56016 research outputs found
Sort by
SPHARM-Net: Spherical Harmonics-based Convolution for Cortical Parcellation
Get PDFDepartment of Computer Science and EngineeringWe present a spherical harmonics-based convolutional neural network (CNN) for cortical parcellation, which we call SPHARM-Net. Recent advances in CNNs offer cortical parcellation on a fine-grained triangle mesh of the cortex. Yet, most CNNs designed for cortical parcellation employ spatial convolution that depends on extensive data augmentation and allows only predefined neighborhoods of specific spherical tessellation. On the other hand, a rotation-equivariant convolutional filter avoids data augmentation, and rotational equivariance can be achieved in spectral convolution independent of a neighborhood definition. Nevertheless, the limited resources of a modern machine enable only a finite set of spectral components that might lose geometric details. In this work, we propose (1) a constrained spherical convolutional filter that supports an infinite set of spectral components and (2) an end-to-end framework without data augmentation. The proposed filter encodes all the spectral components without the full expansion of spherical harmonics. We show that rotational equivariance drastically reduces the training time while achieving accurate cortical parcellation. Furthermore, the proposed convolution is fully composed of matrix transformations, which offers efficient and fast spectral processing. In the experiments, we validate SPHARM-Net on two public datasets with manual labels: Mindboggle-101 (N=101) and NAMIC (N=39). The experimental results show that the proposed method outperforms the state-of-the-art methods on both datasets even with fewer learnable parameters without rigid alignment and data augmentation.
Our code is publicly available at https://github.com/Shape-Lab/SPHARM-Net.ope
Strategies of developing Transition Metal-Based Heterogeneous electrocatalyst for efficient energy conversion reactions
No full textSchool of Energy and Chemical Engineering (Energy Engineering)clos
The Kinetics of Dissolution of Single Crystal Diamond (100) and (110) in Nickel and Cobalt Films and Vapor-Liquid-Solid Growth of Graphene Ribbons on Single Crystal and Polycrystalline Cu Foils
Get PDFDepartment of Materials Science and EngineeringThe kinetics of dissolution of single crystal diamond (with surface orientation of 100 or 110, named as D(100) and D(110)) into nickel or cobalt films was measured, to the best of our knowledge, for the first time. This was possible through our discovery that at sufficiently high partial pressure of water vapor the rate determining step was breaking of C-C bonds at the diamond-metal interfaceat lower partial pressures of water vapor, the rate determining step was found to be removal of carbon from the surface of the metal, rather than C-C bond breaking at the diamond-metal interface. The rate of diffusion from the diamond-metal interface to the surface of the metal film was found to never be rate-limiting. We found that single crystal diamond with surface orientation 111 could not be dissolved in either cobalt (Co) or nickel (Ni) films in the temperature range we studied. The details of this study are provided in Chapter 3. Time-of-flight-secondary ion mass spectrometry depth profiles show a concentration gradient of C from a certain depth into the metal film surface down to the M???D(100) interface, and residual gas analyzer measurements show that the gas products formed in the presence of water vapor by reaction of C atoms diffusing to, and thus present at, the metal surface are CO and H2. As mentioned, we discovered two different regimes (we name them I and II) for the kinetics of dissolution of both D(100) and D(110), in which the rate-determining step was the removal of carbon atoms on the open metal surface (regime I, lower partial pressure of water vapor) or dissolution of diamond at the metal???diamond interface (regime II, higher partial pressure of water vapor) that yielded different Arrhenius parameters. We found that the rate of dissolution of diamond in Co was higher than that in Ni for both D(100) and D(110) and for both regimes I and II, and possible reasons are suggested. As mentioned, we also found that D(111) could not be dissolved at the Ni/D(111) and Co/D(111) interface in the presence of water vapor (over the same range of sample temperatures). The reaction paths for dissolution of C at the M???D(100) or M???D(110) interface and for removal of C from the free surfaces of Ni and Co were assessed through density functional theory modeling at 1273 K by colleagues Yongchul Kim and Prof. Geunsik Lee.
