Chapter

Deep Learning-Based Classification of Plant Xylem Tissue from Light Micrographs

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

Anatomical studies of plant hydraulic traits have traditionally been conducted by manual measurements of light micrographs. An automated process could expedite analysis and broaden the scope of questions that can be asked, but such an approach would require the ability to accurately classify plant cells according to their type. Our research evaluates a deep learning-based model which accepts a cropped cell image input alongside a broader cropped image which incorporates contextual information of that cell type’s original cropped image, and learns to segregate these plant cells based off of the features of both inputs. Whilst a single cropped image classification yielded adequate results with outputs matching the ground-truth labels, we discovered that a second image input significantly bolstered the model’s learning and accuracy (98.1%), indicating that local context provides important information needed to accurately classify cells. Finally our results imply a future application of our classifier to automatic cell-type detection in xylem tissue image cross sections.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

Article
Full-text available
The heart is an essential organ in the human body. It contains various types of cells, such as cardiomyocytes, mesothelial cells, endothelial cells, and fibroblasts. The interactions between these cells determine the vital functions of the heart. Therefore, identifying the different cell types and revealing the expression rules in these cell types are crucial. In this study, multiple machine learning methods were used to analyze the heart single-cell profiles with 11 different heart cell types. The single-cell profiles were first analyzed via light gradient boosting machine method to evaluate the importance of gene features on the profiling dataset, and a ranking feature list was produced. This feature list was then brought into the incremental feature selection method to identify the best features and build the optimal classifiers. The results suggested that the best decision tree (DT) and random forest classification models achieved the highest weighted F1 scores of 0.957 and 0.981, respectively. The selected features, such as NPPA, LAMA2, DLC1, and the classification rules extracted from the optimal DT classifier played a crucial role in cardiac structure and function in recent research and enrichment analysis. In particular, some lncRNAs (LINC02019, NEAT1) were found to be quite important for the recognition of different cardiac cell types. In summary, these findings provide a solid academic foundation for the development of molecular diagnostics and biomarker discovery for cardiac diseases.
Article
Full-text available
The recent developments in artificial intelligence have the potential to facilitate new research methods in ecology. Especially Deep Convolutional Neural Networks (DCNNs) have been shown to outperform other approaches in automatic image analyses. Here we apply a DCNN to facilitate quantitative wood anatomical (QWA) analyses, where the main challenges reside in the detection of a high number of cells, in the intrinsic variability of wood anatomical features, and in the sample quality. To properly classify and interpret features within the images, DCNNs need to undergo a training stage. We performed the training with images from transversal wood anatomical sections, together with manually created optimal outputs of the target cell areas. The target species included an example for the most common wood anatomical structures: four conifer species; a diffuse-porous species, black alder (Alnus glutinosa L.); a diffuse to semi-diffuse-porous species, European beech (Fagus sylvatica L.); and a ring-porous species, sessile oak (Quercus petraea Liebl.). The DCNN was created in Python with Pytorch, and relies on a Mask-RCNN architecture. The developed algorithm detects and segments cells, and provides information on the measurement accuracy. To evaluate the performance of this tool we compared our Mask-RCNN outputs with U-Net, a model architecture employed in a similar study, and with ROXAS, a program based on traditional image analysis techniques. First, we evaluated how many target cells were correctly recognized. Next, we assessed the cell measurement accuracy by evaluating the number of pixels that were correctly assigned to each target cell. Overall, the “learning process” defining artificial intelligence plays a key role in overcoming the issues that are usually manually solved in QWA analyses. Mask-RCNN is the model that better detects which are the features characterizing a target cell when these issues occur. In general, U-Net did not attain the other algorithms’ performance, while ROXAS performed best for conifers, and Mask-RCNN showed the highest accuracy in detecting target cells and segmenting lumen areas of angiosperms. Our research demonstrates that future software tools for QWA analyses would greatly benefit from using DCNNs, saving time during the analysis phase, and providing a flexible approach that allows model retraining.
