PolypMixNet: Enhancing semi-supervised polyp segmentation with polyp-aware augmentation.

BACKGROUND: AI-assisted polyp segmentation in colonoscopy plays a crucial role in enabling prompt diagnosis and treatment of colorectal cancer. However, the lack of sufficient annotated data poses a significant challenge for supervised learning approaches. Existing semi-supervised learning methods also suffer from performance degradation, mainly due to task-specific characteristics, such as class imbalance in polyp segmentation. PURPOSE: The purpose of this work is to develop an effective semi-supervised learning framework for accurate polyp segmentation in colonoscopy, addressing limited annotated data and class imbalance challenges. METHODS: We proposed PolypMixNet, a semi-supervised framework, for colorectal polyp segmentation, utilizing novel augmentation techniques and a Mean Teacher architecture to improve model performance. PolypMixNet introduces the polyp-aware mixup (PolypMix) algorithm and incorporates dual-level consistency regularization. PolypMix addresses the class imbalance in colonoscopy datasets and enhances the diversity of training data. By performing a polyp-aware mixup on unlabeled samples, it generates mixed images with polyp context along with their artificial labels. A polyp-directed soft pseudo-labeling (PDSPL) mechanism was proposed to generate high-quality pseudo labels and eliminate the dilution of lesion features caused by mixup operations. To ensure consistency in the training phase, we introduce the PolypMix prediction consistency (PMPC) loss and PolypMix attention consistency (PMAC) loss, enforcing consistency at both image and feature levels. Code is available at https://github.com/YChienHung/PolypMix. RESULTS: PolypMixNet was evaluated on four public colonoscopy datasets, achieving 88.97% Dice and 88.85% mIoU on the benchmark dataset of Kvasir-SEG. In scenarios where the labeled training data is limited to 15%, PolypMixNet outperforms the state-of-the-art semi-supervised approaches with a 2.88-point improvement in Dice. It also shows the ability to reach performance comparable to the fully supervised counterpart. Additionally, we conducted extensive ablation studies to validate the effectiveness of each module and highlight the superiority of our proposed approach. CONCLUSION: PolypMixNet effectively addresses the challenges posed by limited annotated data and unbalanced class distributions in polyp segmentation. By leveraging unlabeled data and incorporating novel augmentation and consistency regularization techniques, our method achieves state-of-the-art performance. We believe that the insights and contributions presented in this work will pave the way for further advancements in semi-supervised polyp segmentation and inspire future research in the medical imaging domain.

Anomalous Water Wetting on a Hydrophilic Substrate under a High Electric Field.

Macroscopically, the traditional Young-Lippmann equation is used to describe the water contact angle under a weak electric field. Here we report a new wetting mechanism of deionized water under a strong electric field that defies the conventional Young-Lippmann equation. The contact angle of the deionized water droplet on a model hexagonal lattice with a different initial wettability is extensively modulated by the vertical electric field. The cosine of water contact angle on a hydrophilic substrate displays an anomalous linear relationship with the field, in contrast to the hydrophobic case, which shows an inverse parabolic relationship. Such anomalous wetting is verified by experimental measurements of water droplets on a pyroelectric substrate. Moreover, we identify that this anomaly arises from the linear modulation of the solid-liquid interfacial tension of hydrophilic substrates by the electric field. Our findings provide atomistic insight into the fundamental laws and new phenomena of water-surface interactions under extreme electric fields.

Boundary-enhanced semi-supervised retinal layer segmentation in optical coherence tomography images using fewer labels.

