Research in methodologies for modelling the oral cavity.

The paper aims to explore the current state of understanding surrounding in silico oral modelling. This involves exploring methodologies, technologies and approaches pertaining to the modelling of the whole oral cavity; both internally and externally visible structures that may be relevant or appropriate to oral actions. Such a model could be referred to as a 'complete model' which includes consideration of a full set of facial features (i.e. not only mouth) as well as synergistic stimuli such as audio and facial thermal data. 3D modelling technologies capable of accurately and efficiently capturing a complete representation of the mouth for an individual have broad applications in the study of oral actions, due to their cost-effectiveness and time efficiency. This review delves into the field of clinical phonetics to classify oral actions pertaining to both speech and non-speech movements, identifying how the various vocal organs play a role in the articulatory and masticatory process. Vitaly, it provides a summation of 12 articulatory recording methods, forming a tool to be used by researchers in identifying which method of recording is appropriate for their work. After addressing the cost and resource-intensive limitations of existing methods, a new system of modelling is proposed that leverages external to internal correlation modelling techniques to create a more efficient models of the oral cavity. The vision is that the outcomes will be applicable to a broad spectrum of oral functions related to physiology, health and wellbeing, including speech, oral processing of foods as well as dental health. The applications may span from speech correction, designing foods for the aging population, whilst in the dental field we would be able to gain information about patient's oral actions that would become part of creating a personalised dental treatment plan.

Semantic Segmentation of Spontaneous Intracerebral Hemorrhage, Intraventricular Hemorrhage, and Associated Edema on CT Images Using Deep Learning.

This study evaluated deep learning algorithms for semantic segmentation and quantification of intracerebral hemorrhage (ICH), perihematomal edema (PHE), and intraventricular hemorrhage (IVH) on noncontrast CT scans of patients with spontaneous ICH. Models were assessed on 1732 annotated baseline noncontrast CT scans obtained from the Tranexamic Acid for Hyperacute Primary Intracerebral Haemorrhage (ie, TICH-2) international multicenter trial (ISRCTN93732214), and different loss functions using a three-dimensional no-new-U-Net (nnU-Net) were examined to address class imbalance (30% of participants with IVH in dataset). On the test cohort (n = 174, 10% of dataset), the top-performing models achieved median Dice similarity coefficients of 0.92 (IQR, 0.89-0.94), 0.66 (0.58-0.71), and 1.00 (0.87-1.00), respectively, for ICH, PHE, and IVH segmentation. U-Net-based networks showed comparable, satisfactory performances on ICH and PHE segmentations (P > .05), but all nnU-Net variants achieved higher accuracy than the Brain Lesion Analysis and Segmentation Tool for CT (BLAST-CT) and DeepLabv3+ for all labels (P < .05). The Focal model showed improved performance in IVH segmentation compared with the Tversky, two-dimensional nnU-Net, U-Net, BLAST-CT, and DeepLabv3+ models (P < .05). Focal achieved concordance values of 0.98, 0.88, and 0.99 for ICH, PHE, and ICH volumes, respectively. The mean volumetric differences between the ground truth and prediction were 0.32 mL (95% CI: -8.35, 9.00), 1.14 mL (-9.53, 11.8), and 0.06 mL (-1.71, 1.84), respectively. In conclusion, U-Net-based networks provide accurate segmentation on CT images of spontaneous ICH, and Focal loss can address class imbalance. International Clinical Trials Registry Platform (ICTRP) no. ISRCTN93732214 Supplemental material is available for this article. © RSNA, 2022 Keywords: Head/Neck, Brain/Brain Stem, Hemorrhage, Segmentation, Quantification, Convolutional Neural Network (CNN), Deep Learning Algorithms, Machine Learning Algorithms.

Opsin knockdown specifically slows phototransduction in broadband and UV-sensitive photoreceptors in Periplaneta americana.

