Determination of Trace Organic Contaminant Concentration via Machine Classification of Surface-Enhanced Raman Spectra.

Surface-enhanced Raman spectroscopy (SERS) has been well explored as a highly effective characterization technique that is capable of chemical pollutant detection and identification at very low concentrations. Machine learning has been previously used to identify compounds based on SERS spectral data. However, utilization of SERS to quantify concentrations, with or without machine learning, has been difficult due to the spectral intensity being sensitive to confounding factors such as the substrate parameters, orientation of the analyte, and sample preparation technique. Here, we demonstrate an approach for predicting the concentration of sample pollutants from SERS spectra using machine learning. Frequency domain transform methods, including the Fourier and Walsh-Hadamard transforms, are applied to spectral data sets of three analytes (rhodamine 6G, chlorpyrifos, and triclosan), which are then used to train machine learning algorithms. Using standard machine learning models, the concentration of the sample pollutants is predicted with >80% cross-validation accuracy from raw SERS data. A cross-validation accuracy of 85% was achieved using deep learning for a moderately sized data set (∼100 spectra), and 70-80% was achieved for small data sets (∼50 spectra). Performance can be maintained within this range even when combining various sample preparation techniques and environmental media interference. Additionally, as a spectral pretreatment, the Fourier and Hadamard transforms are shown to consistently improve prediction accuracy across multiple data sets. Finally, standard models were shown to accurately identify characteristic peaks of compounds via analysis of their importance scores, further verifying their predictive value.

High-throughput sperm DNA analysis at the single-cell and population levels.

Clinical semen quality assessment is critical to the treatment of infertility. Sperm DNA integrity testing provides critical information that can steer treatment and influence outcomes and offspring health. Flow cytometry is the gold standard approach to assess DNA integrity, but it is not commonly applied at the clinical level. The sperm chromatin dispersion (SCD) assay provides a simpler and cheaper alternative. However, SCD is low-throughput and non-quantitative - sperm assessment is serial, manual and suffers inter- and intra-observer variations. Here, an automated SCD analysis method is presented that enables quantitative sperm DNA quality assessment at the single-cell and population levels. Levering automated optical microscopy and a chromatin diffusion-based analysis, a sample of thousands of sperm that would otherwise require 5 hours is assessed in under 10 minutes - a clinically viable workflow. The sperm DNA diffusion coefficient (DDNA) measurement correlates (R2 = 0.96) with DNA fragmentation index (DFI) from the cytometry-based sperm chromatin structure assay (SCSA). The automated measurement of population-level sperm DNA fragmentation (% sDF) prevents inter-observer variations and shows a good agreement with the SCSA % DFI (R2 = 0.98). This automated approach standardizes and accelerates SCD-based sperm DNA analysis, enabling the clinical application of sperm DNA integrity assessment.

High-Rate and Selective CO2 Electrolysis to Ethylene via Metal-Organic-Framework-Augmented CO2 Availability.

High-rate conversion of carbon dioxide (CO2 ) to ethylene (C2 H4 ) in the CO2 reduction reaction (CO2 RR) requires fine control over the phase boundary of the gas diffusion electrode (GDE) to overcome the limit of CO2 solubility in aqueous electrolytes. Here, a metal-organic framework (MOF)-functionalized GDE design is presented, based on a catalysts:MOFs:hydrophobic substrate materials layered architecture, that leads to high-rate and selective C2 H4 production in flow cells and membrane electrode assembly (MEA) electrolyzers. It is found that using electroanalysis and operando X-ray absorption spectroscopy (XAS), MOF-induced organic layers in GDEs augment the local CO2 concentration near the active sites of the Cu catalysts. MOFs with different CO2 adsorption abilities are used, and the stacking ordering of MOFs in the GDE is varied. While sputtering Cu on poly(tetrafluoroethylene) (PTFE) (Cu/PTFE) exhibits 43% C2 H4 Faradaic efficiency (FE) at a current density of 200 mA cm- 2 in a flow cell, 49% C2 H4 FE at 1 A cm- 2 is achieved on MOF-augmented GDEs in CO2 RR. MOF-augmented GDEs are further evaluated in an MEA electrolyzer, achieving a C2 H4 partial current density of 220 mA cm-2 for CO2 RR and 121 mA cm-2 for the carbon monoxide reduction reaction (CORR), representing 2.7-fold and 15-fold improvement in C2 H4 production rate, compared to those obtained on bare Cu/PTFE.

