Search

Diabetic retinopathy (DR) is an important cause of blindness worldwide. However, DR is hard to be detected in the early stages, and the diagnostic procedure can be time-consuming even for the experienced experts. Therefore, a computer-aided diagnosis method based on deep learning algorithms is proposed to automatedly diagnose the referable diabetic retinopathy by classifying color retinal fundus photographs into two grades. In this paper, a novel convolutional neural network model with the Siamese-like architecture is trained with a transfer learning technique. Different from the previous works, the proposed model accepts binocular fundus images as inputs and learns their correlation to help to make a prediction. In the case with a training set of only 28 104 images and a test set of 7024 images, an area under the receiver operating curve of 0.951 is obtained by the proposed binocular model, which is 0.011 higher than that obtained by the existing monocular model. To further verify the effectiveness of the binocular design, a binocular model for five-class DR detection is also trained and evaluated on a 10% validation set. The result shows that it achieves a kappa score of 0.829 which is higher than that of the existing non-ensemble model.

In this paper, we present a novel one-dimensional deep convolutional neural network (1D-DCNN) with a multi-label-way-based algorithm for comprehensively and automatically extracting features and classifying mixture gases. Although a number of pattern recognition methods have been used to analyze the mixed gases, the performances of these methods highly depend on the hand-crafted feature engineering. By contrast, the proposed implementation, based on one-dimensional convolution, is capable of automatically extracting features and distinguishing the individual component of binary mixture gases composed of ethylene, CO, and methane. To the best of our knowledge, the proposed 1D-DCNN algorithm is first applied in the mixture gases’ recognition. In addition, the proposed 1D-DCNN with multi-label way not only significantly reduces the label dimension but also quantifies the probability of each component in mixed gases. Compared with the conventional pattern recognition algorithms including support vector machine, artificial neural network, k-nearest neighbor, and random forest, the proposed 1D-DCNN exhibits a higher recognition accuracy (96.30%) based on our extensive experimental results using ten-fold cross validation.

To keep pace with the upcoming big-data era, the development of a device-level neuromorphic system with highly efficient computing paradigms is underway with numerous attempts. Synaptic transistors based on an all-solution processing method have received growing interest as building blocks for neuromorphic computing based on spikes. Here, we propose and experimentally demonstrated the dual operation mode in poly{2,2-(2,5-bis(2-octyldodecyl)-3,6-dioxo-2,3,5,6-tetrahydropyrrolo[3,4-c]pyrrole-1,4-diyl)dithieno[3,2-b]thiophene-5,5-diyl-alt-thiophen-2,5-diyl}(PDPPBTT)/ZnO junction-based synaptic transistor from ambipolar charge-trapping mechanism to analog the spiking interfere with synaptic plasticity. The heterojunction formed by PDPPBTT and ZnO layers serves as the basis for hole-enhancement and electron-enhancement modes of the synaptic transistor. Distinctive synaptic responses of paired-pulse facilitation (PPF) and paired-pulse depression (PPD) were configured to achieve the training/recognition function for digit image patterns at the device-to-system level. The experimental results indicate the potential application of the ambipolar transistor in future neuromorphic intelligent systems.

The scaled boundary finite element method (SBFEM) is introduced for the investigation of the liquid sloshing in an annular cylindrical container with the coaxial porous structures. The main advantages of the SBFEM for this problem is that only the outer wall of the annular tank is discretized while the inner wall and the porous structures need not be treated, which is not only convenient for the generation of mesh, but also reduces the spatial dimension by one. Meanwhile, the solutions of the velocity potential are analytical in the radical direction of the scaled boundary coordinate system. In this paper, two types of the porous structures, that are, the circular and the arc-shaped porous structures, are taken into account. For the arc-shaped system, a virtual circular by extending the arc-shaped structure porous is introduced and one can set the porous-effect parameter of the virtual section to infinity, so that the whole flow domain can be divided into two sub-domains by the porous structure as the same approach with the circular system. By the assumption of the incompressible, inviscid, and irrotational flow and using the variational principle, the SBFEM governing equations with respect to the radical coordinate of the velocity potential in each sub-domain are obtained. Then, the governing equations can be solved analytically by introducing the Bessel functions and the modified Bessel functions as the base solutions. The excellent efficiency and accuracy of the proposed formulations are demonstrated by comparing the SBFEM numerical results with the analytical solutions. In addition, the effects of the different parameters for sloshing characteristics, such as the porous-effect parameter, radius of porous structure, standing wave number, opening degree and location of the arc-shaped porous structure are further studied.

In this paper, we present a novel K times singular value decomposition (K-SVD) based denoising algorithm dedicated to the division-of-focal-plane (DoFP) polarization image sensors. The proposed method is based on sparse representation over trained dictionary. Using the proposed K-SVD algorithm to update the dictionary, the image content can be more effectively expressed. Compared with the previous denoising algorithms, the proposed implementation is capable of decomposing the input DoFP image as the optimum sparse combination of the dictionary elements, which are generated by orthogonal matching pursuit. This not only separates the Gaussian noise from the target DoFP image with a significantly elevated peak signal-to-noise ratio (PSNR) but also well-preserves ythe details of the original image. According to our extensive experimental results on various test images, the proposed algorithm outperforms the state-of-the-art principal component analysis based denoising algorithm by 3 dB in terms of PSNR. Moreover, visual comparison results, which show excellent agreement with the PSNR results, are presented as well.

