Search

Computer-aided automatic pancreas segmentation is essential for early diagnosis and treatment of pancreatic diseases. However, the annotation of pancreas images requires professional doctors and considerable expenditure. Due to imaging differences among various institution population, scanning devices, imaging protocols, and so on, significant degradation in the performance of model inference results is prone to occur when models trained with domain-specific (usually institution-specific) datasets are directly applied to new (other centers/institutions) domain data. In this paper, we propose a novel unsupervised domain adaptation method based on adversarial learning to address pancreas segmentation challenges with the lack of annotations and domain shift interference.A 3D semantic segmentation model with attention module and residual module is designed as the backbone pancreas segmentation model. In both segmentation model and domain adaptation discriminator network, a multiscale progressively weighted structure is introduced to acquire different field of views. Features of labeled data and unlabeled data are fed in pairs into the proposed multiscale discriminator to learn domain-specific characteristics. Then the unlabeled data features with pseudodomain label are fed to the discriminator to acquire domain-ambiguous information. With this adversarial learning strategy, the performance of the segmentation network is enhanced to segment unseen unlabeled data.Experiments were conducted on two public annotated datasets as source datasets, respectively, and one private dataset as target dataset, where annotations were not used for the training process but only for evaluation. The 3D segmentation model achieves comparative performance with state-of-the-art pancreas segmentation methods on source domain. After implementing our domain adaptation architecture, the average dice similarity coefficient (DSC) of the segmentation model trained on the NIH-TCIA source dataset increases from 58.79% to 72.73% on the local hospital dataset, while the performance of the target domain segmentation model transferred from the medical segmentation decathlon (MSD) source dataset rises from 62.34% to 71.17%.Correlations of features across data domains are utilized to train the pancreas segmentation model on unlabeled data domain, improving the generalization of the model. Our results demonstrate that the proposed method enables the segmentation model to make meaningful segmentation for unseen data of the training set. In the future, the proposed method has the potential to apply segmentation model trained on public dataset to clinical unannotated CT images from local hospital, effectively assisting radiologists in clinical practice.

Ethylene epoxidation is one of the most critical industrial reactions. Although there have been many studies on the reaction, the details of the reaction kinetics remain elusive. In this work, an extensive microkinetic modelling based on DFT calculations was conducted to study ethylene epoxidation on Ag(111). Coverage-dependent and coverage-independent microkinetic modellings were performed with rigorous energetics calculated on adsorption and transition states. The detailed comparisons between the two modellings show that the coverage-independent modellings are inadequate even though the coverage is very low at the steady state. It is also found that the activity and selectivity trends from the coverage-dependent modelling are consistent with the experimental values under the experimental conditions. A comprehensive kinetic analysis under industrial conditions was conducted based on the coverage-dependent modelling, which shows that increasing the temperature and pressure can effectively enhance the activity and ethylene epoxide (EO) selectivity.

With the development of radiology and computer technology, diagnosis by medical imaging is heading toward precision and automation. Due to complex anatomy around the pancreatic tissue and high demands for clinical experience, the assisted pancreas segmentation system will greatly promote clinical efficiency. However, the existing segmentation model suffers from poor generalization among images from multiple hospitals. In this paper, we propose an end-to-end data-adaptive pancreas segmentation system to tackle the problems of lack of annotations and model generalizability. The system employs adversarial learning to transfer features from labeled domains to unlabeled domains, seeking a dynamic balance between domain discrimination and unsupervised segmentation. The image quality control toolbox is embedded in the system, which standardizes image quality in terms of intensity, field of view, and so on, to decrease heterogeneity among image domains. In addition, the system implements a data-adaptive process end-to-end without complex operations by doctors. The experiments are conducted on an annotated public dataset and an unannotated in-hospital dataset. The results indicate that after data adaptation, the segmentation performance measured by the dice similarity coefficient on unlabeled images improves from 58.79% to 75.43%, with a gain of 16.64%. Furthermore, the system preserves quantitatively structured information such as the pancreas’ size and volume, as well as objective and accurate visualized images, which assists clinicians in diagnosing and formulating treatment plans in a timely and accurate manner.

Water-alkaline electrolysis holds a great promise for industry-scale hydrogen production but is hindered by the lack of enabling hydrogen evolution reaction electrocatalysts to operate at ampere-level current densities under low overpotentials. Here, we report the use of hydrogen spillover-bridged water dissociation/hydrogen formation processes occurring at the synergistically hybridized Ni

To date, the understanding of reactions at solid-liquid interfaces has proven challenging, mainly because of the inaccessible nature of such systems to current experimental techniques with atomic resolution. This has meant that many important features, including free energy barriers and the atomistic structure of intermediates, remain unknown. To tackle these issues, we construct and utilize a high-dimensional neural network (HDNN) potential for the simulation of hydrogen evolution at the HCl(aq)/Pt(111) interface, taking into consideration the influence of adsorbate-adsorbate, adsorbate-solvent interactions, and ion solvation explicitly. Long time scale MD simulations reveal coadsorbed Had/H2Oad on the surface. The free energy profiles for the Tafel and Heyrovsky type hydrogen coupling are extracted using umbrella sampling. It is found that the preferential mechanism can change depending on the surface coverage, highlighting the dual mechanistic nature for HER on Pt(111). Our work demonstrates the importance of controlling the solvent-substrate interactions in developing catalysts beyond Pt.

