Contrast-enhanced 3D micro-CT of plant tissues using different impregnation techniques

Automatic 3D cell segmentation of fruit parenchyma tissue from X-ray micro CT images using deep learning

BACKGROUND

High quality 3D information of the microscopic plant tissue morphology-the spatial organization of cells and intercellular spaces in tissues-helps in understanding physiological processes in a wide variety of plants and tissues. X-ray micro-CT is a valuable tool that is becoming increasingly available in plant research to obtain 3D microstructural information of the intercellular pore space and individual pore sizes and shapes of tissues. However, individual cell morphology is difficult to retrieve from micro-CT as cells cannot be segmented properly due to negligible density differences at cell-to-cell interfaces. To address this, deep learning-based models were trained and tested to segment individual cells using X-ray micro-CT images of parenchyma tissue samples from apple and pear fruit with different cell and porosity characteristics.

RESULTS

The best segmentation model achieved an Aggregated Jaccard Index (AJI) of 0.86 and 0.73 for apple and pear tissue, respectively, which is an improvement over the current benchmark method that achieved AJIs of 0.73 and 0.67. Furthermore, the neural network was able to detect other plant tissue structures such as vascular bundles and stone cell clusters (brachysclereids), of which the latter were shown to strongly influence the spatial organization of pear cells. Based on the AJIs, apple tissue was found to be easier to segment, as the porosity and specific surface area of the pore space are higher and lower, respectively, compared to pear tissue. Moreover, samples with lower pore network connectivity, proved very difficult to segment.

CONCLUSIONS

The proposed method can be used to automatically quantify 3D cell morphology of plant tissue from micro-CT instead of opting for laborious manual annotations or less accurate segmentation approaches. In case fruit tissue porosity or pore network connectivity is too low or the specific surface area of the pore space too high, native X-ray micro-CT is unable to provide proper marker points of cell outlines, and one should rely on more elaborate contrast-enhancing scan protocols.

Analysis of fixation materials in micro-CT: It doesn't always have to be styrofoam

Good fixation of filigree specimens for micro-CT examinations is often a challenge. Movement artefacts, over-radiation or even crushing of the specimen can easily occur. Since different specimens have different requirements, we scanned, analysed and compared 19 possible fixation materials under the same conditions in the micro-CT. We focused on radiodensity, porosity and reversibility of these fixation materials. Furthermore, we have made sure that all materials are cheap and easily available. The scans were performed with a SkyScan 1173 micro-CT. All dry fixation materials tested were punched into 5 mm diameter cylinders and clamped into 0.2 ml reaction vessels. A voxel size of 5.33 μm was achieved in a 180° scan in 0.3° steps. Ideally, fixation materials should not be visible in the reconstructed image, i.e., barely binarised. Besides common micro-CT fixation materials such as styrofoam (-935 Hounsfield Units) or Basotect foam (-943 Hounsfield Units), polyethylene air cushions (-944 Hounsfield Units), Micropor foam (-926 Hounsfield Units) and polyurethane foam, (-960 Hounsfield Units to -470 Hounsfield Units) have proved to be attractive alternatives. Furthermore, more radiopaque materials such as paraffin wax granulate (-640 Hounsfield Units) and epoxy resin (-190 Hounsfield Units) are also suitable as fixation materials. These materials often can be removed in the reconstructed image by segmentation. Sample fixations in the studies of recent years are almost all limited to fixation in Parafilm, Styrofoam, or Basotect foam if the fixation type is mentioned at all. However, these are not always useful, as styrofoam, for example, dissolves in some common media such as methylsalicylate. We show that micro-CT laboratories should be equipped with various fixation materials to achieve high-level image quality.