Conventional capacitance-to-digital converters (CDCs) suffer limitations either on narrow capacitance range or low resolution for jitter-induced noise and high power consumption. In order to avoid these limitations, a 13-b 1 pF–10 nF generic CDC is presented. In the proposed CDC with the oversampled $\Delta \Sigma $ modulation, the zero-crossing-based circuits (ZCBCs) are used to replace the operational transconductance amplifier to avoid feedback loop stability issues. However, the ZCBCs inevitably incur the non-idealities and thus, a novel calibration scheme is presented for efficient non-ideality-error cancellation. In addition, for the purpose of making the proposed CDC sufficiently intelligent to adapt to a wide capacitance-sensing range, an adaptive auto-range mechanism is proposed. The above three techniques complement each other and work as a whole leading to the proposed CDC with wide range, high resolution, high linearity, and low power consumption. A prototype fabricated using 0.18- $\mu \text{m}$ CMOS technology is experimentally verified using a MEMS capacitive humidity sensor. The measurement results show that the CDC achieves a 13-b root-mean-square noise equivalent resolution with a 128- $\mu \text{s}$ conversion time and a 230 fJ/conversion-step figure of merit. The calibration scheme enhances the linearity from 7 to 11.4 b in the 1 pF–10 nF compatible capacitance range.

The interfacial electronic structure of layer-by-layer 4-(dicyanomethylene)-2-t-butyl-6-(1,1,7,7-tetramethyljulolidyl-9-enyl)-4H-pyran (DCJTB)/fullerene C 60 was investigated using ultra-violet photoemission spectroscopy (UPS). Photoemission data of period of DCJTB/C 60 /DCJTB films suggested the formation of surface dipole and interfacial band bending across the interfaces, which greatly facilitates the charge transfer from DCJTB to C 60 and from C 60 to DCJTB layer as well. When applied this layer-by-layer structure to a near-infrared photodetector, a maximum of photocurrent was achieved by the device with 3 periods of DCJTB/C 60 thin films. Finally, the detailed work mechanism of this detector was discussed.

In the daily life, there is a tremendous need of fast monitoring of mixture gases and corresponding concentration estimation. For example, in the petrochemical industry, toxic and flammable gases, such as carbon monoxide, methanol and ethanol always co-exist. Without fast and accurate gas species identification and concentration detection system, great security risks will exist. Here, we propose to use an electronic nose system consisting of attention mechanism equipped recurrent neural network with double loss function (2L-ARNN) algorithm to achieve above tasks simultaneously. Firstly, an end-to-end encoder and decoder frame is applied to offer the flexibility to process variable length of the input. Then a novel gated recurrent unit network facilitates the automatic feature extraction from temporal dynamic behaviour. Based on this, attention mechanism dynamically assigns weight vectors of the gas features, which strengthens the robustness for the concentration fluctuation scenarios. Finally, with double loss function, both targets of mixture gas identification and concentration calculation are achieved using the same network. Consequently, within 5 s, 2L-ARNN algorithm has probably an accuracy of 97.67 % for air, CO, ethylene and methane mixture gases, which is significantly improved compared to classical methods. Simultaneously, the estimated normalized root mean squared errors for CO, ethylene, methane are 6.20 %,7.86 % and 5.71 %, which is 6% smaller than that of convolutional neural network. Therefore, our proposed algorithm features fast processing rate, superior accuracy for both gas identification and concentration estimation with improved anti-interference capability for wide-range of mixture gases conditions are essential for real-time gas monitoring systems.

Micro- and nano-satellites are attractive due to the low development and launching costs. Carried by micro- and nano-satellites, synthetic aperture radars (SARs) have great potential for urban, oceanography, land use, and agriculture usages. Different than conventional satellites, the payload of micro- and nano-satellites is limited. This imposes great challenges on the SAR system design. Traditional SARs adopt thousands of millimeter-wave integrated circuit (MMIC) components; they are bulky and power hungry. This paper investigates the feasibility to deploy the emerging technologies to shrink the volume, reduce the cost, and improve the power efficiency of the SAR system. Highly integrated radar transceiver integrated circuits (ICs) are reviewed, and an SAR transceiver IC is developed. Design technologies of low-noise amplifiers (LNAs) and high-power amplifiers (HPAs) are compared, and Gallium Nitride (GaN) technology is proposed. A novel microelectromechanical system-based delay line is also pro...

Spartina alterniflora (Spartina) is the only halophyte in the salt marsh. However, the molecular basis of its high salt tolerance remains elusive. In this study, we used Pacific Biosciences (PacBio) full-length single-molecule long-read sequencing and RNA-seq to elucidate the transcriptome dynamics of high salt tolerance in Spartina by salt gradient experiments. High-quality unigenes, transcription factors, non-coding RNA and Spartina-specific transcripts were identified. Co-expression network analysis found that protein kinase-encoding genes (SaOST1, SaCIPK10 and SaLRRs) are hub genes in the salt tolerance regulatory network. High salt stress induced the expression of transcription factors but repressed the expression of long non-coding RNAs. The Spartina transcriptome is closer to rice than Arabidopsis, and a higher proportion of transporter and transcription factor-encoding transcripts have been found in Spartina. Transcriptome analysis showed that high salt stress induced the expression of carbohydrate metabolism, especially cell-wall biosynthesis-related genes in Spartina, and repressed its expression in rice. Compared with rice, high salt stress highly induced the expression of stress response, protein modification and redox-related gene expression and greatly inhibited translation in Spartina. High salt stress also induced alternative splicing in Spartina, while differentially expressed alternative splicing events associated with photosynthesis were overrepresented in Spartina but not in rice. Finally, we built the SAPacBio website for visualizing full-length transcriptome sequences, transcription factors, ncRNAs, salt-tolerant genes and alternative splicing events in Spartina. Overall, this study suggests that the salt tolerance mechanism in Spartina is different from rice in many aspects and is far more complex than expected.