Machine learning for sperm selection.

Infertility rates and the number of couples seeking fertility care have increased worldwide over the past few decades. Over 2.5 million cycles of assisted reproductive technologies are being performed globally every year, but the success rate has remained at ~33%. Machine learning, an automated method of data analysis based on patterns and inference, is increasingly being deployed within the health-care sector to improve diagnostics and therapeutics. This technique is already aiding embryo selection in some fertility clinics, and has also been applied in research laboratories to improve sperm analysis and selection. Tremendous opportunities exist for machine learning to advance male fertility treatments. The fundamental challenge of sperm selection - selecting the most promising candidate from 108 gametes - presents a challenge that is uniquely well-suited to the high-throughput capabilities of machine learning algorithms paired with modern data processing capabilities.

CO2 electrolysis to multicarbon products at activities greater than 1 A cm-2.

Electrolysis offers an attractive route to upgrade greenhouse gases such as carbon dioxide (CO2) to valuable fuels and feedstocks; however, productivity is often limited by gas diffusion through a liquid electrolyte to the surface of the catalyst. Here, we present a catalyst:ionomer bulk heterojunction (CIBH) architecture that decouples gas, ion, and electron transport. The CIBH comprises a metal and a superfine ionomer layer with hydrophobic and hydrophilic functionalities that extend gas and ion transport from tens of nanometers to the micrometer scale. By applying this design strategy, we achieved CO2 electroreduction on copper in 7 M potassium hydroxide electrolyte (pH ≈ 15) with an ethylene partial current density of 1.3 amperes per square centimeter at 45% cathodic energy efficiency.

Molecular tuning of CO2-to-ethylene conversion.

The electrocatalytic reduction of carbon dioxide, powered by renewable electricity, to produce valuable fuels and feedstocks provides a sustainable and carbon-neutral approach to the storage of energy produced by intermittent renewable sources1. However, the highly selective generation of economically desirable products such as ethylene from the carbon dioxide reduction reaction (CO2RR) remains a challenge2. Tuning the stabilities of intermediates to favour a desired reaction pathway can improve selectivity3-5, and this has recently been explored for the reaction on copper by controlling morphology6, grain boundaries7, facets8, oxidation state9 and dopants10. Unfortunately, the Faradaic efficiency for ethylene is still low in neutral media (60 per cent at a partial current density of 7 milliamperes per square centimetre in the best catalyst reported so far9), resulting in a low energy efficiency. Here we present a molecular tuning strategy-the functionalization of the surface of electrocatalysts with organic molecules-that stabilizes intermediates for more selective CO2RR to ethylene. Using electrochemical, operando/in situ spectroscopic and computational studies, we investigate the influence of a library of molecules, derived by electro-dimerization of arylpyridiniums11, adsorbed on copper. We find that the adhered molecules improve the stabilization of an 'atop-bound' CO intermediate (that is, an intermediate bound to a single copper atom), thereby favouring further reduction to ethylene. As a result of this strategy, we report the CO2RR to ethylene with a Faradaic efficiency of 72 per cent at a partial current density of 230 milliamperes per square centimetre in a liquid-electrolyte flow cell in a neutral medium. We report stable ethylene electrosynthesis for 190 hours in a system based on a membrane-electrode assembly that provides a full-cell energy efficiency of 20 per cent. We anticipate that this may be generalized to enable molecular strategies to complement heterogeneous catalysts by stabilizing intermediates through local molecular tuning.

Deep learning-based selection of human sperm with high DNA integrity.

Despite the importance of sperm DNA to human reproduction, currently no method exists to assess individual sperm DNA quality prior to clinical selection. Traditionally, skilled clinicians select sperm based on a variety of morphological and motility criteria, but without direct knowledge of their DNA cargo. Here, we show how a deep convolutional neural network can be trained on a collection of ~1000 sperm cells of known DNA quality, to predict DNA quality from brightfield images alone. Our results demonstrate moderate correlation (bivariate correlation ~0.43) between a sperm cell image and DNA quality and the ability to identify higher DNA integrity cells relative to the median. This deep learning selection process is directly compatible with current, manual microscopy-based sperm selection and could assist clinicians, by providing rapid DNA quality predictions (under 10 ms per cell) and sperm selection within the 86th percentile from a given sample.

