The NWRD Dataset: An Open-Source Annotated Segmentation Dataset of Diseased Wheat Crop.

Wheat stripe rust disease (WRD) is extremely detrimental to wheat crop health, and it severely affects the crop yield, increasing the risk of food insecurity. Manual inspection by trained personnel is carried out to inspect the disease spread and extent of damage to wheat fields. However, this is quite inefficient, time-consuming, and laborious, owing to the large area of wheat plantations. Artificial intelligence (AI) and deep learning (DL) offer efficient and accurate solutions to such real-world problems. By analyzing large amounts of data, AI algorithms can identify patterns that are difficult for humans to detect, enabling early disease detection and prevention. However, deep learning models are data-driven, and scarcity of data related to specific crop diseases is one major hindrance in developing models. To overcome this limitation, in this work, we introduce an annotated real-world semantic segmentation dataset named the NUST Wheat Rust Disease (NWRD) dataset. Multileaf images from wheat fields under various illumination conditions with complex backgrounds were collected, preprocessed, and manually annotated to construct a segmentation dataset specific to wheat stripe rust disease. Classification of WRD into different types and categories is a task that has been solved in the literature; however, semantic segmentation of wheat crops to identify the specific areas of plants and leaves affected by the disease remains a challenge. For this reason, in this work, we target semantic segmentation of WRD to estimate the extent of disease spread in wheat fields. Sections of fields where the disease is prevalent need to be segmented to ensure that the sick plants are quarantined and remedial actions are taken. This will consequently limit the use of harmful fungicides only on the targeted disease area instead of the majority of wheat fields, promoting environmentally friendly and sustainable farming solutions. Owing to the complexity of the proposed NWRD segmentation dataset, in our experiments, promising results were obtained using the UNet semantic segmentation model and the proposed adaptive patching with feedback (APF) technique, which produced a precision of 0.506, recall of 0.624, and F1 score of 0.557 for the rust class.

Real-Time Energy Efficient Hand Pose Estimation: A Case Study.

The estimation of human hand pose has become the basis for many vital applications where the user depends mainly on the hand pose as a system input. Virtual reality (VR) headset, shadow dexterous hand and in-air signature verification are a few examples of applications that require to track the hand movements in real-time. The state-of-the-art 3D hand pose estimation methods are based on the Convolutional Neural Network (CNN). These methods are implemented on Graphics Processing Units (GPUs) mainly due to their extensive computational requirements. However, GPUs are not suitable for the practical application scenarios, where the low power consumption is crucial. Furthermore, the difficulty of embedding a bulky GPU into a small device prevents the portability of such applications on mobile devices. The goal of this work is to provide an energy efficient solution for an existing depth camera based hand pose estimation algorithm. First, we compress the deep neural network model by applying the dynamic quantization techniques on different layers to achieve maximum compression without compromising accuracy. Afterwards, we design a custom hardware architecture. For our device we selected the FPGA as a target platform because FPGAs provide high energy efficiency and can be integrated in portable devices. Our solution implemented on Xilinx UltraScale+ MPSoC FPGA is 4.2× faster and 577.3× more energy efficient than the original implementation of the hand pose estimation algorithm on NVIDIA GeForce GTX 1070.

Synthesis and functionalization of protease-activated nanoparticles with tissue plasminogen activator peptides as targeting moiety and diagnostic tool for pancreatic cancer.