Automatic segmentation of multiple layers in retinal optical coherence tomography (OCT) images is crucial for eye disease diagnosis and treatment. Despite the success of deep learning algorithms, it still remains a challenge due to the blurry layer boundaries and lack of adequate pixel-wise annotations. To tackle these issues, we propose a Boundary-Enhanced Semi-supervised Network (BE-SemiNet) that exploits an auxiliary distance map regression task to improve retinal layer segmentation with scarce labeled data and abundant unlabeled data. Specifically, a novel Unilaterally Truncated Distance Map (UTDM) is firstly introduced to alleviate the class imbalance problem and enhance the layer boundary learning in the regression task. Then for the pixel-wise segmentation and UTDM regression branches, we impose task-level and data-level consistency regularization on unlabeled data to enrich the diversity of unsupervised information and improve the regularization effects. Pseudo supervision is incorporated in consistency regularization to bridge the task prediction spaces for consistency and expand training labeled data. Experiments on two public retinal OCT datasets show that our method can greatly improve the supervised baseline performance with only 5 annotations and outperform the state-of-the-art methods. Since it is difficult and labor-expensive to obtain adequate pixel-wise annotations in practice, our method has a promising application future in clinical retinal OCT image analysis.

Synergistic effect of Ag and Cu on improvingin vitrobiological properties of K2Ti6O13nanowires for potential biomedical applications.

The development of novel antibacterial nano-materials with synergistic biological effects has attracted extensive interest of the researchers. In the study, 0.5 mol% Ag and 0.5 mol% Cu co-doped K2Ti6O13(0.5 Ag-0.5 Cu-KTO) nanomaterial was successfully synthesized using two-step method of sol-gel and hydrothermal synthesis. The crystal structure of 0.5 Ag-0.5 Cu-KTO was the same as that of monoclinic K2Ti6O13. Ag ions and Cu ions were uniformly loaded on K2Ti6O13by replacing partial Ti ions, so that these antibacterial ions could be slowly released. High specific surface area of 0.5 Ag-0.5 Cu-KTO (337.6 m2g-1) provided more surface active sites for Ag-Cu doping and adsorption. More negative surface zeta potential (-32.83 mV in phosphate buffer solution and -21.45 mV in physiological saline solution, respectively) would be beneficial to prevent the aggregation of the nanowires in physiological environment. Under the same doping amount, compared to 1.0 mol% Cu doped K2Ti6O13, 0.5 Ag-0.5 Cu-KTO exhibited better antibacterial performance against gram-positive and gram-negative bacteria at only 100 µg ml-1dose concentration, near to 1.0 mol% Ag doped K2Ti6O13(1.0 Ag-KTO). And 0.5 Ag-0.5 Cu-KTO showed more excellent biocompatibility than 1.0 Ag-KTO, which was attribute to the introduction of Cu ions effectively decreasing the hemolytic and cytotoxic risks from Ag ions. As expected, the synthesized 0.5 Ag-0.5 Cu-KTO nanowires demonstrated excellent structural stability, high antibacterial activity, good hemocompatibility and cytocompatibility owing to the synergistic effects of Cu and Ag ions. 0.5 Ag-0.5 Cu-KTO nanowires will be a promising antimicrobial candidate for biomedical applications.

Durably Self-Sustained Droplet on a Fully Miscible Liquid Film.

Droplets impacting onto a solid or liquid surface inducing wetting, floatation, splash, coalescence, etc. is ubiquitous in nature and industrial processes. Here, we report that liquid droplets exhibit spherical caps upon contact with a fully miscible liquid film of lower surface tension, despite the spontaneous mixing of the two liquids. Such a spherical cap on a continuous liquid surface sustains a long lifespan up to minutes before ultimately merging into the film. Benefiting from large viscous forces in a thin film as a result of spatial confinement, the surface flow is substantially suppressed. Therefore, the surface tension gradient responsible for this phenomenon is maintained because the normal diffusion of film liquid into the droplet can timely dilute film liquid supplied by uphill Marangoni flow at the droplet surface. The present finding removes the conventional cognition that droplet coalescence is prompt on fully miscible continuous liquid surfaces, thus benefiting design of new types of microfluidic devices.

Effects of different selenium sources on duodenum and jejunum tight junction network and growth performance of broilers in a model of fluorine-induced chronic oxidative stress.