Flip-Flap: A Simple Dual-View Imaging Method for 3D Reconstruction of Thick Plant Samples

Plant development is a complex process that relies on molecular and cellular events being co-ordinated in space and time. Microscopy is one of the most powerful tools available to investigate this spatiotemporal complexity. One step towards a better understanding of complexity in plants would be the acquisition of 3D images of entire organs. However, 3D imaging of intact plant samples is not always simple and often requires expensive and/or non-trivial approaches. In particular, the inner tissues of thick samples are challenging to image. Here, we present the Flip-Flap method, a simple imaging protocol to produce 3D images of cleared plant samples at the organ scale. This method allows full 3D reconstruction of plant organs suitable for 3D segmentation and further related analysis and can be easily handled by relatively inexperienced microscopists.

Maternal-filial transfer structures in endosperm: A nexus of nutritional dynamics and seed development

Although the ultimate purpose of a seed is the successful establishment of the next generation, seed development involves more than embryo growth. In angiosperms, seed development requires the intimate coordination of three distinct entities - maternal tissue and two offspring, embryo and embryo-nourishing endosperm. Although seeds are cornerstones of many terrestrial ecosystems and human diets, we are only beginning to understand the interactions among seed tissues and the molecular processes and genes that determine them. Recent studies of gene expression and function in distantly related angiosperms, combined with over 100 years of embryological research, have repeatedly highlighted the endosperm associated with maternal-filial boundaries as a central point in seed developmental dynamics. In this review, we highlight evidence that links this zone with nutritional dynamics, developmental signaling, and imprinted gene expression. We suggest that the underappreciated diversity of this specialized endosperm across angiosperms deserves further study from developmental, molecular, and genetic perspectives.

Microstructural changes enhance oxygen transport in tomato (Solanum lycopersicum) fruit during maturation and ripening

Climacteric ripening of tomato fruit is initiated by a characteristic surge of the production rate of ethylene, accompanied by an increase in respiration rate. As both activities consume O₂ and produce CO₂ , gas concentration gradients develop in the fruit that cause diffusive transport. This may, in turn, affect respiration and ethylene biosynthesis. Gas diffusion in fruit depends on the amount and connectivity of cells and intercellular spaces in 3D. We investigated micromorphological changes in different tomato tissues during development and ripening by visualizing cells and pores based on high-resolution micro-computed tomography, and computed effective O₂ diffusivity coefficients based on microstructural features of the tissues. We demonstrated that mesocarp and septa tissues have larger cells but small and more disconnected pores than the placenta and columella, resulting in relatively lower effective O₂ diffusivity coefficients. Cell disintegration occurred in the mesocarp and septa during ripening, indicating lysigenous air pore formation and resulting in a gradual increase of the effective O₂ diffusivity. The results suggest that hypoxic conditions caused by the increasing size and, hence, diffusion resistance of the growing fruit may induce an increase of tissue porosity that results in a greatly enhanced O₂ diffusivity and, thus, helps to alleviate them.

X-ray computed tomography for 3D plant imaging

X-ray computed tomography (CT) is a valuable tool for 3D imaging of plant tissues and organs. Applications include the study of plant development and organ morphogenesis, as well as modeling of transport processes in plants. Some challenges remain, however, including attaining higher contrast for easier quantification, increasing the resolution for imaging subcellular features, and decreasing image acquisition and processing time for high-throughput phenotyping. In addition, phase contrast, multispectral, dark-field, soft X-ray, and time-resolved imaging are emerging. At the same time, a large amount of 3D image data are becoming available, posing challenges for data management. We review recent advances in the area of X-ray CT for plant imaging, and describe opportunities for using such images for studying transport processes in plants.