Deep learning for the classification of human sperm.

BACKGROUND: Infertility is a global health concern, and couples are increasingly seeking medical assistance to achieve reproduction. Semen analysis is a primary assessment performed by a clinician, in which the morphology of the sperm population is evaluated. Machine learning algorithms that automate, standardize, and expedite sperm classification are the subject of ongoing research. METHOD: We demonstrate a deep learning method to classify sperm into one of several World Health Organization (WHO) shape-based categories. Our method uses VGG16, a deep convolutional neural network (CNN) initially trained on ImageNet, a collection of human-annotated everyday images, which we retrain for sperm classification using two freely-available sperm head datasets (HuSHeM and SCIAN). RESULTS: Our deep learning approach classifies sperm at high accuracy and performs well in head-to-head comparisons with earlier approaches using identical datasets. We demonstrate improvement in true positive rate over a classifier approach based on a cascade ensemble of support vector machines (CE-SVM) and show similar true positive rates as compared to an adaptive patch-based dictionary learning (APDL) method. Retraining an off-the-shelf VGG16 network avoids excessive neural network computation or having to learn and use the massive dictionaries required for sparse representation, both of which can be computationally expensive. CONCLUSIONS: We show that our deep learning approach to sperm head classification represents a viable method to automate, standardize, and accelerate semen analysis. Our approach highlights the potential of artificial intelligence technologies to eventually exceed human experts in terms of accuracy, reliability, and throughput.

Live sperm trap microarray for high throughput imaging and analysis.

There is a growing appreciation and understanding of cell-to-cell variability in biological samples. However, research and clinical practice in male fertility has relied on population, or sample-based characteristics. Single-cell resolution is particularly important given the winner-takes-all nature of both natural and in vitro fertilization: it is the properties of a single cell, not the population, that are passed to the next generation. While there are a range of methods for single cell analysis, arraying a larger number of live sperm has not been possible due to the strong locomotion of the cells. Here we present a 103-trap microarray that traps, aligns and arrays individual live sperm. The method enables high-resolution imaging of the aligned cell head, the application of dye-based DNA and mitochondrial analyses, and the quantification of motility characteristics, such as tail beat. In testing, a 2400-post array trapped ∼400 sperm for individual analyses of tail beating frequency and amplitude, DNA integrity via acridine orange staining, and mitochondrial activity via staining. While literature results are mixed regarding a possible correlation between motility and DNA integrity of sperm at sample-level, results here find no statistical correlation between tail beat characteristics and DNA integrity at the cell-level. The trap array uniquely enables the high-throughput study of individual live sperm in semen samples - assessing the inherently single-cell selection process of fertilization, with single-cell resolution.

Confinement effects on DNA hybridization in electrokinetic micro- and nanofluidic systems.

Spatial confinement, within cells or micro- and nanofabricated devices, impacts the conformation and binding kinetics of biomolecules. Understanding the role of spatial confinement on molecular behavior is important for comprehending diverse biological phenomena, as well as for designing biosensors. Specifically, the behavior of molecular binding under an applied electric field is of importance in the development of electrokinetic biosensors. Here, we investigate whether confinement of DNA oligomers in capillary electrophoresis impacts the binding kinetics of the DNA. To infer the role of confinement on hybridization dynamics, we perform capillary electrophoresis measurements on DNA oligomers within micro- and nanochannels, then apply first-order reaction dynamics theory to extract kinetic parameters from electropherogram data. We find that the apparent dissociation constants at the nanoscale (i.e., within a 100 nm channel) are lower than at the microscale (i.e., within a 1 µm channel), indicating stronger binding with increased confinement. This confirms, for the first time, that confinement-based enhancement of DNA hybridization persists under application of an electric field.

An Experimental Approach to Systematically Probe Charge Inversion in Nanofluidic Channels.

