We non-rigidly register immunohistological serial sections of human specimen.
Using feature detection and matching we iteratively compute non-rigid deformations.
Vascular networks in spleen and bone marrow are shown on a level not possible before.
A quantitative evaluation of our method shows its efficiency.