Microstructure investigation of plant architecture with X-ray microscopy

In recent years, the plant morphology has been well studied by multiple approaches at cellular and subcellular levels. Two-dimensional (2D) microscopy techniques offer imaging of plant structures on a wide range of magnifications for researchers. However, subcellular imaging is still challenging in plant tissues like roots and seeds. Here we use a three-dimensional (3D) imaging technology based on the X-ray microscope (XRM) and analyze several plant tissues from different plant species. The XRM provides new insights into plant structures using non-destructive imaging at high-resolution and high contrast. We also utilized a workflow aiming to acquire accurate and high-quality images in the context of the whole specimen. Multiple plant samples including rice, tobacco, Arabidopsis and maize were used to display the differences of phenotypes. Our work indicates that the XRM is a powerful tool to investigate plant microstructure in high-resolution scale. Our work also provides evidence that evaluate and quantify tissue specific differences for a range of plant species. We also characterize novel plant tissue phenotypes by the XRM, such as seeds in Arabidopsis, and utilize them for novel observation measurement. Our work represents an evaluated spatial and temporal resolution solution on seed observation and screening.

The formation of interlocked grain in African mahogany (Khaya spp.) analysed by X-ray computed microtomography

Interlocked grain occurs when the orientation of xylem fibres oscillates, alternating between left- and right-handed spirals in successive wood layers. The cellular mechanisms giving rise to interlocked grain, thought to involve the slow rotation of fusiform initials within the vascular cambium, remain unclear. We suggest that observations of wood structure at the cellular level, but over large areas, might reveal these mechanisms. We assayed timber from several commercially important tropical angiosperms from the genus Khaya (African mahogany) that exhibit interlocked grain using X-ray computed microtomography followed by orthogonal slicing and image processing in ImageJ. Reconstructed tangential longitudinal sections were processed with the ImageJ directionality plug-in to directly measure fibre orientation and showed grain deviations of more than 10° from vertical in both left- and right-handed directions. Grain changed at locally constant rates, separated by locations where the direction of grain change sharply reversed. Image thresholding and segmentation conducted on reconstructed cross sections allowed the identification of vessels and measurement of their location, with vessel orientations then calculated in Matlab and, independently, in recalculated tangential longitudinal sections with the directionality plug-in. Vessel orientations varied more than fibre orientations, and on average deviated further from vertical than fibres at the locations where the direction of grain change reversed. Moreover, the reversal location for vessels was shifted ~400 μm towards the pith compared with the fibres, despite both cell types arising from the same fusiform initials within the vascular cambium. We propose a simple model to explain these distinct grain patterns. Were an auxin signal to control both the reorientation of cambial initials, as well as coordinating the end-on-end differentiation and linkage of xylem vessel elements, then it would be possible for fibres and vessels to run at subtly different angles, and to show different grain reversal locations.

Spatial heterogeneity of biochar (segregation) in biochar-amended media: An overlooked phenomenon, and its impact on saturated hydraulic conductivity

While the use of biochar as a soil amendment is gaining popularity for environmental and agricultural purposes, spatial heterogeneity of biochar (segregation) in biochar-amended media and its underlying causes have been overlooked. In this study, for the first time particle segregation in biochar-amended media and its impact on the media's saturated hydraulic conductivity (K_sat) were investigated. Two uniformly graded media were amended with different sizes of a wood-based biochar under dry and wet conditions. While the intended biochar volume fraction (b_f) was 17.5%, in dry-packed columns biochar was often segregated and the measured b_f ranged from 7.5 ± 0.8 SE% (SE = standard error) to 23.6 ± 1.8 SE% across all spatial locations. If, however, 20% water (volume of water/bulk volume of packed media) was added to the mixtures during mixing, homogeneous packings were achieved. In dry-packing, segregation was governed by the difference in the physical properties of the media and the biochar: particle size, density, and shape. In wet-packing, segregation was prevented due to the inter-particle adhesion forces associated with water. Although X-ray computed tomography images showed that the presence of segregation altered particle distributions and pore morphologies, the K_sat for wet-packed and dry-packed columns were statistically identical. The results of this study suggest that laboratory methods for packing biochar-amended media should include moisturizing the mixture to inhibit particle segregation. Mixing under wet conditions is recommended for any type of soil and biochar and for any scale of application, in both the laboratory and field.