The potential distribution at the electrode interface is a core factor in electrochemistry, and it is usually treated by the classic Gouy-Chapman-Stern (G-C-S) model. Yet the G-C-S model is not applicable to nanosized particles collision electrochemistry as it describes steady-state electrode potential distribution. Additionally, the effect of single nanoparticles (NPs) on potential should not be neglected because the size of a NP is comparable to that of an electrode. Herein, a theoretical model termed as Metal-Solution-Metal Nanoparticle (M-S-MNP) is proposed to reveal the dynamic electrode potential distribution at the single-nanoparticle level. An explicit equation is provided to describe the size/distance-dependent potential distribution in single NPs stochastic collision electrochemistry, showing the potential distribution is influenced by the NPs. Agreement between experiments and simulations indicates the potential roles of the M-S-MNP model in understanding the charge transfer process at the nanoscale.

Photocatalytic water splitting provides an economically feasible way for converting solar energy into hydrogen. Great efforts have been devoted to developing efficient photocatalysts; however, the surface catalytic reactions, especially for the sluggish oxygen evolution reaction (OER), still remain a challenge, which limits the overall photocatalytic energy efficiency. Herein, we design a Rhn cluster cocatalyst, with Rh0-Rh3+ sites anchoring the Mo-doped BiVO4 model photocatalytic system. The resultant photocatalyst enables a high visible-light photocatalytic oxygen production activity of 7.11 mmol g−1 h−1 and an apparent quantum efficiency of 29.37 % at 420 nm. The turnover frequency (TOF) achieves 416.73 h−1, which is 378 times higher than that of the photocatalyst only with Rh3+ species. Operando X-ray absorption characterization shows the OER process on the Rh0-Rh3+ sites. The DFT calculations further illustrate a bifunctional OER mechanism over the Rh0-Rh3+ sites, in which the oxygen intermediate attacks the Rh3+ sites with assistance of a hydrogen atom transfer to the Rh0 sites, thus breaking the scaling relationship of various oxygen intermediates.

Ceria-based catalysts exhibit great activity in catalyzing selective hydrogenation of CO2 to methanol. However, the underlying mechanism of this reaction, especially the generation of active H species, remains unclear. In this work, we performed extensive density functional theory calculations corrected by on-site Coulomb interaction (DFT + U) to investigate the H2 dissociation and the reaction between the active H species and CO2 on the pristine and Cu-doped CeO2(111) (denoted as Cu/CeO2(111)) surfaces. Our calculations evidenced that the heterolytic H2 dissociation for hydride generation can more readily occur on the Cu/CeO2(111) surface than on the pristine CeO2(111) surface. We also found that the Cu dopant can facilitate the formation of surface oxygen vacancies, further promoting the generation of hydride species. Moreover, the adsorption of CO2 and the hydrogenation of CO2 to HCOO* can be greatly promoted on the Cu/CeO2(111) surface with hydride species, which can lead to the high activity and selectivity toward CO2 hydrogenation to methanol.

Methane (CH4) controllable activation is the key process for CH4 upgrading, which is sensitive to the surface oxygen species. The high thermal conductivity and superb thermal stability of the hexagonal boron nitride (h-BN) sheet makes a single transition metal atom doped hexagonal boron nitride monolayer (TM-BN) possible to be a promising material for catalyzing methane partial oxidation. The performances of 24 TM-BNs for CH4 activation are systematically investigated during the CH4 oxidation by means of first-principles computation. The calculation results unravel the periodic variation trends for the stability of TM-BN, the adsorption strength and the kind of O2 species, and the resulting CH4 activation performance on TM-BNs. The formed peroxide O22− of which the O O bond could be broken and O− anions are found to be reactive oxygen species for CH4 activation under the mild conditions. It is found that the redox potential of TM center, including its valence electron number, coordination environment, and the work function of TM-BN, is the underlying reason for the formation of different oxygen species and the resulting activity for CH4 oxidative dehydrogenation.

