R2D2: A scalable deep learning toolkit for medical imaging segmentation

Deep learning has gained a significant popularity in recent years thanks to its tremendous success across a wide range of relevant fields of applications, including medical image analysis domain in particular. Although convolutional neural networks (CNNs) based medical applications have been providing powerful solutions and revolutionizing medicine, efficiently training of CNNs models is a tedious and challenging task. It is a computationally intensive process taking long time and rare system resources, which represents a significant hindrance to scientific research progress. In order to address this challenge, we propose in this article, R2D2, a scalable intuitive deep learning toolkit for medical imaging semantic segmentation. To the best of our knowledge, the present work is the first that aims to tackle this issue by offering a novel distributed versions of two well-known and widely used CNN segmentation architectures [ie, fully convolutional network (FCN) and U-Net]. We introduce the design and the core building blocks of R2D2. We further present and analyze its experimental evaluation results on two different concrete medical imaging segmentation use cases. R2D2 achieves up to 17.5× and 10.4× speedup than single-node based training of U-Net and FCN, respectively, with a negligible, though still unexpected segmentation accuracy loss. R2D2 offers not only an empirical evidence and investigates in-depth the latest published works but also it facilitates and significantly reduces the effort required by researchers to quickly prototype and easily discover cutting-edge CNN configurations and architectures.     

Energy harvesting with pairs of variable capacitors without control circuits

This work describes several electronic electrostatic generators that can be built using two pairs of complementary variable capacitors, in a way that dispenses control circuits for their operation. In all cases, a basic unstable generator based on an electrostatic charge multiplier is used to bias variable capacitors, and the current generated by these capacitors is then rectified and sent to the load. Experimental versions of the proposed generators were built using variable capacitors made with 3D-printing techniques, moved in a back-and-forth way by a small motor. As these generators operate at high impedance level, always with small currents, they are insensitive to the resistivity of the conductive plastic used to make the variable capacitors. Several new structures are presented, and their properties are evaluated theoretically and experimentally.     

Dual‐Layer Hollow Fibers for Sulfur Removal from Fuels

Emission of sulfur compounds to the atmosphere is universally recognized as one key target to be reduced. For membrane pervaporation which is considered as a potential purification process of fuels, dual‐layer polyurethane (PU)/polyethersulfone hollow‐fiber membranes were prepared. A novel fabrication technique is proposed using a quadruple spinneret to produce the fiber with such morphology by simultaneous spinning of two polymer solutions in the presence of two corresponding precipitation media. Activated carbon was added into the PU solution to improve the transport properties of the selective layer. Resulting hollow‐fiber membranes showed very good adhesion between the selective layer and its support, in addition to an effective removal of a sulfur compound such as 2‐methyl thiophene from a typical model fuel, an indication of good prospects for both the fabrication technique and for sulfur removal by pervaporation of fuels.     

Silicene Nanoribbons on an Insulating Thin Film

Silicene, a new 2D material has attracted intense research because of the ubiquitous use of silicon in modern technology. However, producing free-standing silicene has proved to be a huge challenge. Until now, silicene could be synthesized only on metal surfaces where it naturally forms strong interactions with the metal substrate that modify its electronic properties. Here, the authors report the first experimental evidence of silicene nanoribbons on an insulating NaCl thin film. This work represents a major breakthrough, for the study of the intrinsic properties of silicene, and by extension to other 2D materials that have so far only been grown on metal surfaces.     

Influence of pressurised cryogenic nitrogen technology on Arthrospira platensis (spirulina) preservation during storage

In this study, preservation of spirulina using the new pressurised cryogenic nitrogen technology (PCN) was compared to classical methods used in laboratories and industry. Spirulina morphology was better preserved by PCN compared to unpressurised cryogeny and classical freezing at −20 °C that led to cells fragmentation. A 25% loss of Phycocyanin-C content against 60% were measured after 98 days storage for 6-Bar PCN process and frozen samples, respectively. The Total AntiOxidant Power (PAOT Liquid Technology^®) was used for determination of total antioxidant and oxidant power of spirulina extracts. PAOT value of PCN samples was 50% higher than the frozen sample. From ABTS measurements on PCN spirulina fractions sonicated or not, it was suggested that pressurisation at 6 bars allowed a better preservation of free antioxidants (outside the cells) due to replacement of oxygen by nitrogen in the frozen beads. After dehydration, phycocyanin-C content variation during storage at 20 °C and 33% RH showed higher loss for freeze-dried spirulina treated at 0 compared to 6 Bars.     

