Development and Validation of a Deep Learning Based Automated Minirhizotron Image Analysis Pipeline

Multi-year aboveground data of minirhizotron facilities in Selhausen

Improved understanding of crops' response to soil water stress is important to advance soil-plant system models and to support crop breeding, crop and varietal selection, and management decisions to minimize negative impacts. Studies on eco-physiological crop characteristics from leaf to canopy for different soil water conditions and crops are often carried out at controlled conditions. In-field measurements under realistic field conditions and data of plant water potential, its links with CO2 and H2O gas fluxes, and crop growth processes are rare. Here, we presented a comprehensive data set collected from leaf to canopy using sophisticated and comprehensive sensing techniques (leaf chlorophyll, stomatal conductance and photosynthesis, canopy CO2 exchange, sap flow, and canopy temperature) including detailed crop growth characteristics based on destructive methods (crop height, leaf area index, aboveground biomass, and yield). Data were acquired under field conditions with contrasting soil types, water treatments, and different cultivars of wheat and maize. The data from 2016 up to now will be made available for studying soil/water-plant relations and improving soil-plant-atmospheric continuum models.

Deep learning-based association analysis of root image data and cucumber yield

The root system is important for the absorption of water and nutrients by plants. Cultivating and selecting a root system architecture (RSA) with good adaptability and ultrahigh productivity have become the primary goals of agricultural improvement. Exploring the correlation between the RSA and crop yield is important for cultivating crop varieties with high-stress resistance and productivity. In this study, 277 cucumber varieties were collected for root system image analysis and yield using germination plates and greenhouse cultivation. Deep learning tools were used to train ResNet50 and U-Net models for image classification and segmentation of seedlings and to perform quality inspection and productivity prediction of cucumber seedling root system images. The results showed that U-Net can automatically extract cucumber root systems with high quality (F1_score ≥ 0.95), and the trained ResNet50 can predict cucumber yield grade through seedling root system image, with the highest F1_score reaching 0.86 using 10-day-old seedlings. The root angle had the strongest correlation with yield, and the shallow- and steep-angle frequencies had significant positive and negative correlations with yield, respectively. RSA and nutrient absorption jointly affected the production capacity of cucumber plants. The germination plate planting method and automated root system segmentation model used in this study are convenient for high-throughput phenotypic (HTP) research on root systems. Moreover, using seedling root system images to predict yield grade provides a new method for rapidly breeding high-yield RSA in crops such as cucumbers.

The State of the Art in Root System Architecture Image Analysis Using Artificial Intelligence: A Review

Roots are essential for acquiring water and nutrients to sustain and support plant growth and anchorage. However, they have been studied less than the aboveground traits in phenotyping and plant breeding until recent decades. In modern times, root properties such as morphology and root system architecture (RSA) have been recognized as increasingly important traits for creating more and higher quality food in the "Second Green Revolution". To address the paucity in RSA and other root research, new technologies are being investigated to fill the increasing demand to improve plants via root traits and overcome currently stagnated genetic progress in stable yields. Artificial intelligence (AI) is now a cutting-edge technology proving to be highly successful in many applications, such as crop science and genetic research to improve crop traits. A burgeoning field in crop science is the application of AI to high-resolution imagery in analyses that aim to answer questions related to crops and to better and more speedily breed desired plant traits such as RSA into new cultivars. This review is a synopsis concerning the origins, applications, challenges, and future directions of RSA research regarding image analyses using AI.

