Daegu Gyeongbuk Institute of Science and Technology
DGIST Library Institutional RepositoryNot a member yet
12664 research outputs found
Sort by
Federated learning with knowledge distillation for multi-organ segmentation with partially labeled datasets
No full textThe state-of-the-art multi-organ CT segmentation relies on deep learning models, which only generalize when trained on large samples of carefully curated data. However, it is challenging to train a single model that can segment all organs and types of tumors since most large datasets are partially labeled or are acquired across multiple institutes that may differ in their acquisitions. A possible solution is Federated learning, which is often used to train models on multi-institutional datasets where the data is not shared across sites. However, predictions of federated learning can be unreliable after the model is locally updated at sites due to ‘catastrophic forgetting’. Here, we address this issue by using knowledge distillation (KD) so that the local training is regularized with the knowledge of a global model and pre-trained organ-specific segmentation models. We implement the models in a multi-head U-Net architecture that learns a shared embedding space for different organ segmentation, thereby obtaining multi-organ predictions without repeated processes. We evaluate the proposed method using 8 publicly available abdominal CT datasets of 7 different organs. Of those datasets, 889 CTs were used for training, 233 for internal testing, and 30 volumes for external testing. Experimental results verified that our proposed method substantially outperforms other state-of-the-art methods in terms of accuracy, inference time, and the number of parameters. © 2024 Elsevier B.V.FALSEsciescopu
Progress and Perspective in harnessing MXene–carbon-based composites (0–3D): Synthesis, performance, and applications
No full textMXene is recognized as a promising catalyst for versatile applications due to its abundant metal sites, physicochemical properties, and structural formation. This comprehensive review offers an in-depth analysis of the incorporation of carbon into MXene, resulting in the formation of MXene–carbon-based composites (MCCs). Pristine MXene exhibits numerous outstanding characteristics, such as its atomically thin 2D structure, hydrophilic surface nature, metallic electrical conductivity, and substantial specific surface area. The introduction of carbon guides the assembly of MCCs through electrostatic self-assembly, pairing positively charged carbon with negatively charged MXene. These interactions result in increased interlayer spacing, reduced ion/electron transport distances, and enhanced surface hydrophilicity. Subsequent sections delve into the synthesis methods for MCCs, focusing on MXene integrated with various carbon structures, including 0D, 1D, 2D, and 3D carbon. Comprehensive discussions explore the distinctive properties of MCCs and the unique advantages they offer in each application domain, emphasizing the contributions and advancements they bring to specific fields. Furthermore, this comprehensive review addresses the challenges encountered by MCCs across different applications. Through these analyses, the review promotes a deeper understanding of exceptional characteristics and potential applications of MCCs. Insights derived from this review can serve as guidance for future research and development efforts, promoting the widespread utilization of MCCs across a broad spectrum of disciplines and spurring future innovations. © 2024 Elsevier LtdFALSEsciescopu
PHARMACEUTICAL COMPOSITION FOR PREVENTING OR TREATING OBESITY CONTAINING BIGLYCAN AS ACTIVE INGREDIENT
No full textRepulsive Sema3E-Plexin-D1 signaling coordinates both axonal extension and steering via activating an autoregulatory factor, Mtss1
Get PDFAxon guidance molecules are critical for neuronal pathfinding because they regulate directionality and growth pace during nervous system development. However, the molecular mechanisms coordinating proper axonal extension and turning are poorly understood. Here, metastasis suppressor 1 (Mtss1), a membrane protrusion protein, ensured axonal extension while sensitizing axons to the Semaphorin 3E (Sema3E)-Plexin-D1 repulsive cue. Sema3E-Plexin-D1 signaling enhanced Mtss1 expression in projecting striatonigral neurons. Mtss1 localized to the neurite axonal side and regulated neurite outgrowth in cultured neurons. Mtss1 also aided Plexin-D1 trafficking to the growth cone, where it signaled a repulsive cue to Sema3E. Mtss1 ablation reduced neurite extension and growth cone collapse in cultured neurons. Mtss1-knockout mice exhibited fewer striatonigral projections and irregular axonal routes, and these defects were recapitulated in Plxnd1-or Sema3e-knockout mice. These findings demonstrate that repulsive axon guidance activates an exquisite autoregulatory program coordinating both axonal extension and steering during neuronal pathfinding. © Kim, Li et al.TRUEsciescopu
METHOD AND APPARATUS FOR ESTIMATING DISTURBANCE OF CONTROL SYSTEM BASED ON INPUT/OUTPUT DATA
No full textAchieving high durability in all-solid-state lithium metal batteries using metal-organic framework solid polymer electrolytes
No full textSolid-state polymer electrolytes (SPEs) possess several favorable properties, such as high flexibility, easy processability, and better safety for batteries. Thus, SPEs are attracting considerable attention for the development of safer Li ion batteries. However, SPEs typically lack the required ionic conductivity for improved Li+ transportation. Therefore, suitable fillers are often used to prepare composite polymer electrolytes (CPEs) with improved ionic conductivity. In this direction, a novel CPE with increased electrochemical stability and durability was prepared by incorporating highly porous Zr-based metal-organic frameworks (MOFs), herein referred to as ZR8, as fillers in poly(ethylene oxide) (PEO)/Li salt systems. PEO in the prepared CPE showed low crystallization ratios, and the CPE showed enhanced Li+ ion transportation, thereby improving mechanical and fire-retardant properties. It was also able to hinder Li dendrite formation and eventually facilitated efficient transport of Li+ ions through the electrolyte. A high specific surface area, the presence of Lewis acid-base sites, and good thermal and chemical stability of the ZR8 filler contributed to the improved electrolyte performance. CPE prepared with 7.5% ZR8 fillers showed a good ionic conductivity of 2.53 × 10−4 S cm−1 at 30 °C and 1.35 × 10−3 S cm−1 at 60 °C and an electrochemical window of ∼5.58 V. The [Li|ZR8-7.5|Li] cell showed an excellent stability in 8000-h operation without any flaw (at 60 °C and a current density of 100 μA cm−2). Additionally, the [Li|ZR8-7.5|LFP] cell was able to retain over 80% of its initial capacity (144.9 mA h g−1 at 0.5C) even after 800 cycles. © 2024 The Royal Society of Chemistry.FALSEsciescopu
Emerging Challenges in Textile Energy Electrodes: Interfacial Engineering for High-Performance Next-Generation Flexible Energy Storage Devices
No full textThe development of highly conductive fibril-type textile electrodes is crucial for the advancement of various smart wearable electronics including high-performance energy storage devices. To achieve this goal, it is essential to convert insulating textiles into conductive counterparts while maintaining flexibility and porosity. Additionally, the incorporation of electrochemically active components into textile conductors enables tailor-made textile energy electrodes for specific applications. Thus, textile conductors act not only as conductors but also as energy reservoirs for energy-active components, providing a facile electron transfer network. However, textile conductors fabricated by most existing methods face challenges such as low conductivity, blockage, and brittleness. One approach to overcome these problems is to utilize interfacial interactions between individual components and textiles. Conductive nanoparticle assembly and electrodeposition based on such rational design result in highly conductive, flexible, and large surface area textile conductors. The subsequent guided assembly of active components creates high-performance textile energy electrodes. This perspective describes how interfacial interaction-based assembly can enhance the performance of textile conductors and textile energy electrodes. It also explores various conductor preparation approaches and recent advances in the field for applications in supercapacitors and lithium-ion batteries. © 2023 The Authors. Small Structures published by Wiley-VCH GmbH. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.TRUEsciescopu