NuSEA: Nuclei Segmentation With Ellipse Annotations.

OBJECTIVE: Nuclei segmentation is a crucial pre-task for pathological microenvironment quantification. However, the acquisition of manually precise nuclei annotations for improving the performance of deep learning models is time-consuming and expensive. METHODS: In this paper, an efficient nuclear annotation tool called NuSEA is proposed to achieve accurate nucleus segmentation, where a simple but effective ellipse annotation is applied. Specifically, the core network U-Light of NuSEA is lightweight with only 0.86 M parameters, which is suitable for real-time nuclei segmentation. In addition, an Elliptical Field Loss and a Texture Loss are proposed to enhance the edge segmentation and constrain the smoothness simultaneously. RESULTS: Extensive experiments on three public datasets (MoNuSeg, CPM-17, and CoNSeP) demonstrate that NuSEA is superior to the state-of-the-art (SOTA) methods and better than existing algorithms based on point, rectangle, and text annotations. CONCLUSIONS: With the assistance of NuSEA, a new dataset called NuSEA-dataset v1.0, encompassing 118,857 annotated nuclei from the whole-slide images of 12 organs is released. The codes and the new dataset are publicly available at https://github.com/dreambamboo/NuSEA/. SIGNIFICANCE: NuSEA provides a rapid and effective annotation tool for nuclei in histopathological images, benefiting future explorations in deep learning algorithms.

Generalizable and Discriminative Representations for Adversarially Robust Few-Shot Learning.

Few-shot image classification (FSIC) is beneficial for a variety of real-world scenarios, aiming to construct a recognition system with limited training data. In this article, we extend the original FSIC task by incorporating defense against malicious adversarial examples. This can be an arduous challenge because numerous deep learning-based approaches remain susceptible to adversarial examples, even when trained with ample amounts of data. Previous studies on this problem have predominantly concentrated on the meta-learning framework, which involves sampling numerous few-shot tasks during the training stage. In contrast, we propose a straightforward but effective baseline via learning robust and discriminative representations without tedious meta-task sampling, which can further be generalized to unforeseen adversarial FSIC tasks. Specifically, we introduce an adversarial-aware (AA) mechanism that exploits feature-level distinctions between the legitimate and the adversarial domains to provide supplementary supervision. Moreover, we design a novel adversarial reweighting training strategy to ameliorate the imbalance among adversarial examples. To further enhance the adversarial robustness without compromising discriminative features, we propose the cyclic feature purifier during the postprocessing projection, which can reduce the interference of unforeseen adversarial examples. Furthermore, our method can obtain robust feature embeddings that maintain superior transferability, even when facing cross-domain adversarial examples. Extensive experiments and systematic analyses demonstrate that our method achieves state-of-the-art robustness as well as natural performance among adversarially robust FSIC algorithms on three standard benchmarks by a substantial margin.

Customizing precise, tunable, and universal cascade charge transfer chains towards versatile photoredox catalysis.

The core factors dictating the photocatalysis efficiency are predominantly centered on controllable modulation of anisotropic spatial charge transfer/separation and regulating vectorial charge transport pathways. Nonetheless, the sluggish charge transport kinetics and incapacity of precisely tuning interfacial charge flow at the nanoscale level are still the primary dilemma. Herein, we conceptually demonstrate the elaborate design of a cascade charge transport chain over transition metal chalcogenide-insulating polymer-cocatalyst (TIC) photosystems via a progressive self-assembly strategy. The intermediate ultrathin non-conjugated insulating polymer layer, i.e., poly(diallyl-dimethylammonium chloride) (PDDA), functions as the interfacial electron relay medium, and simultaneously, outermost co-catalysts serve as the terminal "electron reservoirs", synergistically contributing to the charge transport cascade pathway and substantially boosting the interfacial charge separation. We found that the insulating polymer mediated unidirectional charge transfer cascade is universal for a large variety of metal or non-metal reducing co-catalysts (Au, Ag, Pt, Ni, Co, Cu, NiSe2, CoSe2, and CuSe). More intriguingly, such peculiar charge flow characteristics endow the self-assembled TIC photosystems with versatile visible-light-driven photoredox catalysis towards photocatalytic hydrogen generation, anaerobic selective organic transformation, and CO2-to-fuel conversion. Our work would provide new inspiration for smartly mediating spatial vectorial charge transport towards emerging solar energy conversion.

Polaronic Nonlinear Optical Response and All-Optical Switching Based on an Ionic Metal Oxide.