Article
Full-text available
We present a new large-scale three-fold annotated microscopy image dataset, aiming to advance the plant cell biology research by exploring different cell microstructures including cell size and shape, cell wall thickness, intercellular space, etc. in deep learning (DL) framework. This dataset includes 9,811 unstained and 6,127 stained (safranin-o, toluidine blue-o, and lugol’s-iodine) images with three-fold annotation including physical, morphological, and tissue grading based on weight, different section area, and tissue zone respectively. In addition, we prepared ground truth segmentation labels for three different tuber weights. We have validated the pertinence of annotations by performing multi-label cell classification, employing convolutional neural network (CNN), VGG16, for unstained and stained images. The accuracy has been achieved up to 0.94, while, F2-score reaches to 0.92. Furthermore, the ground truth labels have been verified by semantic segmentation algorithm using UNet architecture which presents the mean intersection of union up to 0.70. Hence, the overall results show that the data are very much efficient and could enrich the domain of microscopy plant cell analysis for DL-framework.
Article
Full-text available
Quantitative analysis of plant and animal morphogenesis requires accurate segmentation of individual cells in volumetric images of growing organs. In the last years, deep learning has provided robust automated algorithms that approach human performance, with applications to bio-image analysis now starting to emerge. Here, we present PlantSeg, a pipeline for volumetric segmentation of plant tissues into cells. PlantSeg employs a convolutional neural network to predict cell boundaries and graph partitioning to segment cells based on the neural network predictions. PlantSeg was trained on 1xed and live plant organs imaged with confocal and light sheet microscopes. PlantSeg delivers accurate results and generalizes well across different tissues, scales, acquisition settings even on non plant samples. We present results of PlantSeg applications in diverse developmental contexts. PlantSeg is free and open-source, with both a command line and a user-friendly graphical interface (https://github.com/hci-unihd/plant-seg).
Conference Paper
Full-text available
Traditionally, data mining algorithms and machine learning algorithms are engineered to approach the problems in isolation. These algorithms are employed to train the model in separation on a specific feature space and same distribution. Depending on the business case, a model is trained by applying a machine learning algorithm for a specific task. A widespread assumption in the field of machine learning is that training data and test data must have identical feature spaces with the underlying distribution. On the contrary, in real world this assumption may not hold and thus models need to be rebuilt from the scratch if features and distribution changes. It is an arduous process to collect related training data and rebuild the models. In such cases, Transferring of Knowledge or transfer learning from disparate domains would be desirable. Transfer learning is a method of reusing a pre-trained model knowledge for another task. Transfer learning can be used for classification, regression and clustering problems. This paper uses one of the pre-trained models – VGG - 16 with Deep Convolutional Neural Network to classify images.
Article
Full-text available
Xylem is a vascular tissue that conducts sap (water and dissolved minerals) from the roots to the rest of the plant while providing physical support and resources. Sap is conducted within dead hollow cells (called vessels in flowering plants) arranged to form long pipes. Once formed, vessels do not change their structure and last from years to millennia. Vessels’ configuration (size, abundance, and spatial pattern) constitutes a record of the plant–environment relationship, and therefore, a tool for monitoring responses at the plant and ecosystem level. This information can be extracted through quantitative anatomy; however, the effort to identify and measure hundreds of thousands of conductive cells is an inconvenience to the progress needed to have solid assessments of the anatomical–environment relationship. In this paper, we propose an automatic methodology based on convolutional neural networks to segment xylem vessels. It includes a post-processing stage based on the use of redundant information to improve the performance of the outcome and make it useful in different sample configurations. Three different neural networks were tested obtaining similar results (pixel accuracy about 90%), which indicates that the methodology can be effectively used for segmentation of xylem vessels into images with non-homogeneous variations of illumination. The development of accurate automatic tools using CNNs would reduce the entry barriers associated with quantitative xylem anatomy expanding the use of this technique by the scientific community.
