Hybrid network with difference degree and attention mechanism combined with radiomics (H-DARnet) for MVI prediction in HCC

MVI is a risk assessment factor related to hepatocellular carcinoma (HCC) recurrence after hepatectomy or liver transplantation. The goal of this paper is to study the preoperative diagnosis of microvascular invasion (MVI) by using a deep learning algorithm in non-contrast T2 weighted magnetic resonance imaging (MRI) images instead of pathological images. Herein, an ensemble learning algorithm named H-DARnet—based on the difference degree and attention mechanism, combined with radiomics, for MVI prediction—is proposed. Our hybrid network combines the fine-grained, high-level semantic, and radiomics features and exhibits a rich multilevel-feature architecture composed of global-local-prior knowledge with suitable complementarity. The total loss function comprises two regularization items––the triplet and the cross-entropy loss function––which are selected for the triplet network and SE-DenseNet, respectively. The hard triplet sample selection strategy for a triplet network and data augmentation for small-scale liver image datasets in convolutional neural network (CNN) training is indispensable. For 200 patch level test samples (135 positive samples and 65 negative samples), our method can obtain the best prediction results, the AUC, sensitivity, and specificity were 0.826, 79.5% and 73.8%, respectively. The experiment results show that MVI can be predicted by using MRI images, and the proposed method is better than other deep learning algorithms and hand-crafted feature algorithms. The proposed ensemble learning algorithm is proved to be an effective method for MVI prediction.     

Tough and transparent blends of polylactide with block copolymers of ethylene glycol and propylene glycol

In order to modify its physical properties, particularly the drawability and toughness, polylactide (PLA) was melt blended with a set of PEG-b-PPG-b-PEG block copolymers with varying ratio of the hydrophilic (PEG) and hydrophobic (PPG) blocks. Miscibility of the copolymers with PLA depended on their molar mass and PEG content. Interestingly, the best drawability was achieved in the case of partially miscible blends, in which fine liquid inclusions of the modifier were dispersed in PLA rich matrix with glass transition temperature only moderately decreased, to about 50 °C. About 37 fold increase of the elongation at break and about 1.5 fold increase of the tensile impact strength with respect to neat PLA were reached at small content (10 wt.%) of the modifier. Despite the phase separation, the blends remained transparent. In addition, the barrier properties for oxygen were improved.     

High performance hybrid reinforcement of nitrile rubber using short pineapple leaf fiber and carbon black

Rubber composites with very high moduli at low elongation, high elongation at break and high ultimate breaking strength have been developed. The matrix was acrylonitrile butadiene rubber (NBR) and the hybrid (fibrous and particulate) reinforcements were short, fine pineapple leaf fiber (PALF) and carbon black. The amount of PALF was fixed at 10 parts (by weight) per hundred of rubber (phr) while that of carbon black was varied from 0 to 30 phr. Uniaxial NBR composites were prepared. Tensile strength, elongation at break, modulus and tear strength of the hybrid composites were characterized in both longitudinal (parallel to the fiber axis) and transverse (perpendicular to the fiber axis) directions. The addition of carbon black causes the slope of the early part of the stress–strain curve to increase and also extends breaking to greater strains. At carbon black contents of 20 phr and above, the stress–strain relation displays an upturn at high elongations, providing greater ultimate strength. Comparison with the usual carbon black filled rubber shows that the composite behavior at low strains is determined by the PALF, and at high strains by the carbon black. This high performance PALF-carbon black reinforced NBR shows great promise for engineering applications.     

壁面温度在多参数下对平板边界层的影响研究

为了得到壁面温度在不同来流速度、不同湍流强度条件下对边界层转捩与减阻的影响规律,本文采用Transitionk-kl-ω模型对低来流速度下无压力梯度的光滑平板进行了数值模拟。结果表明,随着来流速度的升高,壁温升高所起到的减阻效果更好,即高来流速度对壁面温度更为敏感。当来流处于中高湍流强度下时,壁温升高能起到推迟转捩的作用,且随着湍流强度的升高,转捩推迟的效果越好,但减阻效果正好相反;当来流处于低湍流强度下时,壁温升高会使得转捩提前发生。壁温升高抑制了边界层内流体的脉动程度,使得层流的稳态不易被破坏,流动更加稳定;同时,壁温升高使得边界层内流体的速度梯度减小,从而降低了壁面摩擦系数,故壁温升高能起到推迟边界层转捩与减阻的作用。     

原位合成Ag/Ag2MoO4纳米复合润滑剂对YSZ涂层摩擦学性能的影响

通过大气等离子喷涂工艺制备了氧化钇稳定氧化锆(YSZ)涂层，采用真空浸渍技术和水热合成的方法，将含有反应物离子或分子的前驱体溶液引入YSZ涂层内部固有的微裂纹和孔洞等缺陷中，并在缺陷中原位合成了直径约78~111 nm的Ag/Ag₂MoO₄类球形纳米颗粒，首次制备出了YSZ-Ag/Ag₂MoO₄复合涂层. 摩擦试验结果表明:与YSZ涂层相比，YSZ-Ag/Ag₂MoO₄复合涂层由于在室温和600 ℃下形成了润滑层，抑制了YSZ涂层摩擦表面的脆性断裂和磨粒磨损，从而显著降低了涂层的摩擦系数和磨损率，有效提高了涂层的摩擦学性能.      

