Alternative to microscopy in time of Corona – virtual microscopy with Biigle 2.0
by Andrea Burfeid-Castellanos |
As you all are daily aware, Corona has (hopefully temporarily) changed all of our lives. And our need for diatom identification has not abated even though access to microscopes has been difficult for most. But there is an option. Biigle 2.0 (Biigle.de) developed by the group of Tim W. Nattkemper of the university of Bielefeld (Schoening et al. 2016, Langenkämper et al. 2017) is a free browser based annotation software in which you can upload single diatom (or, indeed, any organism) images or complete slide scans and annotate away.
After creating a user, different projects, subprojects (Volumes) can be prepared (fig. 1). There, images and slide scans can be stored according to project, sampling effort or whatever category is needed. All images or slides can be labelled to maintain clarity.
Figure 1 Dashboard containing volumes, information on label trees and members with permission to annotate.
In the Phycology Research Group led by Bánk Beszteri at the University Duisburg-Essen we are working with slide scans made by a semi-automatic slide scanner (Olympus VS200). The scans are made with 1 pixel being equivalent to 0.1 x 0.1 μm2 and end up in 1 GB virtual slides (4.02 mm²). These virtual slides are uploaded onto Biigle 2.0. This way our results remain openly accessible for scientific enquiry while making collaborative annotation very easy.
The images can be zoomed into using the scroll function on the mouse. If a scale is added to the image, the scale-bar will change accordingly, and the measuring function will give information on μm instead of pixels. When annotating, using tools similar to those known from Photoshop or GIMP, the annotations are made following an already existing label tree that can either be made from scratch, downloaded from WoRMS (World Regist) or shared. Our label tree on Freshwater diatoms, for instance, contains all diatom taxa found in Freshwater Benthic Diatoms of Central Europe (Hofmann et al. 2017) and some provided by the Rhône-Alpes Atlas (Bey & Ector 2013) and diatoms.org (Bahls et al. 2010). The structure of our tree follows the one proposed by Diatoms of North America (DONA, diatoms.org). Of course, marine and brackish diatom label trees can also be added, and one about southern planktonic diatoms already exists.
Figure 2 Overview of diatom preparation with label tree on display.
Biigle for all
When identifying diatoms, one can always find a particular diatom that is difficult (or impossible) to identify. Biigle makes it very easy to share an image grab through e-mail or a direct widget built into the website. The scrolling into a slide modifies the URL-address to fit the topography, so that this url-address can easily be shared by the collaborators in the project.
Furthermore, as you can see in figure 1, the request report button facilitates the download of your annotations in any format necessary to ease data handling and analysis. This, for instance could be used with an “OMNIDIA” code (Lecointe et al. 1993) label tree to directly upload onto the biomonitoring software.
For control, Biigle 2.0 has a “Largo re-evaluation” function, in which single annotation labels (=taxa) can be selected to re-evaluate the annotated valves using thumbnails (fig. 2). These thumbnails are linked to the parent image, so that the measuring and zooming into the valve are facilitated. This would be similar to the morpho-group voucher floras that were explained in the DONA webinar (https://www.youtube.com/watch?v=PRLBBbGduwM), but they are alive and growing with every single annotation made for instance for diatom counts.
In case of changes in nomenclature or erroneous identification, the “Largo-re-evaluation” function also permits a relabelling with a few clicks.
Figure 3 Largo re-evaluation tool. When selecting thumbnails and clicking “continue”, a new annotation can be added
Biigle in academia
During the recent lockdown, I started using Biigle 2.0 for teaching Bachelor students the ropes of diatom identification on-line. The students can measure valves, identify them taxonomically or pinpoint unknown taxa, while the teacher can check and correct the annotations as needed. The handling of the browser is very user friendly and students do not have a steep learning curve to add to the taxonomic one.
The same images can be annotated by different students, masking the annotations of the other students, which can give information on inter- and intra-personal variation of annotations. Since different categories can be selected for users (editor, admin, guest, expert), the annotations of higher ranking categories should have prevalence over the lower ranking annotations.
When used in combination with a conference software (zoom, Skype, BBB), the students can share their desktops and ask questions and comment on difficulties and actively learn to apply the tools discussed.
The resulting and corrected annotations are being used to feed deep learning algorithms for the identification of diatoms
If this short introduction has piqued your interest, you can find a thorough tutorial made by Michael Kloster on this website: https://www.uni-due.de/phykologie/sherpa and contact me for further questions.
Bahls L. L., Edlund M. B., Kociolek J. P., Lowe R. L., Potapova M. G., Spaulding S. A., Bishop I., Burge D. R. L., English J., Furey P., Graeff C. L., Hamilton P. B., Jovanovska E., Laliberté G., Lee S., Manoylov K. M., Morales E. A., Phillips N., Rushforth S., Stancheva R., Stepanek J., Tyree M., Vaccarino M., Veselá J. & Winter D. 2010. Diatoms of North America.
Bey M.-Y. & Ector L. 2013. Atlas des diatomées des cours d’eau de la région Rhône-Alpes.
Hofmann G., Werum M., Lange-Bertalot H. & Cantonati M. 2017. Freshwater Benthic Diatoms of Central Europe. In: (M. Cantonati, M. G. Kelly, and H. Lange-Bertalot (eds))p.. Koeltz Botanical Books.
Langenkämper D., Zurowietz M., Schoening T. & Nattkemper T. W. 2017. BIIGLE 2.0 – Browsing and Annotating Large Marine Image Collections. Frontiers in Marine Science 4.
Lecointe C., Coste M. & Prygiel J. 1993. “Omnidia”: software for taxonomy, calculation of diatom indices and inventories management. Hydrobiologia 269: 509–513.
Schoening T., Osterloff J. & Nattkemper T. W. 2016. RecoMIA—Recommendations for Marine Image Annotation: Lessons Learned and Future Directions. Frontiers in Marine Science 3. Frontiers.