Article
Full-text available
High-resolution microscopy images of tissue specimens provide detailed information about the morphology of normal and diseased tissue. Image analysis of tissue morphology can help cancer researchers develop a better understanding of cancer biology. Segmentation of nuclei and classification of tissue images are two common tasks in tissue image analysis. Development of accurate and efficient algorithms for these tasks is a challenging problem because of the complexity of tissue morphology and tumor heterogeneity. In this paper we present two computer algorithms; one designed for segmentation of nuclei and the other for classification of whole slide tissue images. The segmentation algorithm implements a multiscale deep residual aggregation network to accurately segment nuclear material and then separate clumped nuclei into individual nuclei. The classification algorithm initially carries out patch-level classification via a deep learning method, then patch-level statistical and morphological features are used as input to a random forest regression model for whole slide image classification. The segmentation and classification algorithms were evaluated in the MICCAI 2017 Digital Pathology challenge. The segmentation algorithm achieved an accuracy score of 0.78. The classification algorithm achieved an accuracy score of 0.81. These scores were the highest in the challenge.
Article
Full-text available
Blood and its components have an important place in human life and are the best indicator tool in determining many pathological conditions. In particular, the classification of white blood cells is of great importance for the diagnosis of hematological diseases. In this study, 350 microscopic blood smear images were tested with 6 different machine learning algorithms for the classification of white blood cells and their performances were compared. 35 different geometric and statistical (texture) features have been extracted from blood images for training and test parameters of machine learning algorithms. According to the results, the Multinomial Logistic Regression (MLR) algorithm performed better than the other methods with an average 95% test success. The MLR can be used for automatic classification of white blood cells. It can be used especially as a source for diagnosis of diseases for hematologists and internal medicine specialists.
Chapter
Full-text available
Chaparral ecosystems represent the iconic vegetation of California, and in particular southern California, where it forms the dominant vegetation cover over broad areas of the foothills of the Coast, Transverse, and Peninsular ranges. Evergreen sclerophyll shrubs which makeup the characteristic component of chaparral communities parallel a similar dominance of this life-form in the Mediterranean Basin, central Chile, the Cape Region of South Africa, and Southwest Australia, regions of the world with a Mediterranean-type climate of warm dry summers and cool wet winters. The Mediterranean Biome comprised of these five regions are biodiversity hotspots that contain about one-sixth of the vascular plant species in the world in just 2.2% of the world’s land area. Despite this global significance, these regions continue to be heavily impacted by urbanization, land-use change, climate change, and invasions by non-native species. Chaparral floras include not just the dominant woody shrubs but a diverse assemblage of annual and herbaceous perennial species, many of which have life histories linked to postfire succession. Fire is a natural component of the disturbance regime of chaparral and burns broad portions of the landscape in a coarse-grained manner, but with fine-grained differences in fuel composition and slope aspects. Short fire-return intervals of less than 10–15 years present an increasing threat to chaparral ecosystems by eliminating shrub regeneration and leading to type-conversion to non-native annual grasslands. Water availability and associated adaptive traits of drought tolerance are major factors in partitioning chaparral community composition. Nutrient availability is also important, as are, to a lesser extent, extremes of winter temperature. Although often maligned as a useless or even dangerous because of concerns over fire hazard, chaparral ecosystems provide critical ecosystem services through their roles in erosion control, hydrology, biomass sequestration, and preservation of biodiversity.
Article
Full-text available
The secondary xylem of woody plants transports water, mechanically supports the plant body, and stores resources. These three functions are interdependent giving rise to tradeoffs in function. Understanding the relationships among these functions and their structural basis forms the context in which to interpret xylem evolution. The tradeoff between xylem transport efficiency and safety from cavitation has been carefully examined with less focus on other functions, particularly storage. Here, we synthesize data on all three xylem functions in angiosperm branch xylem in the context of tradeoffs. Species that have low safety and efficiency, examined from a resource economics perspective, are predicted to be adapted for slow resource acquisition and turnover as characterizes some environments. Tradeoffs with water storage primarily arise due to differences in fiber traits, while tradeoffs in carbohydrate storage are driven by parenchyma content of tissue. We find support for a tradeoff between safety from cavitation and storage of both water and starch in branch xylem tissue and between water storage capacity and mechanical strength. Living fibers may facilitate carbohydrate storage without compromising mechanical strength. The division of labor between different xylem cell types allows for considerable functional and structural diversity at multiple scales.
