ChronoRoot: High-throughput phenotyping by deep segmentation networks reveals novel temporal parameters of plant root system architecture

Abstract

Background Deep learning methods have outperformed previous techniques in most computer vision tasks, including image-based plant phenotyping. However, massive data collection of root traits and the development of associated artificial intelligence approaches have been hampered by the inaccessibility of the rhizosphere. Here we present ChronoRoot, a system that combines 3D-printed open-hardware with deep segmentation networks for high temporal resolution phenotyping of plant roots in agarized medium. Results We developed a novel deep learning–based root extraction method that leverages the latest advances in convolutional neural networks for image segmentation and incorporates temporal consistency into the root system architecture reconstruction process. Automatic extraction of phenotypic parameters from sequences of images allowed a comprehensive characterization of the root system growth dynamics. Furthermore, novel time-associated parameters emerged from the analysis of spectral features derived from temporal signals. Conclusions Our work shows that the combination of machine intelligence methods and a 3D-printed device expands the possibilities of root high-throughput phenotyping for genetics and natural variation studies, as well as the screening of clock-related mutants, revealing novel root traits.

Root System Architecture and Environmental Flux Analysis in Mature Crops using 3D Root Mesocosms

“…This method could provide a sensitivity analysis for empirical sampling strategies using actual, rather than virtual (Burridge et al, 2020; Morandage et al, 2019), groundtruths. Such information could also be used as a valuable resource for improving root structure-function simulations (Kalogiros et al, 2016; Postma et al, 2017; Schnepf et al, 2018), or for the development of artificial intelligence approaches to complement missing data (Falk et al, 2020; Gaggion et al, 2021; Ruiz-Munoz et al, 2020).…”

High-throughput and automatic structural and developmental root phenotyping on Arabidopsis seedlings

“…One approach to quantify root growth and geometry under a wide range of growth conditions is to use a transparent medium, such as agarose gels, a common cultivation practice for the model plant Arabidopsis and other species. Several semi-automatic tools have been developed to assist root phenotyping on agar plates [ 10 , 23 , 25 , 27 , 35 ]. However, these devices are either limited in the number of plates imaged or restricted in temporal resolution.…”

“…This feature is required to distinguish local responses from systemic responses (elongation speed, branching density) [ 30 , 31 , 22 ], but offering such guarantees remains a challenge. There are several difficulties in current solutions, mainly due to the capacity to deal with root intersections and root contacts [ 10 , 33 , 35 ]. Each new crossing doubles the number of possible root paths, which causes a combinatorial explosion of the reconstruction problem (see Fig.…”

“…These time points are processed using the conventional two-step pipeline before exploring the correspondences between the extracted architectures over time. This operation requires advanced matching methods [ 2 , 10 , 33 ] and extra processing steps that could induce a performance drop when cumulating the errors of the separate reconstructions. Some authors attempted to integrate temporal information in the reconstruction process to address this issue.…”

“…Some authors attempted to integrate temporal information in the reconstruction process to address this issue. Gaggion et al [ 10 ] sidestep the geometric alignment task by fixing the specimen on the imaging device during the time course. With this setup guaranteeing a proper alignment of the successive observations, the authors enhance the temporal consistency of the obtained segmentation using a weighted average over the time series.…”