At the 8th International Plant Phenotyping Symposium taking place in October 2024 in Lincoln, Nebraska, USA, Dr. Marcus Jansen, Application Scientist at LemnaTec is presenting a poster – it’s poster number 39

You can find our poster in the poster presenting area in the Embassy Exhibition Hall. Poster presentations are scheduled for October 8th and 9th, between 18 h and 19 h.

Looking forward to meeting you at IPPS!

Application-oriented root system phenotyping with AI-based algorithms

Marcus Jansen, Rüdiger Goetz, LemnaTec GmbH, Nerscheider Weg 170, 52076 Aachen, Germany

Root functions are essential in adapting plants to climatic conditions and future developments. Addressing the challenges of root phenotyping, root growth systems, corresponding imaging technologies, and image analysis algorithms are constantly being developed and improved. From a technology provider’s perspective, imaging and image processing methods and technology must be ready to be operated by customers, meaning that the demand for user interaction should be minimized, particularly in the analysis part.

Imaging technology has been developed to work with many typical root analysis methods, including transparent pots, rhizotron boxes, paper-based root displays, or transparent media in petri-dishes. All such growth systems for root analysis have different optical properties so that dedicated imaging systems were established to acquire high-quality images of the growing roots. Thereby the focus was set on imaging systems that use standard snapshot cameras combined with application-adapted LED light to limit the technical effort of imaging. Consequently, a range of imaging technologies is offered for different applications in plant research, breeding, or seed production.

With a common set of image processing algorithms, images from all image recording systems can be processed for phenotypic features of roots and root systems. This comprises root length and area, branching and angels of roots, geometrical features of root systems, or root distributions across the rooting space. With data on root curvature and tangential angles between roots, the analytical algorithms provide in-depth information on how roots explore the substrate space. As analytical algorithms are trained using machine learning, the analysis of customized analysis targets can be enabled. Machine learning not only helps to cope with overlaps or occlusions, but also can serve to recognize stress- or disease-induced damages.

Using these hard- and software tools, many applications can be addressed. Genetic control of root development, root responses to environmental factors, or nutrient-dependent root growth are some of the examples where root phenotypic data, combined with data on shoot features provide deeper insights into biological functions.

A digital copy of the poster will be available here!