3D cell segmentation
Encadrants : Somoballi Ghoshal
Email : somoballi.ghoshal@pasteur.fr
Disponible : OUI
Nombre d'étudiants : 1
Description : Our aim is to accurately segment the cells in 3D from lightsheet microscopy images of mouse embryo heart. Initially we will look into all the existing state of the art works for 3D cell segmentation, find the drawbacks and try to improve on them, with respect to accuracy for our dataset. We plan to improve the UNET architecture for 3D cell segmentation with proper identification of individual cells as well as nucleus.
Pré-requis : knowledge of deep learning
Travail demandé : Implementing UNET with modified loss functions and attention UNET for 3D cell segmentation from lightsheet microscopy images of mouse embryo heart