Article
Full-text available
Chaparral is the most abundant vegetation type in California and current climate change models predict more frequent and severe droughts that could impact plant community structure. Understanding the factors related to species-specific drought mortality is essential to predict such changes. We predicted that life history type, hydraulic traits, and plant size would be related to the ability of species to survive drought. We evaluated the impact of these factors in a mature chaparral stand during the drought of 2014, which has been reported as the most severe in California in the last 1,200 years. We measured tissue water potential, native xylem specific conductivity, leaf specific conductivity, percentage loss in conductivity, and chlorophyll fluorescence for 11 species in February 2014, which was exceptionally dry following protracted drought. Mortality among the 11 dominant species ranged from 0 to 93%. Total stand density was reduced 63.4% and relative dominance of species shifted after the drought. Mortality was negatively correlated with water potential, native xylem specific conductivity, and chlorophyll fluorescence, but not with percent loss in hydraulic conductivity and leaf specific conductivity. The model that best explained mortality included species and plant size as main factors and indicated that larger plants had greater survival for 2 of the species. In general, species with greater resistance to water-stress induced cavitation showed greater mortality levels. Despite adult resprouters typically being more vulnerable to cavitation, results suggest that their more extensive root systems enable them to better access soil moisture and avoid harmful levels of dehydration. These results are consistent with the hypothesis that short-term high intensity droughts have the strongest effect on mature plants of shallow-rooted dehydration tolerant species, whereas deep-rooted dehydration avoiding species fare better in the short-term. Severe droughts can drive changes in chaparral structure as a result of the differential mortality among species.
Article
Full-text available
Just as people with the same weight can have different body builds, woods with the same wood density can have different anatomies. Here, our aim was to assess the magnitude of anatomical variation within a restricted range of wood density and explore its potential ecological implications. Twig wood of 69 angiosperm tree and shrub species was analyzed. Species were selected so that wood density varied within a relatively narrow range (0.38-0.62 g cm-3). Anatomical traits quantified included wood tissue fractions (fibres, axial parenchyma, ray parenchyma, vessels, and conduits with maximum lumen diameter below 15 μm), vessel properties, and pith area. To search for potential ecological correlates of anatomical variation the species were sampled across rainfall and temperature contrasts, and several other ecologically-relevant traits were measured (plant height, leaf area to sapwood area ratio, and modulus of elasticity). Despite the limited range in wood density, substantial anatomical variation was observed. Total parenchyma fraction varied from 0.12 to 0.66 and fibre fraction from 0.20 to 0.74, and these two traits were strongly inversely correlated (r = -0.86, P < 0.001). Parenchyma was weakly (0.24 ≤|r|≤ 0.35, P < 0.05) or not associated with vessel properties nor with height, leaf area to sapwood area ratio, and modulus of elasticity (0.24 ≤|r|≤ 0.41, P < 0.05). However, vessel traits were fairly well correlated with height and leaf area to sapwood area ratio (0.47 ≤|r|≤ 0.65, all P < 0.001). Modulus of elasticity was mainly driven by fibre wall plus vessel wall fraction rather than by the parenchyma component. Overall, there seem to be at least three axes of variation in xylem, substantially independent of each other: a wood density spectrum, a fibre-parenchyma spectrum, and a vessel area spectrum. The fibre-parenchyma spectrum does not yet have any clear or convincing ecological interpretation.
