Skip to Main content Skip to Navigation
Theses

Monitoring of premature newborns by video and audio analyses

Abstract : The objective of this work, conducted as part of the European project Digi-NewB and a CIFRE thesis, was to propose a new noninvasive approach to monitoring in neonatal intensive care units (NICUs). This new monitoring should make possible a continuous evaluation of the neuro-behavioural evolution of premature newborns using non-invasive modalities such as video and audio. After a bibliographical study of more than 150papers, a first study was carried out on a semiautomatic estimation of sleep stages. The proposed approach combined for the first time video and audio analyses. The limitations identified during this study led to the proposition of a new audio-video system. Its integration into NICU was studied and evaluated. Then, methods, based on video and audio processing techniques and classification (Random Forest, KNN, Multi-layer Perceptron ...), were proposed. They allow a continuous characterization of the newborn behaviour in terms of movement quantification and cry analysis. The difficulties related to the constraints of the real NICU conditions were studied and solutions to avoid irrelevant periods (e.g.,parents or medical staff in the camera field of view, alarms coming from medical devices) were developed. The results are encouraging and show that it is now possible to imagine a new generation of monitoring based on noninvasive analyses to characterize the neurobehavioural development of the newborn.
Complete list of metadatas

Cited literature [234 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02476000
Contributor : Abes Star :  Contact
Submitted on : Wednesday, February 12, 2020 - 2:04:43 PM
Last modification on : Tuesday, October 20, 2020 - 10:50:41 AM
Long-term archiving on: : Wednesday, May 13, 2020 - 3:30:53 PM

File

CABON_Sandie.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02476000, version 1

Collections

Citation

Sandie Cabon. Monitoring of premature newborns by video and audio analyses. Other. Université Rennes 1, 2019. English. ⟨NNT : 2019REN1S055⟩. ⟨tel-02476000⟩

Share

Metrics

Record views

65

Files downloads

221