Digital Diatomology: Combining slide scanning microscopy, web-based collaborative annotation, and deep learning for diatom analysis

Description: Light microscopic analysis is an important tool for diatomists and the way we do it has hardly changed for the past 100 years. Our group has been interested in how digital methods could enable upscaling or enhancement of light microscopy. It is now possible to obtain high quality, high resolution microscopic scans of full diatom slides. This makes it possible to identify diatom taxa on slide scans through a web browser, in collaboration with colleagues in different locations and, to some extent, also to apply digital image analysis methods including deep learning to assist us in locating, measuring, identifying, and counting diatoms. We would like to give you a few insights into where we currently stand with establishing these methods and what they can do for diatom researchers. The year 2020 has shown the benefit of such methods even beyond our original intentions (which was mainly focused on digitization of the Hustedt diatom collection), allowing us to conduct research, collaboration, and teaching even under the present need for physical distancing.

Target Audience: Anyone interested in using new tools and learning about the potential digital evolution of diatom taxonomy.