Article
Full-text available
Resistance to xylem cavitation depends on the size of xylem pit membrane pores and the strength of vessels to resist collapse or, in the case of freezing-induced cavitation, conduit diameter. Altering these traits may impact plant biomechanics or water transport efficiency. The evergreen sclerophyllous shrub species, collectively referred to as chaparral, which dominate much of the mediterranean-type climate region of southern California, have been shown to display high cavitation resistance (pressure potential at 50% loss of hydraulic conductivity; P 50). We examined xylem functional and structural traits associated with more negative P 50 in stems of 26 chaparral species. We correlated raw-trait values, without phylogenetic consideration, to examine current relationships between P 50 and these xylem traits. Additionally, correlations were examined using phylogenetic independent contrasts (PICs) to determine whether evolutionary changes in these xylem traits correlate with changes in P 50 . Co-occurring chaparral species widely differ in their P 50 (À0.9 to À11.0 MPa). Species experiencing the most negative seasonal pressure potential (P min) had the highest resistance to xylem cavitation (lowest P 50). Decreased P 50 was associated with increased xylem density, stem mechanical strength (modulus of rupture), and transverse fiber wall area when both raw values and PICs were analyzed. These results support a functional and evolutionary relationship among these xylem traits and cavitation resistance. Chaparral species that do not sprout following fire but instead recruit post-fire from seed had the greatest resistance to cavitation, presumably because they rely on post-fire survival of seedlings during the summer dry period to persist in the landscape. Raw values of hydraulic vessel diameter (d h), maximum vessel length, and xylem-specific hydraulic conductivity (K s) were correlated to P 50 ; however, d h , maximum vessel length, and K s were not correlated to P 50 when analyzed using PICs, suggesting that these traits have not undergone correlated evolutionary change. We found no difference in xylem traits between species occurring at freezing vs. nonfreezing sites, although freezing has been shown to affect the survival and distributions of some chaparral species. Stem mechanical strength, fiber properties, and post-fire regeneration type appear to be key factors in the evolution of cavitation resistance among chaparral shrubs.
Article
Full-text available
Climate change in South Africa may threaten the sclerophyllous evergreen shrubs of this region. Available data suggest that they are not as tolerant of water stress as chaparral shrubs occurring in climatically similar California, USA. Seventeen species from nine angiosperm families, including both fynbos and succulent karoo species, were studied at a field site in Western Cape Province, South Africa. Minimum seasonal pressure potential ( P min ), xylem specific conductivity ( K s ), stem strength against breakage (modulus of rupture, MOR), xylem density, theoretical vessel implosion resistance ( ) and several fibre and vessel anatomical traits were measured. Species displayed great variability in P min , similar to the range reported for chaparral and karoo shrub species, but in contrast to previous reports for fynbos shrubs. More negative P min was associated with having greater xylem density, MOR and . There was no relationship between P min and traits associated with increased water transport efficiency. Xylem density integrates many xylem traits related to water stress tolerance, including P min , MOR and , as well as percentage fibre wall, parenchyma, vessel area and fibre lumen diameter. Xylem density may be an integral trait for predicting the impact of climate change on evergreen shrubs.
Article
Full-text available
Progressive diebacks of outer canopy branchlets of Ceanothus crassifolius were repeatedly observed after rainless periods up to 9 mo in duration in the Santa Monica Mountains of southern California. Mean xylem pressures of branchlets near the end of drought were as low as -11.2 MPa (N = 22) with a mean of about 60 dead branchlets per shrub. Inoculation (N = 15) with three species of fungi previously isolated from the same population of C. crassifolius did not promote dieback, suggesting that the observed decline was not fungal induced, as had been proposed. Further, at least 50% of healthy-appearing twigs, without symptoms of dieback, contained isolatible endophytic fungi. We used a centrifugal force method to determine the range of xylem pressure causing cavitation (vulnerability curves) for branchlets (N = 12) and roots (N = 16). We combined vulnerability curves with soil texture data (N = 6) into a water transport model that estimated the critical values (P(Lcrit)) of leaf xylem pressure associated with the loss of water from soil to foliage. Maximum P(Lcrit) was between -10 and -11 MPa and within the range of minimum measured xylem pressures of branchlets during drought and dieback. Branchlet dieback correlated with seasonal declines in xylem pressure in concert with declining safety margins from hydraulic failure. Symptoms of dieback were duplicated in the field by partially severing stem xylem that normally supplied branchlets with water. Taken together, these results indicate that loss of hydraulic conductance to foliage was the probable cause of the observed dieback in C. crassifolius. Partial dieback of peripheral branchlets, and its attendant reduction in evaporative surface area, may be a last-resort mechanism for whole-plant water conservation and drought survival in this species.