Photoreceptors with different spectral sensitivities serve different physiological and behavioral roles. We hypothesized that such functional evolutionary optimization could also include differences in phototransduction dynamics. We recorded elementary responses to light, quantum bumps (QBs), of broadband green-sensitive and ultraviolet (UV)-sensitive photoreceptors in the cockroach, Periplaneta americana, compound eyes using intracellular recordings. In addition to control photoreceptors, we used photoreceptors from cockroaches whose green opsin 1 (GO1) or UV opsin expression was suppressed by RNA interference. In the control broadband and UV-sensitive photoreceptors average input resistances were similar, but the membrane capacitance, a proxy for membrane area, was smaller in the broadband photoreceptors. QBs recorded in the broadband photoreceptors had comparatively short latencies, high amplitudes and short durations. Absolute sensitivities of both opsin knockdown photoreceptors were significantly lower than in wild type, and, unexpectedly, their latency was significantly longer while the amplitudes were not changed. Morphologic examination of GO1 knockdown photoreceptors did not find significant differences in rhabdom size compared to wild type. Our results differ from previous findings in Drosophila melanogaster rhodopsin mutants characterized by progressive rhabdomere degeneration, where QB amplitudes were larger but phototransduction latency was not changed compared to wild type.

Temporal refinement of 3D CNN semantic segmentations on 4D time-series of undersampled tomograms using hidden Markov models.

Recently, several convolutional neural networks have been proposed not only for 2D images, but also for 3D and 4D volume segmentation. Nevertheless, due to the large data size of the latter, acquiring a sufficient amount of training annotations is much more strenuous than in 2D images. For 4D time-series tomograms, this is usually handled by segmenting the constituent tomograms independently through time with 3D convolutional neural networks. Inter-volume information is therefore not utilized, potentially leading to temporal incoherence. In this paper, we attempt to resolve this by proposing two hidden Markov model variants that refine 4D segmentation labels made by 3D convolutional neural networks working on each time point. Our models utilize not only inter-volume information, but also the prediction confidence generated by the 3D segmentation convolutional neural networks themselves. To the best of our knowledge, this is the first attempt to refine 4D segmentations made by 3D convolutional neural networks using hidden Markov models. During experiments we test our models, qualitatively, quantitatively and behaviourally, using prespecified segmentations. We demonstrate in the domain of time series tomograms which are typically undersampled to allow more frequent capture; a particularly challenging problem. Finally, our dataset and code is publicly available.

Domain Adaptation of Synthetic Images for Wheat Head Detection.

Wheat head detection is a core computer vision problem related to plant phenotyping that in recent years has seen increased interest as large-scale datasets have been made available for use in research. In deep learning problems with limited training data, synthetic data have been shown to improve performance by increasing the number of training examples available but have had limited effectiveness due to domain shift. To overcome this, many adversarial approaches such as Generative Adversarial Networks (GANs) have been proposed as a solution by better aligning the distribution of synthetic data to that of real images through domain augmentation. In this paper, we examine the impacts of performing wheat head detection on the global wheat head challenge dataset using synthetic data to supplement the original dataset. Through our experimentation, we demonstrate the challenges of performing domain augmentation where the target domain is large and diverse. We then present a novel approach to improving scores through using heatmap regression as a support network, and clustering to combat high variation of the target domain.

GANana: Unsupervised Domain Adaptation for Volumetric Regression of Fruit.

3D reconstruction of fruit is important as a key component of fruit grading and an important part of many size estimation pipelines. Like many computer vision challenges, the 3D reconstruction task suffers from a lack of readily available training data in most domains, with methods typically depending on large datasets of high-quality image-model pairs. In this paper, we propose an unsupervised domain-adaptation approach to 3D reconstruction where labelled images only exist in our source synthetic domain, and training is supplemented with different unlabelled datasets from the target real domain. We approach the problem of 3D reconstruction using volumetric regression and produce a training set of 25,000 pairs of images and volumes using hand-crafted 3D models of bananas rendered in a 3D modelling environment (Blender). Each image is then enhanced by a GAN to more closely match the domain of photographs of real images by introducing a volumetric consistency loss, improving performance of 3D reconstruction on real images. Our solution harnesses the cost benefits of synthetic data while still maintaining good performance on real world images. We focus this work on the task of 3D banana reconstruction from a single image, representing a common task in plant phenotyping, but this approach is general and may be adapted to any 3D reconstruction task including other plant species and organs.

