Landmark-constrained 3-D Histological Imaging: A Morphology-preserving Approach

Abstract

The inspection of histological image sequences to gain knowledge about the original three-dimensional (3-D) morphological structure is a standard method in medical research. Its main advantage is that light microscopes feature high resolution enhanced visibility due to staining. In many cases this imaging technology could immensely profit from 3-D reconstructions of the slice images. For volumetric stacking, however, the tissue deformations due to slice preparation require an unwarping strategy to restore the original morphology. The challenge is to reverse the artificial deformations while preserving the natural morphological changes. In particular, unintentional straightening of curved structures across multiple slices has to be avoided. In this article, we propose a novel way to incorporate landmarks representing the morphological progression. They are used as additional regularization for intensity based non-rigid registration which is capable to exactly match the landmarks. Our approach is tested on synthetical and histological data sets. We show that it delivers smooth contours while preserving the morphological structure, and is a promising addition to existing methods.