Article
Full-text available
The centrifuge method for measuring the resistance of xylem to cavitation by water stress was modified to also account for any additional cavitation that might occur from a freeze-thaw cycle. A strong correlation was found between cavitation by freezing and mean conduit diameter. On the one extreme, a tracheid-bearing conifer and diffuse-porous angiosperms with small-diameter vessels (mean diameter <30 μm) showed no freezing-induced cavitation under modest water stress (xylem pressure = -0.5 MPa), whereas species with larger diameter vessels (mean >40 μm) were nearly completely cavitated under the same conditions. Species with intermediate mean diameters (30-40 μm) showed partial cavitation by freezing. These results are consistent with a critical diameter of 44 μm at or above which cavitation would occur by a freeze-thaw cycle at -0.5 MPa. As expected, vulnerability to cavitation by freezing was correlated with the hydraulic conductivity per stem transverse area. The results confirm and extend previous reports that small-diameter conduits are relatively resistant to cavitation by freezing. It appears that the centrifuge method, modified to include freeze-thaw cycles, may be useful in separating the interactive effects of xylem pressure and freezing on cavitation.
Article
Full-text available
Possible mechanical and hydraulic costs to increased cavitation resistance were examined among six co-occurring species of chaparral shrubs in southern California. We measured cavitation resistance (xylem pressure at 50% loss of hydraulic conductivity), seasonal low pressure potential (P(min)), xylem conductive efficiency (specific conductivity), mechanical strength of stems (modulus of elasticity and modulus of rupture), and xylem density. At the cellular level, we measured vessel and fiber wall thickness and lumen diameter, transverse fiber wall and total lumen area, and estimated vessel implosion resistance using (t/b)(h)(2), where t is the thickness of adjoining vessel walls and b is the vessel lumen diameter. Increased cavitation resistance was correlated with increased mechanical strength (r(2) = 0.74 and 0.76 for modulus of elasticity and modulus of rupture, respectively), xylem density (r(2) = 0.88), and P(min) (r(2) = 0.96). In contrast, cavitation resistance and P(min) were not correlated with decreased specific conductivity, suggesting no tradeoff between these traits. At the cellular level, increased cavitation resistance was correlated with increased (t/b)(h)(2) (r(2) = 0.95), increased transverse fiber wall area (r(2) = 0.89), and decreased fiber lumen area (r(2) = 0.76). To our knowledge, the correlation between cavitation resistance and fiber wall area has not been shown previously and suggests a mechanical role for fibers in cavitation resistance. Fiber efficacy in prevention of vessel implosion, defined as inward bending or collapse of vessels, is discussed.
