Literatura académica sobre el tema "Fine-structure Preservation"

Successful immunolabeling in electron microscopy of animal and plant tissues requires a combination of excellent antigen preservation while maintaining the original structure of the tissue. One important element is tissue embedding which accomplishes two goals for the immunohistochemist, the preservation of tissue specimen structure and maintenance of biological antigenicity. Tissue embedding in plastic resins is a common method in which several important elements must be considered.1.Fine tissue structure must not be damaged by the polymerization.2.The plastic must be stable to the electron beam.3.Light scattering properties of the plastic should be minimal.4.The plastic should cut easily.5.The plastic must be of sufficiently low viscosity to infiltrate the tissue.

A problem in immunocytochemistry is obtaining acceptable fixation of tissue while retaining antigenicity. Two concentrations (1% and 2.5%) of glutaraldehyde, with and without secondary fixation in 1% Osmium Tetroxide (OsO4) and varying fixation times were used. Fixation in 1% glutanddehyde for 3 h was adequate to preserve the tissue. Some loss of fine structure was visible under an electron microscope. A solution of 2.5% glutaraldehyde was more effective in preserving fine structure. At 2 h fixation the tissue was well preserved and only slight loss of tine detail was observed. A longer fixation results in better ultrastructural preservation, but can cause loss of antigenicity. OsO4 fixes lipids and acts as an electron dense stain. OsO4 has a negative effect on antigenicity. The use of OsO4 had little effect upon the preservation of ultrastructural detail and-did not improve staining; therefore, it was omitted in later fixations. Based on this experimental evidence, initial localization experiments will utilize tissue fixed in 1% glutaraldehyde for 3 h.

the article discusses the possibilities of improving the strength characteristics of fine-grained concrete. Modification of compositions and production technology of fine-grained high-strength concrete is possible with the use of natural and man-made raw materials of various chemical and mineral composition. It is shown that it is possible to increase the economic feasibility of high-strength fine-grained concretes with the preservation of performance characteristics due to the use of man-made raw materials and production waste. The issues of controlling the processes of structure formation and identifying a potentially stable state of hardening compositions are considered, possibly on the basis of modification and design methods for the composition of construction objects with improved properties.

We are working to improve methods for the study of cellular fine structure. Our approach is to advance each of the key steps in the preparation of specimens for EM: high quality fixation that will preserve both structure and antigenicity; methods for specific labeling; efficient acquisition of 3-D electron microscopic data; and software for 3-D reconstruction and display.Our work on high quality structure preservation has focused on methods for fast freezing and freeze substitution. Both plunge freezing of specimens grown on coated gold grids and high pressure freezing of either cultured cells or tissue specimens have yielded well preserved material. These samples are suitable for freeze substitution fixation with either anhydrous aldehydes in acetone at -90°C, for the preservation of antigens, or aldehydes, tannic acid, OsO4, and uranyl acetate for optimal preservation the structure.We have used a JEOL JEM-1,000 high voltage microscope to image sections about 250nm thick, employing a goniometer stage to perform dual axis tomography for 3-D reconstruction with approximately isotropic resolution at ∼7nm.

Somatic embryogenesis is one of the newest technology that applied for the mass production of cocoa. This research aims to evaluate the regeneration rate of somatic embryos through somatic embryogenesis propagation techniques on java fine flavor cocoa. Cultivars in this study are ICCRI 01, ICCRI 02, DR 1, DR 2, DRC 16, DR 38, PNT 16, and PNT 30. Observations include parameters to determine the percentage of primary callus and embryogenic callus formation and the number of somatic embryos produced. Based on data, the ability of callus to produce primary embryos is highly dependent on plant cultivars and explant sources. Five cultivars showed a higher regeneration rate using explants from the petal part, while the rest showed a higher regeneration rate using explants from the staminode section. Embryogenic callus from each cacao cultivar has the same basic structure: a nodular friable structure consisting of many embryonic cells. Some fine flavor cacao cultivars that were able to produce callus and primary somatic embryos could not produce secondary somatic embryos and plantlets. However, two cultivars, which had low potential in producing primary embryos, had the high ability to produce secondary somatic embryos and develop into plantlets.