Forecasting water temperature in lakes and reservoirs using seasonal climate prediction.

Seasonal climate forecasts produce probabilistic predictions of meteorological variables for subsequent months. This provides a potential resource to predict the influence of seasonal climate anomalies on surface water balance in catchments and hydro-thermodynamics in related water bodies (e.g., lakes or reservoirs). Obtaining seasonal forecasts for impact variables (e.g., discharge and water temperature) requires a link between seasonal climate forecasts and impact models simulating hydrology and lake hydrodynamics and thermal regimes. However, this link remains challenging for stakeholders and the water scientific community, mainly due to the probabilistic nature of these predictions. In this paper, we introduce a feasible, robust, and open-source workflow integrating seasonal climate forecasts with hydrologic and lake models to generate seasonal forecasts of discharge and water temperature profiles. The workflow has been designed to be applicable to any catchment and associated lake or reservoir, and is optimized in this study for four catchment-lake systems to help in their proactive management. We assessed the performance of the resulting seasonal forecasts of discharge and water temperature by comparing them with hydrologic and lake (pseudo)observations (reanalysis). Precisely, we analysed the historical performance using a data sample of past forecasts and reanalysis to obtain information about the skill (performance or quality) of the seasonal forecast system to predict particular events. We used the current seasonal climate forecast system (SEAS5) and reanalysis (ERA5) of the European Centre for Medium Range Weather Forecasts (ECMWF). We found that due to the limited predictability at seasonal time-scales over the locations of the four case studies (Europe and South of Australia), seasonal forecasts exhibited none to low performance (skill) for the atmospheric variables considered. Nevertheless, seasonal forecasts for discharge present some skill in all but one case study. Moreover, seasonal forecasts for water temperature had higher performance in natural lakes than in reservoirs, which means human water control is a relevant factor affecting predictability, and the performance increases with water depth in all four case studies. Further investigation into the skillful water temperature predictions should aim to identify the extent to which performance is a consequence of thermal inertia (i.e., lead-in conditions).

Gene transcription changes in a locust model of noise-induced deafness.

Locusts have auditory structures called Müller's organs attached to tympanic membranes on either side of the abdomen. We measured the normalized abundances of 500 different mRNA transcripts in 320 Müller's organs obtained from 160 locusts (Schistocerca gregaria) that had been subjected to a loud continuous 3-kHz tone for 24 h. Abundance ratios were then measured relative to transcripts from 360 control organs. A histogram of the number of observed transcripts versus their abundance ratios (noise exposed/control) was well fitted by a Cauchy distribution with median value near one. Transcripts below 5% and above 95% of the cumulative distribution function of the fitted Cauchy distribution were selected as putatively different from the expected values of an untreated preparation. This yielded eight transcripts with ratios increased by noise exposure (ratios 1.689-3.038) and 18 transcripts with reduced ratios (0.069-0.457). Most of the transcripts with increased abundance represented genes responsible for cuticular construction, suggesting extensive remodeling of some or all the cuticular components of the auditory structure, whereas the reduced abundance transcripts were mostly involved in lipid and protein storage and metabolism, suggesting a profound reduction in metabolic activity in response to the overstimulation.NEW & NOTEWORTHY Locust ears have functional and genetic similarities to human ears, including loss of hearing from age or noise exposure. We measured transcript abundances in transcriptomes of noise-exposed and control locust ears. The data indicate remodeling of the ear tympanum and profound reductions in metabolism that may explain reduced sound transduction. These findings advance our understanding of this useful model and suggest further experiments to elucidate mechanisms that ears use to cope with excessive stimulation.

Mechanotransduction channel Piezo is widely expressed in the spider, Cupiennius salei, mechanosensory neurons and central nervous system.