Division of focal plane (DoFP) polarimeter is widely used in polarization imaging sensors. The periodically arranged micro-polarizers integrated on the focal plane ensure its outstanding real-time performance, but reduce the spatial resolution of output images and further affect the calculation of polarization parameters. In this paper, a four-layer, end-to-end fully convolutional neural network called Fork-Net is proposed, which aims to directly improve the imaging quality of three polarization properties: intensity (i.e., S

Motivation Single-cell RNA-sequencing (scRNA-seq) is fast and becoming a powerful technique for studying dynamic gene regulation at unprecedented resolution. However, scRNA-seq data suffer from problems of extremely high dropout rate and cell-to-cell variability, demanding new methods to recover gene expression loss. Despite the availability of various dropout imputation approaches for scRNA-seq, most studies focus on data with a medium or large number of cells, while few studies have explicitly investigated the differential performance across different sample sizes or the applicability of the approach on small or imbalanced data. It is imperative to develop new imputation approaches with higher generalizability for data with various sample sizes. Results We proposed a method called scHinter for imputing dropout events for scRNA-seq with special emphasis on data with limited sample size. scHinter incorporates a voting-based ensemble distance and leverages the synthetic minority oversampling technique for random interpolation. A hierarchical framework is also embedded in scHinter to increase the reliability of the imputation for small samples. We demonstrated the ability of scHinter to recover gene expression measurements across a wide spectrum of scRNA-seq datasets with varied sample sizes. We comprehensively examined the impact of sample size and cluster number on imputation. Comprehensive evaluation of scHinter across diverse scRNA-seq datasets with imbalanced or limited sample size showed that scHinter achieved higher and more robust performance than competing approaches, including MAGIC, scImpute, SAVER and netSmooth. Availability and implementation Freely available for download at https://github.com/BMILAB/scHinter. Supplementary information Supplementary data are available at Bioinformatics online.

This paper is devoted to study the bending response, free vibration and mechanical buckling of functionally graded material (FGM) plates based on the scaled boundary finite element method (SBFEM) for the first time. On the basis of the three-dimensional (3D) theory of elasticity, the SBFEM governing equations of the FGM plate are derived from the principle of virtual work and solved analytically in the thickness direction to obtain the displacements, stresses, natural frequencies and critical buckling loads. A high order spectral element only with three degrees of freedom per node that is able to satisfy the high order continuity of the displacement fields while facilitating the numerical computation is applied. In the present method, only the bottom surface of plate structures needs to be discretized while numerical approximations in the thickness direction are no longer required, which leads to an accurate solution of the displacement in the thickness direction and a considerable reduction of the computational cost. Furthermore, the model strictly follows 3D theory of elasticity without employing any kinematic assumptions of plate theory, so that it is able to eliminate the shear locking problem for numerical simulations of FGM plates when the thickness becomes thinner. Accuracy and superior computational efficiency of the present formulations have been verified by comparing with the available analytical and numerical solutions achieved by other researchers.

Functionally graded materials have widespread applications for the practical engineering and industry of design and development. In the present work, free vibration and transient dynamic behaviors of functionally graded material (FGM) sandwich plates are semi-analytically investigated by the scaled boundary finite element method (SBFEM). A layerwise approach based on the three-dimensional (3D) theory of elasticity is adopted for the simulation of FGM sandwich plates, and the material properties involving Young’s modulus and mass density are assumed to be continuously graded in the thickness direction according to a power law function while Poisson’s ratio is taken to be constant in each individual layer. In order to obtain the global equilibrium equations for the dynamic problem of FGM sandwich plates, the SBFEM governing equation associated with a single layer is derived based on the principle of virtual work in the first step. Subsequently, they are assembled for the whole FGM sandwich plate by imposing the continuity conditions of the layer interface in the displacement fields. Natural frequencies are determined by eigenvalue analysis, and the transient dynamic equilibrium equation is solved on the basis of the Newmark’s method. One of the advantages of present technique is the capability of reducing the spatial dimension and no discretizations in the thickness direction are needed, which results in a considerable reduction in the computational costs of the project. Furthermore, dynamic responses can be obtained directly from the semi-analytical solutions without shear correction factors or special treatments for the shear-locking effect as some high-order shear deformation plate theories. Excellent adaptability in mesh distortion, thickness ratio and boundary condition of proposed formulations for the dynamic analysis of FGM sandwich plates are confirmed and comprehensive parametric investigations are also carried out. Numerical results show that the dynamic responses obtained by the proposed approach converge rapidly and excellent agreement can be achieved between SBFEM results and the available solutions with a small number of computational costs, which indicates high accuracy and efficiency of the proposed formulations in dynamic analyses of FGM sandwich plates.