Expanding the Toolbox of R-Selective Amine Transaminases by Identification and Characterization of New Members

Amine transaminases (ATAs) are used to synthesize enantiomerically pure amines, which are building blocks for pharmaceuticals and agrochemicals. R-selective ATAs belong to the fold type IV PLP-dependent enzymes, and different sequence-, structure- and substrate scope-based features have been identified in the past decade. However, our knowledge is still restricted due to the limited number of characterized (R)-ATAs, with additional bias towards fungal origin. We aimed to expand the toolbox of (R)-ATAs and contribute to the understanding of this enzyme subfamily. We identified and characterized four new (R)-ATAs. The ATA from Exophiala sideris contains a motif characteristic for d -ATAs, which was previously believed to be a disqualifying factor for (R)-ATA activity. The crystal structure of the ATA from Shinella is the first from a Gram-negative bacterium. The ATAs from Pseudonocardia acaciae and Tetrasphaera japonica are the first characterized (R)-ATAs with a shortened/missing N-terminal helix. The active-site charges vary significantly between the new and known ATAs, correlating with their diverging substrate scope.     

Pairwise testing for software product lines: comparison of two approaches

Software Product Lines (SPL) are difficult to validate due to combinatorics induced by variability, which in turn leads to combinatorial explosion of the number of derivable products. Exhaustive testing in such a large products space is hardly feasible. Hence, one possible option is to test SPLs by generating test configurations that cover all possible t feature interactions (t-wise). It dramatically reduces the number of test products while ensuring reasonable SPL coverage. In this paper, we report our experience on applying t-wise techniques for SPL with two independent toolsets developed by the authors. One focuses on generality and splits the generation problem according to strategies. The other emphasizes providing efficient generation. To evaluate the respective merits of the approaches, measures such as the number of generated test configurations and the similarity between them are provided. By applying these measures, we were able to derive useful insights for pairwise and t-wise testing of product lines.     

The design and evaluation of a tool to support software designers at the whiteboard

Despite the availability of advanced software design tools, software developers are known to frequently turn to the whiteboard when faced with a specific design problem. Free from any restrictions, they engage in an informal form of software design that relies heavily on sketching. However, whereas whiteboards afford flexibility and fluidity, they at the same time limit a designer in only being able to draw and erase content. This paper presents Calico, a novel software design tool that leverages electronic whiteboards to enhance the design experience with explicit support for the creative, exploratory aspects of design. Specifically, Calico offers a grid to partition the drawing space, scraps to organize and manipulate the content, and a reusable palette together with gesture-based input to address several natural behaviors exhibited by software designers, including frequent shifts in focus, use of low-detail models, use of a variety of notations, and a tendency to refine representations. To evaluate Calico, we performed a laboratory experiment involving the design of an educational traffic simulator. Through a three-pronged analysis examining the use of Calico’s features, design behaviors exhibited by the participants, and the nature of the design conversations, we show how Calico benefits the design process as it takes place at the whiteboard.     

Evaluation of temperature field and heat flux by inverse analysis during steel strip rolling

Knowledge of the temperature field in the roll is a critical factor of modern, high-speed rolling mills. In this paper, an inverse analytical method is developed to determine the temperature field and especially the temperature (and heat flux) at the surface of the roll by measuring the temperature with a thermocouple (fully embedded) at only one point inside the roll. Iterative methods are not studied because short computation times are desired. Some assumptions are done to resolve analytically the unsteady heat equation, taking into account the restrictions of the measurement system (e.g., measurement according to successive times). The solution is validated by comparing the outputs of the method and prescribed analytical temperature fields. Good agreement is obtained. Noise sensitivity is estimated by adding artificial random numbers to the inputs. Good accuracy is observed. A 10% error of the temperature sensor depth is also considered and does not compromise the method. On the other hand, the computation time (around 0.05 s by cycle) is studied to rapidly optimise the industrial parameters during the rolling process.