Mechanosensory neurons use mechanotransduction (MET) ion channels to detect mechanical forces and displacements. Proteins that function as MET channels have appeared multiple times during evolution and occur in at least four different families: the DEG/ENaC and TRP channels, as well as the TMC and Piezo proteins. We found twelve putative members of MET channel families in two spider transcriptomes, but detected only one, the Piezo protein, by in situ hybridization in their mechanosensory neurons. In contrast, probes for orthologs of TRP, ENaC or TMC genes that code MET channels in other species did not produce any signals in these cells. An antibody against C. salei Piezo detected the protein in all parts of their mechanosensory cells and in many neurons of the CNS. Unspecific blockers of MET channels, Ruthenium Red and GsMTx4, had no effect on the mechanically activated currents of the mechanosensory VS-3 neurons, but the latter toxin reduced action potential firing when these cells were stimulated electrically. The Piezo protein is expressed throughout the spider nervous system including the mechanosensory neurons. It is possible that it contributes to mechanosensory transduction in spider mechanosensilla, but it must have other functions in peripheral and central neurons.

Wild deer in the United Kingdom are a potential reservoir for the livestock parasite Babesia divergens.

Redwater fever is an economically important disease of cattle in the United Kingdom caused by the protozoan parasite Babesia divergens. Control efforts are dependent on accurate local historic knowledge of disease occurrence, together with an accurate appreciation of current underlying risk factors. Importantly, the involvement of red deer in the transmission of this pathogen in the UK remains unclear. We employed a polymerase chain reaction approach combined with DNA sequencing to investigate Babesia infections in livestock and red deer at a UK farm with a history of tick-borne disease. This revealed several B. divergens-infected cattle that were not displaying overt clinical signs. Additionally, 11% of red deer on the farmland and surrounding areas were infected with this parasite. We also found that 16% of the red deer were infected with Babesia odocoilei, the first time this parasite has been detected in the UK. The finding of B. divergens in the red deer population updates our knowledge of epidemiology in the UK and has implications for the effective control of redwater fever.

Concentration and origin of lead (Pb) in liver and bone of Eurasian buzzards (Buteo buteo) in the United Kingdom.

Ingestion of lead (Pb) derived from ammunition used in the hunting of game animals is recognised to be a significant potential source of Pb exposure of wild birds, including birds of prey. However, there are only limited data for birds of prey in Europe regarding tissue concentrations and origins of Pb. Eurasian buzzards (Buteo buteo) found dead in the United Kingdom during an 11-year period were collected and the concentrations of Pb in the liver and femur were measured. Concentrations in the liver consistent with acute exposure to Pb were found in 2.7% of birds and concentration in the femur consistent with exposure to lethal levels were found in 4.0% of individuals. Pb concentration in the femur showed no evidence of consistent variation among or within years, but was greater for old than for young birds. The Pb concentration in the liver showed no effect of the birds' age, but varied markedly among years and showed a consistent tendency to increase substantially within years throughout the UK hunting season for gamebirds. The resemblance of the stable isotope composition of Pb from buzzard livers to that of Pb from the types of shotgun ammunition most widely-used in the UK increased markedly with increasing Pb concentration in the liver. Stable isotope results were consistent with 57% of the mass of Pb in livers of all of the buzzards sampled being derived from shotgun pellets, with this proportion being 89% for the birds with concentrations indicating acute exposure to Pb. Hence, most of the Pb acquired by Eurasian buzzards which have liver concentrations likely to be associated with lethal and sublethal effects is probably obtained when they prey upon or scavenge gamebirds and mammals shot using Pb shotgun pellets.

Improving Atraumatic Chest Pain Evaluation in an Urban, Safety-net Hospital Through Incorporation of a Modified HEART Score.

