3D segmentation of plant root systems using spatial pyramid pooling and locally adaptive field-of-view inference

Abstract

BackgroundThe non-invasive 3D-imaging and successive 3D-segmentation of plant root systems has gained interest within fundamental plant research and selectively breeding resilient crops. Currently the state of the art consists of computed tomography (CT) scans and reconstruction followed by an adequate 3D-segmentation process.ChallengeGenerating an exact 3D-segmentation of the roots becomes challenging due to inhomogeneous soil composition, as well as high scale variance in the root structures themselves.Approach(1) We address the challenge by combining deep convolutional neural networks (DCNNs) with a weakly supervised learning paradigm. Furthermore, (2) we apply a spatial pyramid pooling (SPP) layer to cope with the scale variance of roots. (3) We generate a fine-tuned training data set with a specialized sub-labeling technique. (4) Finally, to yield fast and high-quality segmentations, we propose a specialized iterative inference algorithm, which locally adapts the field of view (FoV) for the network.ExperimentsWe compare our segmentation results against an analytical reference algorithm for root segmentation (RootForce) on a set of roots from Cassava plants and show qualitatively that an increased amount of root voxels and root branches can be segmented.ResultsOur findings show that with the proposed DCNN approach combined with the dynamic inference, much more, and especially fine, root structures can be detected than with a classical analytical reference method.ConclusionWe show that the application of the proposed DCNN approach leads to better and more robust root segmentation, especially for very small and thin roots.

“Chamber #8” – a holistic approach of high-throughput non-destructive assessment of plant roots

“…In comparison to previously conducted manual experiments, using the automation system installed in “chamber #8”, we can now scan and measure up to 42 plants within 8 hours (compared to 20 hours using manual labor). Furthermore, the analysis of various root systems such as maize roots is now possible using the automated post-processing step for 3D root segmentation ( Gerth et al., 2021 ; Alle et al., 2023 ). For the analysis of maize roots 16 genotypes ( A554, B104, Lo1261, IDT, Oh02, W23, Lo1106, Lo1282, A554, B104, Lo1261, B73, Oh02, W23, Lo1106, Lo1282 ) were used with five replicates each under combined nitrogen-water stress conditions.…”

“…The selection and parameterization of the individual processing steps can be customized. Currently, the segmentation of roots ( Gerth et al., 2021 ; Alle et al., 2023 ) and the segmentation of “grain ears” ( Schmidt et al., 2020 ) are supported by the automated workflow, although other algorithms can easily be inserted into the concept with an easy Python API.…”

Convolutional neural networks combined with conventional filtering to semantically segment plant roots in rapidly scanned X-ray computed tomography volumes with high noise levels

“…However, 3D segmentation reconstructed from 2D segmentation processed slice-by-slice tends to overlook 3D connections; therefore, a model utilizing 3D images as input is superior [ 39 ]. Semantic segmentation in 3D is used to extract roots of various thicknesses [ 40 ]. A conventional filtering method, such as shape-based filtering [ 26 , 27 , 41 ], can be used if the root thickness variation is minimal, whereas alternative approaches are required for large variations [ 40 ].…”

“…Semantic segmentation in 3D is used to extract roots of various thicknesses [ 40 ]. A conventional filtering method, such as shape-based filtering [ 26 , 27 , 41 ], can be used if the root thickness variation is minimal, whereas alternative approaches are required for large variations [ 40 ]. This variation can be addressed by adding a spatial pyramid pooling layer [ 42 ] or multi-resolution encoder–decoder networks [ 43 ] to a CNN extracting multi-scale features.…”

Blazing a trail towards higher photosynthetic efficiency : A multi-dimensional study of Hirschfeldia incana

“…X-ray Computed Tomography is emerging as the approach of choice for 3D imaging of the above-and below-ground organs of plants (Teramoto et al, 2020;Piovesan et al, 2021), and would allow for a first investigation aimed at identifying where the major sinks for photosynthate lie in H. incana and its relatives. While the integration of this technology is still limited in high-throughput phenotyping platforms, recent developments (Gerth et al, 2021;Alle et al, 2023) suggest that an experiment similar as what described above could be performed in the near future with focus on sink traits as well.…”