4. Segmentation stage with U-Net

4.1 Choice of architecture

Segmentation is the first step in our processing chain, illustrated in figure 1 . It is applied by considering the volume slice by slice, so as to process 2D images. A convolutional U-Net was chosen for its ability to operate with relatively few training images and to enable precise segmentation. This type of neural network was developed in 2015 for biomedical image segmentation at the Department of Computer Science at the University of Freiburg in...