Removal of oxygen from silicon by electron beam melting

Small amounts of multicrystalline silicon were melted in an electron beam furnace in different experimental conditions in order to investigate the oxygen evaporation behavior during the electron beam melting (EBM) process. The oxygen content level before and after EBM was determined by secondary ion mass spectroscopy. The oxygen content was reduced from 6.177 to 1.629 ppmw when silicon was melted completely at 15 kW with removal efficiency up to 73.6 %. After that, it decreased continually to <0.0517 ppmw when the refining time exceeded 600 s with a removal efficiency of more than 99.08 %. During the melting process, the evaporation rate of silicon is 1.10 × 10^?5 kg/s. The loss of silicon could be reduced up to 1.7 % during oxygen removal process to a desirable figure, indicating EBM is an effective method to remove oxygen from silicon and decrease the loss of silicon.     

Doped Zero-Dimensional Cesium Zinc Halides for High-Efficiency Blue Light Emission

All-inorganic zero-dimensional (0D) metal halides have recently received increasing attention due to their excellent photoluminescence (PL) performance and high stability. Herein, we present the successful doping of copper(I) into 0D Cs₂ZnBr₄. The incorporating of Cu⁺ cations enables the originally weakly luminescent Cs₂ZnBr₄ to exhibit an efficient blue emission centered at around 465 nm, with a high photoluminescence quantum yield (PLQY) of 65.3 %. Detailed spectral characterizations, including ultrafast transient absorption (TA) techniques, were carried out to investigate the effect of Cu⁺ dopants and the origin of blue emission in Cs₂ZnBr₄:Cu. To further study the role of the A-site cation and halogen, A₂ZnCl₄:Cu (A=Cs, Rb) were also synthesized and found to generate intense sky-blue emission (PLQY≈73.1 %). This work represents an effective strategy for the development of environmentally friendly, low-cost and high-efficiency blue-emitting 0D all-inorganic metal halides.     

Synthesis of the chain-shaped carbon fibers decorated with SiO<Subscript><Emphasis Type="Italic">x</Emphasis></Subscript> beads by chemical vapor deposition

Chain-shaped carbon fibers decorated with SiO _x beads had been synthesized by using methyltriehlorosilane and normal carbon fibers as the raw materials, during which the chemical vapor deposition technique was applied. The products were characterized by scanning electron microscopy (SEM), energy dispersive spectrum (EDS), X-ray diffraction (XRD) and Raman spectroscopy. The as-received products show chain-shaped morphologies that have the normal carbon fibers as “strings” and possess regular SiO _x beads of 30–40 μm with relative smooth surface. The formation of the chain-shaped morphology is considered to involve the shrinkage of the absorbed silica layer and the possible growth mechanism is further proposed. It is thought that the chain-shaped carbon fibers are formed by two main stages.     

紫外分光光谱法快速测定原料油中芳香烃的含量

应用紫外分光光谱技术建立了快速测定原料油中芳香烃含量的方法。分别以色谱分离法提纯的蜡油、柴油芳香烃组分为标准物,绘制标准曲线。实验表明蜡油和柴油中芳香烃的含量与吸光度具有良好的线性关系(相关系数r2分别为0.999 73,0.999 44),蜡油和柴油中芳烃的加标回收率分别为96.83%,98.97%。蜡油和柴油中芳烃测定结果的相对标准偏差分别为0.32%～0.98%,0.89%～1.52%(n=6)。该方法进行原料油可磺化有效组分的检测快速可靠,能更好地应用于实际生产。     

Identification of a fault zone ahead of the tunnel excavation face using the extended Kalman filter

Simulation of mechanized tunneling and on-site excavation require very good knowledge of the geomechanical and material properties. Identification of the material must be fast and continuously performed during tunnel excavation for the best possible strategies for advancing the tunnel boring machine. We present in this work the use of the extended Kalman filter (EKF) for identification of the inclined fault zone ahead of the face. The EKF showed fast and stable convergence of the model parameters under study. In comparison with the particle swarm optimization technique applied to the same back analysis problem, faster convergence of the identified parameters as well as high robustness with respect to the choice of the initial parameter values have been observed.