Skip to main content
. Author manuscript; available in PMC: 2023 May 22.
Published in final edited form as: Proc Mach Learn Res. 2022 Jul;172:508–520.

Figure 2:

Figure 2:

TopoFit architecture: a series of graph-convolutional (GCN) blocks iteratively warp a template surface m0 to fit the anatomy of an input image x. Each block gt predicts a deformation ϕt by sampling spatial image features from a volumetric U-Net h at the positions of mt−1. The template mesh is up-sampled between each deformation step.