Atraumatic chest pain is a common emergency department (ED) presentation and the American College of Cardiology and American Heart Association recommends stress testing within 72 hours. The HEART score predicts major adverse cardiac events (MACE) in ED populations and does not require universal stress testing. An evaluation based solely on history, electrocardiography, and biomarkers, therefore, is an attractive approach to risk stratification in resource-limited settings. The HEART score has not been previously evaluated in a safety net hospital setting. We therefore implemented an interdisciplinary clinical care guideline utilizing the HEART score to stratify patients presenting to our inner-city hospital. During a 6-month study period, 1170 patients were evaluated (521 before and 649 after implementation). Among the 998 patients with confirmed follow-up 6-weeks after the index ED encounter, the prevalence of MACE (all-cause mortality, acute myocardial infarction, or coronary revascularization) was 0% [95% confidence interval (CI), 0%-1%] for low, 9% (95% CI, 7%-12%) for moderate, and 52% (95% CI, 39%-65%) for high-risk groups. Guideline implementation significantly increased admissions (+12%, 95% CI, 7%-17%) primarily in the moderate risk group (+38%, 95% CI, 29%-47%), but significantly decreased median ED length of stay (-37 minutes, 95% CI, 17-58). It also led to an increase in stress testing among moderate and high-risk patients (+10%, 95% CI, 0%-19%). In conclusion, the HEART score effectively stratified risk of MACE in a safety net population, improved evaluation consistency, and decreased ED length of stay. However, implementation was associated with an increase in hospitalizations and stress testing. Although the American Heart Association/American College of Cardiology guideline regarding atraumatic chest pain in the ED recommends universal noninvasive testing, the value of this approach, particularly in conjunction with the HEART score is uncertain in safety net hospitals. Further evaluation of the costs and clinical advantages of this approach are warranted.

Volumetric Segmentation of Cell Cycle Markers in Confocal Images Using Machine Learning and Deep Learning.

Understanding plant growth processes is important for many aspects of biology and food security. Automating the observations of plant development-a process referred to as plant phenotyping-is increasingly important in the plant sciences, and is often a bottleneck. Automated tools are required to analyze the data in microscopy images depicting plant growth, either locating or counting regions of cellular features in images. In this paper, we present to the plant community an introduction to and exploration of two machine learning approaches to address the problem of marker localization in confocal microscopy. First, a comparative study is conducted on the classification accuracy of common conventional machine learning algorithms, as a means to highlight challenges with these methods. Second, a 3D (volumetric) deep learning approach is developed and presented, including consideration of appropriate loss functions and training data. A qualitative and quantitative analysis of all the results produced is performed. Evaluation of all approaches is performed on an unseen time-series sequence comprising several individual 3D volumes, capturing plant growth. The comparative analysis shows that the deep learning approach produces more accurate and robust results than traditional machine learning. To accompany the paper, we are releasing the 4D point annotation tool used to generate the annotations, in the form of a plugin for the popular ImageJ (FIJI) software. Network models and example datasets will also be available online.

Suppression of Gq and PLC gene expression has a small effect on quantum bumps in vivo in Periplaneta americana.

Visual signal transmission by Drosophila melanogaster photoreceptors is mediated by a Gq protein that activates a phospholipase C (PLC). Mutations and deficiencies in expression of either of these proteins cause severe defects in phototransduction. Here we investigated whether these proteins are also involved in the cockroach, Periplaneta americana, phototransduction by silencing Gq α-subunit (Gqα) and phosphoinositide-specific phospholipase C (PLC) by RNA interference and observing responses to single photons (quantum bumps, QB). We found (1) non-specific decreases in membrane resistance, membrane capacitance and absolute sensitivity in the photoreceptors of both Gqα and PLC knockdowns, and (2) small changes in QB statistics. Despite significant decreases in expressions of Gq and PLC mRNA, the changes in QB properties were surprisingly modest, with mean latencies increasing by ~ 10%, and without significant decrease in their amplitudes. To better understand our results, we used a mathematical model of the phototransduction cascade. By modifying the Gq and PLC abundances, and diffusion rates for Gq, we found that QB latencies and amplitudes deteriorated noticeably only after large decreases in the protein levels, especially when Gq diffusion was slow. Also, reduction in Gq but not PLC lowered quantum efficiency. These results suggest that expression of the proteins may be redundant.

Physiological Basis of Noise-Induced Hearing Loss in a Tympanal Ear.