In Chapter 4, we describe the bottom-up direct growth of graphene ribbons catalyzed by deliberate introduction of silica particles onto a Cu(111) foil surface, and we ascribe their growth to a combination of vapor-liquid-solid (VLS) growth (longitudinal growth) and vapor-solid (VS) growth (lateral growth onto the already existing ribbon that extends longitudinally from the VLS growth). Micrometer-long single crystal graphene ribbons (tapered when grown above 900 ???, but uniform width when grown in the range 850 to 900 ???, as this latter temperature range is too low for VS growth) using silica particle seeds were synthesized on single crystal Cu(111) foil. We discovered that tapered and uniform-width graphene ribbons grew strictly along the Cu direction on Cu(111) and polycrystalline copper (Cu) foils. Silica particles on both the single crystal and polycrystalline Cu foils formed (semi-)molten Cu-Si-O droplets at growth temperatures, then catalyzed nucleation and drove the longitudinal growth of graphene ribbons. Longitudinal growth is likely by a vapor-liquid-solid (VLS) mechanism, but edge growth (above 900 ???) is due to catalytic activation of ethylene (C2H4) and attachment of C atoms or species (???vapor solid??? or VS growth) at the edges. We found that the taper angle is determined only by the growth temperature and that the growth rates were independent of the particle size. A surface-diffusion vapor-liquid-solid growth model thus seems most appropriate for rationalizing the longitudinal growth. According to our kinetics study, we found the activation enthalpy (1.73 ?? 0.03 eV) for longitudinal ribbon growth on Cu(111) from ethylene is lower than that for VS growth at the edges of the GRs (2.78 eV ?? 0.15 eV) and for the graphene island growth (2.85 ?? 0.07 eV) that occurs concurrently. (That is, the Cu(111) surface has both GRs and hexagonal graphene islands. The graphene islands nucleate and grow on the regions of the Cu(111) surface where there is not a silica particle.ope
On-site colorimetric detection of adulterated gasoline using highly reflective 1D photonic crystal sensors based on photo-crosslinked polymer-titania hybrids
No full textThe adulteration and distribution of pure commercial gasoline remain a global problem because of the lack of a simple method for the rapid on-site detection of adulterated gasoline. Therefore, herein, we propose a simple method, based on highly reflective one-dimensional photonic crystals (1D PCs), for the on-site rapid colorimetric detection of adulterated gasoline by the naked eye. A photo-crosslinked copolymer-titania hybrid (Ti70) with a high refractive index and a photo-crosslinked methyl acrylate (MA)-based polymeric material containing 4-benzoylphenyl acrylate (BPA) (P(MA-co-BPA)) with a low refractive index were synthesized, characterized, and used in the fabrication of an 11-layered 1D PC sensor (Ti70/P(MA-co-BPA) PC). Ti70/P(MA-co-BPA) PC exhibited no color change in pure commercial gasoline. However, when Ti70/P(MA-co-BPA) PC was immersed in methanol, xylene, toluene, and benzene, it rapidly turned from bluish-purple to blue, sky-blue, greenish-yellow, and reddish-orange, respectively, and had high reflectivities. Additionally, when immersed in binary (gasoline/toluene and gasoline/methanol) and ternary (gasoline/toluene/methanol) mixtures, Ti70/P(MA-co-BPA) PC demonstrated significant color changes within short periods (similar to 2 min). Ti70/P(MA-co-BPA) PC exhibited excellent recyclability, which was verified via repeated swelling/deswelling tests for 10 cycles. These results indicate the applicability of the proposed novel Ti70/P(MA-co-BPA) PC for the rapid on-site colorimetric detection of adulterated gasoline by the naked-eye. We believe that the findings of this study will provide considerable insight for developing simple and high-performance 1D PC sensors for practical use and will also serve as a basis for exploring their potential applications
Nonlinear MHD modeling of n=1 RMP-induced pedestal transport and mode coupling effects on ELM suppression in KSTAR
No full textFully suppressing edge-localized modes (ELMs), e.g., with resonant magnetic perturbations (RMPs), is essential to reach and sustain high-performance steady-state H-mode plasmas because large ELMs can significantly reduce the lifetime of divertor components in future tokamak reactors. RMP-driven ELM suppression in KSTAR has been modeled by coupling the neoclassical transport code PENTRC to the nonlinear 3D MHD code JOREK. We have found that the radial transport from the combined effects of the kink-peeling, tearing response, and neoclassical toroidal viscosity can explain the pedestal degradation observed in experiments. In addition, it has been found that the RMP response can increase the inter-ELM heat flux on the lower outer divertor by redistributing the heat transport between the divertor plates. In addition to the degraded pedestal, ELM suppression is also attributable to the RMP-induced mode interactions. While the linear stability of peeling-ballooning mode (PBMs) improves owing to the degraded pedestal, the PBM and RMP interaction increases the spectral transfer between edge harmonics, preventing catastrophic growth and the crash of unstable modes. Here, it turns out that the magnetic islands near the pedestal top can play a vital role in mediating the mode interactions
Delineating transcriptional crosstalk between Mycobacterium avium subsp. paratuberculosis and human THP-1 cells at the early stage of infection via dual RNA-seq analysis
Get PDFMycobacterium avium subsp. paratuberculosis (MAP) is the causative agent of Johne's disease, a chronic debilitating disease in ruminants. To control this disease, it is crucial to understand immune evasion and the mechanism of persistence by analyzing the early phase interplays of the intracellular pathogens and their hosts. In the present study, host-pathogen interactions at the transcriptomic level were investigated in an in vitro macrophage infection model. When differentiated human THP-1 cells were infected with MAP, the expression of various genes associated with stress responses and metabolism was altered in both host and MAP at 3 h post-infection. MAP upregulates stress-responsive global gene regulators, such as two-component systems and sigma factors, in response to oxidative and cell wall stress. Downstream genes involved in type VII secretion systems, cell wall synthesis (polyketide biosynthesis proteins), and iron uptake were changed in response to the intracellular environment of macrophages. On the host side, upregulation of inflammatory cytokine genes was observed along with pattern recognition receptor genes. Notably, alterations in gene sets involved in arginine metabolism were observed in both the host and MAP, along with significant downregulation of NOS2 expression. These observations suggest that the utilization of metabolites such as arginine by intracellular MAP might affect host NO production. Our dual RNA-seq data can provide novel insights by capturing the global transcriptome with higher resolution, especially in MAP, thus enabling a more systematic understanding of host-pathogen interactions
Highly Vanadium Substitutional Doping in 2D Semiconductor at Wafer-Scale Using Reaction Promoter
No full textIntimate Contact with Perovskite Layer via Mesoporous Structured MoS2 Toward High-Performance Perovskite Solar Cells with Photostability
No full textReal quadratic fields admitting universal lattices of rank 7
No full textIn this paper, we prove that if d is sufficiently large square free positive rational integer, then there is no positive definite & RADIC; universal quadratic O-F-lattice of rank 7 where F = Q(root d). (C)& nbsp;2021 Elsevier Inc. All rights reserved