In the design of multiplierless finite impulse response (FIR) filters, tremendous efforts have been made to reduce the number of adders of the multiplier block for the reduction of overall chip area and power consumption. However, fewer in the multiplier block do not necessarily lead to lower power consumption, since the structural adders dominate the power consumption of an FIR filter circuit. In this paper, we propose a power-oriented optimization method for linear phase FIR filters. In the proposed algorithm, the power index, which is the average adder depth of the structural adders, is used as the optimization objective in the discrete coefficients search. A gate-level simulation of benchmark filters shows that the proposed technique designs filters consuming less power than those obtained by the best available algorithms, which aim to minimize the number of adders. The power savings over existing designs can be as much as 19.6%.

This study aimed to investigate the effect of AOPPs on apoptosis in human chondrocytes. Chondrocytes were treated with AOPPs. Cell death, nicotinamide adenine dinucleotide phosphate (NADPH) oxidase activity, reactive oxygen species (ROS) generation, and the expression of apoptotic proteins were detected in vitro. AOPPs levels were detected by colorimetric method. The results in vitro demonstrated that AOPPs induced cell death in human chondrocyte through a redox-dependent pathway, including RAGE-mediated, NADPH oxidase-dependent ROS generation, and poly (ADP-ribose) polymerase-1 (PARP-1) activation. Targeting AOPPs-induced cellular mechanisms might emerge as a promising therapeutic option for patients with RA.

Application of the scaled boundary finite element method (SBFEM) in free vibration and transient dynamic problems of the composite magneto-electro-elastic (MEE) cylindrical shell is considered in this paper for the first time. The mathematic model of the composite MEE cylindrical shell is governed by the three-dimensional (3D) theory of magneto-electro-elasticity in the cylindrical coordinate system, and the dynamic equilibrium equations are then derived into a second-order ordinary differential equation by using the SBFEM. In this method, discretizations are confined only to the surfaces or boundaries of the computational domain but no fundamental solutions and singular integrals are required, which reduces the spatial dimension of problem by one and making mesh generation much easier. At the same time, the transverse shear locking can be successfully avoided due to the analyticity of the SBFEM governing equation in the radial direction of the scaled boundary coordinate system. Based on the state space approach, the basic equations of the composite MEE cylindrical shell in the scaled boundary coordinate system are reduced to a system of first-order ordinary differential equation in terms of the independent variable involving the generalized displacement and internal nodal force fields, and then the free vibration and transient dynamic responses of the system are solved by using the generalized eigenvalue technique and Newmark's integration scheme, respectively. Numerical results illustrate that the convergence of the proposed formulations is satisfactory, and excellent agreement can be achieved for limiting cases with the solutions available in published work or that obtained using the commercial software ANSYS. In addition, the influences of the structural parameters (including thickness ratio, aspect ratio and stacking configuration), material properties and boundary conditions on the vibration and dynamic behaviors of the composite MEE cylindrical shells are discussed.

A three-dimensional (3D) theory of elasticity is presented for the solution of the generalized displacements and stresses in the composite laminated and sandwich cylindrical shell structures with covered or embedded piezoelectric layers based on the scaled boundary finite element method (SBFEM). The SBFEM is a weak-form differential technique, which can lead to accurate results performing discretization only on the middle plane of shell structures so that the considered model can be treated as a two-dimensional (2D) discretizated problem. A new type 2D high order spectral element shape function is introduced both for approximations of geometry model and basis variables in the scaled boundary coordinate system. Employing the weighted residual principle in conjunction with Green’s theorem to the normalized static equilibrium equations of each layer leads to the SBFEM governing equation with respect to the radial generalized displacement fields. Then a system of a first-order ordinary differential equation is obtained by introducing a dual variable and analytically solved by using the precise integration technique (PIT). As a result, the major advantage of the proposed formulations for solving the bending problem of the composite laminated and sandwich cylindrical piezoelectric shell is no need to discretize the 3D model with a great deal of degrees of freedom while ensures the computational accuracy. Two test examples are carried out to demonstrate the adaptability and reliability of the present method and illustrate very good agreements and rapid convergence with the solutions based on the finite element approach only using a small number of elements. Numerical examples about the sandwich cylindrical piezoelectric shell are presented to investigate the parametric effects of the thickness ratio as well as the stacking sequence on the variation of generalized displacements and stresses. The results show that both the thickness ratio and stacking sequence have significant influence on the bending behaviors of the sandwich cylindrical piezoelectric shells.

Supplemental Figures. This file contains all the Supplemental Figures. (PPTX 206 kb)

Figure S1. Illustration of the raw and imputed data of two randomly selected cells from the cortex fetal-quiescent of the Darmanis data. Figure S2. Violin plots showing expression profiles of three marker genes (a) and numbers of expressed genes (b) in the nine cell types of the Darmains data before or after imputation. Figure S3. Benchmarking of scNPF-propagation on eight published scRNA-seq data sets. Figure S4. Evaluation of the effect of paramter r of scNPF-propagation on two data sets, Darmanis (A) and Baron (B). Figure S5. Characteristics of three priori gene-gene interaction networks. Figure S6. Characteristics of imputed expression matrices for the Darmanis data obtained by the context mode or the priori mode with different interaction networks. Figure S7. Benchmarking of scNPF-propagation on eight published scRNA-seq data sets using the context mode and the priori mode with different priori networks including String, HumanNet, and INet. Figure S8. Benchmark results of scNPF-fusion on the Baron data. Figure S9. Performance comparison of the five similarity measurements on eight published scRNA-seq data sets. Figure S10. Benchmarking of scNPF-fusion on eight published scRNA-seq data sets. Figure S11. Benchmarking of scNPF-fusion on eight published scRNA-seq data sets by applying hierarchical clustering on the similarity matrices. Figure S12. Benchmarking of scNPF-fusion on eight published scRNA-seq data sets by applying spectral clustering on the similarity matrices. Figure S13. Benchmarking of scNPF-fusion on eight published scRNA-seq data sets by applying partitioning around medoids clustering on the similarity matrices. Figure S14. Evaluation of the effect of parameters of scNPF-fusion on two data sets, Darmanis (A) and Baron (B). Figure S15. Visualization of results from scNPF-fusion with different network combinations on the Darmanis data. Figure S16. Performance comparison of similarities learned from scNPF-fusion with different network combinations on eight published scRNA-seq data sets. Figure S17. Benchmarking of scNPF-fusion with different network combinations on eight published scRNA-seq data sets. (PPTX 6626 kb)