It has been well-established that light-matter interactions, as manifested by diverse linear and nonlinear optical (NLO) processes, are mediated by real and virtual particles, such as electrons, phonons, and excitons. Polarons, often regarded as electrons dressed by phonons, are known to contribute to exotic behaviors of solids, from superconductivity to photocatalysis, while their role in materials' NLO response remains largely unexplored. Here, the NLO response mediated by polarons supported by a model ionic metal oxide, TiO2, is examined. It is observed that the formation of polaronic states within the bandgap results in a dramatic enhancement of NLO absorption coefficient by over 130 times for photon energies in the sub-bandgap regions, characterized by a 100 fs scale ultrafast response that is typical for thermalized electrons in metals. The ultrafast polaronic NLO response is then exploited for the development of all-optical switches for ultrafast pulse generation in near-infrared (NIR) fiber lasers and modulation of optical signal in the telecommunication band based on evanescent interaction on a planar waveguide chip. These results suggest that the polarons supported by dielectric ionic oxides can fill the gaps left by dielectric and metallic materials and serve as a novel platform for nonlinear photonic applications.

Tailoring Broadband Nonlinear Optical Characteristics and Ultrafast Photocarrier Dynamics of Bi2O2S Nanosheets by Defect Engineering.

Low-dimensional bismuth oxychalcogenides have shown promising potential in optoelectronics due to their high stability, photoresponse, and carrier mobility. However, the relevant studies on deep understanding for Bi2O2S is quite limited. Here, comprehensive experimental and computational investigations are conducted in the regulated band structure, nonlinear optical (NLO) characteristics, and carrier dynamics of Bi2O2S nanosheets via defect engineering, taking O vacancy (OV) and substitutional Se doping as examples. As the OV continuously increased to ≈35%, the optical bandgaps progressively narrow from ≈1.21 to ≈0.81 eV and NLO wavelengths are extended to near-infrared regions with enhanced saturable absorption. Simultaneously, the relaxation processes are effectively accelerated from tens of picoseconds to several picoseconds, as the generated defect energy levels can serve as both additional absorption cross-sections and fast relaxation channels supported by theoretical calculations. Furthermore, substitutional Se doping in Bi2O2S nanosheets also modulate their optical properties with the similar trends. As a proof-of-concept, passively mode-locked pulsed lasers in the ≈1.0 µm based on the defect-rich samples (≈35% OV and ≈50% Se-doping) exhibit excellent performance. This work deepens the insight of defect functions on optical properties of Bi2O2S nanosheets and provides new avenues for designing advanced photonic devices based on low-dimensional bismuth oxychalcogenides.

Toward Intrinsic Adversarial Robustness Through Probabilistic Training.

Modern deep neural networks have made numerous breakthroughs in real-world applications, yet they remain vulnerable to some imperceptible adversarial perturbations. These tailored perturbations can severely disrupt the inference of current deep learning-based methods and may induce potential security hazards to artificial intelligence applications. So far, adversarial training methods have achieved excellent robustness against various adversarial attacks by involving adversarial examples during the training stage. However, existing methods primarily rely on optimizing injective adversarial examples correspondingly generated from natural examples, ignoring potential adversaries in the adversarial domain. This optimization bias can induce the overfitting of the suboptimal decision boundary, which heavily jeopardizes adversarial robustness. To address this issue, we propose Adversarial Probabilistic Training (APT) to bridge the distribution gap between the natural and adversarial examples via modeling the latent adversarial distribution. Instead of tedious and costly adversary sampling to form the probabilistic domain, we estimate the adversarial distribution parameters in the feature level for efficiency. Moreover, we decouple the distribution alignment based on the adversarial probability model and the original adversarial example. We then devise a novel reweighting mechanism for the distribution alignment by considering the adversarial strength and the domain uncertainty. Extensive experiments demonstrate the superiority of our adversarial probabilistic training method against various types of adversarial attacks in different datasets and scenarios.

Layer-Tunable Nonlinear Optical Characteristics and Photocarrier Dynamics of 2D PdSe2 in Broadband Spectra.

