Slide scans of diatom preparations from river Menne

This archive contains the slide scans used for the deep learning experiments published in Kloster et al. 2022: Improving deep learning-based segmentation of diatoms in gigapixel-sized virtual slides by object-based tile positioning and object integrity constraint. The diatom material was sampled at six different locations from the river Menne (Germany), cleaned/oxidized and mounted in Naphrax. Please refer to the manuscript for details.

Each scan represents area of ca. 5 × 3.5 mm² scanned by an VS200 slide scanning microscope (Olympus Europa SE & Co. KG, Hamburg, Germany). 61 focal planes were imaged in bright-field mode at a distance of 0.28 µm each using a UPLXAPO60XO 60x/1.42 oil immersion objective. Focus stacking and stitching were performed with the built-in functions of the VS200 ASW software (v3.2.1). Subsequently the RGB color data was reduced to 8bit monochrome grayscale images. This process resulted in the six virtual slide images, each roughly 55,000 × 39,000 pixel.

These images depict benthic diatoms Menne river comprising more than 110 different taxa (estimated by identifying a subset of specimens). The sampling of this stream was authorized by the Kreis Paderborn authorities through the 4032-20-450 permit.