Single-cell RNA sequencing (scRNA-seq) has been an important inspiration for the study of biomolecules through its reveal of cell heterogeneity. However, due to the low capture efficiency and frequent drop-out events in the single-cell sequencing process, the scRNA-seq data often has high sparsity and random missing values, which brings great difficulties to the subsequent analysis. The network propagation method based on random walk with restart (RWR) effectively fills in the missing values in the scRNA-seq data and reduces noise by referring to the prior information of gene interaction. Dimensionality reduction is also a commonly used pre-processing method for high-dimensional and sparse scRNA-seq data, which can be combined with the RWR-based data imputation to achieve noise reduction and feature extraction of scRNA-seq data. This article compares the performance of the commonly used single-cell data dimension reduction methods combined with the RWR network smoothing in different type of scRNA-seq data sets, and analyzes their applicability and stability.

Diabetic retinopathy (DR) is a common complication of diabetes which is one of the leading causes of blindness worldwide. However, DR is hard to detect in the early stages and the diagnostic procedure can be time-consuming and abundant expertise is needed. Therefore, we proposed a computer-aided diagnosis method based on a deep learning algorithm to automatically diagnose DR and divide color retinal fundus photographs in to five grades. Besides, a novel pre-processing algorithm is adopted to enhance the quality and uniformity of input retinal images and a transfer learning method to achieve better performance. Finally, the system is evaluated based on a test set with 7023 images and an accuracy of 80.0% and a kappa score of 0.64 are achieved.

In this paper, we present four denoising algorithms dedicated to the division of focal plane (DoFP) polarization image sensors, including the average filtering, median filtering, Wiener filtering and wavelet threshold denoising algorithms. Compared with the previous implementations solely based on the Gaussian noise model, this paper, for the first time, covers the non-Gaussian noises, such as the salt & pepper noise and Poisson noise. According to our extensive experimental results, the wavelet threshold denoising outperforms for suppressing the Gaussian noise; while the median filtering and the Wiener filtering outperform for suppressing the low-density salt & pepper noise and Poisson noise, respectively.

In this paper, we present a novel one-dimensional deep convolutional neural network (1D-DCNN) based algorithm for classifying mixture gases. Compared with the previously reported electronic nose system that can only recognize pure gas, the proposed implementation is capable of distinguishing the individual component of binary mixture gases composed of Ethylene, CO and Methane. To the best of our knowledge, the proposed 1D-DCNN algorithm is firstly applied in the mixture gases recognition. Compared with the conventional pattern recognition algorithms including support vector machine (SVM) and artificial neural network (ANN), the proposed 1D-DCNN exhibits higher average accuracy (96.15%) based on extensive experimental results using 10-fold cross validation.

In this paper, we present a novel fast gas recognition algorithm based on convolutional recurrent neural network. Rectangular kernels and long short-term memory (LSTM) are adopted to calculate the gas-label distribution from relationship between sensors and relationship among time as well. Compared with the widely adopted algorithms, such as support vector machine (SVM), K nearest neighbor (KNN), random forest (RF), the proposed implementation features an ultra-short detection time of 4 seconds. Moreover, according to our extensive experimental results, a high accuracy of 98.28% is achieved, which greatly outperforms the abovementioned algorithms.

This paper presents a novel K times singular value decomposition (K-SVD) based denoising algorithm for the division-of-focal-plane (DoFP) polarization image sensors. In the proposed implementation, the input DoFP image can be expressed by the optimum sparse combination of the dictionary elements via K-SVD and orthogonal matching pursuit (OMP) algorithms. As a result, this implementation is capable of eliminating the Gaussian noise significantly and well-preserving the details and edges of the target DoFP image. Our extensive experimental results on various test images show that the proposed algorithm yields better visual quality and maintains a lower PSNR value while compared with a wide range of previous implementations.

Conventional resistive sensor readout ICs (ROICs) suffer limitations, either because of their high resistive sensor-biasing current consumption or low resolution for jitter-induced noise. To overcome these limitations, a 13-bit resistive sensor ROIC is presented in this work. In the proposed ROIC, which relies on zero-crossing-based circuits (ZCBCs), the current source is used not only to settle the output of the ΔΣ integrator, but also to bias the resistive sensor, and thus avoid implementing a dedicated power-hungry bias current branch. In addition, the oversampled ΔΣ modulation technique guarantees high resolution for the proposed ROIC. A prototype was fabricated using 0.18 μm CMOS technology. The measurement results show the ROIC achieved a 13-bit root-mean-square noise equivalent resolution with an 8 kS/s conversion speed and a 480 fJ/conversion-step figure of merit. Moreover, it has good nonlinearity of less than 250 ppm and a power supply sensitivity of 1%/V.

