Histopathology-Based Diagnosis of Oral Squamous Cell Carcinoma Using Deep Learning.

Oral squamous cell carcinoma (OSCC) is prevalent around the world and is associated with poor prognosis. OSCC is typically diagnosed from tissue biopsy sections by pathologists who rely on their empirical experience. Deep learning models may improve the accuracy and speed of image classification, thus reducing human error and workload. Here we developed a custom-made deep learning model to assist pathologists in detecting OSCC from histopathology images. We collected and analyzed a total of 2,025 images, among which 1,925 images were included in the training set and 100 images were included in the testing set. Our model was able to automatically evaluate these images and arrive at a diagnosis with a sensitivity of 0.98, specificity of 0.92, positive predictive value of 0.924, negative predictive value of 0.978, and F1 score of 0.951. Using a subset of 100 images, we examined whether our model could improve the diagnostic performance of junior and senior pathologists. We found that junior pathologists were able to delineate OSCC in these images 6.26 min faster when assisted by the model than when working alone. When the clinicians were assisted by the model, their average F1 score improved from 0.9221 to 0.9566 in the case of junior pathologists and from 0.9361 to 0.9463 in the case of senior pathologists. Our findings indicate that deep learning can improve the accuracy and speed of OSCC diagnosis from histopathology images.

Comparison of long non-coding RNAs in adipose and muscle tissues between seven indigenous Chinese and the Yorkshire pig breeds.

lncRNAs play crucial roles in fat metabolism in animals. Previously, we have compared the mRNA transcriptome profiles between seven fat-type Chinese pig breeds and one lean-type Western breed (Yorkshire, YY). The associations between differentially expressed (DE) genes and phenotypical traits were investigated. In the present study, to further explore the underlying regulatory mechanisms, lncRNAs were sequenced and compared between YY and Chinese indigenous breeds. The results showed 9114 and 7538 DE lncRNAs between at least one Chinese breed and the YY breed in the adipose and muscle tissue respectively. KEGG enrichment analysis revealed that the target genes of these DE lncRNAs mainly influenced the glucolipid metabolism, which is an important process affecting meat quality. Correlation analyses between the DE lncRNA and DE mRNA genes related to meat quality and growth traits were performed. The results showed that LTCONS_00073280 was associated with intramuscular fat content. Four lncRNAs (LTCONS_00101781, LTCONS_00037879, LTCONS_00088260 and LTCONS-00128343) might mediate backfat thickness. Overall, this study provides candidate lncRNAs that potentially affect meat quality, which might be useful for molecular breeding of pig breeds in future.

Automated Radiographic Evaluation of Adenoid Hypertrophy Based on VGG-Lite.

Adenoid hypertrophy is a pathological hyperplasia of the adenoids, which may cause snoring and apnea, as well as impede breathing during sleep. The lateral cephalogram is commonly used by dentists to screen for adenoid hypertrophy, but it is tedious and time-consuming to measure the ratio of adenoid width to nasopharyngeal width for adenoid assessment. The purpose of this study was to develop a screening tool to automatically evaluate adenoid hypertrophy from lateral cephalograms using deep learning. We proposed the deep learning model VGG-Lite, using the largest data set (1,023 X-ray images) yet described to support the automatic detection of adenoid hypertrophy. We demonstrated that our model was able to automatically evaluate adenoid hypertrophy with a sensitivity of 0.898, a specificity of 0.882, positive predictive value of 0.880, negative predictive value of 0.900, and F1 score of 0.889. The comparison of model-only and expert-only detection performance showed that the fully automatic method (0.07 min) was about 522 times faster than the human expert (36.6 min). Comparison of human experts with or without deep learning assistance showed that model-assisted human experts spent an average of 23.3 min to evaluate adenoid hypertrophy using 100 radiographs, compared to an average of 36.6 min using an entirely manual procedure. We therefore concluded that deep learning could improve the accuracy, speed, and efficiency of evaluating adenoid hypertrophy from lateral cephalograms.

Honeycomb and triangular domain wall networks in heteroepitaxial systems.

A comprehensive study is presented for the influence of misfit strain, adhesion strength, and lattice symmetry on the complex Moiré patterns that form in ultrathin films of honeycomb symmetry adsorbed on compact triangular or honeycomb substrates. The method used is based on a complex Ginzburg-Landau model of the film that incorporates elastic strain energy and dislocations. The results indicate that different symmetries of the heteroepitaxial systems lead to distinct types of domain wall networks and phase transitions among various surface Moiré patterns and superstructures. More specifically, the results show a dramatic difference between the phase diagrams that emerge when a honeycomb film is adsorbed on substrates of honeycomb versus triangular symmetry. It is also shown that in the small deformation limit, the complex Ginzburg-Landau model reduces to a two-dimensional sine-Gordon free energy form. This free energy can be solved exactly for one dimensional patterns and reveals the role of domains walls and their crossings in determining the nature of the phase diagrams.

