Segmentation of neonates cerebral ventricles with 2D CNN in 3D US data: suitable training-set size and data augmentation strategies - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

Segmentation of neonates cerebral ventricles with 2D CNN in 3D US data: suitable training-set size and data augmentation strategies

Matthieu Martin
Bruno Sciolla
  • Fonction : Auteur
Philippe Quetin
  • Fonction : Auteur
Philippe Delachartre

Résumé

For its clinical potential, segmentation of the preterm neonate's Cerebral Ventricular System (CVS) in 3D ultrasound (US) data using convolutional neural networks (CNN) is an emerging field. Nevertheless, gathering manually annotated data to efficiently train a CNN is difficult. In this paper, we address the question of how many training volumes and which kind of artificial data augmentation strategies would be suitable for this application. We explored this topic by training a U-net with different training-set size and by using several artificial data augmentation strategies. In our setup , accurate segmentation results (Dice ≥ 0.8) were obtained with only 9 training volumes. The use of artificial data augmentation improved significantly (p < 0.05) the accuracy obtained without data augmentation when performing horizontal flips (between the right and the left brain hemispheres). The other types of data augmentation that we tried did not significantly improve U-net accuracy.
Fichier principal
Vignette du fichier
Martin_IUS_2019.pdf (974.69 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03148378 , version 1 (22-02-2021)

Identifiants

Citer

Matthieu Martin, Bruno Sciolla, Michaël Sdika, Philippe Quetin, Philippe Delachartre. Segmentation of neonates cerebral ventricles with 2D CNN in 3D US data: suitable training-set size and data augmentation strategies. 2019 IEEE International Ultrasonics Symposium (IUS), Oct 2019, Glasgow, United Kingdom. pp.2122-2125, ⟨10.1109/ULTSYM.2019.8925799⟩. ⟨hal-03148378⟩
47 Consultations
86 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More