Fast and Efficient Root Phenotyping via Pose Estimation

Image segmentation is commonly used to estimate the location and shape of plants and their external structures. Segmentation masks are then used to localize landmarks of interest and compute other geometric features that correspond to the plant's phenotype. Despite its prevalence, segmentation-based approaches are laborious (requiring extensive annotation to train) and error-prone (derived geometric features are sensitive to instance mask integrity). Here, we present a segmentation-free approach that leverages deep learning-based landmark detection and grouping, also known as pose estimation. We use a tool originally developed for animal motion capture called SLEAP (Social LEAP Estimates Animal Poses) to automate the detection of distinct morphological landmarks on plant roots. Using a gel cylinder imaging system across multiple species, we show that our approach can reliably and efficiently recover root system topology at high accuracy, few annotated samples, and faster speed than segmentation-based approaches. In order to make use of this landmark-based representation for root phenotyping, we developed a Python library (sleap-roots) for trait extraction directly comparable to existing segmentation-based analysis software. We show that pose-derived root traits are highly accurate and can be used for common downstream tasks including genotype classification and unsupervised trait mapping. Altogether, this work establishes the validity and advantages of pose estimation-based plant phenotyping. To facilitate adoption of this easy-to-use tool and to encourage further development, we make sleap-roots, all training data, models, and trait extraction code available at: https://github.com/talmolab/sleap-roots and https://osf.io/k7j9g/.

Automatic Root Length Estimation from Images Acquired In Situ without Segmentation

Image-based root phenotyping technologies, including the minirhizotron (MR), have expanded our understanding of the in situ root responses to changing environmental conditions. The conventional manual methods used to analyze MR images are time-consuming, limiting their implementation. This study presents an adaptation of our previously developed convolutional neural network-based models to estimate the total (cumulative) root length (TRL) per MR image without requiring segmentation. Training data were derived from manual annotations in Rootfly, commonly used software for MR image analysis. We compared TRL estimation with 2 models, a regression-based model and a detection-based model that detects the annotated points along the roots. Notably, the detection-based model can assist in examining human annotations by providing a visual inspection of roots in MR images. The models were trained and tested with 4,015 images acquired using 2 MR system types (manual and automated) and from 4 crop species (corn, pepper, melon, and tomato) grown under various abiotic stresses. These datasets are made publicly available as part of this publication. The coefficients of determination (R2), between the measurements made using Rootfly and the suggested TRL estimation models were 0.929 to 0.986 for the main datasets, demonstrating that this tool is accurate and robust. Additional analyses were conducted to examine the effects of (a) the data acquisition system and thus the image quality on the models' performance, (b) automated differentiation between images with and without roots, and (c) the use of the transfer learning technique. These approaches can support precision agriculture by providing real-time root growth information.

Multi-year belowground data of minirhizotron facilities in Selhausen

The production of crops secure the human food supply, but climate change is bringing new challenges. Dynamic plant growth and corresponding environmental data are required to uncover phenotypic crop responses to the changing environment. There are many datasets on above-ground organs of crops, but roots and the surrounding soil are rarely the subject of longer term studies. Here, we present what we believe to be the first comprehensive collection of root and soil data, obtained at two minirhizotron facilities located close together that have the same local climate but differ in soil type. Both facilities have 7m-long horizontal tubes at several depths that were used for crosshole ground-penetrating radar and minirhizotron camera systems. Soil sensors provide observations at a high temporal and spatial resolution. The ongoing measurements cover five years of maize and wheat trials, including drought stress treatments and crop mixtures. We make the processed data available for use in investigating the processes within the soil-plant continuum and the root images to develop and compare image analysis methods.

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.

Automatic segmentation of cotton roots in high-resolution minirhizotron images based on improved OCRNet

Root phenotypic parameters are the important basis for studying the growth state of plants, and root researchers obtain root phenotypic parameters mainly by analyzing root images. With the development of image processing technology, automatic analysis of root phenotypic parameters has become possible. And the automatic segmentation of roots in images is the basis for the automatic analysis of root phenotypic parameters. We collected high-resolution images of cotton roots in a real soil environment using minirhizotrons. The background noise of the minirhizotron images is extremely complex and affects the accuracy of the automatic segmentation of the roots. In order to reduce the influence of the background noise, we improved OCRNet by adding a Global Attention Mechanism (GAM) module to OCRNet to enhance the focus of the model on the root targets. The improved OCRNet model in this paper achieved automatic segmentation of roots in the soil and performed well in the root segmentation of the high-resolution minirhizotron images, achieving an accuracy of 0.9866, a recall of 0.9419, a precision of 0.8887, an F1 score of 0.9146 and an Intersection over Union (IoU) of 0.8426. The method provided a new approach to automatic and accurate root segmentation of high-resolution minirhizotron images.