Conventional capacitance-to-digital converters (CDCs) suffer limitations either on narrow capacitance range or low resolution for jitter-induced noise and high power consumption. In order to overcome these limitations, a 13-bit 1 pF-to-10 nF generic CDC is presented. In the proposed CDC with the oversampled ΔΣ modulation, the zero-crossing-based circuits (ZCBCs) are used to replace the operational transconductance amplifier to avoid feedback loop stability issues. However, the ZCBCs inevitably incur the non-idealities and thus a novel calibration scheme is presented for efficient non-ideality-error cancellation. A prototype fabricated using 0.18 μm CMOS technology is experimentally verified using a MEMS capacitive humidity sensor. The measurement results show the CDC achieves a 13-bit rms noise equivalent resolution with a 128 μs conversion time and a 230 fJ/conversion-step figure of merit. The calibration scheme enhances the linearity from 7 bits to 11.4 bits in the 1 pF-to-10 nF compatible capacitance range.

Hydrogen sensor based on nanoporous palladium film with a short response time at room temperature has been investigated. Nanoporous palladium film was deposited on treated porous anodic aluminum oxide template by DC magnetron sputtering. We have researched the detection performance of the hydrogen sensor depending on different post-deposition annealing temperatures ranging from 150 to 400 °C. In this sensor, we obtained a sensor response time as short as 30 s at 1 % hydrogen concentration with an anneal temperature of 200 °C. The sensor exhibited good performance under hydrogen concentrations from 0.1 to 1 %.

In this work, a greedy algorithm for the design of sparse linear-phase finite impulse response filters wherein the coefficients are successively fixed to zero individually is proposed. To meet the filter specifications, the coefficient for which the middle value of its feasible range is closest to zero is selected to be set to zero, whereas all the other unfixed coefficients are free to change. Design examples show that the proposed technique can design FIR filters with higher sparsity than that obtained by existing nonexhaustive algorithms for given specifications. To show the optimality of the algorithm, we design 100 filters, with results showing that the global optimal solution, i.e., the sparsest solution found by exhaustive search, can be achieved in most cases, but with much less computation time.

Deterministic tree search algorithms for the design of multiplierless linear phase finite impulse response filters are generally time consuming. Many researches therefore focus on how to restrict the number of discrete values assigned to each coefficient during a tree search. In this paper, a two-step tree search algorithm is proposed. In the first step, a polynomial-time tree search algorithm where each coefficient is fixed to a single one discrete value is introduced. Since the synthesis of large coefficients is dominant in the hardware cost over small coefficients, in the second step optimization, the small coefficients obtained in the first step is kept unaltered and the large coefficients are further divided into several groups and the coefficients are optimized group by group alternatingly. Such a two-step search strategy maximally utilizes the limited computational resources and can achieve lower hardware cost design in a shorter design time, compared with existing algorithms.

Alternative polyadenylation (APA) can lead to various mRNA isforms differing in their 3' UTR, which contributes to the dynamics of gene regulation, including stability, localization and translation of mRNA. However, clustering of genes using poly(A) site data has not been extensively studied. Here we constructed a two-layer model based on canonical correlation analysis (CCA) to explore the clustering of genes with the consideration of the distribution and abundance of APA sites. We adopted the widely used hierarchical clustering with Average-Linkage (Hc-A), Complete-Linkage (Hc-C), and Single-Linkage (Hc-S) to comprehensively compare the proposed method with four distance measures including Pearson correlation coefficient (PCC), Euclidean distance (EUC), Manhattan distance (MAN), and Maximum distance (MAX) based on three performance indexes. Results showed that the proposed method performed better than other four distance measures and could generate clusters with more biological meaning and underlying molecular mechanisms.

For multiplierless FIR filter design and implementation, optimization of the product accumulation block (PAB) in transposed direct form structure has been ignored by most of the research. In this work, the power consumption of the PABs of FIR filters is studied theoretically and experimentally. It is shown that the PAB contributes most of the total power consumption in multiplierless FIR filters. Two implementation methods are proposed to reduce the power consumption of the PAB. Simulation results show that the total power can be reduced significantly at the cost of some area overhead.

In this paper, we present a novel smoothness-based interpolation algorithm for the division of focal plane Polarimeters (DoFP). By calculating the divided blocks' variance that represents their local smoothness, the proposed algorithm well-balances between the traditional bilinear and bicubic interpolation algorithms. In addition, compared with the previously reported gradient-based interpolation algorithm which only indicates the image's directional change along 0°, 90°, 45° and 135°, the presented smoothness-based algorithm covers the variations along all the possible directions, leading to more accurate selection between the bilinear and bicubic algorithms. According to our extensive simulation results, the proposed implementation exhibits the lowest mean square error (MSE) for the test images among all the previously reported algorithms, including bilinear, bicubic and gradient-based interpolation algorithms.

This paper presents a frequency-mode capacitive sensor readout front-end operating in a wide temperature range. First, a biased varactor temperature compensation (BVTC) is proposed to compensate the aggregate temperature gradients from the sensor and the oscillator circuit, achieving a nullified temperature coefficient for the oscillation frequency. Second, a temperature coherent quantization (TCQ) approach is proposed to enhance the sensor' sensitivity and to provide a self-referenced clock for digitization, whereby the influence from the temperature effect of the clock is minimized through a hybrid down-conversion time-to-digital converter (DTDC). A prototype chip was fabricated using the 0.18-μm CMOS process and it was verified using a commercial 5.92-to-6.53-pF capacitive pressure sensor over a temperature range of −20°C to 120°C. Proven by experiments, the prototype presented as good as ±0.4% full-scale pressure error in the 140°C temperature range.

