Semi-Supervised Medical Image Segmentation with Co-Distribution Alignment.

Medical image segmentation has made significant progress when a large amount of labeled data are available. However, annotating medical image segmentation datasets is expensive due to the requirement of professional skills. Additionally, classes are often unevenly distributed in medical images, which severely affects the classification performance on minority classes. To address these problems, this paper proposes Co-Distribution Alignment (Co-DA) for semi-supervised medical image segmentation. Specifically, Co-DA aligns marginal predictions on unlabeled data to marginal predictions on labeled data in a class-wise manner with two differently initialized models before using the pseudo-labels generated by one model to supervise the other. Besides, we design an over-expectation cross-entropy loss for filtering the unlabeled pixels to reduce noise in their pseudo-labels. Quantitative and qualitative experiments on three public datasets demonstrate that the proposed approach outperforms existing state-of-the-art semi-supervised medical image segmentation methods on both the 2D CaDIS dataset and the 3D LGE-MRI and ACDC datasets, achieving an mIoU of 0.8515 with only 24% labeled data on CaDIS, and a Dice score of 0.8824 and 0.8773 with only 20% data on LGE-MRI and ACDC, respectively.

Class-Specific Distribution Alignment for semi-supervised medical image classification.

Despite the success of deep neural networks in medical image classification, the problem remains challenging as data annotation is time-consuming, and the class distribution is imbalanced due to the relative scarcity of diseases. To address this problem, we propose Class-Specific Distribution Alignment (CSDA), a semi-supervised learning framework based on self-training that is suitable to learn from highly imbalanced datasets. Specifically, we first provide a new perspective to distribution alignment by considering the process as a change of basis in the vector space spanned by marginal predictions, and then derive CSDA to capture class-dependent marginal predictions on both labeled and unlabeled data, in order to avoid the bias towards majority classes. Furthermore, we propose a Variable Condition Queue (VCQ) module to maintain a proportionately balanced number of unlabeled samples for each class. Experiments on three public datasets HAM10000, CheXpert and Kvasir show that our method provides competitive performance on semi-supervised skin disease, thoracic disease, and endoscopic image classification tasks.

Animal-Protein-Based and Synthetic-Based Foamed Mixture Lightweight Soil Doped with Bauxite Tailings: Macro and Microscopic Properties.

In order to explore the effect of the foaming agent type on the properties of foamed mixture lightweight soil mixed with bauxite tailings (FMLSB), low-density (437.5 kg/m3 and 670 kg/m3) and high-density (902.5 kg/m3 and 1170 kg/m3) FMLSB were prepared using protein-based and synthetic-based foaming agents (AF and SF, respectively). The foam stability, micro characteristics, compressive strength, fluidity, and volume of water absorption of the FMLSB were investigated. The results showed that the foam made from AF had better strength and stability compared to SF. The internal pore sizes of both AF- and SF-FMLSB at low density were large, but at high density the internal pore sizes and area porosity of AF-FMLSB were smaller than those of SF-FMLSB. In terms of compressive strength, the compressive strength of AF-FMLSB was improved by 17.5% to 43.2% compared to SF-FMLSB. At low density, the fluidity of AF- and SF-FMLSB is similar, while at high density the fluidity of AF-FMLSB is much higher than that of SF-FMLSB. In addition, the stable volume of water absorption of SF-FMLSB is smaller than that of AF-FMLSB at low density, and the corresponding water resistance is better, but the situation is reversed at high density.

Interfacial Gated Graphene Photodetector with Broadband Response.

A much stronger interfacial gating effect was observed in the graphene/HfO2/Si photodetector when compared with that in the graphene/SiO2/Si photodetector. We found that this improvement was due to the higher interface state density at the HfO2/Si interface and the higher dielectric constant of the HfO2 layer. The photoresponsivity of the graphene/HfO2/Si photodetector is as high as 45.8 A W-1. Germanium and amorphous MoS2 (a-MoS2) were used to prepare graphene/HfO2/Ge and graphene/HfO2/a-MoS2 photodetectors, further demonstrating the high efficiency of the interfacial gating mechanism for photodetection. Because of the 0.196 eV bandgap of a-MoS2, which was verified in our previous report, the graphene/HfO2/a-MoS2 photodetector realized ultrabroadband photodetection over the range from 473 nm (visible) to 2712 nm (mid-infrared) at room temperature with photoresponsivity as high as 5.36 A W-1 and response time as fast as 68 µs, which represent significant improvements from the corresponding properties of the pure a-MoS2 photodetectors in our previous report and are comparable with those of state-of-the-art broadband photodetectors. By taking full advantage of the interfacial gating mechanism, a fast response, high photoresponsivity and ultrabroadband photodetection were achieved simultaneously. These interfacial gated graphene photodetectors also offer simple fabrication and full semiconductor process compatibility. The advantages described here indicate that the proposed photodetectors have significant potential for use in electronic and optoelectronic applications and offer a new path toward the development of ultrabroadband photodetectors.

Emission Enhancement from CdSe/ZnS Quantum Dots Induced by Strong Localized Surface Plasmonic Resonances without Damping.

A high-performance exciton-localized surface plasmon (LSP) coupling system consisting of well-designed plasmonic nanostructures and CdSe/ZnS quantum dots (QDs) was fabricated by first introducing a Ta2O5 layer as both an adhesive coating and coupling medium. It is shown that a larger emission enhancement factor of 6 from CdSe/ZnS QDs can be obtained from the strong coupling effect between QDs and triprism Au nanoarrays and the high scattering efficiency of LSPs without damping. This can be attributed to the matching conditions and a low extinction coefficient with little damping absorption of the Ta2O5 layer in the system. The radiative scattering rate of ΓLSPs can make a contribution to the spontaneous emission rate Γ and thus improve the internal quantum yield of the QDs. This strategy could be promising for practical application of metal-modified fluorescence enhancement.