Figure 2.

A schematic illustration of our proposed deep learning, fully convolutional networks (DFCN)‐CoSeg network with feature fusion for positron emission tomography (PET)‐computed tomography (CT) cosegmentation. Two parallel 3D‐UNets are built for CT and PET respectively. In DFCN‐CoSeg, all feature maps produced by all the encoders of either the CT or PET branch are concatenated in the corresponding decoders, as depicted by the dotted arrow lines. [Color figure can be viewed at wileyonlinelibrary.com]