Skip to main content
. 2019 Dec 28;6(4):044009. doi: 10.1117/1.JMI.6.4.044009

Fig. 5.

Fig. 5

(a) Mean Dice scores on the validation dataset as functions of training iterations using U-Net with different resolution levels, (b) U-Net with different filter numbers, (c) different convolutional block structures in U-Net, and (d) different network structures.