MnO2-based catalysts exhibit remarkable activities in a room-temperature NO oxidation reaction (RT-NOR) but are challenged by water-induced deactivation. To improve the water-resistance driven by molecular-level mechanistic understanding, we herein report a systematic first-principles study and microkinetic simulation on elucidating the water-modulated RT-NOR mechanism on the notable β-MnO2 catalyst. Results show that RT-NOR preferentially follows the Mars-van Krevelen mechanism on β-MnO2(110) and that the coupling of lattice O (Olat) with NO sitting on the surface Mncus site is the rate-limiting step. The presence of H2O results in a competitive effect against NO adsorbing on Mncus and inhibits NO2 formation. More crucially, the H2O-dissociated H species eventually accumulate on Olat and are difficult to remove, which seriously poison the catalyst, although the Mncus-anchored OH intermediate can oxidize NO to NO2 via the nitrite/nitrate route. Weakening the H adsorption on Olat, as well as the H2O adsorption on Mncus, is thus proposed as conductive to alleviating MnO2 deactivation. More essentially, we find that the vacancy formation energy of Olat, Ef(Ovac), can serve as a unified descriptor for balancing H2O-resistance and catalytic activity; the excessively low Ef(Ovac) is the intrinsic origin of MnO2 susceptible to deactivation by H2O. By doping engineering guided by such a descriptor, we identify that the Cr-, Nb-, Ru-, or Rh-substituted doping can properly increase the Ef(Ovac) of MnO2 and weaken the H2O-induced deactivation in RT-NOR. These results may provide a theoretical basis on optimizing MnO2-based catalysts with improved H2O-resistance.

Epoxidation is an efficient chemical process for manufacturing styrene oxide; however, it remains a huge challenge to develop cost-effective and environment-friendly catalysts for facilitating this...

Selective acetylene hydrogenation as a vital industrial reaction has been studied for decades. However, some key issues remain elusive. Herein, a detailed microkinetic model using density functiona...

Due to the low solubility and diffusion coefficients of carbon dioxide in aqueous solutions, it is difficult to achieve high conversion current density using carbon dioxide electrolytic cells with aqueous electrolytes. The appearance of alkaline polymer electrolyte electrolytic cells (APEECs) brings hope to solve this problem. However, the veil on the details of CO2 reduction in APEECs has not been lifted, which hinders their performance improvements. In this study, the electrochemical reduction of carbon dioxide at the aqueous solution/copper interface and polymer electrolyte/copper interface were compared by first principles calculations to present a fundamental understanding of the superiority of APEECs. Through advanced molecular dynamics simulations, full reaction energetics at the operation potential are obtained. It was identified that the polymer electrolyte/copper interface exhibited a high turnover frequency mainly because of breaking away from the restrictions of aqueous-based systems. Moreover, our microkinetic model results show that by adjusting the hydroxyl ion concentration and water content of the polymer electrolyte, the conversion current density of APEECs can be further increased to two orders of magnitude higher than that of aqueous-based electrolytic cells.

Rational catalyst design is one of the most challenging tasks in heterogeneous catalysis, mainly due to the lack of optimization strategies. In this work, we proposed a gradient-based optimization strategy for rational catalyst design. In this strategy, the catalyst design is treated as an optimization process of surface structures toward maximum activity. The bonding contribution equation, a quantitative relationship between the surface structures and the adsorption energies, is utilized to predict the adsorption in the optimization. A traditional catalyst for CO oxidation, platinum, is optimized using this strategy. After eight step optimization, the activity of CO oxidation increases dramatically. It is found that the origin of the high activity of the designed catalyst is not only from the suitable bonding strength of all the species and reaction barriers of all elementary steps, but also from circumventing the BEP relation. Our design strategy offers a powerful tool to understand and design novel catalytic materials.

The higher reactivity of the methanol product over the methane reactant for the direct oxidation of methane to methanol is explored. C–H activation, C–O coupling, and C–OH coupling are investigated as key steps in the selective oxidation of methane using DFT. These elementary steps are initially considered in the gas phase for a variety of fcc (111) pristine metal surfaces. Methanol is found to be consistently more reactive for both C–H activation and subsequent oxidation steps. With an aqueous environment being understood experimentally to have a profound effect on the selectivity of this process, these steps are also considered in the aqueous phase by ab initio molecular dynamics calculations. The water solvent is modelled explicity, with each water molecule given the same level of theory as the metal surface and surface species. Free energy profiles for these steps are generated by umbrella sampling. It is found that an aqueous environment has a considerable effect on the kinetics of the elementary steps yet has little effect on the methane/methanol selectivity-conversion limit. Despite this, we find that the aqueous phase promotes the C–OH pathway for methanol formation, which could enhance the selectivity for methanol formation over that of other oxygenates., Consideration of the selectivity of methane to methanol in the aqueous phase with AIMD.