Acoustic overexposure, such as listening to loud music too often, results in noise-induced hearing loss. The pathologies of this prevalent sensory disorder begin within the ear at synapses of the primary auditory receptors, their postsynaptic partners and their supporting cells. The extent of noise-induced damage, however, is determined by overstimulation of primary auditory receptors, upstream of where the pathologies manifest. A systematic characterization of the electrophysiological function of the upstream primary auditory receptors is warranted to understand how noise exposure impacts on downstream targets, where the pathologies of hearing loss begin. Here, we used the experimentally-accessible locust ear (male, Schistocerca gregaria) to characterize a decrease in the auditory receptor's ability to respond to sound after noise exposure. Surprisingly, after noise exposure, the electrophysiological properties of the auditory receptors remain unchanged, despite a decrease in the ability to transduce sound. This auditory deficit stems from changes in a specialized receptor lymph that bathes the auditory receptors, revealing striking parallels with the mammalian auditory system.SIGNIFICANCE STATEMENT Noise exposure is the largest preventable cause of hearing loss. It is the auditory receptors that bear the initial brunt of excessive acoustic stimulation, because they must convert excessive sound-induced movements into electrical signals, but remain functional afterward. Here we use the accessible ear of an invertebrate to, for the first time in any animal, characterize changes in auditory receptors after noise overexposure. We find that their decreased ability to transduce sound into electrical signals is, most probably, due to changes in supporting (scolopale) cells that maintain the ionic composition of the ear. An emerging doctrine in hearing research is that vertebrate primary auditory receptors are surprisingly robust, something that we show rings true for invertebrate ears too.

Deep convolutional neural networks for image-based Convolvulus sepium detection in sugar beet fields.

BACKGROUND: Convolvulus sepium (hedge bindweed) detection in sugar beet fields remains a challenging problem due to variation in appearance of plants, illumination changes, foliage occlusions, and different growth stages under field conditions. Current approaches for weed and crop recognition, segmentation and detection rely predominantly on conventional machine-learning techniques that require a large set of hand-crafted features for modelling. These might fail to generalize over different fields and environments. RESULTS: Here, we present an approach that develops a deep convolutional neural network (CNN) based on the tiny YOLOv3 architecture for C. sepium and sugar beet detection. We generated 2271 synthetic images, before combining these images with 452 field images to train the developed model. YOLO anchor box sizes were calculated from the training dataset using a k-means clustering approach. The resulting model was tested on 100 field images, showing that the combination of synthetic and original field images to train the developed model could improve the mean average precision (mAP) metric from 0.751 to 0.829 compared to using collected field images alone. We also compared the performance of the developed model with the YOLOv3 and Tiny YOLO models. The developed model achieved a better trade-off between accuracy and speed. Specifically, the average precisions (APs@IoU0.5) of C. sepium and sugar beet were 0.761 and 0.897 respectively with 6.48 ms inference time per image (800 × 1200) on a NVIDIA Titan X GPU environment. CONCLUSION: The developed model has the potential to be deployed on an embedded mobile platform like the Jetson TX for online weed detection and management due to its high-speed inference. It is recommendable to use synthetic images and empirical field images together in training stage to improve the performance of models.

RNA interference supports a role for Nanchung-Inactive in mechanotransduction by the cockroach, Periplaneta americana, tactile spine.

Proteins encoded by nanchung, inactive, nompC and piezo genes have been shown to play crucial roles in the initial detection of mechanical force by various insect auditory neurons, nociceptors and touch receptors. Most of this previous research has been performed on the larval and adult fruit fly, Drosophila melanogaster. We identified and assembled all four homologous genes in transcriptomes from the cockroach, Periplaneta americana. Injection of long double-stranded RNA (dsRNA) into the adult cockroach abdomen successfully reduced the expression of each gene, as measured by quantitative PCR (RT-qPCR). A simple electrophysiological assay was used to record action potential firing in afferent nerves of cockroach femoral tactile spines in response to a standardized mechanical step displacement. Responses of nanchung knockdown animals were significantly reduced compared to matched sham-injected animals at 14 and 21 days after injection, and inactive knockdowns similarly at 21 days. In contrast, responses of nompC and piezo knockdowns were unchanged. Our results support a model in which Nanchung and Inactive proteins combine to form a part of the mechanotransduction mechanism in the cockroach tactile spine.