We present a novel nonlocal mean (NLM) algorithm using an anisotropic structure tensor to achieve higher accuracy of imaging denoising and better preservation of fine image details. Instead of using the intensity to identify the pixel, the proposed algorithm uses the structure tensor to characterize the boundary information around the pixel more comprehensively. Meanwhile, similarity of the structure tensor is computed in a Riemannian space for more rigorous comparison, and the similarity weight of the pixel (or patch) is determined by the intensity and structure tensor simultaneously. The proposed algorithm is compared with the original NLM algorithm and a modified NLM algorithm that is based on the principle component analysis. Quantitative and qualitative comparisons of the three NLM algorithms are presented as well.

L’imagerie thoracique présente des défis importants, chaque modalité d’imagerie ayant ses propres limitations. Le scanner (CT), référence en imagerie pulmonaire, offre une haute résolution spatiale, mais utilise des rayonnements ionisants, posant des risques pour les patients nécessitant des examens fréquents. À l’inverse, l’IRM pulmonaire offre une alternative sans radiation, mais est limitée par des problèmes techniques tels qu’un faible contraste et des artefacts, limitant son adoption clinique à grande échelle. Récemment, l’IRM à temps d’écho ultracourt (UTE-MRI) a montré un potentiel pour surmonter certaines de ces limitations, mais elle ne parvient toujours pas à atteindre la haute résolution et la qualité d’image du scanner, notamment pour l’évaluation détaillée des structures. L’objectif principal de cette thèse est de développer et valider des modèles basés sur l’apprentissage profond pour synthétiser des images de scanner TDM à partir d’IRM UTE. Plus précisément, nous visons à évaluer la qualité des images, la reconstruction anatomique et l’applicabilité clinique de ces images de synthèse par rapport aux IRM UTE originales et aux scanners CT correspondants. Dans un premier temps, nous avons exploré les bases de la synthèse d’images médicales, en établissant le socle pour la conversion de l’IRM vers le scanner. Nous avons mis en œuvre un modèle de GAN 2D basé sur le cadre pix2pixHD, optimisé avec la normalisation SPADE et affiné les techniques de prétraitement telles que le rééchantillonnage et l’enregistrement. L’évaluation clinique par des radiologues experts a montré des résultats prometteurs en comparant les images synthétiques aux scans réels. La synthèse a été ensuite améliorée par l’introduction de la perte perceptuelle, qui a affiné les détails structurels et la qualité visuelle, et par l’intégration de stratégies 2.5D pour équilibrer la synthèse 2D et 3D. De plus, nous avons mis l’accent sur un processus de validation rigoureux utilisant des métriques spécifiques à la tâche, remettant en question les métriques globales traditionnelles basées sur l’intensité, en nous concentrant sur la reconstruction précise des structures anatomiques. Dans la dernière étape, nous avons développé un cadre de synthèse 3D robuste et évolutif en adaptant nnU-Net pour la génération de scanner, accompagné d’une fonction de perte priorisant les caractéristiques anatomiques, permettant une meilleure reconstruction des structures critiques telles que les voies respiratoires et les vaisseaux. Notre travail met en évidence le potentiel des modèles basés sur l’apprentissage profond pour générer des images synthétiques de haute qualité de type scanner à partir d’IRM UTE, offrant une amélioration significative de l’imagerie pulmonaire non invasive. Ces avancées pourraient considérablement améliorer l’applicabilité clinique de l’IRM UTE, en offrant une alternative plus sûre au scanner pour le suivi des maladies pulmonaires chroniques. De plus, un brevet est actuellement en préparation pour l’adoption de notre méthode, ouvrant la voie à une utilisation potentielle en milieu clinique
Thoracic imaging faces significant challenges, with each imaging modality presenting its own limitations. CT, the gold standard for lung imaging, delivers high spatial resolution but relies on ionizing radiation, posing risks for patients requiring frequent scans. Conversely, lung MRI, offers a radiation-free alternative but is hindered by technical issues such as low contrast and artifacts, limiting its broader clinical use. Recently, UTE-MRI shows promise in addressing some of these limitations, but still lacks the high resolution and image quality of CT, particularly for detailed structural assessment. The primary objective of this thesis is to develop and validate deep learning-based models for synthesizing CT-like images from UTE-MRI. Specifically, we aim to assess the image quality, anatomical accuracy, and clinical applicability of these synthetic CT images in comparison to the original UTE-MRI and real CT scans in thoracic imaging. Initially, we explored the fundamentals of medical image synthesis, establishing the groundwork for MR to CT translation. We implemented a 2D GAN model based on the pix2pixHD framework, optimizing it using SPADE normalization and refining preprocessing techniques such as resampling and registration. Clinical evaluation with expert radiologists showed promising results in comparing synthetic images to real CT scans. Synthesis was further enhanced by introducing perceptual loss, which improved structural details and visual quality, and incorporated 2.5D strategies to balance between 2D and 3D synthesis. Additionally, we emphasized a rigorous validation process using task-specific metrics, challenging traditional intensity-based and global metrics by focusing on the accurate reconstruction of anatomical structures. In the final stage, we developed a robust and scalable 3D synthesis framework by adapting nnU-Net for CT generation, along with an anatomical feature-prioritized loss function, enabling superior reconstruction of critical structures such as airways and vessels. Our work highlights the potential of deep learning-based models for generating high-quality synthetic CT images from UTE-MRI, offering a significant improvement in non-invasive lung imaging. These advances could greatly enhance the clinical applicability of UTE-MRI, providing a safer alternative to CT for the follow-up of chronic lung diseases. Furthermore, a patent is currently in preparation for the adoption of our method, paving the way for potential clinical use

