Stain independent segmentation of whole slide images: A case study in renal histology

Most automated image segmentation approaches in histological analyses are either developed or trained for a specific histological stain. In this paper, we propose and investigate a stain independent segmentation framework, which extends the application of an existing segmentation routine developed for a specific stain to arbitrary stains. The idea is to bring consecutive slides with different stains into spatial alignment to be able to transfer the location information obtained by segmenting images with a particular stain onto the successively registered images dyed with other stains. In a case study with renal whole slide images, we perform experiments and compare the results of the proposed method with straight-forward segmentation (i.e. segmenting the stain, the method had been developed for) and also evaluated the impact of the remaining alignment error.