Towards infield, live plant phenotyping using a reduced-parameter CNN.

There is an increase in consumption of agricultural produce as a result of the rapidly growing human population, particularly in developing nations. This has triggered high-quality plant phenotyping research to help with the breeding of high-yielding plants that can adapt to our continuously changing climate. Novel, low-cost, fully automated plant phenotyping systems, capable of infield deployment, are required to help identify quantitative plant phenotypes. The identification of quantitative plant phenotypes is a key challenge which relies heavily on the precise segmentation of plant images. Recently, the plant phenotyping community has started to use very deep convolutional neural networks (CNNs) to help tackle this fundamental problem. However, these very deep CNNs rely on some millions of model parameters and generate very large weight matrices, thus making them difficult to deploy infield on low-cost, resource-limited devices. We explore how to compress existing very deep CNNs for plant image segmentation, thus making them easily deployable infield and on mobile devices. In particular, we focus on applying these models to the pixel-wise segmentation of plants into multiple classes including background, a challenging problem in the plant phenotyping community. We combined two approaches (separable convolutions and SVD) to reduce model parameter numbers and weight matrices of these very deep CNN-based models. Using our combined method (separable convolution and SVD) reduced the weight matrix by up to 95% without affecting pixel-wise accuracy. These methods have been evaluated on two public plant datasets and one non-plant dataset to illustrate generality. We have successfully tested our models on a mobile device.

Active Vision and Surface Reconstruction for 3D Plant Shoot Modelling.

Plant phenotyping is the quantitative description of a plant's physiological, biochemical, and anatomical status which can be used in trait selection and helps to provide mechanisms to link underlying genetics with yield. Here, an active vision- based pipeline is presented which aims to contribute to reducing the bottleneck associated with phenotyping of architectural traits. The pipeline provides a fully automated response to photometric data acquisition and the recovery of three-dimensional (3D) models of plants without the dependency of botanical expertise, whilst ensuring a non-intrusive and non-destructive approach. Access to complete and accurate 3D models of plants supports computation of a wide variety of structural measurements. An Active Vision Cell (AVC) consisting of a camera-mounted robot arm plus combined software interface and a novel surface reconstruction algorithm is proposed. This pipeline provides a robust, flexible, and accurate method for automating the 3D reconstruction of plants. The reconstruction algorithm can reduce noise and provides a promising and extendable framework for high throughput phenotyping, improving current state-of-the-art methods. Furthermore, the pipeline can be applied to any plant species or form due to the application of an active vision framework combined with the automatic selection of key parameters for surface reconstruction.

Deployment of Lidar from a Ground Platform: Customizing a Low-Cost, Information-Rich and User-Friendly Application for Field Phenomics Research.

Using sensors and electronic systems for characterization of plant traits provides valuable digital inputs to support complex analytical modeling in genetics research. In field applications, frequent sensor deployment enables the study of the dynamics of these traits and their interaction with the environment. This study focused on implementing lidar (light detection and ranging) technology to generate 2D displacement data at high spatial resolution and extract plant architectural parameters, namely canopy height and cover, in a diverse population of 252 maize (Zea mays L.) genotypes. A prime objective was to develop the mechanical and electrical subcomponents for field deployment from a ground vehicle. Data reduction approaches were implemented for efficient same-day post-processing to generate by-plot statistics. The lidar system was successfully deployed six times in a span of 42 days. Lidar data accuracy was validated through independent measurements in a subset of 75 experimental units. Manual and lidar-derived canopy height measurements were compared resulting in root mean square error (RMSE) = 0.068 m and r2 = 0.81. Subsequent genome-wide association study (GWAS) analyses for quantitative trait locus (QTL) identification and comparisons of genetic correlations and heritabilities for manual and lidar-based traits showed statistically significant associations. Low-cost, field-ready lidar of computational simplicity make possible timely phenotyping of diverse populations in multiple environments.