PhD Defense of  Nafissa DIA from PRETA team on friday, the sixth of december at 9:30 am:

" Non-invasive monitoring of fetal heart rate: non-negative matrix factorization of electrocardiographic and phonocardiographic signals. "

Place: Amphithéâtre de l’IAB, site Santé, Allée des Alpes, 38700 La Tronche
 

Thesis supervision:

• Monsieur Pierre-Yves Guméry, Professeur des Universités, Université Grenoble Alpes, Laboratoire TIMC, Directeur
• Madame Julie Fontecave-Jallon, Maître de conférences, Université Grenoble Alpes, Laboratoire TIMC, Co-directrice
• Monsieur Bertrand Rivet, Maître de conférences, Grenoble-INP, Laboratoire GIPSA-Lab, Co-directeur

Jury:

• Madame Anne Heurtier, Professeur des Universités, Université d'Angers, Laboratoire Angevin de Recherche en Ingénierie des Systèmes (LARIS), France, Rapporteuse
• Monsieur Vicente Zarzoso, Professeur des Universités, Université de Nice Sophia-Antipolis, Laboratoire d'Informatique, Signaux et Systèmes de Sophia Antipolis, France, Rapporteur
• Monsieur Ahmed Kora, Professeur, Ecole Supérieure Multinationale des Télécommunications, Dakar, Sénégal, Examinateur
• Monsieur Norbert Noury, Professeur des Universités, Université Claude Bernard Lyon, Institut des Neurosciences de Lyon, France, Examinateur
• Madame Véronique Equy, Praticien hospitalier, CHU Grenoble-Alpes, France, Invitée

Abstract:

With more than 200,000 births per day in the world, fetal well-being monitoring during labor and birth is a major clinical challenge. This monitoring aims to analyze the fetal heart rate (FHR) and its variability, and it has to be robust while minimizing the number of non-invasive sensors to lay on the mother's abdomen.

In this context, electrocardiogram (ECG) and phonocardiogram (PCG) signals are of interest since they bring cardiac information, both redundant and complementary. This multimodality as well as some features of ECG and PCG signals, as quasi-periodicity, have been exploited in this thesis. Several propositions, based on non-negative matrix factorization (NMF),  have been put in competition.

The first approach proposed was to detect heart beats to estimate FHR. The second approach exploited the quasi-periodicity of ECG and PCG signals so as to directly follow FHR and appeared to be the most promising for FHR estimation. This is based on a source-filter modeling of fetal ECG or PCG signals extracted from the abdominal signals. The NMF decomposition of the fetal signals allows to identify the filter and source parts of the model. FHR is carried by the source part.

This approach was evaluated on a clinical database of ECG and PCG signals recorded, during the thesis, on pregnant women in late pregnancy. The estimated FHR were validated by comparison with cardiotocography (CTG), which is the clinical reference technique for FHR monitoring. The performances show the interest of the proposed methodology as a potential alternative to CTG.

Keywords:

Electrocardiogram, Phonocardiogram, Fetal heart rate, Multimodality, Biomedical signals processing, Non-negative matrix factorization