BACKGROUND: Functionalized nanoparticles (NPs) are one promising tool for detecting specific molecular targets and combine molecular biology and nanotechnology aiming at modern imaging. We aimed at ligand-directed delivery with a suitable target-biomarker to detect early pancreatic ductal adenocarcinoma (PDAC). Promising targets are galectins (Gal), due to their strong expression in and on PDAC-cells and occurrence at early stages in cancer precursor lesions, but not in adjacent normal tissues. RESULTS: Molecular probes (10-29 AA long peptides) derived from human tissue plasminogen activator (t-PA) were selected as binding partners to galectins. Affinity constants between the synthesized t-PA peptides and Gal were determined by microscale thermophoresis. The 29 AA-long t-PA-peptide-1 with a lactose-functionalized serine revealed the strongest binding properties to Gal-1 which was 25-fold higher in comparison with the native t-PA protein and showed additional strong binding to Gal-3 and Gal-4, both also over-expressed in PDAC. t-PA-peptide-1 was selected as vector moiety and linked covalently onto the surface of biodegradable iron oxide nanoparticles (NPs). In particular, CAN-doped maghemite NPs (CAN-Mag), promising as contrast agent for magnetic resonance imaging (MRI), were selected as magnetic core and coated with different biocompatible polymers, such as chitosan (CAN-Mag-Chitosan NPs) or polylactic co glycolic acid (PLGA) obtaining polymeric nanoparticles (CAN-Mag@PNPs), already approved for drug delivery applications. The binding efficacy of t-PA-vectorized NPs determined by exposure to different pancreatic cell lines was up to 90%, as assessed by flow cytometry. The in vivo targeting and imaging efficacy of the vectorized NPs were evaluated by applying murine pancreatic tumor models and assessed by 1.5 T magnetic resonance imaging (MRI). The t-PA-vectorized NPs as well as the protease-activated NPs with outer shell decoration (CAN-Mag@PNPs-PEG-REGAcp-PEG/tPA-pep1Lac) showed clearly detectable drop of subcutaneous and orthotopic tumor staining-intensity indicating a considerable uptake of the injected NPs. Post mortem NP deposition in tumors and organs was confirmed by Fe staining of histopathology tissue sections. CONCLUSIONS: The targeted NPs indicate a fast and enhanced deposition of NPs in the murine tumor models. The CAN-Mag@PNPs-PEG-REGAcp-PEG/tPA-pep1Lac interlocking steps strategy of NPs delivery and deposition in pancreatic tumor is promising.

Three-dimensional polymer coated 45S5-type bioactive glass scaffolds seeded with human mesenchymal stem cells show bone formation in vivo.

45S5-type bioactive glasses are a promising alternative to established substitutes for the treatment of bone defects. Because the three-dimensional (3D) structure of bone substitutes is crucial for bone ingrowth and formation, we evaluated the osteoinductive properties of different polymer coated 3D-45S5 bioactive glass (BG) scaffolds seeded with human mesenchymal stem cells (hMSC) in vivo. BG scaffolds coated with gelatin, cross-linked gelatin, and poly(3-hydroxybutyrate-co-3-hydroxyvalerate) were seeded with hMSC prior to implantation into severe combined immunodeficiency mice. Newly formed bone was evaluated with histomorphometry and micro-computed tomography. Bone formation was detectable in all groups, whereas the gelatin-coated BG scaffolds showed the best results and should be considered in further studies.

Labeling of cancer cells with magnetic nanoparticles for magnetic resonance imaging.

PURPOSE: The process of invasion and metastasis formation of tumor cells can be studied by following the migration of labeled cells over prolonged time periods. This report investigates the applicability of iron oxide nanoparticles as a magnetic resonance imaging (MRI) contrast agent for cell labeling. METHODS: γFe2 O3 nanoparticles prepared with direct flame spray pyrolysis are biofunctionalized with poly-l-lysine (PLL). The nanoparticles within the cells were observed with transmission electron microscopy, bright-field microscopy, and magnetorelaxometry. MRI of labeled cells suspended in agarose was used to estimate the detection limit. RESULTS: PLL-coated particles are readily taken up, stored in intracellular clusters, and gradually degraded by the cells. During cell division, the nanoparticle clusters are divided and split between daughter cells. The MRI detection limit was found to be 25 cells/mm(3) for R2*, and 70 cells/mm(3) for R2. The iron specificity, however, was higher for R2 images. Due to the degradation of intracellular γFe2 O3 to paramagnetic iron ions within 13 days, the R1, R2, and R2* contrast gradually decreased over this time period to approximately 50% of its initial value. CONCLUSIONS: These results suggest that PLL-coated γFe2 O3 nanoparticles can be used as an MRI contrast agent for long-term studies of cell migration. Magn Reson Med 71:1896-1905, 2014. © 2013 Wiley Periodicals, Inc.