Layered 2D transition metal dichalcogenides (TMDCs) exhibited fascinating nonlinear optical (NLO) properties for constructing varied promising optoelectronics. However, exploring the desired 2D materials with both superior nonlinear absorption and ultrafast response in broadband spectra remain the key challenges to harvest their greatest potential. Here, based on synthesizing 2D PdSe2 films with the controlled layer number, the authors systematically demonstrated the broadband giant NLO performance and ultrafast excited carrier dynamics of this emerging material under femtosecond visible-to-near-infrared laser-pulse excitation (400-1550 nm). Layer-dependent and wavelength-dependent evolution of optical bandgap, nonlinear absorption, and photocarrier dynamics in the obtained 2D PdSe2 are clearly revealed. Specially, the transition from semiconducting to semimetallic PdSe2 induced dramatic changes of their interband absorption-relaxation process. This work makes 2D PdSe2 more competitive for future ultrafast photonics and also opens up a new avenue for the optical performance optimization of various 2D materials by rational design of these materials.

Transition from steady shear to oscillatory shear rheology of dense suspensions.

Recent studies have highlighted that oscillatory and time-dependent shear flows might help increase the flowability of dense suspensions. While most focus has been on cross-flows we here study a simple two-dimensional suspensions where we apply simultaneously oscillatory and stationary shear along the same direction. We first show that the dissipative viscosities in this set-up significantly decrease with an increasing shear-rate magnitude of the oscillations and given that the oscillatory strain is small, in a similar fashion as found previously for cross-flow oscillations. As for cross-flow oscillations, the decrease can be attributed to the large decrease in the number of contacts and an altered microstructure as one transitions from a steady shear to an oscillatory shear dominated rheology. As subresults we find both an extension to the µ(J) rheology, a constitutive relationship between the shear stresses and the shear rate, valid for oscillatory shear flows and that shear-jamming of frictional particles at oscillatory shear dominated flows occurs at higher packing fractions compared to steady shear dominated flows.

Enhancing flashover performance of alumina/epoxy spacers by adaptive surface charge regulation using graded conductivity coating.

Surface flashover properties of alumina/epoxy spacers, involving a surface charge accumulation process, are critical for the safe and reliable operation of a high-voltage direct-current (HVDC) gas-insulated transmission line (GIL). This study reports surface charging behavior and flashover performance of alumina/epoxy spacers with different surface conductivity graded coating (SCGC) schemes in SF6/N2 mixtures under DC stress. Four kinds of SCGC schemes, i.e. localized coating near high voltage (HV-coating), near grounded electrode (GND-coating), at the middle of spacer surface (SPM-coating) and near both high voltage and grounded electrode (HV-GND-coating), are designed by partially spraying SiC/epoxy composites on the spacer surface. Surface charge distribution patterns exhibit varied features with different SCGC schemes. The HV-coating and GND-coating schemes lead to aggravated homo-charge and hetero-charge accumulation respectively, whereas in the SPM-coating scheme surface charge shows a multi-tier distribution pattern with alternating polarity. A transition of the dominant surface charge mechanism from bulk conductivity to surface conductivity with increasing conductivity on the coated area is found. Flashover performance differs a lot with different SCGC schemes: the HV-coating and HV-GND-coating schemes increase the flashover voltage while the SPM-coating and GND-coating schemes degrade it. The optimal surface insulation strength is achieved in the HV-coating scheme with a coating width of about 10 mm. The impact of different SCGC schemes on flashover performance is revealed based on the electric field analysis by considering the effect of surface charges.

Computational and experimental analysis of short peptide motifs for enzyme inhibition.

The metabolism of living systems involves many enzymes that play key roles as catalysts and are essential to biological function. Searching ligands with the ability to modulate enzyme activities is central to diagnosis and therapeutics. Peptides represent a promising class of potential enzyme modulators due to the large chemical diversity, and well-established methods for library synthesis. Peptides and their derivatives are found to play critical roles in modulating enzymes and mediating cellular uptakes, which are increasingly valuable in therapeutics. We present a methodology that uses molecular dynamics (MD) and point-variant screening to identify short peptide motifs that are critical for inhibiting ß-galactosidase (ß-Gal). MD was used to simulate the conformations of peptides and to suggest short motifs that were most populated in simulated conformations. The function of the simulated motifs was further validated by the experimental point-variant screening as critical segments for inhibiting the enzyme. Based on the validated motifs, we eventually identified a 7-mer short peptide for inhibiting an enzyme with low µM IC50. The advantage of our methodology is the relatively simplified simulation that is informative enough to identify the critical sequence of a peptide inhibitor, with a precision comparable to truncation and alanine scanning experiments. Our combined experimental and computational approach does not rely on a detailed understanding of mechanistic and structural details. The MD simulation suggests the populated motifs that are consistent with the results of the experimental alanine and truncation scanning. This approach appears to be applicable to both natural and artificial peptides. With more discovered short motifs in the future, they could be exploited for modulating biocatalysis, and developing new medicine.