Deep chest X‐ray: Detection and classification of lesions based on deep convolutional neural networks

We investigated whether a convolutional neural network (CNN) can enhance the usability of computer‐aided detection (CAD) of chest radiographs for various pulmonary abnormal lesions. The numbers of normal and abnormal patients were 6055 and 3463, respectively. Two radiologists delineated regions of interest for lesions and labeled the disease types as ground truths. The datasets were split into training, tuning, and testing as 7:1: 2. Total test sets were randomly selected in 1214 normal and 690 abnormal. A 5‐fold, cross‐validation was performed on our datasets. For the classification of normal and abnormal, we developed a CNN based on DenseNet169; for abnormal detection, The You Only Look Once (YOLO) v2 with DenseNet was used. Detection and classification of normal and five classes of diseases (nodule[s], consolidation, interstitial opacity, pleural effusion, and pneumothorax) on chest radiographs were analyzed. Our CNN model classified chest radiographs as normal or abnormal with an accuracy of 97.8%. For the results of the abnormal, F1 score, was 75.2 ± 2.28% for nodules, 55.0 ± 4.3% for consolidation, 78.2 ± 7.85% for interstitial opacity, 81.6 ± 2.07% for pleural effusion, and 70.0 ± 7.97% for pneumothorax, respectively. In addition, we conducted the experiments between our method and RetinaNet with only nodules. The results of our method and RetinaNet at cutoff‐0.5 in the free response operating characteristic curve were 83.45% and 80.55%, respectively. Our algorithm demonstrated viable detection and disease classification capacity and could be used for CAD of lung diseases on chest radiographs.     

Phase Engineering of Transition Metal Dichalcogenides with Unprecedentedly High Phase Purity,Stability, and Scalability via Molten-Metal-Assisted Intercalation

The crystalline phase of layered transition metal dichalcogenides (TMDs) directly determines their material property. The most thermodynamically stable phase structures in TMDs are the semiconducting 2H and metastable metallic 1T phases. To overcome the low phase purity and instability of 1T-TMDs, which limits the utilization of their intrinsic properties, various synthesis strategies for 1T-TMDs have been proposed in phase-engineering studies. Herein, a facile and scalable synthesis of 1T-phase molybdenum disulfide (MoS₂) via the molten-metal-assisted intercalation (MMI) approach is introduced, which exploits the capillary action of molten potassium and the difference between the electron affinity of MoS₂ and the ionization potential of potassium. Highly reactive molten potassium metal can readily intercalate into the MoS₂ interlayers, inducing an efficient phase transition from the 2H to 1T crystal structure. The ionic bonding between the intercalated potassium and sulfur lowers the energy barrier of the 1T-phase transition, enhancing the phase stability of the 1T crystals. Owing to the high purity and stability of the 1T phase, the electrocatalytic performance for the hydrogen evolution reaction is significantly higher in 1T-MoS₂ (MMI) than in 2H-MoS₂ and even in 1T-MoS₂ synthesized using n-butyllithium.     

Origins of High Performance and Degradation in the Mixed Perovskite Solar Cells

The origins of the high device performance and degradation in the air are the greatest issues for commercialization of perovskite solar cells. Here this study investigates the possible origins of the mixed perovskite cells by monitoring defect states and compositional changes of the perovskite layer over the time. The results of deep‐level transient spectroscopy analysis reveal that a newly identified defect formed by Br atoms exists at deep levels of the mixed perovskite film, and its defect state shifts when the film is aged in the air. The change of the defect state is originated from loss of the methylammonium molecules of the perovskite layer, which results in decreased J_SC, deterioration of the power conversion efficiency and long‐term stability of perovskite solar cells. The results provide a powerful strategy to diagnose and manage the efficiency and stability of perovskite solar cells.     