Copper-based materials are efficient electrocatalysts for the conversion of CO2 to C2+ products, and most these materials are reconstructed in situ to regenerate active species. It is a challenge to precisely design precatalysts to obtain active sites for the CO2 reduction reaction (CO2 RR). Herein, we develop a strategy based on local sulfur doping of a Cu-based metal-organic framework precatalyst, in which the stable Cu-S motif is dispersed in the framework of HKUST-1 (S-HKUST-1). The precatalyst exhibits a high ethylene selectivity in an H-type cell with a maximum faradaic efficiency (FE) of 60.0 %, and delivers a current density of 400 mA cm-2 with an ethylene FE up to 57.2 % in a flow cell. Operando X-ray absorption results demonstrate that Cuδ+ species stabilized by the Cu-S motif exist in S-HKUST-1 during CO2 RR. Density functional theory calculations indicate the partially oxidized Cuδ+ at the Cu/Cux Sy interface is favorable for coupling of the *CO intermediate due to the modest distance between coupling sites and optimized adsorption energy.

The localization of photogenerated electrons in photovoltaic and photocatalytic materials is crucial for reducing the electron–hole recombination rate. Here, the photogenerated electron localization is systematically investigated on the CH3NH3PbI3 (MAPbI3) perovskite using first-principles calculations. It is found that under vacuum conditions, the photogenerated electron is delocalized in the MAPbI3 bulk as well as on the stochiometric MAPbI3(001) surface with the CH3NH3I (MAI) termination, while it is trapped on the defect-free PbI2-terminated surface. Our ab initio molecular dynamics simulations reveal that the introduction of solutions will prompt the formation of localized electronic states. The photogenerated electron is discovered to be localized on both the MAI- and PbI2-terminated surfaces in the presence of solutions with different concentrations of HI, from pure water to the saturated solution. We demonstrate that the Pb–I bond weakening or breaking resulting in an unsaturated coordination of a Pb site is the prerequisite to trap the photogenerated electron.

Single-atom catalysts have been applied in many processes recently. The difference of their kinetic behavior compared to the traditional heterogeneous catalysts has not been extensively discussed yet. Herein a complete catalytic cycle of CH4 combustion assuming to be confined at isolated single sites of the Co3O4(110) surface is computationally compared with that on multi sites. The macroscopic kinetic behaviors of CH4 combustion on Co3O4(110) is systematically and quantitatively compared between those on the single site and multi sites utilizing kinetic Monte Carlo simulations upon the energetic information from the PBE + U calculation and statistic mechanics. The key factors governing the kinetics of CH4 combustion are disclosed for both the catalytic cycles respectively following the single-site and multi-site mechanisms. It is found that cooperation of multi active sites can promote the activity of complete CH4 combustions substantially in comparison to separated single-site catalyst whereas the confinement of active sites could regulate the selectivity of CH4 oxidation. The quantitative understanding of catalytic mechanism paves the way to improve the activity and selectivity for CH4 oxidation.

Ethanol dry reformation (EDR) is a chemical process for syngas production, which consumes a greenhouse gas and reduces carbon footprint. We present a mechanistic study of EDR over Rh catalyst based...

The empirical optimization of the preparation of catalytically active copper-containing catalysts is far more advanced than the fundamental understanding of the catalyst performance because of the ...

Single-atom catalysts not only maximize metal atom efficiency, they also display properties that are considerably different to their more conventional nanoparticle equivalents, making them a promising family of materials to investigate. Herein we developed a general host–guest strategy to fabricate various metal single-atom catalysts on nitrogen-doped carbon (M1/CN, M = Pt, Ir, Pd, Ru, Mo, Ga, Cu, Ni, Mn). The iridium variant Ir1/CN electrocatalyses the formic acid oxidation reaction with a mass activity of 12.9 $${{{\rm{A}}\,{\rm{mg}}^{-1}_{{\rm{Ir}}}}}$$ whereas an Ir/C nanoparticle catalyst is almost inert (~4.8 × 10−3 $${{{\rm{A}}\,{\rm{mg}}^{-1}_{{\rm{Ir}}}}}$$). The activity of Ir1/CN is also 16 and 19 times greater than those of Pd/C and Pt/C, respectively. Furthermore, Ir1/CN displays high tolerance to CO poisoning. First-principle density functional theory reveals that the properties of Ir1/CN stem from the spatial isolation of iridium sites and from the modified electronic structure of iridium with respect to a conventional nanoparticle catalyst. Single-atom catalysts maximize metal atom efficiency and exhibit properties that can be considerably different to their nanoparticle equivalent. Now a general host–guest strategy to make various single-atom catalysts on nitrogen-doped carbon has been developed; the iridium variant electrocatalyses the formic acid oxidation reaction with high mass activity and displays high tolerance to CO poisoning.

Identifying the activity origin of noble-metal-based catalysts is extremely important for fundamental research and practical application. We report that single Au atoms on Co3O4 (Au1/Co3O4 single-a...