Charge inversion of the surfaces of nanofluidic channels occurs in systems with high-surface charge and/or highly charged ions and is of particular interest because of applications in biological and energy conversion systems. However, the details of such charge inversion have not been clearly elucidated. Specifically, although we can experimentally and theoretically show charge inversion, understanding at what conditions charge inversion occurs, as well how much the charge-inverting ions change the surface, are not known. Here, we show a novel experimental approach for uniquely finding both the ζ-potential and adsorption time of charge inverting ions in aqueous nanofluidic systems.

Two-Dimensional Electric Double Layer Structure with Heterogeneous Surface Charge.

In this work we present a systematic study of the lateral (parallel to the wall) and normal (perpendicular to the wall) nanostructure of the electric double layer at a heterogeneous interface between two regions of different surface charges, often found in nanoscale electrochemical devices. Specifically, classical density functional theory (DFT) is used to probe a cation concentration range of 10 mM to 1 M, for valences of +1, + 2, and +3, and a diameter range of 0.15-0.9 nm over widely varying surface charges (between -0.15 and +0.15 C/m2). The DFT results predict significant lateral and normal nanostructure in the form of ion concentration oscillations. These results are directly compared with those from Poisson-Boltzmann theory, showing significant deviation between the two theories, not only in the concentration profiles, but also in the sign of the electrostatic potential.

(Almost) Stationary Isotachophoretic Concentration Boundary in a Nanofluidic Channel Using Charge Inversion.

The present work is an experimental study of a new means to induce a quasi-stationary boundary for concentration or separation in a nanochannel induced by charge inversion. Instead of using pressure-driven counter-flow to keep the front stationary, we exploit charge inversion by a highly charged electrolyte, Ru(bpy)3Cl2, that changes the sign of the zeta potential in part of the channel from negative to positive. Having a non-charge inverting electrolyte (MgCl2) in the other part of the channel and applying an electric field can create a standing front at the interface between them without added dispersion due to an externally applied pressure-driven counterflow. The resulting slow moving front position can be easily imaged optically since Ru(bpy)3Cl2 is fluorescent. A simple analytical model for the velocity field and front axial position that reproduces the experimental location of the front shows that the location can be tuned by changing the concentration of the electrolytes (and thus local zeta potential). Both of these give the charge inversion-mediated boundary significant advantages over current methods of concentration and separation and the method is, therefore, of particular importance to chemical and biochemical analysis systems such as chromatography and separations and for enhancing the stacking performance of field amplified sample injection and isotachophoresis. By choosing a non-charge inverting electrolyte other than MgCl2, either this electrolyte or the Ru(bpy)3Cl2 solution can be made to be the leading or trailing electrolyte.

Label free detection of nucleic acids by modulating nanochannel surfaces.

We develop surface-modified 100 nm silica nanofluidic channels that change in measured conductivity upon exposure to single- or double-stranded DNA. Through careful monitoring of both electromigrative and advective current in the channel, we can detect nanomolar concentrations of DNA. These results can be exploited for inexpensive, all-electronic DNA sensors.

Hybridization thermodynamics of DNA oligonucleotides during microchip capillary electrophoresis.

Capillary electrophoresis (CE) is a powerful analytical tool for performing separations and characterizing properties of charged species. For reacting species during a CE separation, local concentrations change leading to nonequilibrium conditions. Interpreting experimental data with such nonequilibrium reactive species is nontrivial due to the large number of variables involved in the system. In this work we develop a COMSOL multiphysics-based numerical model to simulate the electrokinetic mass transport of short interacting ssDNAs in microchip capillary electrophoresis. We probe the importance of the dissociation constant, K(D), and the concentration of DNA on the resulting observed mobility of the dsDNA peak, µ(w), by using a full sweep of parametric simulations. We find that the observed mobility is strongly dependent on the DNA concentration and K(D), as well as ssDNA concentration, and develop a relation with which to understand this dependence. Furthermore, we present experimental microchip capillary electrophoresis measurements of interacting 10 base ssDNA and its complement with changes in buffer ionic strength, DNA concentration, and DNA sequence to vary the system equilibria. We then compare our results to thermodynamically calculated K(D) values.