Image Registration of Differently Stained Histological Sections

We have focused our interest on the registration of brightfield transmitted light microscopy images with respect to different histological stainings. For this kind of registration problem we have developed a new segmentation procedure. Based on the obtained consistent segmentations, a nonlinear registration transformation is computed. The applied registration procedure uses a curvature-based nonlinear partial differential equation in order to find the appropriate mapping between the images. We present an example for the registration of images of two consecutive histological sections from a uterine cervix specimen, where one section stained with p16^INK4a was mapped onto another with H&E staining. Finally, we give an illustrative example for a tissue volume reconstruction based on an alternately stained large serial section.