ROSE-X: an annotated data set for evaluation of 3D plant organ segmentation methods

BACKGROUND

The production and availability of annotated data sets are indispensable for training and evaluation of automatic phenotyping methods. The need for complete 3D models of real plants with organ-level labeling is even more pronounced due to the advances in 3D vision-based phenotyping techniques and the difficulty of full annotation of the intricate 3D plant structure.

RESULTS

We introduce the ROSE-X data set of 11 annotated 3D models of real rosebush plants acquired through X-ray tomography and presented both in volumetric form and as point clouds. The annotation is performed manually to provide ground truth data in the form of organ labels for the voxels corresponding to the plant shoot. This data set is constructed to serve both as training data for supervised learning methods performing organ-level segmentation and as a benchmark to evaluate their performance. The rosebush models in the data set are of high quality and complex architecture with organs frequently touching each other posing a challenge for the current plant organ segmentation methods. We report leaf/stem segmentation results obtained using four baseline methods. The best performance is achieved by the volumetric approach where local features are trained with a random forest classifier, giving Intersection of Union (IoU) values of 97.93% and 86.23% for leaf and stem classes, respectively.

CONCLUSION

We provided an annotated 3D data set of 11 rosebush plants for training and evaluation of organ segmentation methods. We also reported leaf/stem segmentation results of baseline methods, which are open to improvement. The data set, together with the baseline results, has the potential of becoming a significant resource for future studies on automatic plant phenotyping.

Using Pixel-Based Microscope Images to Generate 3D Reconstructions of Frozen and Thawed Plant Tissue

Histological analysis of frozen and thawed plants has been conducted for many years but the observation of individual sections only provides a two-dimensional representation of a three-dimensional phenomenon. Currently available optical sectioning techniques for viewing internal structures in three dimensions are either low in resolution or the instrument cannot penetrate deep enough into the tissue to visualize the whole plant. Methods using higher resolution equipment are expensive and often require time-consuming training. In addition, conventional stains cannot be used for optical sectioning techniques. We present a relatively simple and less expensive technique using pixel-based (JPEG) images of conventionally stained histological sections of an Arabidopsis thaliana plant. The technique uses commercially available software to generate a 3D representation of internal structures.

Discrete element modelling of tomato tissue deformation and failure at the cellular scale

Bruise damage in fruit results from cell wall failure and inter-cellular separation. Despite the importance of the micro-mechanics of plant tissue with respect to its integrity, it remains largely unquantified and poorly understood, due to many difficulties during experimental characterization. In this article, a 3D micro-mechanical plant tissue model that is able to model cell rupture and inter-cellular debonding and thus provide more insight into the micro-mechanics was developed. The model is based on the discrete element method (DEM) and represents the tissue as a mass-spring system. Each plant cell is represented as a deformable visco-elastoplastic triangulated mesh under turgor pressure. To model cell wall rupture, it is assumed that a spring connection in the wall breaks at a certain critical stretch ratio and that a ruptured cell is turgorless. The inter-cellular contact model assumes brittle fracture between a cell's node and an adjacent cell's triangle when their bond distance exceeds a critical value. A high-speed tomato fruit cell compression test was simulated and the modelled force-strain curve compares well with the experimental data, including for strains above the elastic limit. By varying the shape of the cell in the compression simulation it was shown that the force-strain curve is highly dependent on the cell shape and thus parameter fitting procedures based on a spherical cell model will be inaccurate. Furthermore, the wall stiffness and thickness showed a positive linear relationship with the force at cell bursting. Besides simulating compression tests of single cells, we also simulated tensile and compression tests on small tissue specimens. Realistic tissue structures of tomato mesocarp tissue were generated by a novel method using DEM simulations of deformable cells in a shrinking cylinder. The cell area, volume and anisotropy distributions of the virtual tissue compared well with micro-CT images of real tomato mesocarp tissue (normalized root mean square error values smaller than 3%). The tissue compression and tensile test simulations demonstrated an important influence of the inter-cellular bonding energy and tissue porosity on the tissue failure characteristics and elastic modulus.