Abstract Although platelets were one of the first types of cells to be studied using the electron microscope, a clear understanding of normal platelet architecture using electron microscopy had to await improved techniques for the isolation, preservation, and fixation of these cells. Similarly, an understanding of the changes in blood platelet fine structure during clotting required improvements in preparative techniques. The preservation and fixation of certain platelet subcellular structures, notably dense granules and contractile filaments, have presented particular challenges. In this chapter, we describe basic fixation procedures for platelet transmission electron microscopy, and for immunofluorescent and immunoelectron microscopic studies, while providing a brief overview of some other aspects of platelet electron microscopic studies.

Recently, the pretrain-finetuning paradigm has attracted tons of attention in graph learning community due to its power of alleviating the lack of labels problem in many real-world applications. Current studies use existing techniques, such as weight constraint, representation constraint, which are derived from images or text data, to transfer the invariant knowledge from the pre-train stage to fine-tuning stage. However, these methods failed to preserve invariances from graph structure and Graph Neural Network (GNN) style models. In this paper, we present a novel optimal transport-based fine-tuning framework called GTOT-Tuning, namely, Graph Topology induced Optimal Transport fine-Tuning, for GNN style backbones. GTOT-Tuning is required to utilize the property of graph data to enhance the preservation of representation produced by fine-tuned networks. Toward this goal, we formulate graph local knowledge transfer as an Optimal Transport (OT) problem with a structural prior and construct the GTOT regularizer to constrain the fine-tuned model behaviors. By using the adjacency relationship amongst nodes, the GTOT regularizer achieves node-level optimal transport procedures and reduces redundant transport procedures, resulting in efficient knowledge transfer from the pre-trained models. We evaluate GTOT-Tuning on eight downstream tasks with various GNN backbones and demonstrate that it achieves state-of-the-art fine-tuning performance for GNNs.

Relation Extraction (RE) identifies relations between entities in text, typically relying on supervised models that demand abundant high-quality data. Various approaches, including Data Augmentation (DA), have been proposed as promising solutions for addressing low-resource challenges in RE. However, existing DA methods in RE often struggle to ensure consistency and contextual diversity in generated data due to the fine-grained nature of RE. Inspired by the extensive generative capabilities of large language models (LLMs), we introduce a novel framework named ConsistRE, aiming to maintain context consistency in RE. ConsistRE initiates by collecting a substantial corpus from external resources and employing statistical algorithms and semantics to identify keyword hints closely related to relation instances. These keyword hints are subsequently integrated as contextual constraints in sentence generation, ensuring the preservation of relation dependence and diversity with LLMs. Additionally, we implement syntactic dependency selection to enhance the syntactic structure of the generated sentences. Experimental results from the evaluation of SemEval, TACRED, and TACREV datasets unequivocally demonstrate that ConsistRE outperforms other baselines in F1 values by 1.76%, 3.92%, and 2.53%, respectively, particularly when operating under low-resource experimental conditions.