Encapsulation of Mesenchymal Stem Cells Improves Vascularization of Alginate-Based Scaffolds.

Vascularization of bioartificial tissues can be significantly enhanced by the generation of an arteriovenous (AV) loop. Besides the surgical vascularization, the choice of the scaffold and the applied cells are indispensable cofactors. The combination of alginate dialdehyde and gelatin (ADA-GEL) and mesenchymal stem cells (MSCs) is a promising approach with regard to biocompatibility, biodegradation, as well as de novo tissue formation. In this study, we targeted the investigation of the vascularization of ADA-GEL with and in the absence of encapsulated MSCs in the AV loop model. A Teflon chamber filled with ADA-GEL microcapsules was placed in the groin of Lewis rats and an AV loop was placed into the chamber. Group A encompassed the ADA-GEL without MSCs, whereas group B contained 2 × 106 DiI-labeled MSCs/mL ADA-GEL. Four weeks postoperatively, tissue formation and vascularization were investigated by histology and microcomputed tomography. We were able to prove vascularization originating from the AV loop in both groups with statistically significant more vessels in group B containing MSCs. Moreover, encapsulated MSCs promoted biodegradation of the ADA-GEL microcapsules. In the present study, we were able to demonstrate for the first time, the successful vascularization of ADA-GEL microcapsules by means of the AV loop. Furthermore, ADA-GEL displayed a good biocompatibility and encapsulation of MSCs into ADA-GEL microcapsule-enhanced vascularization as well as biodegradation.

In-Vivo Imaging of Cell Migration Using Contrast Enhanced MRI and SVM Based Post-Processing.

The migration of cells within a living organism can be observed with magnetic resonance imaging (MRI) in combination with iron oxide nanoparticles as an intracellular contrast agent. This method, however, suffers from low sensitivity and specificty. Here, we developed a quantitative non-invasive in-vivo cell localization method using contrast enhanced multiparametric MRI and support vector machines (SVM) based post-processing. Imaging phantoms consisting of agarose with compartments containing different concentrations of cancer cells labeled with iron oxide nanoparticles were used to train and evaluate the SVM for cell localization. From the magnitude and phase data acquired with a series of T2*-weighted gradient-echo scans at different echo-times, we extracted features that are characteristic for the presence of superparamagnetic nanoparticles, in particular hyper- and hypointensities, relaxation rates, short-range phase perturbations, and perturbation dynamics. High detection quality was achieved by SVM analysis of the multiparametric feature-space. The in-vivo applicability was validated in animal studies. The SVM detected the presence of iron oxide nanoparticles in the imaging phantoms with high specificity and sensitivity with a detection limit of 30 labeled cells per mm3, corresponding to 19 µM of iron oxide. As proof-of-concept, we applied the method to follow the migration of labeled cancer cells injected in rats. The combination of iron oxide labeled cells, multiparametric MRI and a SVM based post processing provides high spatial resolution, specificity, and sensitivity, and is therefore suitable for non-invasive in-vivo cell detection and cell migration studies over prolonged time periods.

Heidelberg-mCT-Analyzer: a novel method for standardized microcomputed-tomography-guided evaluation of scaffold properties in bone and tissue research.

