Logo of the planktAI project, depicting an algae and the word planktAI on it.


Automatic identification of microscopic algae for biodiversity assessment and trophic status estimation

Within the planktAI project, we aim to provide an API for the automatic identification of microalgal species and the assessment of the trophic status of freshwater bodies, usable by both professionals and citizen scientists. We also want to compare the performance of the image-based and DNA-based approaches to biodiversity assessment.


Goals of the project

a) automatic identification of freshwater microalgae for biodiversity surveys
b) automatic assessment of algal biovolume & trophic status of water bodies

Methods and approach

There were three major steps that needed to be done in order to fulfil the project goal:

a) An extensive database of labelled phytoplankton images was missing.
b) We needed to create a comprehensive understanding of the systematics of ecological indicators. 
c) We needed to run a deep learning detection/classification model for the majority of microalgae.

As a first step, we sample(d) standing water bodies (oligotrophic to hypertrophic) in Central and Northern Europe from spring 2022 to spring 2023 with a mesh size ≥ 10 μm. 

We then compiled a list of  approx. 100 ecological indicators, alongside their accompanying algal flora. From the collected samples, we derived > 200 images per taxon, done as bright-field images and imaging flow cytometry. With those images, we created dalphi, a database of labelled phytoplankton images.

The next step as applying deep learning methods for 
a) the automatic detection & classification of the target taxa in mixed samples and
b) the automatic estimation of the cell/colony count per taxon within a sample.

As a result, we can provide 
a) an API for automatic identification of microalgae
b) integration of image- and DNA-based phytoplankton diversity assessment

Key publications

Susanne Dunker et al. (2022)
The potential of multispectral imaging flow cytometry for environmental monitoring. 
Cytometry. 101( 9): 782– 799. doi: 10.1002/cyto.a.24658

Susanne Dunker (2020)
Imaging Flow Cytometry for Phylogenetic and MorphologicallyBased Functional Group Clustering of a Natural Phytoplankton Community over 1 Year in an Urban Pond.
Cytometry, 97: 727-736. doi: 10.1002/cyto.a.24044

Susanne Dunker (2019)
Hidden Secrets Behind Dots: Improved Phytoplankton Taxonomic Resolution Using High-Throughput Imaging Flow Cytometry
Cytometry, 95: 854-868. doi: 10.1002/cyto.a.23870

Susanne Dunker et al. (2018)
Combining high‑throughput imaging flow cytometry and deep learning for efficient species and life‑cycle stage identification of phytoplankton
BMC Ecology 18, 51. doi: 10.1186/s12898-018-0209-5


If you have questions regarding this project, please reach out to:

Dr. Susanne Dunker 

Dr. Susanne Dunker

Senior scientist
+49 341 9733170

Go to Editor View