Article
Full-text available
Ice formation in the xylem sap produces air bubbles that under negative xylem pressures may expand and cause embolism in the xylem conduits. We used the centrifuge method to evaluate the relationship between freeze-thaw embolism and conduit diameter across a range of xylem pressures (Px) in the conifers Pinus contorta and Juniperus scopulorum. Vulnerability curves showing loss of conductivity (embolism) with Px down to -8 MPa were generated with versus without superimposing a freeze-thaw treatment. In both species, the freeze-thaw plus water-stress treatment caused more embolism than water stress alone. We estimated the critical conduit diameter (Df) above which a tracheid will embolize due to freezing and thawing and found that it decreased from 35 microm at a Px of -0.5 MPa to 6 microm at -8 MPa. Further analysis showed that the proportionality between diameter of the air bubble nucleating the cavitation and the diameter of the conduit (kL) declined with increasingly negative Px. This suggests that the bubbles causing cavitation are smaller in proportion to tracheid diameter in narrow tracheids than in wider ones. A possible reason for this is that the rate of dissolving increases with bubble pressure, which is inversely proportional to bubble diameter (La Place's law). Hence, smaller bubbles shrink faster than bigger ones. Last, we used the empirical relationship between Px and Df to model the freeze-thaw response in conifer species.
Article
Full-text available
Here, hypotheses about stem and root xylem structure and function were assessed by analyzing xylem in nine chaparral Rhamnaceae species. Traits characterizing xylem transport efficiency and safety, mechanical strength and storage were analyzed using linear regression, principal components analysis and phylogenetic independent contrasts (PICs). Stems showed a strong, positive correlation between xylem mechanical strength (xylem density and modulus of rupture) and xylem transport safety (resistance to cavitation and estimated vessel implosion resistance), and this was supported by PICs. Like stems, greater root cavitation resistance was correlated with greater vessel implosion resistance; however, unlike stems, root cavitation resistance was not correlated with xylem density and modulus of rupture. Also different from stems, roots displayed a trade-off between xylem transport safety from cavitation and xylem transport efficiency. Both stems and roots showed a trade-off between xylem transport safety and xylem storage of water and nutrients, respectively. Stems and roots differ in xylem structural and functional relationships, associated with differences in their local environment (air vs soil) and their primary functions.
Article
Premise: The xylem tissue of plants performs three principal functions: transport of water, support of the plant body, and nutrient storage. Tradeoffs may arise because different structural requirements are associated with different functions or because suites of traits are under selection that relate to resource acquisition, use, and turnover. The structural and functional basis of xylem storage is not well established. We hypothesized that greater starch storage would be associated with greater sapwood parenchyma and reduced fibers, which would compromise resistance to xylem tensions during dehydration. Methods: We measured cavitation resistance, minimum water potential, starch content, and sapwood parenchyma and fiber area in 30 species of southern California chaparral shrubs (evergreen and deciduous). Results: We found that species storing greater starch within their xylem tended to avoid dehydration and were less cavitation resistant, and this was supported by phylogenetic independent contrasts. Greater sapwood starch was associated with greater parenchyma area and reduced fiber area. For species without living fibers, the associations with parenchyma were stronger, suggesting that living fibers may expand starch storage capacity while also contributing to the support function of the vascular tissue. Drought-deciduous species were associated with greater dehydration avoidance than evergreens. Conclusions: Evolutionary forces have led to an association between starch storage and dehydration resistance as part of an adaptive suite of traits. We found evidence for a tradeoff between tissue mechanical traits and starch storage; moreover, the evolution of novel strategies, such as starch-storing living fibers, may mitigate the strength of this tradeoff.
Article
We present a method for real-time visualization and automatic processing for detection and classification of untreated cancer cells in blood during stain-free imaging flow cytometry using digital holographic microscopy and machine learning in throughput of 15 cells per second. As a preliminary model for circulating tumor cells in the blood, following a preliminary label-free rapid enrichment stage based on the cell size, we applied our holographic imaging approach, providing the quantitative optical thickness profile of the cells during flow. We automatically classified primary and metastatic colon cancer cells, where the two types of cancer cells were isolated from the same individual, as well as four types of blood cells. We used low-coherence off-axis interferometric phase microscopy and a microfluidic channel to image cells during flow quantitatively. The acquired images were processed and classified based on their morphology and quantitative phase features during the cell flow. We achieved high accuracy of 92.56% for distinguishing between the cells, enabling further automatic enrichment and cancer-cell grading from blood, following preliminary cell purification. This article is protected by copyright. All rights reserved.