The protective effects and underlying molecular mechanisms of sodium selenite (SS) and selenomethionine (SM) against chronic oxidative stress-induced duodenum and jejunum tight junction (TJ) network disturbance and growth inhibition of broilers were investigated in the current experiment. At the age of 1 d, 720 Lingnan Yellow broiler chicks were allocated to 4 experimental diets (with 6 replicates per diet and 30 birds per replicate) and offered either a control diet (fluorine [F] 23 mg/kg, control [CoN] group) or test diets (800 mg/kg F, high F [HF] group; 800 mg/kg F+0.15 mg selenium [Se]/kg as SS [SS group] or SM [SM group]) for 56 d. The results showed that HF group could induce chronic oxidative stress and subsequently increased (P < 0.05) proinflammatory cytokines levels of duodenum and jejunum in comparison with the CoN group. Increased proinflammatory cytokines levels of HF group promoted myosin light chain kinase (MLCK) transcription, thus leading to a decrease (P < 0.05) in TJ proteins expression of duodenum and jejunum when compared with the CoN group. A reduction of TJ proteins expression destroyed the TJ structures in the HF group, which in turn increased intestinal mucosal permeability of duodenum and jejunum and ultimately induced growth inhibition of broilers. Dietary Se supplementation could ameliorate HF-induced duodenum and jejunum TJ network impairment and growth retardation of broilers, potentially by increasing (P < 0.05) the glutathione peroxidase and thioredoxin reductase activities, reducing (P < 0.05) the reactive oxygen species and malondialdehyde levels, regulating the secretion of proinflammatory cytokines, and mediating the transcription level of MLCK in the duodenum and jejunum. Additionally, our data also suggested that the protective effects of SM were superior to those of SS. This study will provide a theoretical basis for developing SM into an efficient protective agent for intestinal mucosal barrier in poultry.

APRNet: Alternative Prediction Refinement Network for Polyp Segmentation.

Colorectal cancer has become the second leading cause of cancer-related death, attracting considerable interest for automatic polyp segmentation in polyp screening system. Accurate segmentation of polyps from colonoscopy is a challenging task as the polyps diverse in color, size and texture while the boundary between polyp and background is sometimes ambiguous. We propose a novel alternative prediction refinement network (APRNet) to more accurately segment polyps. Based on the UNet architecture, our APRNet aims at exploiting all-level features by alternatively leveraging features from encoder and decoder branch. Specifically, a series of prediction residual refinement modules (PRR) learn the residual and progressively refine the segmentation at various resolution. The proposed APRNet is evaluated on two benchmark datasets and achieves new state-of-the-art performance with a dice of 91.33% and an accuracy of 97.31% on the Kvasir-SEG dataset, and a dice of 86.33% and an accuracy of 97.12% on the EndoScene dataset.Clinical relevance- This work proposes an automatic and accurate polyp segmentation algorithm that achieves new state- of-the-art performance, which can potentially act as an observer pointing out polyps in colonoscopy procedure.

Transparent proton transport through a two-dimensional nanomesh material.

Molecular sieving is of great importance to proton exchange in fuel cells, water desalination, and gas separation. Two-dimensional crystals emerge as superior materials showing desirable molecular permeability and selectivity. Here we demonstrate that a graphdiyne membrane, an experimentally fabricated member in the graphyne family, shows superior proton conductivity and perfect selectivity thanks to its intrinsic nanomesh structure. The trans-membrane hydrogen bonds across graphdiyne serve as ideal channels for proton transport in Grotthuss mechanism. The free energy barrier for proton transfer across graphdiyne is ~2.4 kJ mol-1, nearly identical to that in bulk water (2.1 kJ mol-1), enabling "transparent" proton transport at room temperature. This results in a proton conductivity of 0.6 S cm-1 for graphdiyne, four orders of magnitude greater than graphene. Considering its ultimate pore size of 0.55 nm, graphdiyne membrane blocks soluble fuel molecules and exhibits superior proton selectivity. These advantages endow graphdiyne a great potential as proton exchange material.

p53-dependent CD51 expression contributes to characteristics of cancer stem cells in prostate cancer.

