Dear author,
I have one questions in your file surfNet_diff.py. In each epoch, the input of the model is midsurf with corresponding vertices and faces, while you use the vertices from the midsurf and faces from real white/pial surface to organize the predicted white/pial mesh, to calculate the chamfer distance loss.
Do the midsurf and white/pial surface have the same the number of vertices and faces? Since the Midsurf is generated from segmentation map using the Marching cubes algorithm, do you permit the geometric information between mdisurface and white/pial surface is the same?
Looking forward to your response.
Thanks.
Dear author,
I have one questions in your file surfNet_diff.py. In each epoch, the input of the model is midsurf with corresponding vertices and faces, while you use the vertices from the midsurf and faces from real white/pial surface to organize the predicted white/pial mesh, to calculate the chamfer distance loss.
Do the midsurf and white/pial surface have the same the number of vertices and faces? Since the Midsurf is generated from segmentation map using the Marching cubes algorithm, do you permit the geometric information between mdisurface and white/pial surface is the same?
Looking forward to your response.
Thanks.