Skip to main content
. Author manuscript; available in PMC: 2024 Jul 1.
Published in final edited form as: Med Phys. 2021 Nov 17;48(12):7657–7672. doi: 10.1002/mp.15101

Figure 2:

Figure 2:

The encoder-decoder network used in the studies. The resolution remained the same for all the featuremaps in the network, which was 96 × 96 during patch-based training and 640 × 640 during inference. The number of channels for all the blue blocks were 32 except for the input and output layer.