Quantification of Tumour Invasion Fronts Using 3D Reconstructed Histological Serial Sections

Abstract

The analysis of the 3D structure of tumour invasion fronts within the uterine cervix is considered essential for both discovering and understanding inherent architectural-functional relationships. The variation range of the invasion patterns known so far reaches from a smooth tumour-host boundary to more diffusely spreading patterns, which all are supposed to have a different prognostic importance. However, any verbal morphological quantifications in previous studies have been made just on single histological sections. Therefore, the intention of this paper is twofold: to provide reconstructed 3D tumoural tissue data and to apply an algorithmic tumour invasion quantification. Thus, to stay as much as close to routine pathology we as well use HE-stained histological sections but as serial sections of remarkable extent (90-500 slices). Slicing and staining, however, may induce some severe artefacts rarely to avoid, mainly different kinds of distortions.

The paper introduces an extended processing chain doing a robust volume reconstruction starting from stacks of digitised transmitted light microscope colour images resulting in a 3D visualisation of the invasion front of the cervical tumour. For the invasion quantification we refer to digital compactness which is considered to be in tight correspondence to those invasion features pathologists generally are paying attention when verbally assessing 2D sections in routine.