Article
Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.
Article
Cohesion-tension transport of water is an energetically efficient way to carry large amounts of water from the roots up to the leaves. However, the cohesion-tension mechanism places the xylem water under negative hydrostatic pressure (P x), rendering it susceptible to cavitation. There are conflicts among the structural requirements for minimizing cavitation on the one hand vs maximizing efficiency of transport and construction on the other. Cavitation by freeze-thaw events is triggered by in situ air bubble formation and is much more likely to occur as conduit diameter increases, creating a direct conflict between conducting efficiency and sensitivity to freezing induced xylem failure. Temperate ring-porous trees and vines with wide diameter conduits tend to have a shorter growing season than conifers and diffuse-porous trees with narrow conduits. Cavitation by water stress occurs by air seeding at interconduit pit membranes. Pit membrane structure is at least partially uncoupled from conduit size, leading to a much less pronounced trade-off between conducting efficiency and cavitation by drought than by freezing. Although wider conduits are generally more susceptible to drought-induced cavitation within an organ, across organs or species this trend is very weak. Different trade-offs become apparent at the level of the pit membranes that interconnect neighbouring conduits. Increasing porosity of pit membranes should enhance conductance but also make conduits more susceptible to air seeding. Increasing the size or number of pit membranes would also enhance conductance, but may weaken the strength of the conduit wall against implosion. The need to avoid conduit collapse under negative pressure creates a significant trade-off between cavitation resistance and xylem construction cost, as revealed by relationships between conduit wall strength, wood density and cavitation pressure. Trade-offs involving cavitation resistance may explain the correlations between wood anatomy, cavitation resistance, and the physiological range of negative pressure experienced by species in their native habitats.
Conference Paper
Recognition of plants has become an active area of research as most of the plant species are at the risk of extinction. This paper uses an efficient machine learning approach for the classification purpose. This proposed approach consists of three phases such as preprocessing, feature extraction and classification. The preprocessing phase involves a typical image processing steps such as transforming to gray scale and boundary enhancement. The feature extraction phase derives the common DMF from five fundamental features. The main contribution of this approach is the Support Vector Machine (SVM) classification for efficient leaf recognition. 12 leaf features which are extracted and orthogonalized into 5 principal variables are given as input vector to the SVM. Classifier tested with flavia dataset and a real dataset and compared with k-NN approach, the proposed approach produces very high accuracy and takes very less execution time.
Article
Wood density (D t), an excellent predictor of mechanical properties, is typically viewed in relation to support against gravity, wind, snow, and other environmental forces. In contrast, we show the surprising extent to which variation in D t and wood structure is linked to support against implosion by negative pressure in the xylem pipeline. The more drought-tolerant the plant, the more negative the xylem pressure can become without cavitation, and the greater the internal load on the xylem conduit walls. Accordingly, D t was correlated with cavitation resistance. This trend was consistent with the maintenance of a safety factor from implosion by negative pressure: conduit wall span (b) and thickness (t) scaled so that (t/b)2 was proportional to cavitation resistance as required to avoid wall collapse. Unexpectedly, trends in D t may be as much or more related to support of the xylem pipeline as to support of the plant.
Starch storage capacity of sapwood is related to dehydration avoidance during drought
  • R B Pratt
  • RB Pratt
Training data-efficient image transformers & distillation through attention
  • H Touvron
  • M Cord
  • M Douze
  • F Massa
  • A Sablayrolles
  • H Jégou
Touvron, H., Cord, M., Douze, M., Massa, F., Sablayrolles, A., Jégou, H.: Training data-efficient image transformers & distillation through attention. In: International Conference on Machine Learning, PMLR, pp. 10347-10357 (2021)
  • R B Pratt
  • A L Jacobsen
Pratt, R.B., Jacobsen, A.L.: Conflicting demands on angiosperm xylem: tradeoffs among storage, transport and biomechanics. Plant, Cell Environ. 40(6), 897-913 (2017)