A Systematic Review of Effective Hardware and Software Factors Affecting High-Throughput Plant Phenotyping

Plant phenotyping studies the complex characteristics of plants, with the aim of evaluating and assessing their condition and finding better exemplars. Recently, a new branch emerged in the phenotyping field, namely, high-throughput phenotyping (HTP). Specifically, HTP exploits modern data sampling techniques to gather a high amount of data that can be used to improve the effectiveness of phenotyping. Hence, HTP combines the knowledge derived from the phenotyping domain with computer science, engineering, and data analysis techniques. In this scenario, machine learning (ML) and deep learning (DL) algorithms have been successfully integrated with noninvasive imaging techniques, playing a key role in automation, standardization, and quantitative data analysis. This study aims to systematically review two main areas of interest for HTP: hardware and software. For each of these areas, two influential factors were identified: for hardware, platforms and sensing equipment were analyzed; for software, the focus was on algorithms and new trends. The study was conducted following the PRISMA protocol, which allowed the refinement of the research on a wide selection of papers by extracting a meaningful dataset of 32 articles of interest. The analysis highlighted the diffusion of ground platforms, which were used in about 47% of reviewed methods, and RGB sensors, mainly due to their competitive costs, high compatibility, and versatility. Furthermore, DL-based algorithms accounted for the larger share (about 69%) of reviewed approaches, mainly due to their effectiveness and the focus posed by the scientific community over the last few years. Future research will focus on improving DL models to better handle hardware-generated data. The final aim is to create integrated, user-friendly, and scalable tools that can be directly deployed and used on the field to improve the overall crop yield.

High frequency root dynamics: sampling and interpretation using replicated robotic minirhizotrons

Automating dynamic fine root data collection in the field is a longstanding challenge with multiple applications for co-interpretation and synthesis for ecosystem understanding. High frequency root data are only achievable with paired automated sampling and processing. However, automatic minirhizotron (root camera) instruments are still rare and data are often not collected in natural soils or analysed at high temporal resolution. Instruments must also be affordable for replication and robust under variable natural conditions. Here, we show a system built with off-the-shelf parts which samples at sub-daily resolution. We paired this with a neural network to analyse all images collected. We performed two mesocosm studies and two field trials alongside ancillary data collection (soil CO2 efflux, temperature, and moisture content, and 'PhenoCam'-derived above-ground dynamics). We produce robust and replicated daily time series of root dynamics under all conditions. Temporal root changes were a stronger driver than absolute biomass on soil CO2 efflux in the mesocosm. Proximal sensed above-ground dynamics and below-ground dynamics from minirhizotron data were not synchronized. Root properties extracted were sensitive to soil moisture and occasionally to time of day (potentially relating to soil moisture). This may only affect high frequency imagery and should be considered in interpreting such data.

Uncovering natural variation in root system architecture and growth dynamics using a robotics-assisted phenomics platform

The plant kingdom contains a stunning array of complex morphologies easily observed above-ground, but more challenging to visualize below-ground. Understanding the magnitude of diversity in root distribution within the soil, termed root system architecture (RSA), is fundamental in determining how this trait contributes to species adaptation in local environments. Roots are the interface between the soil environment and the shoot system and therefore play a key role in anchorage, resource uptake, and stress resilience. Previously, we presented the GLO-Roots (Growth and Luminescence Observatory for Roots) system to study the RSA of soil-grown Arabidopsis thaliana plants from germination to maturity (Rellán-Álvarez et al., 2015). In this study, we present the automation of GLO-Roots using robotics and the development of image analysis pipelines in order to examine the temporal dynamic regulation of RSA and the broader natural variation of RSA in Arabidopsis, over time. These datasets describe the developmental dynamics of two independent panels of accessions and reveal highly complex and polygenic RSA traits that show significant correlation with climate variables of the accessions’ respective origins.