Skip to Main content Skip to Navigation
Theses

Reconstruction 3D de données échographiques du cerveau du prématuré et segmentation des ventricules cérébraux et thalami par apprentissage supervisé

Abstract : About 15 million children are born prematurely each year worldwide. These patients are likely to suffer from brain abnormalities that can cause neurodevelopmental disorders: cerebral palsy, deafness, blindness, intellectual development delay, … Studies have shown that the volume of brain structures is a good indicator which enables to reduce and predict these risks in order to guide patients through appropriate care pathways during childhood. This thesis aims to show that 3D ultrasound could be an alternative to MRI that would enable to quantify the volume of brain structures in all premature infants. This work focuses more particularly on the segmentation of the lateral ventricles (VL) and thalami. Its four main contributions are: the development of an algorithm which enables to create 3D ultrasound data from 2D transfontanellar ultrasound of the premature brain, the segmentation of thigh quality he lateral ventricles and thalami in clinical time and the learning by a convolutional neural networks (CNN) of the anatomical position of the lateral ventricles. In addition, we have created several annotated databases in partnership with the CH of Avignon. Our reconstruction algorithm was used to reconstruct 25 high-quality ultrasound volumes. It was validated in-vivo where an accuracy 0.69 ± 0.14 mm was obtained on the corpus callosum. The best segmentation results were obtained with the V-net, a 3D CNN, which segmented the CVS and the thalami with respective Dice of 0.828± 0.044 and 0.891±0.016 in a few seconds. Learning the anatomical position of the CVS was achieved by integrating a CPPN (Compositional Pattern Producing Network) into the CNNs. It significantly improved the accuracy of CNNs when they had few layers. For example, in the case of the 7-layer V-net network, the Dice has increased from 0.524± 0.076 to 0.724±0.107. This thesis shows that it is possible to automatically segment brain structures of the premature infant into 3D ultrasound data with precision and in a clinical time. This proves that high quality 3D ultrasound could be used in clinical routine to quantify the volume of brain structures and paves the way for studies to evaluate its benefit to patients.
Complete list of metadatas

Cited literature [159 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02487473
Contributor : Abes Star :  Contact
Submitted on : Thursday, July 16, 2020 - 11:33:33 AM
Last modification on : Tuesday, July 21, 2020 - 4:53:55 PM

File

these.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-02487473, version 2

Citation

Matthieu Martin. Reconstruction 3D de données échographiques du cerveau du prématuré et segmentation des ventricules cérébraux et thalami par apprentissage supervisé. Traitement du signal et de l'image [eess.SP]. Université de Lyon, 2019. Français. ⟨NNT : 2019LYSEI118⟩. ⟨tel-02487473v2⟩

Share

Metrics

Record views

57

Files downloads

26