Search

Currently, convolutional neural networks (CNNs) have made remarkable achievements in skin lesion classification because of their end-to-end feature representation abilities. However, precise skin lesion classification is still challenging because of the following three issues: (1) insufficient training samples, (2) inter-class similarities and intra-class variations, and (3) lack of the ability to focus on discriminative skin lesion parts. To address these issues, we propose a deep metric attention learning CNN (DeMAL-CNN) for skin lesion classification. In DeMAL-CNN, a triplet-based network (TPN) is first designed based on deep metric learning, which consists of three weight-shared embedding extraction networks. TPN adopts a triplet of samples as input and uses the triplet loss to optimize the embeddings, which can not only increase the number of training samples, but also learn the embeddings robust to inter-class similarities and intra-class variations. In addition, a mixed attention mechanism considering both the spatial-wise and channel-wise attention information is designed and integrated into the construction of each embedding extraction network, which can further strengthen the skin lesion localization ability of DeMAL-CNN. After extracting the embeddings, three weight-shared classification layers are used to generate the final predictions. In the training procedure, we combine the triplet loss with the classification loss as a hybrid loss to train DeMAL-CNN. We compare DeMAL-CNN with the baseline method, attention methods, advanced challenge methods, and state-of-the-art skin lesion classification methods on the ISIC 2016 and ISIC 2017 datasets, and test its generalization ability on the PH2 dataset. The results demonstrate its effectiveness.

In the past few years, biomass pyrolysis has attracted interest as a promising way to convert biomass such as cellulose into valuable resources. However, a comprehensive conversion process is still difficult to realize because of the complex physical and chemical processes of cellulose pyrolysis. In this study, a comprehensive model of heat transfer in cellulose pyrolysis was developed. The model includes detailed kinetic reaction scheme and heat-transfer description, accounting for the evolution of the physical properties and condensation of volatiles, which results from mass-transport limitations. The experiments were conducted in a self-developed apparatus by varying the thickness, void fraction, and heating rate of the samples. The predicted results of different samples well agreed with the experimental data. This model enables predicting the product distribution of cellulose pyrolysis and in turn can be used to assess the feasibility, design, and scalability of industrial conversion applications.

A simple method was developed to prepare tannic acid-decorated spongy graphene hydrogel (TAGH) as interconnected, three-dimensional (3D) porous network supercapacitor electrode materials. Tannic acid (TA) not only can be fully competent in reducing graphene oxide and stabilizing reduced graphene oxide, but also can prevent graphene sheets from agglomerating by suppressing the p-p stacking interactions. Such 3D porous TAGH electrodes with low charge transfer resistance of 0.4 O.cm2 displayed a maximum specific capacitance of 533 F/g at 0.5 A/g. The assembled TAGH supercapacitor showed a good cycling stability (84.6% capacitance retention after 8000 charging/discharging cycles) and mechanical flexibility (93.4% capacitance retention after 500 folding/unfolding cycles). Remarkably, the flexible device can achieve superior energy density of 37.0 Wh/kg and higher power density of 4476.7 W/kg. These results demonstrate a great potential for application of 3D TAGH electrodes in flexible, high performance wearable electronics and energy storage devices.

Conductive hydrogels have good prospects in the fields of flexible electronic devices and artificial intelligence due to their biocompatibility, durability, and functional diversity. However, the process of hydrogel polymerization is time-consuming and energy-consuming, and freezing at zero temperature is inevitable, which seriously hinders its applications and working life. Herein, zwitterionic conductive hydrogels with self-adhesive and antifreeze properties were prepared in one minute by introducing two-dimensional (2D) MXene nanosheets into the autocatalytically enhanced system composed of tannic acid-modified cellulose nanofibers and zinc chloride. The system has strong environmental applicability (−60 to 40 °C), good stretchability (ductility ≈ 980%), durable adhesion (even after 30 days of exposure to air), and strong electrical conductivity (20 °C, 30 mS cm–1). By virtue of these advantages, the prepared zwitterionic hydrogels can be developed into flexible strain sensors to monitor large human movements and subtle physiological signals over a wide temperature range and to capture signals from handwriting and voice recognition. In addition, multiple flexible sensors can be assembled into a three-dimensional (3D) array, which can detect the magnitude and spatial distribution of strain or force. These results demonstrate that the prepared zwitterionic hydrogels have promising applications in the fields of medical monitoring and artificial intelligence.

Although chronic spontaneous urticaria (CSU) is a common disease, GWASs of CSU are lacking. We aimed to identify susceptibility SNPs by performing a GWAS in Chinese Han adults with CSU. The discovery cohort included 430 CSU cases and 482 healthy controls. The GWAS findings were validated in 800 CSU cases and 900 healthy controls. Genetic, functional enrichment, and bioinformatic analyses of genome-wide significant SNPs were performed to assess the association between CSU and autoimmunity or atopy. Five genome-wide significant SNPs were identified: rs434124/LILRA3, rs61986182/IGHG1/2, rs73075571/TDGF1, rs9378141/HLA-G, and rs3789612/PTPN22. The first four SNPs were in linkage disequilibrium with autoimmune-related diseases‒associated SNPs and were cis-expression quantitative trait loci in immune cells. The five SNPs-annotated genes were significantly enriched in immune processes. Higher polygenic risk scores and allele frequencies of rs3789612∗T, rs9378141∗C, and rs73075571∗G were significantly associated with autoimmune-related CSU phenotypes, including positive antithyroglobulin IgG, positive anti-FcεRIα IgG, total IgE <40 IU/ml, and positive antithyroid peroxidase IgG but not with atopic or allergic sensitized CSU phenotypes. This GWAS of CSU identifies five risk loci and reveals that CSU shares genetic overlap with autoimmune diseases and that genetic factors predisposing to CSU mainly manifest through associations with autoimmune traits.

A simple method was developed to prepare tannic acid-decorated spongy graphene hydrogel (TAGH) as interconnected, three-dimensional (3D) porous network supercapacitor electrode materials. Tannic acid (TA) not only can be fully competent in reducing graphene oxide and stabilizing reduced graphene oxide, but also can prevent graphene sheets from agglomerating by suppressing the π-π stacking interactions. Such 3D porous TAGH electrodes with low charge transfer resistance of 0.4 Ω.cm2displayed a maximum specific capacitance of 533 F/g at 0.5 A/g. The assembled TAGH supercapacitor showed a good cycling stability (84.6% capacitance retention after 8000 charging/discharging cycles) and mechanical flexibility (93.4% capacitance retention after 500 folding/unfolding cycles). Remarkably, the flexible device can achieve superior energy density of 37.0 Wh/kg and higher power density of 4476.7 W/kg. These results demonstrate a great potential for application of 3D TAGH electrodes in flexible, high performance wearable electronics and energy storage devices.