segmentaCon. However, their remarkable efficacy was on the expense of the learning Cme, model complexity and the number of parameters to be learned. In this sense, looking for alternaCve opCmized DL models is of parCcular interest.

Profile: We are looking for a PhD student strongly mo/vated by new techniques for biomedical signal processing. Applicants should demonstrate background in deep learning, machine learning, numerical analysis and signal processing. Programming proficiency in Python is required.

PhD emplacement: Laboratoire Traitement du Signal et de l’Image (LTSI) – INSERM U1099, Bât. 22, Campus de Beaulieu, Université de Rennes, 35042, Rennes, France.

Star2ng period: October 1st 2023

Reward: Doctoral contract – INSERM

Contact: Applica/ons comprised CV, cover leber and mark transcript are to be submibed to Ahmad Karfoul (ahmad.karfoul@univ-rennes.fr) and Lodi Senhadji (lotfi.senhadji@univ- rennes.fr).