Multilevel Deep Feature Generation Framework for Automated Detection of Retinal Abnormalities Using OCT Images

Optical coherence tomography (OCT) images coupled with many learning techniques have been developed to diagnose retinal disorders. This work aims to develop a novel framework for extracting deep features from 18 pre-trained convolutional neural networks (CNN) and to attain high performance using OCT images. In this work, we have developed a new framework for automated detection of retinal disorders using transfer learning. This model consists of three phases: deep fused and multilevel feature extraction, using 18 pre-trained networks and tent maximal pooling, feature selection with ReliefF, and classification using the optimized classifier. The novelty of this proposed framework is the feature generation using widely used CNNs and to select the most suitable features for classification. The extracted features using our proposed intelligent feature extractor are fed to iterative ReliefF (IRF) to automatically select the best feature vector. The quadratic support vector machine (QSVM) is utilized as a classifier in this work. We have developed our model using two public OCT image datasets, and they are named database 1 (DB1) and database 2 (DB2). The proposed framework can attain 97.40% and 100% classification accuracies using the two OCT datasets, DB1 and DB2, respectively. These results illustrate the success of our model.     

Calculation of molecular weight distribution in a batch thermal polymerization of styrene

Molecular weight averages have long been used as a measure of polymer molecular weight properties in industrial polymer manufacturing processes. With a kinetic model, it is possible to directly calculate the polymer chain length distribution by integrating an infinite number of the polymer population balance equations. However, when the polymer chain length is very large, such a direct integration of polymer population balance equations can be computationally demanding. In this paper, the method of finite molecular weight moments is applied to the calculation of polymer chain length distribution in a batch free radical thermal polymerization of styrene. The weight fraction of a finite chain length interval is directly calculated in conjunction with a kinetic model. The method of calculation is illustrated through model simulations.     

Molecularly imprinted polymers as artificial steroid receptors

The polymers selective to six different steroids (testosterone, Δ⁴-androstene-3,17-dione, 1,4-androstadiene-3,17-dione, β-estradiol, progesterone, testosterone propionate) have been synthesized using molecular imprinting based on noncovalent interactions. Analysis of the influence of structural features of the steroids under study has shown that molecules with a relatively rigid structure and the OH group at C-17 position are the most efficient templates for methacrylic acid-containing imprinted polymers. The chromatographic study of the polymers synthesized has demonstrated a strong dependence of the selectivity and intensity of interaction with analytes on the composition of solvents used both as porogen and chromatographic mobile phase. To obtain polymers with highly selective recognition sites and to create the optimal conditions for molecular recognition, all possible interactions (between template and functional monomer, template and solvent, solvent and functional monomer) should be taken into account. <?TF="palat-i"> The batch rebinding study of testosterone by the imprinted polymer in acetonitrile has revealed some heterogeneity of recognition sites, and permitted determination of K_ass = 1.05 × 10⁴ M ⁻¹, ΔG° = −5.4 kcal/mol and N = 1.2 μmol/g for high-affinity sites and K_ass = 0.33 × 10⁴ M ⁻¹, ΔG° = −4.8 kcal/mol and N = 2.2 μmol/g for low-affinity sites. <?TF="palat-i"> The results obtained show how it is possible to regulate in different modes the molecular recognition by imprinted polymers as well as to fabricate polymers possessing the necessary properties depending on their practical application.© 1998 John Wiley & Sons, Ltd.     

Analysis of phospholipids using an open‐tubular capillary column with a monolithic layer of molecularly imprinted polymer in capillary electrochromatography‐electrospray ionization‐tandem mass spectrometry

In this study, an open‐tubular capillary electrochromatography (OT‐CEC) column with a monolithic layer of molecularly imprinted polymer (MIP) based on methacrylic acid, ethylene glycol dimethacrylate, and 4‐styrenesulfonic acid was utilized for the simultaneous separation and characterization of phospholipid (PL) molecular structures by interfacing with electrospray ionization‐tandem mass spectrometry (ESI‐MS‐MS). Introducing an MIP‐based monolith along with charged species at the OT column made it possible to separate PL molecules based on differences in head groups and acyl chain lengths in CEC. For the interface of OT‐CEC with ESI‐MS‐MS, a simple nanospray interface utilizing a sheath flow was developed and the resulting OT‐CEC‐ESI‐MS‐MS was able to separate PL standards (phosphatidylserines, phosphatidylethanolamines, phosphatidylglycerols, phosphatidic acid, and lysophosphatidylglycerols). The developed method was applied to human urinary lipid extracts, and resulted in the separation and structural identification of 18 molecules by data‐dependent collision‐induced dissociation.     

