FlowCAM | Lifewatch regional portal

You are here

FlowCAM

The FlowCAM (Flow Cytometer And Microscope) is and automated technique for particle enumeration that uses flow cytometry and microscopy, enabling near to automatic classification and quantification of phytoplankton in a water sample.

General

Phytoplankton forms the basis of the food web, contains species that are toxic and is a good bio-indicator of environmental changes. Variations in phytoplankton communities may impact a wide range of species, ranging from zooplankton and fish to seabirds and marine mammals. However, studies of plankton are difficult to carry out because of their high diversity and spatiotemporal variability. In addition, traditional identification techniques are time-consuming and require highly qualified taxonomists.

Infrastructure

The Flanders Marine Institute (VLIZ) installed a FlowCAM as part of the Belgian LifeWatch observatory.

From May 2017 onwards, phytoplankton is sampled at 9 onshore stations every month; and every season, the sampling occurs at 17 on and offshore stations. The samples are fixed with Lugol and stored for its posterior analysis. Once in the Marine Station Ostend (MSO), the samples are processed by the FlowCAM at 4X magnification to obtain good resolution images of size ranges of interest, primarily micro-phytoplankton from 70 to 300 µm (2017) and 50 to 300 µm (2018 onwards). A gray-scale camera is used because it offers higher resolution quality for species identification and because of the fixed samples do not present fluorescent phytoplankton pigments.

     

Left: FlowCAM laboratory set-up at MSO (©VLIZ) - Middle: Phytoplankton sample fixed with Lugol (©VLIZ) - Right: Sample entry in FlowCAM (©VLIZ)

Once the pictures are obtained, they are uploaded to our main image database. An automated image classification model is then used to make predictions regarding the plankton species that occur on each picture. The model we use is a deep learning classifier, more specifically a convolutional neural network. These type of models have to be trained on a large training data set, i.e. a data set where the plankton label for each picture is known. After running the automatic classification a human taxonomist still validates the predicted classification by the model to ensure a high quality taxonomic data set.

          
Left: Processing of a sample with VisualSpreadsheet (©VLIZ) - Right: FlowCAM collage from a sample at 4x magnification (©VLIZ)

Map with LifeWatch stations sampled during monthly 1-day (circles) and seasonal 2-day (squares) sampling campaigns.

 

Data

Data can be accessed via the LifeWatch Data Explorer for flowcam data.

Metadata is available at:

  • LifeWatch observatory data: phytoplankton observations by imaging flow cytometry (FlowCam) in the Belgian Part of the North Sea, More

 

Useful links

 

Selection of recent publications

  • Hablützel, P.I.; Rombouts, I.; Dillen, N.; Lagaisse, R.; Mortelmans, J.; Ollevier, A.; Perneel, M.; Deneudt, K. (2021). Exploring New Technologies for Plankton Observations and Monitoring of Ocean Health, in: Kappel, E.S. et al. Frontiers in ocean observing: Documenting ecosystems, understanding environmental changes, forecasting hazards. Oceanography, Suppl. 34(4): pp. 20-25. [link to IMIS record]
  • Amadei Martínez, L.; Mortelmans, J.; Dillen, N.; Debusschere, E.; Deneudt, K. (2020). LifeWatch observatory data: phytoplankton observations in the Belgian Part of the North Sea. Biodiversity Data Journal 8: e57236. https://hdl.handle.net/10.3897/bdj.8.e57236 [link to IMIS record]
  • Lloret, L.; Heredia, I.; Aguilar, F.; Debusschere, E.; Deneudt, K.; Hernandez, F. (2018). Convolutional Neural Networks for Phytoplankton identification and classification. Biodiversity Information Science and Standards 2: e25762. https://hdl.handle.net/10.3897/biss.2.25762 [link to IMIS record]