Castration-resistant prostate cancer (CRPC), which is considered to contain cancer stem cells (CSCs), leads to a high relapse rate in patients with prostate cancer (PCa). However, the markers of prostate CSCs are controversial. Here we demonstrate that CD51, in part, correlates with the poor prognosis of PCa patients. Further, we find that CD51 is a functional molecule that is able to promote the malignancy of PCa through enhancing tumor initiation, metastatic potential, and chemoresistance. Moreover, we find that elevated CD51 expression in PCa specimens correlates with p53 loss of function. Mechanistically, we demonstrate that p53 acts via Sp1/3 to repress CD51 transcription, and CD51 is required for PCa stemness and metastasis properties, and is downregulated by p53. Taken together, these results indicate that CD51 is a novel functional marker for PCa, which may provide a therapeutic target for the efficiently restricting PCa progression.

A modified Wenzel model for water wetting on van der Waals layered materials with topographic surfaces.

A modified Wenzel model is proposed for describing the wetting behavior of van der Waals layered materials with topographic surfaces, based on the measured linear relationship between water wetting and surface roughness for high quality Bi2Se3 thin films, synthesized using molecular beam epitaxy (MBE) in the optimized temperature window of 180-200 °C. The water contact angles are found to have apparent dependence on the nanoscale surface morphology, enabling film wettability as a new tool to quickly characterize the quality of atomically thin films. The water contact angle of the ideal Bi2Se3 surface is inferred to be ∼98.4°, indicating its intrinsic hydrophobic nature; however, the edge of the terrace on its surface is extremely hydrophilic, leading to easy hydrophobic/hydrophilic transitions. The atomistic mechanism is further revealed by first principles calculations. The regulated wettability is of great importance for electronic applications of Bi2Se3 and other two-dimensional materials with distinctive electronic structures.

In vitro osteogenetic activity of pearl.

In vivo and in vitro studies have shown that shell nacre and hydroxyapatite (HA) are promising bioactive materials for bone repair. In this work, the osteogenetic activity of pearl is evaluated by soaking it in simulated body fluid (SBF) and cell culture, taking shell nacre and HA as control materials at the same time. After soaking in SBF, HA particles were rapidly formed on the surface of pearl, the dissolution of CaCO3 and the binding between organic components and Ca2+ ions in pearl provide favorable conditions for the HA precipitation, and the whole process follows a dissolution-binding-precipitation mechanism. Calcium surplus, not conventional calcium deficiency, is found in HA crystal structure; it implies that type B-HA is formed on pearl surface in this study. HRTEM observation shows that HA is poorly crystallized with so many dislocations and shuttle-like amorphous areas. Cell culture reveals that pearl could stimulate osteoblast proliferation, which proceeded more quickly and smoothly than that on shell nacre and HA, and abundant extracellular matrix occupied the whole pearl surface by 5 days. It is concluded that pearl is a superior osteoinductive material with high osteogenetic activity.

Titanium disulfide nanotubes as hydrogen-storage materials.

TiS2 nanotubes, which were synthesized through a chemical transport reaction, are very effective in reversible hydrogen absorption and desorption with the capacity of 2.5 wt %.

Interaction mechanism of in-situ nano-TiN-AlN particles and solid/liquid interface during solidification.

This paper deals with the interaction mechanism between in situ nanometer-grade TiN-AlN particles and the solid/liquid (S/L) interface during the solidification of an in situ TiN-AlN/Al composite. According to the setting of a force balance for the particles in front of the S/L interface during solidification, F = F(buoyant) + F(repulsive) + F(viscous). We obtained the relationship between the critical cooling velocity of the liquid composite, Vr, and the size of the ceramic particle, rp. By this relationship formula, we can know that the S/L interface engulfs particles or pushes them to the crystal grain boundary during the solidification of a TiN-AlN/Al composite. It is found that Vr is proportional to the radius of ceramic particles by transmission electron microscope (TEM) observation. The TEM test indicates that the smaller the particle is, the more easily the S/L interface engulfs particles.