Digging roots is easier with AI.

The scale of root quantification in research is often limited by the time required for sampling, measurement and processing samples. Recent developments in Convolutional Neural Networks (CNN) have made faster and more accurate plant image analysis possible which may significantly reduce the time required for root measurement, but challenges remain in making these methods accessible to researchers without an in-depth knowledge of Machine Learning. We analyzed root images acquired from three destructive root samplings using the RootPainter CNN-software that features an interface for corrective annotation for easier use. Root scans with and without non-root debris were used to test if training a model, i.e., learning from labeled examples, can effectively exclude the debris by comparing the end-results with measurements from clean images. Root images acquired from soil profile walls and the cross-section of soil cores were also used for training and the derived measurements were compared with manual measurements. After 200 minutes of training on each dataset, significant relationships between manual measurements and RootPainter-derived data were noted for monolith (R 2=0.99), profile wall (R 2=0.76) and core-break (R 2=0.57). The rooting density derived from images with debris was not significantly different from that derived from clean images after processing with RootPainter. Rooting density was also successfully calculated from both profile wall and soil core images, and in each case the gradient of root density with depth was not significantly different from manual counts. Differences in root-length density (RLD: cm cm -3) between crops with contrasting root systems were captured using automatic segmentation at soil profiles with high RLD (1 to 5 cm cm -3) as well as at low RLD (0.1 to 0.3 cm cm -3). Our results demonstrate that the proposed approach using CNN can lead to substantial reductions in root sample processing workloads, increasing the potential scale of future root investigations.

[1]  Miriam Athmann,et al.  Biopore-Induced Deep Root Traits of Two Winter Crops , 2020, Agriculture.

[2]  K. Thorup-Kristensen,et al.  Identifying new deep-rooted plant species suitable as undersown nitrogen catch crops , 2015, Journal of Soil and Water Conservation.

[3]  R. Richards,et al.  Soil coring at multiple field environments can directly quantify variation in deep root traits to select wheat genotypes for breeding , 2014, Journal of experimental botany.

[4]  J. V. van Aardt,et al.  Stem and root assessment in mangrove forests using a low-cost, rapid-scan terrestrial laser scanner , 2020, Wetlands Ecology and Management.

[5]  R. Horn,et al.  Root growth dynamics inside and outside of soil biopores as affected by crop sequence determined with the profile wall method , 2015, Biology and Fertility of Soils.

[7]  Michael Gygli,et al.  Continuous Adaptation for Interactive Object Segmentation by Learning from Corrections , 2019, ECCV.

[8]  W. Böhm,et al.  In situ estimation of root length at natural soil profiles , 1976, The Journal of Agricultural Science.

[9]  John A Kirkegaard,et al.  The distribution and abundance of wheat roots in a dense, structured subsoil--implications for water uptake. , 2010, Plant, cell & environment.

[10]  M. Quemada,et al.  Intercropping effect on root growth and nitrogen uptake at different nitrogen levels , 2015 .

[11]  Jens Petersen,et al.  Segmentation of roots in soil with U-Net , 2019, Plant Methods.

[12]  S. Teramoto,et al.  A Deep Learning-Based Phenotypic Analysis of Rice Root Distribution from Field Images , 2020, Plant phenomics.

[13]  J. Couwenberg,et al.  Digital, Three-Dimensional Visualization of Root Systems in Peat , 2020, Soil Systems.

[14]  T. Döring,et al.  Vertical Root Distribution of Different Cover Crops Determined with the Profile Wall Method , 2020 .

[15]  John A. Kirkegaard,et al.  Benefits of increased soil exploration by wheat roots , 2011 .

[16]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[17]  T. Kautz,et al.  Dynamics of plant nutrient uptake as affected by biopore-associated root growth in arable subsoil , 2017, Plant and Soil.

[18]  W. R. Whalley,et al.  Root growth in field-grown winter wheat: Some effects of soil conditions, season and genotype , 2017, European journal of agronomy : the journal of the European Society for Agronomy.

[19]  M. R. G. Oliveira,et al.  Trench Profile Techniques and Core Break Methods , 2001 .

[20]  R. Richards,et al.  Deeper roots associated with cooler canopies, higher normalized difference vegetation index, and greater yield in three wheat populations grown on stored soil water , 2019, Journal of experimental botany.

[21]  Changying Li,et al.  Convolutional Neural Networks for Image-Based High-Throughput Plant Phenotyping: A Review , 2020, Plant phenomics.

[22]  Mohammadreza Soltaninejad,et al.  Three Dimensional Root CT Segmentation Using Multi-Resolution Encoder-Decoder Networks , 2020, IEEE Transactions on Image Processing.

[23]  Lina J. Karam,et al.  Understanding how image quality affects deep neural networks , 2016, 2016 Eighth International Conference on Quality of Multimedia Experience (QoMEX).

[24]  Kristian Thorup-Kristensen,et al.  RootPainter: deep learning segmentation of biological images with corrective annotation , 2020, bioRxiv.

[25]  S. Peth,et al.  Root-length densities of various annual crops following crops with contrasting root systems , 2014 .

[26]  Miriam Athmann,et al.  Can precrops uplift subsoil nutrients to topsoil? , 2021, Plant and Soil.

[27]  Haichao Guo,et al.  RhizoVision Crown: An Integrated Hardware and Software Platform for Root Crown Phenotyping , 2020, Plant phenomics.