关于曲线的L（2，1）-标注

假设曲线G=(V，E)，G的L(2，1)-标注是方程式f：V(G)-[0，∞]，那么如果(x，y)EE，则f(x)-f(y)1≥2，如果dc(x，y)=2，则|f(x)-f(y)|≥1，此处的dc(x，y)是G曲线中x和y之间最短的距离。L(2，1)-标注数字λ(G)是最小数字m，那么G则有最大{f(v)|v∈V}=m的L(2，1)-标注f。格里戈斯和叶[6]及山凯[2]曾通过各种曲线对这个问题进行过研究。本文中我们提高了弦曲线λ(G)的已知上界并提供了曲线λ(G)的第一个上界。     

Integrated design of workcells and unidirectional flowpath layout

In manufacturing cells layout design with a unidirectional flow system, the accurate distance between two workcells can be uncovered with both the determination of IO port locations after the layout design of the cell with its orientation and the unidirectional flowpath layout design. This paper presents the method to obtain a global solution for manufacturing workcells and unidirectional flowpath layout design (ICFLD) with consideration of IO ports of workcells. The flow distance between two workcells is calculated from output port of one workcell to input port of the other workcell through the unidirectional flowpath layout. A zero-one integer programming model is developed for the ICFLD problem. And a heuristic algorithm for the ICFLD problem is developed by decomposing the ICFLD problem into two subproblems and iteratively and alternately solving the decomposed subproblems. Computational experiments are performed and its results are analyzed.     

Controlling general multi-echelon distribution supply chains with improved reorder decision policy utilizing real-time shared stock information

The objective of this paper is to introduce an improved reorder decision policy for controlling general multi-echelon distribution systems utilizing shared stock information. Since traditional reorder policies sometimes show poor performances when applied to distribution systems, the order risk policy, which utilizes shared stock information more accurately, was developed for the two-echelon distribution system in the previous research. In this paper, the order risk policy is extended to general multi-echelon systems. Since the calculation of the exact order risk for general multi-echelon systems is complex, a practical approximation method is developed. Through computational experiments, the superior performance of the order risk policy, compared to the existing reorder policies, is shown and the value of shared stock information varying with the characteristics of the supply chain is analyzed.     

Development of a catalytic combustor with heat exchanger

Catalytic combustion is thought to be a considerable improvement on the traditional one under specific conditions. Due to its special features, catalytic combustion has two strong points compared to flame: no NO_x emission and high reaction efficiency. However, the preheating process of catalytic combustion is an obstacle that deteriorates profitability in operation. So the HTHE (High Temperature Heat Exchanger) is adapted to the system to reinforce the preheating process, and we show that the catalytic combustion is maintained steadily without exceptional heat injection. As a result, the stability on the catalytic surface is the most important operational factor. To achieve it, both mixture gas property and temperature distribution should be controlled.     

J-substitution algorithm in magnetic resonance electrical impedance tomography (MREIT): phantom experiments for static resistivity images

Recently, a new static resistivity image reconstruction algorithm is proposed utilizing internal current density data obtained by magnetic resonance current density imaging technique. This new imaging method is called magnetic resonance electrical impedance tomography (MREIT). The derivation and performance of J-substitution algorithm in MREIT have been reported as a new accurate and high-resolution static impedance imaging technique via computer simulation methods. In this paper, we present experimental procedures, denoising techniques, and image reconstructions using a 0.3-tesla (T) experimental MREIT system and saline phantoms. MREIT using J-substitution algorithm effectively utilizes the internal current density information resolving the problem inherent in a conventional EIT, that is, the low sensitivity of boundary measurements to any changes of internal tissue resistivity values. Resistivity images of saline phantoms show an accuracy of 6.8%-47.2% and spatial resolution of 64 x 64. Both of them can be significantly improved by using an MRI system with a better signal-to-noise ratio.     

Solidification behaviors of the emulsified Al-4wt%Fe alloy powders

The emulsified Al-4wt%Fe alloy powders showed several different microstructures depending on the amount of undercooling such as Al₁₃Fe₄, Al _x Fe primary intermetallic, Al-Al₃Fe or Al-Al₆Fe eutectics and -Al cellular structure. The presence of these phases depends on a competitive growth mechanism, which was determined by the undercooling prior to solidification. The amount of undercooling of the powders was monitored by differential thermal analysis and was matched with the microstructures. The difference of the amount of undercooling which corresponds to the microstructures between the experiment and the previous value converted by the LKT theory was rationalized through the solidification behaviors. A microstructure selection map of Al-4Fe alloy powders for tailored solidification was also suggested.