基于深度学习的秀丽隐杆线虫显微图像分割方法

利用计算机实现自动、准确的秀丽隐杆线虫(C.elegans)的各项形态学参数分析,至关重要的是从显微图像上分割出线虫体态,但由于显微镜下的图像噪声较多,线虫边缘像素与周围环境相似,而且线虫的体态具有鞭毛和其他附着物需要分离,多方面因素导致设计一个鲁棒性的C.elegans分割算法仍然面临着挑战。针对这些问题,提出了一种基于深度学习的线虫分割方法,通过训练掩模区域卷积神经网络(Mask R-CNN)学习线虫形态特征实现自动分割。首先,通过改进多级特征池化将高级语义特征与低级边缘特征融合,结合大幅度软最大损失(LMSL)损失算法改进损失计算;然后,改进非极大值抑制;最后,引入全连接融合分支等方法对分割结果进行进一步优化。实验结果表明,相比原始的Mask R-CNN,该方法平均精确率(AP)提升了4.3个百分点,平均交并比(mIOU)提升了4个百分点。表明所提出的深度学习分割方法能够有效提高分割准确率,在显微图像中更加精确地分割出线虫体。     

Recent progress of the gas diffusion layer in proton exchange membrane fuel cells: Material and structure designs of microporous layer

Proton exchange membrane fuel cells (PEMFCs) have become the most attractive power supply units for stationary and mobile applications. The operation, design characteristics, as well as performance of PEMFCs, are closely related to the multiphase transport of mass, heat, and electricity in the cell, a critical of which is the gas diffusion layer (GDL). It is very important to guarantee the transmission of water and gasses under high current density, and which is the weakness of PEMFCs at present. Microporous layer (MPL) is considered to be the key variable for mass transfer, so varieties of works focus on modification of MPL materials and its structure design. However, there is still a lack of special review to summarize and prospect the progress of MPL in recent years. This review article therefore focuses on the insights and comprehensive understanding of four critical issues of the MPL, the porosity, pore size distribution, wettability, structural design and the durability of MPL. At last, the conclusion and recommendations section summarized the future prospects and recommendations for possible research opportunities.     

Rate equation theory for the hydrogenation kinetics of Mg-based materials

A uniform solid product layer normally assumed in the shrinking-core model cannot predict the kinetic transition behavior of the H₂ adsorption reactions. In this study, the concept of a uniform solid product layer has been replaced by that of the inward growth of solid products on the solid surface. A rate equation is established to calculate the inward growth of the solid product and was implemented into the shrinking-core model to calculate the H₂ adsorption kinetics for various shapes of Mg-based materials. The prediction accuracy of the developed model is verified from the detailed experimental data. To account for the external gas diffusion around the particle and the intraparticle gas diffusion, an analytical equation is derived using the Thiele modulus method. This model can be used to analyze various kinetic aspects and to analyze the effect of change in the particle microstructure on intraparticle diffusion.     

Modeling the effect of deep traps on the capacitance–voltage characteristics of p-type Si-doped GaAs Schottky diodes grown on high index GaAs substrates

Numerical simulation, using SILVACO-TCAD, is carried out to explain experimentally observed effects of different types of deep levels on the capacitance–voltage characteristics of p-type Si-doped GaAs Schottky diodes grown on high index GaAs substrates. Two diodes were grown on (311)A and (211)A oriented GaAs substrates using Molecular Beam Epitaxy (MBE). Although, deep levels were observed in both structures, the measured capacitance–voltage characteristics show a negative differential capacitance (NDC) for the (311)A diodes, while the (211)A devices display a usual behaviour. The NDC is related to the nature and spatial distribution of the deep levels, which are characterized by the Deep Level Transient Spectroscopy (DLTS) technique. In the (311)A structure only majority deep levels (hole traps) were observed while both majority and minority deep levels were present in the (211)A diodes. The simulation, which calculates the capacitance–voltage characteristics in the absence and presence of different types of deep levels, agrees well with the experimentally observed behaviour.     

改进HSR-FCN的服装图像识别分类算法研究

目前网络上的服装图像数量增长迅猛，对于大量服装图像实现智能分类的需求日益增加。将基于区域的全卷积网络（Region-Based Fully Convolutional Networks，R-FCN）引入到服装图像识别中，针对服装图像分类中网络训练时间长、形变服装图像识别率低的问题，提出一种新颖的改进框架HSR-FCN。新框架将R-FCN中的区域建议网络和HyperNet网络相融合，改变图片特征学习方式，使得HSR-FCN可以在更短的训练时间内达到更高的准确率。在模型中引入了空间转换网络，对输入服装图像和特征图进行了空间变换及对齐，加强了对多角度服装和形变服装的特征学习。实验结果表明，改进后的HSR-FCN模型有效地加强了对形变服装图像的学习，且在训练时间更短的情况下，比原来的网络模型R-FCN平均准确率提高了大约3个百分点，达到96.69%。     

Prompting structure stability of O3–NaNi0.5Mn0.5O2 via effective surface regulation based on atomic layer deposition

