Comparison of root image analysis software and their potential effects on root studies

Crop science, Data Science, Image Analysis

Acquiring information about plant roots is key when trying to understand processes involved in the water and nutrient uptake of crops. Compared to the analysis of above ground plant organs, accessing and analyzing plant roots is a work intense process with many uncertainties involved based on the methodologies used for analysis. One important step of root analysis is the use of machine learning based image analysis software to access data such as the total root length, root diameter or root architectural traits. As more and more software to analyze scanned root images gets published and used by the scientific community, the uncertainty of the acquired data might increase further, resulting in a loss of comparability between studies. With this work we would like to identify and compare some of the most widely used root image analysis software out there and identify their potential impact on the uncertainty of the results.

1. Get an overview of the most widely used image analysis software for root analysis.
2. Compare the most widely used image analysis programs for root image analysis with scanned images of different plant species that have already been collected throughout the last years and identify the uncertainties that result out of the choice of the image analysis software.

Conduct a small literature review to identify the most widely used image analysis software (a study that lists some of the tools already existing and could be used as a starting point: https://link.springer.com/article/10.1186/1746-4811-9-38). Compare the application of the previously chosen programs and identify benefits and disadvantages. Apply the different software on images of scanned roots of different crop species that have already been acquired in previous studies.

We expect that the software used will impact the results differently, depending on the analyzed species and methodology used to acquire the images.

6 Month

English or German

As some of the software commonly used might not be accessible through a graphical user interface, basic programming knowledge in R and Phython will be helpful.

Dominik Behrend, Lukas Krusenbaum, Sabine Seidel

dbehrend@uni-bonn.de
lkrusen1@uni-bonn.de