Support vector machine (SVM) now attracts increasing attention in gas classification due to its high performance towards small samples and nonlinearity problems of the dataset. Previously, the probable mismatch between the dataset and the training parameters determined by trial and error or grid search may hinder the exploration of the best result. In this paper, we propose a novel approach to estimate the most suitable training parameters, based on the inbreeding prevention of genetic algorithm (GA) by assigning the training model parameters of SVM as its chromosome. Treating the k-fold cross validation of SVM training as the objective function, our new method makes the population on the whole evolve towards the values that are more appropriate for the dataset. The inbreeding prevention mechanism (IPM) can protect the population from converging over-rapidly before reaching the optimum value. Compared with the standard SVM, the proposed method has greatly improved the prediction accuracy in both training data and testing data.

Near infrared (NIR) synaptic devices offer a remote-control approach to implement neuromorphic computing for data safety and artificial retinal system applications. In upconverting nanoparticles (UCNPs)-mediated optogenetics biosystems, NIR regulation of membrane ion channels allows remote and selective control of the Ca2+ flux to modulate synaptic plasticity behaviors. Inspired by the upconversion optogenetics, we proposed a NIR artificial synapse based on a UCNPs-MoS2 floating gate phototransistor in which MoS2 acts as light-sensitive ion channels to reabsorb the visible light emitted from UCNPs under NIR illumination. As a result, the synaptic device exhibits stable persistent photocurrent (PPC) effect up to 353 K and ultrahigh photogain (~108 electrons per photon), ensuring the long-term potentiation (LTP) behavior. Simulations using the handwritten digit data sets indicate good recognition accuracy of the light-controlled artificial neuron network. Overall, this design concept combining biology, optics and electronics opens up a new avenue for developing optogenetics-inspired neuromorphic technology in the future.

This paper develops a high-performance semi-analytical numerical model to analyze the bending responses of the angle-ply composite laminated cylindrical shells with the fiber reinforced layers using the scaled boundary finite element method (SBFEM). As the thin-walled structures, the angle-ply composite laminated shells are assumed to be made of orthotropic materials in the cylindrical coordinate system. Both the geometric and basic variables are discretized by utilizing the two-dimensional (2D) high order spectral elements in the curved surface domain of the shells. According to the exact three-dimensional (3D) theory of elasticity rather than the approximate shell theories, the weak form of the partial differential governing equations for each layer of the composite laminated cylindrical shells in the cylindrical coordinate system are transformed into ordinary differential equations using the SBFEM. In the circumstances, there are no variables about the curved surface of shell in the SBFEM governing equations for each lamina, so that it can be analytically solved on the basis of the dual variable approach and the precise integration technique (PIT). Employing the interface continuity conditions of displacement between the layers, the complete SBFEM model with respect to the global stiffness matrix of the composite laminated cylindrical shell can be obtained. Unlike the general layerwise theories, which supposes that the basic variables varied linearly with the thickness coordinate, the through-thickness distributions of the displacement field in each discrete layer is assumed to be a quadratic polynomial with respect to the radial coordinate in this paper, thus the through-thickness stress field can be described more accurate. Numerical examples for solving the bending problem of composite laminated cylindrical shells are presented. As a result, the numerical efficiency, accuracy and applicability of the proposed formulations are confirmed by the comparison of the published results involving the distributions of the displacements and stresses through the thickness and along the circumferential direction for different staking configurations, geometric properties and boundary conditions.

The main purpose of this study is to develop a high performance numerical model for the analysis of the coupled electromechanical effects of the fiber-reinforced piezoelectric laminated composite plate under the mechanical or electric loading. Considering that the piezoelectric laminated plate is made of N orthotropic layers with the material axes oriented arbitrarily with respect to the laminate coordinate, the three-dimensional (3D) theory of elasticity for the kth orthotropic (piezoelectric) lamina is applied for the establishment of the mathematic model by using the elastic displacements and the electric potential as the nodal degree of freedom. The scaled boundary finite element method (SBFEM) coupling with the precise integration technique (PIT) are used to obtain the solution. Under the formulations of the SBFEM, only the in-plane of the plate structure is discretized into two-dimensional (2D) elements thus the spatial dimension is reduced by one. By introducing a dual vector of the internal nodal force, the SBFEM governing equation is described as an analytic formulation along the through-thickness with a matrix exponential function which can be estimated with high accuracy using the PIT. In addition, the high order spectral element is applied for the description of the basic variables so that the laminate with the complicated curved boundaries can be accurately determined. At the same time, assuming that the generalized displacement field is a quadratic function with respect to the thickness coordinate of each layer, then the generalized strain and stress fields can be determined directly based on the strain-displacement relations and the constitutive relations in the scaled boundary coordinate system. Numerical results show that a good agreement can be achieved with few discrete nodes for both the thin and moderately thick piezoelectric laminated plate under different types of loads through comparison with the 3D exact solutions and good performance has also been validated. Moreover, two numerical examples including a square piezoelectric laminated plate with the complex configuration (involving the fiber orientation and the distribution of piezoelectric layers) and a circular/annular piezoelectric laminated plate are further discussed to demonstrate the applicability of the present formulations.