Layered O3 type oxides exhibit promising prospects as high-performance cathodes for sodium-ion batteries (SIBs) due to their low cost and high theoretical capacities. Nevertheless, the intrinsic surface composition and bulk structure degradation upon cycling presents a huge obstacle to stable sodium-ion storage/transportation. Besides, the effective surface decoration on layered O3 oxides is still challenging through conventional wet chemical route owing to their extraordinarily high surface sensitivities. Herein, a typical O3 type layered oxide of NaNi_0.5Mn_0.5O₂ (NNMO) was selected and successfully encapsulated by precisely controlled Al₂O₃ layers via atomic layer deposition (ALD) technology. With the optimally controlled Al₂O₃ thickness of 3 nm, the surface regulated NNMO delivers a highly reversible capacity of 73.6 mA h g^-1, with a significantly improved capacity retention of 68.0% after 300 cycles at 0.5 C, and a superior rate capability of 65.8 mA h g^-1 at 10 C. Further air sensitivity tests demonstrate that the protective layer could effectively mitigate the generation of sodium-based impurities on NNMO, and reduce the surface sensitivities. Both chemical and electrochemical aging tests confirm the contribution of Al₂O₃ coating layer in alleviating ion dissolution as well as stabilizing the structure and morphology of NNMO. Based on regulating the surface of O3 type layered oxides utilizing ALD technique, this work supplies an effective and facile strategy to overcome the challenges from fast structure degradation and electrochemical property decay, which not only highlights the significance and effectiveness of surface engineering in secondary batteries, but also sheds light on accurate interface construction and regulation for active electrode materials, particularly for those sensitive to ambient atmosphere.     

Deep Learning Based Process Analytics Model for Predicting Type 2 Diabetes Mellitus

Process analytics is one of the popular research domains that advanced in the recent years. Process analytics encompasses identification, monitoring, and improvement of the processes through knowledge extraction from historical data. The evolution of Artificial Intelligence (AI)-enabled Electronic Health Records (EHRs) revolutionized the medical practice. Type 2 Diabetes Mellitus (T2DM) is a syndrome characterized by the lack of insulin secretion. If not diagnosed and managed at early stages, it may produce severe outcomes and at times, death too. Chronic Kidney Disease (CKD) and Coronary Heart Disease (CHD) are the most common, long-term and life-threatening diseases caused by T2DM. Therefore, it becomes inevitable to predict the risks of CKD and CHD in T2DM patients. The current research article presents automated Deep Learning (DL)-based Deep Neural Network (DNN) with Adagrad Optimization Algorithm i.e., DNN-AGOA model to predict CKD and CHD risks in T2DM patients. The paper proposes a risk prediction model for T2DM patients who may develop CKD or CHD. This model helps in alarming both T2DM patients and clinicians in advance. At first, the proposed DNN-AGOA model performs data preprocessing to improve the quality of data and make it compatible for further processing. Besides, a Deep Neural Network (DNN) is employed for feature extraction, after which sigmoid function is used for classification. Further, Adagrad optimizer is applied to improve the performance of DNN model. For experimental validation, benchmark medical datasets were used and the results were validated under several dimensions. The proposed model achieved a maximum precision of 93.99%, recall of 94.63%, specificity of 73.34%, accuracy of 92.58%, and F-score of 94.22%. The results attained through experimentation established that the proposed DNN-AGOA model has good prediction capability over other methods.     

黄沙坪铅锌多金属矿的成矿规律及深部找矿远景

众所周知,矿物质的成分是多种多样的,社会的发展和科学的进步需要运用到多重金属矿物质,在专业人员的勘探与挖掘中,发现了黄沙坪铅锌多金属矿,这个矿区中有丰富的有色金属,这个矿的发掘为研究成矿规律提供了物质基础,同时也为深部找矿提供了可靠的依据。本文主要分析黄沙坪铅锌多金属矿的成矿规律及深部找矿远景。     

Charge injection in metal/organic/metal structures with ZnO:Al/organic interface modified by Zn1−xMgxO:Al layer

Aluminum-doped zinc oxide (ZnO:Al, AZO) electrodes were covered with very thin (∼6 nm) Zn_1−xMg_xO:Al (AMZO) layers grown by atomic layer deposition. They were tested as hole blocking/electron injecting contacts to organic semiconductors. Depending on the ALD growth conditions, the magnesium content at the film surface varied from x = 0 to x = 0.6. Magnesium was present only at the ZnO:Al surface and subsurface regions and did not diffuse into deeper parts of the layer. The work function of the AZO/AMZO (x = 0.3) film was 3.4 eV (based on the ultraviolet photoelectron spectroscopy). To investigate carrier injection properties of such contacts, single layer organic structures with either pentacene or 2,4-bis[4-(N,N-diisobutylamino)-2,6-dihydroxyphenyl] squaraine layers were prepared. Deposition of the AMZO layers with x = 0.3 resulted in a decrease of the reverse currents by 1–2 orders of magnitude and an improvement of the diode rectification. The AMZO layer improved hole blocking/electron injecting properties of the AZO electrodes. The analysis of the current-voltage characteristics by a differential approach revealed a richer injection and recombination mechanisms in the structures containing the additional AMZO layer. Among those mechanisms, monomolecular, bimolecular and superhigh injection were identified.