3D U-Net Segmentation Improves Root System Reconstruction from 3D MRI Images in Automated and Manual Virtual Reality Work Flows.

Magnetic resonance imaging (MRI) is used to image root systems grown in opaque soil. However, reconstruction of root system architecture (RSA) from 3-dimensional (3D) MRI images is challenging. Low resolution and poor contrast-to-noise ratios (CNRs) hinder automated reconstruction. Hence, manual reconstruction is still widely used. Here, we evaluate a novel 2-step work flow for automated RSA reconstruction. In the first step, a 3D U-Net segments MRI images into root and soil in super-resolution. In the second step, an automated tracing algorithm reconstructs the root systems from the segmented images. We evaluated the merits of both steps for an MRI dataset of 8 lupine root systems, by comparing the automated reconstructions to manual reconstructions of unaltered and segmented MRI images derived with a novel virtual reality system. We found that the U-Net segmentation offers profound benefits in manual reconstruction: reconstruction speed was doubled (+97%) for images with low CNR and increased by 27% for images with high CNR. Reconstructed root lengths were increased by 20% and 3%, respectively. Therefore, we propose to use U-Net segmentation as a principal image preprocessing step in manual work flows. The root length derived by the tracing algorithm was lower than in both manual reconstruction methods, but segmentation allowed automated processing of otherwise not readily usable MRI images. Nonetheless, model-based functional root traits revealed similar hydraulic behavior of automated and manual reconstructions. Future studies will aim to establish a hybrid work flow that utilizes automated reconstructions as scaffolds that can be manually corrected.

Investigating Soil-Root Interactions with the Numerical Model R-SWMS.

In this chapter, we present the Root and Soil Water Movement and Solute transport model R-SWMS, which can be used to simulate flow and transport in the soil-plant system. The equations describing water flow in soil-root systems are presented and numerical solutions are provided. An application of R-SWMS is then briefly discussed, in which we combine in vivo and in silico experiments in order to decrypt water flow in the soil-root domain. More precisely, light transmission imaging experiments were conducted to generate data that can serve as input for the R-SWMS model. These data include the root system architecture, the soil hydraulic properties and the environmental conditions (initial soil water content and boundary conditions, BC). Root hydraulic properties were not acquired experimentally, but set to theoretical values found in the literature. In order to validate the results obtained by the model, the simulated and experimental water content distributions were compared. The model was then used to estimate variables that were not experimentally accessible, such as the actual root water uptake distribution and xylem water potential.

Call for Participation: Collaborative Benchmarking of Functional-Structural Root Architecture Models. The Case of Root Water Uptake.

Three-dimensional models of root growth, architecture and function are becoming important tools that aid the design of agricultural management schemes and the selection of beneficial root traits. However, while benchmarking is common in many disciplines that use numerical models, such as natural and engineering sciences, functional-structural root architecture models have never been systematically compared. The following reasons might induce disagreement between the simulation results of different models: different representation of root growth, sink term of root water and solute uptake and representation of the rhizosphere. Presently, the extent of discrepancies is unknown, and a framework for quantitatively comparing functional-structural root architecture models is required. We propose, in a first step, to define benchmarking scenarios that test individual components of complex models: root architecture, water flow in soil and water flow in roots. While the latter two will focus mainly on comparing numerical aspects, the root architectural models have to be compared at a conceptual level as they generally differ in process representation. Therefore, defining common inputs that allow recreating reference root systems in all models will be a key challenge. In a second step, benchmarking scenarios for the coupled problems are defined. We expect that the results of step 1 will enable us to better interpret differences found in step 2. This benchmarking will result in a better understanding of the different models and contribute toward improving them. Improved models will allow us to simulate various scenarios with greater confidence and avoid bugs, numerical errors or conceptual misunderstandings. This work will set a standard for future model development.

CRootBox: a structural-functional modelling framework for root systems.

Background and Aims: Root architecture development determines the sites in soil where roots provide input of carbon and take up water and solutes. However, root architecture is difficult to determine experimentally when grown in opaque soil. Thus, root architecture models have been widely used and been further developed into functional-structural models that simulate the fate of water and solutes in the soil-root system. The root architecture model CRootBox presented here is a flexible framework to model root architecture and its interactions with static and dynamic soil environments. Methods: CRootBox is a C++-based root architecture model with Python binding, so that CRootBox can be included via a shared library into any Python code. Output formats include VTP, DGF, RSML and a plain text file containing coordinates of root nodes. Furthermore, a database of published root architecture parameters was created. The capabilities of CRootBox for the unconfined growth of single root systems, as well as the different parameter sets, are highlighted in a freely available web application. Key results: The capabilities of CRootBox are demonstrated through five different cases: (1) free growth of individual root systems; (2) growth of root systems in containers as a way to mimic experimental setups; (3) field-scale simulation; (4) root growth as affected by heterogeneous, static soil conditions; and (5) coupling CRootBox with code from the book Soil physics with Python to dynamically compute water flow in soil, root water uptake and water flow inside roots. Conclusions: CRootBox is a fast and flexible functional-structural root model that is based on state-of-the-art computational science methods. Its aim is to facilitate modelling of root responses to environmental conditions as well as the impact of roots on soil. In the future, this approach will be extended to the above-ground part of the plant.