Elemental Analysis of Stone Fruits by Inductively Coupled Plasma Mass Spectrometry and Direct Mercury Analysis

This study reports the concentrations of eight trace essential (Zn, Mn, Cu, Ni, Cr, Co, V, and Se) and four toxic elements (Pb, As, Cd, and Hg) in commonly consumed stone fruits from South Korea. The samples were digested by microwave-induced combustion and analyzed by inductively coupled plasma mass spectrometry (ICP-MS). The concentrations of mercury were analyzed by direct mercury analysis (DMA). The analytical techniques were validated by linearity, limits of detection and quantification, precision, recovery, and for accuracy by analyzing a spinach leave-certified reference material; satisfactory results were obtained in all cases. The concentrations of essential trace elements varied considerably among the stone fruits. Generally stone fruits contained comparatively high concentrations of Zn (0.946 to 7.86?µg/g) and Mn (below the limit of detection to 1.66?µg/g), while lower contents of Cu (0.214 to 1.24?µg/g), Cr (0.032 to 0.114?µg/g), Ni (0.006 to 0.091?µg/g), Co (0.004 to 0.016?µg/g), V (below the limit of detection to 0.023?µg/g), and Se (0.0002 to 0.005?µg/g) were obtained. The concentrations (µg/g) of toxic metals were 0.007 (peach) to 0.016 (cherry) for Pb, 0.001 (plum) to 0.007 (cherry) for As, 0.002 (apricot and cherry) to 0.003 (peach) for Cd, and 0.0003 (peach) to 0.0016 (jujube) for Hg. The values for the estimated dietary intakes, target hazard quotients, and hazard indices were lower than the recommended safety limits by World Health Organization. Therefore, the analyzed stone fruits were deemed to be safe for human consumption.     

Ultrathin Fe‐N‐C Nanosheets Coordinated Fe‐Doped CoNi Alloy Nanoparticles for Electrochemical Water Splitting

The development of highly active and cost‐effective catalyst materials toward electrochemical water splitting is of great importance for converting and storing the intermittent solar energy in the form of hydrogen. Herein, for the first time, an ultrathin Fe and N‐co‐doped carbon nanosheet encapsulated Fe‐doped CoNi alloy nanoparticle (FeCoNi@FeNC) composite is obtained and applied as a bifunctional catalyst for hydrogen evolution reaction (HER) and oxygen evolution reaction (OER). This catalyst exhibits prominent catalytic performances for both HER and OER, which only requires overpotentials of 102 and 330 mV, respectively, to reach a current density of 10 mA cm^?2 in alkaline media. The high catalytic activity is intrinsically associated with the presence of Fe in both nanosheets and nanoparticles, which has triggered the occurrence of coordinative effects between Fe‐N‐C and FeCoNi that are beneficial for HER and OER, as revealed by electrochemical techniques. In an overall water splitting electrolyzer, FeCoNi@FeNC is employed as both the cathode and anode catalysts, achieving 12 mA cm^?2 at 1.63 V for a duration of more than 12 h.     

Mid-infrared heterodyne phase-sensitive dispersion spectroscopy in flame measurements

We present the first demonstration of heterodyne phase-sensitive dispersion spectroscopy (HPSDS) for in situ, non-intrusive and quantitative CO₂ concentration measurements in flames. Dispersion spectroscopy retrieves gas properties by measuring the refractive index in the vicinity of a molecular resonance. The HPSDS scheme features a significant diagnostic advantage of the intrinsic immunity to laser power fluctuations caused by beam steering, thermal radiation and soot scattering in combustion environments, and thus no extra calibration process is required. In this work, we described the spectroscopic fundamentals for measuring heterodyne phase signals in flames. As a proof of principle, we used a mid-infrared interband cascade laser (ICL) near 4183?nm to exploit the strong CO₂ transitions in the R-branch of the v₃ fundamental band. The HPSDS signals of four CO₂ lines, R(76), R(78), R(80) and R(82), were measured in CH₄/air flames to obtain CO₂ concentrations at different equivalence ratios (Φ?=?0.8–1.2), yielding a good agreement with the simultaneous laser absorption measurements using the same ICL. With its immunity to laser power fluctuations verified experimentally, the HPSDS sensor was successfully implemented to measure CO₂ concentrations in C₂H₄/air sooting flames (Φ?=?1.78–2.38). Laser dispersion spectroscopy proves to be a promising and alternative diagnostic tool for combustion measurements.