Division of focal plane (DoFP) polarimeter is widely used in polarization imaging sensors. The periodically arranged micro-polarizers integrated on the focal plane ensure its outstanding real-time performance, but reduce the spatial resolution of output images and further affect the calculation of polarization parameters. In this paper, a four-layer, end-to-end fully convolutional neural network called Fork-Net is proposed, which aims to directly improve the imaging quality of three polarization properties: intensity (i.e., S0), degree of linear polarization (DoLP), and angle of polarization (AoP), rather than focusing on reducing the interpolation error of intensity images of different polarization orientations. The Fork-Net accepts raw mosaic images as input and directly outputs S0, DoLP, and AoP. It is also trained with a customized loss function. The experimental results show that compared with existing methods, the proposed one achieves the highest peak signal-to-noise ratio (PSNR) and prominent visual quality on output images.

In this work, a novel genetic algorithm (GA) is proposed for the design of multiplierless linear phase finite impulse response (FIR) filters. The filters under consideration are of high order and wide coefficient wordlength. Both the single-stage and cascade form are considered. In a practical filter design problem, when the filter specification is stringent, requiring high filter order and wide coefficient wordlength, GAs often fail to find feasible solutions, because the discrete search space thus constructed is huge and the majority of the solution candidates therein can not meet the specification. In the proposed GA, the discrete search space is partitioned into smaller ones. Each small space is constructed surrounding a base discrete coefficient set which is obtained by a proposed greedy algorithm. The partition of the search space increases the chances for the GA to find feasible solutions, but does not sacrifice the coverage of the search. The proposed GA applies to the design of single-stage filters. When a cascade form filter is designed, for each single-stage filter meeting the filter specification generated during the course of GA, an integer polynomial factorization is applied. Design examples show that the proposed GA significantly outperforms existing algorithms dealing with the similar problems in terms of design time, and the hardware cost is saved in most cases. Accepted version

Background Paeoniflorin, a monoterpene glycoside, exerts protective vascular effects, showing good antioxidant properties. However, whether Paeoniflorin has protective effect against the oxidative damage induced by advanced oxidation protein products (AOPPs) in Human umbilical vein endothelial cells (HUVECs) is unknown, as is the underlying mechanism. Purpose The present study was designed to investigate the effect of Paeoniflorin on oxidative damage of HUVECs and elucidate its underlying molecular mechanisms. Methods The fluorescence intensity of 2′, 7′-dichlorofluorescein-diacetate (DCFH-DA) staining was detected for intracellular reactive oxygen species (ROS) production. The increases mitochondrial membrane potential (MMP) was measured via flow cytometry and confocal microscopy using MitoTracker® Deep Red/ MitoTracker® Green staining. The intracellular adenosine triphosphate (ATP) was measured by ATP Determination Kit according to the manufacturer's protocol. Nox2, Nox4, hypoxia inducible factor-1α (HIF-1α), vascular endothelial growth factor (VEGF) and nuclear factor-κB (NF-κB) p65 expressions were detected by western blot. Results Our results showed that Paeoniflorin increases MMP and ATP levels of HUVECs induced by AOPPs, and attenuates NF-κB p65 expression on HUVECs might mainly result from its antioxidant capability by suppressing ROS production. Moreover, we also found that Paeoniflorin can suppress HIF-1α and VEGF protein expression through a decrease of ROS production via down-regulation of Nox2/Nox4 expression in HUVECs. AOPP-induced RAGE mRNA up-regulation was blocked by Paeoniflorin treatment in HUVECs. Conclusion Our results provided the first experimental that Paeoniflorin protects against AOPP-induced oxidative damage in HUVECs, mainly through a mechanism involving a decrease in ROS production by the inhibition of Nox2/Nox4 and RAGE expression; restored ATP depletion and mitochondria dysfunction via ROS suppression; and down-regulated HIF-1α/VEGF, possibly via the ROS-NF-κB axis.

This study aimed to investigate the effect of AOPPs on apoptosis in human chondrocytes. Chondrocytes were treated with AOPPs. Cell death, nicotinamide adenine dinucleotide phosphate (NADPH) oxidase activity, reactive oxygen species (ROS) generation, and the expression of apoptotic proteins were detected in vitro. AOPPs levels were detected by colorimetric method. The results in vitro demonstrated that AOPPs induced cell death in human chondrocyte through a redox-dependent pathway, including RAGE-mediated, NADPH oxidase-dependent ROS generation, and poly (ADP-ribose) polymerase-1 (PARP-1) activation. Targeting AOPPs-induced cellular mechanisms might emerge as a promising therapeutic option for patients with RA.

The level shifters are crucial primitives in the multi-supply voltage circuits and systems. In this paper, an energy-efficient level shifter is proposed to achieve the conversion from the subthreshold voltage to the above threshold voltage. It is a hybrid structure consisting of the Wilson current mirror and the cross-coupled level shifter. By addressing the voltage drop issue of the level shifter based on the Wilson current mirror, the leakage power is significantly reduced, with the advantage of wide input voltage range for the Wilson current mirror level shifter well-preserved. In addition, the multi-threshold CMOS (MTCMOS) technology is employed to provide more flexibility for our ultra-low power design. The reported simulation results using 65 nm CMOS process validate our proposed implementation and an ultra-low power consumption of 19.44 fJ per conversion from 0.2 V to 1.2 V at 1 MHz is achieved without the need of any intermediate power supply.