Bone tissue engineering and bone scaffold development represent two challenging fields in orthopaedic research. Micro-computed tomography (mCT) allows non-invasive measurement of these scaffolds' properties in vivo. However, the lack of standardized mCT analysis protocols and, therefore, the protocols' user-dependency make interpretation of the reported results difficult. To overcome these issues in scaffold research, we introduce the Heidelberg-mCT-Analyzer. For evaluation of our technique, we built 10 bone-inducing scaffolds, which underwent mCT acquisition before ectopic implantation (T0) in mice, and at explantation eight weeks thereafter (T1). The scaffolds' three-dimensional reconstructions were automatically segmented using fuzzy clustering with fully automatic level-setting. The scaffold itself and its pores were then evaluated for T0 and T1. Analysing the scaffolds' characteristic parameter set with our quantification method showed bone formation over time. We were able to demonstrate that our algorithm obtained the same results for basic scaffold parameters (e.g. scaffold volume, pore number and pore volume) as other established analysis methods. Furthermore, our algorithm was able to analyse more complex parameters, such as pore size range, tissue mineral density and scaffold surface. Our imaging and post-processing strategy enables standardized and user-independent analysis of scaffold properties, and therefore is able to improve the quantitative evaluations of scaffold-associated bone tissue-engineering projects.

Combination of BMP2 and MSCs significantly increases bone formation in the rat arterio-venous loop model.

INTRODUCTION: In this study the induction of bone formation in an axially vascularized bone matrix using mesenchymal stem cells (MSCs) and application of bone morphogenetic protein 2 (BMP2) was analyzed in the arteriovenous loop (AVL) model. MATERIALS AND METHODS: An AVL was created in the medial thigh of 42 rats and placed in a porous titanium chamber filled with a particulated porous hydroxyapatite and beta-tricalcium phosphate matrix and fibrin. In group A the fibrin was loaded with 5×10(6) DiI-stained fibrin gel-immobilized primary MSCs from syngenic Lewis rats, in group B the matrix was loaded with 60 µg/mL BMP2 and in group C both, BMP2 and MSCs were applied at implantation time point. After 6 and 12 weeks, specimens were investigated by means of histological, morphometrical, and micro-computed tomography analysis. RESULTS: After implantation of an AVL a dense vascular network was visible in all groups. In group A, newly generated bone islands were detected in the periphery of the main vascular axis. Using BMP2 alone (group B), small islands of newly formed bone were visible evenly distributed in all parts of the constructs. In group C nearly the whole matrix was interspersed with bone formations. In all groups there was an increase of bone formation between the 6 and 12 weeks explantation time points. CONCLUSIONS: This study demonstrates for the first time the successful generation of axially vascularized bone substitutes using MSCs and BMP2 in the AVL rat model using a one step procedure. Using the combination of BMP2 and MSCs there was a significant increase of bone formations detectable compared to the BMP2 or MSCs alone groups.

High flow conditions increase connexin43 expression in a rat arteriovenous and angioinductive loop model.

Gap junctions are involved in vascular growth and their expression pattern is modulated in response to hemodynamic conditions. They are clusters of intercellular channels formed by connexins (Cx) of which four subtypes are expressed in the cardiovascular system, namely Cx37, Cx40, Cx43 and Cx45. We hypothesize that high flow conditions affect vascular expression of Cx in vivo. To test this hypothesis, flow hemodynamics and subsequent changes in vascular expression of Cx were studied in an angioinductive rat arteriovenous (AV) loop model. Fifteen days after interposition of a femoral vein graft between femoral artery and vein encased in a fibrin-filled chamber strong neovascularization was evident that emerged predominantly from the graft. Blood flow through the grafted vessel was enhanced ∼4.5-fold accompanied by increased pulsatility exceeding arterial levels. Whereas Cx43 protein expression in the femoral vein is negligible at physiologic flow conditions as judged by immunostaining its expression was enhanced in the endothelium of the venous graft exposed to these hemodynamic changes for 5 days. This was most likely due to enhanced transcription since Cx43 mRNA increased likewise, whereas Cx37 mRNA expression remained unaffected and Cx40 mRNA was reduced. Although enhanced Cx43 expression in regions of high flow in vivo has already been demonstrated, the arteriovenous graft used in the present study provides a reliable model to verify an association between Cx43 expression and high flow conditions in vivo that was selective for this Cx. We conclude that enhancement of blood flow and its oscillation possibly associated with the transition from laminar to more turbulent flow induces Cx43 expression in a vein serving as an AV loop. It is tempting to speculate that this upregulation is involved in the vessel formation occuring in this model as Cx43 was suggested to be involved in angiogenesis.