Indigenous arsenic(V)-reducing microbial communities in redox-fluctuating near-surface sediments of the Mekong Delta.

Arsenic (As) cycling within soils and sediments of the Mekong Delta of Cambodia is affected by drastic redox fluctuations caused by seasonal monsoons. Extensive flooding during monsoon seasons creates anoxic soil conditions that favor anaerobic microbial processes, including arsenate [As(V)] respiration-a process contributing to the mobilization of As. Repeated oxidation and reduction in near-surface sediments, which contain 10-40 mg kg(-1) As, lead to the eventual downward movement of As to the underlying aquifer. Amplification of a highly conserved functional gene encoding dissimilatory As(V) reductase, arrA, can be used as a molecular marker to detect the genetic potential for As(V) respiration in environmental samples. However, few studies have successfully amplified arrA from sediments without prior enrichment, which can drastically shift community structure. In the present study, we examine the distribution and diversity of arrA genes amplified from multiple sites within the Cambodian Mekong Delta as a function of near-surface depth (10, 50, 100, 200, and 400 cm), where sediments undergo seasonal redox fluctuations. We report successful amplification of 302 arrA gene sequences (72 OTUs) from near-surface Cambodian soils (without prior enrichment or stimulation with carbon amendments), where a large majority (>70%) formed a well-supported clade that is phylogenetically distinct from previously reported sequences from Cambodia and other South and Southeast Asian sediments, with highest sequence similarity to known Geobacter species capable of As(V) respiration, further supporting the potentially important role of Geobacter sp. in arsenic mobilization in these regions.

Anomalous fast dynamics of adsorbate overlayers near an incommensurate structural transition.

We investigate the dynamics of a compressively strained adsorbed layer on a periodic substrate via a simple two-dimensional model that admits striped and hexagonal incommensurate phases. We show that the mass transport is superfast near the striped-hexagonal phase boundary and in the hexagonal phase. For an initial step profile separating a bare substrate region (or "hole") from the rest of a striped incommensurate phase, the superfast domain wall dynamics leads to a bifurcation of the initial step profile into two interfaces or profiles propagating in opposite directions with a hexagonal phase in between. This yields a theoretical understanding of the recent experiments for the Pb/Si(111) system.

Patterning of heteroepitaxial overlayers from nano to micron scales.

Thin heteroepitaxial overlayers have been proposed as templates to generate stable, self-organized nanostructures at large length scales, with a variety of important technological applications. However, modeling strain-driven self-organization is a formidable challenge due to different length scales involved. In this Letter, we present a method for predicting the patterning of ultrathin films on micron length scales with atomic resolution. We make quantitative predictions for the type of superstructures (stripes, honeycomb, triangular) and length scale of pattern formation of two metal-metal systems, Cu on Ru(0001) and Cu on Pd(111). Our findings are in excellent agreement with previous experiments and call for future experimental investigations of such systems.

Glassy phases and driven response of the phase-field-crystal model with random pinning.

We study the structural correlations and the nonlinear response to a driving force of a two-dimensional phase-field-crystal model with random pinning. The model provides an effective continuous description of lattice systems in the presence of disordered external pinning centers, allowing for both elastic and plastic deformations. We find that the phase-field crystal with disorder assumes an amorphous glassy ground state, with only short-ranged positional and orientational correlations, even in the limit of weak disorder. Under increasing driving force, the pinned amorphous-glass phase evolves into a moving plastic-flow phase and then, finally, a moving smectic phase. The transverse response of the moving smectic phase shows a vanishing transverse critical force for increasing system sizes.

Diffusion in periodic potentials with path integral hyperdynamics.

We consider the diffusion of brownian particles in one-dimensional periodic potentials as a test bench for the recently proposed stochastic path integral hyperdynamics (PIHD) scheme [Chen and Horing, J. Chem. Phys. 126, 224103 (2007)]. First, we consider the case where PIHD is used to enhance the transition rate of activated rare events. To this end, we study the diffusion of a single brownian particle moving in a spatially periodic potential in the high-friction limit at low temperature. We demonstrate that the boost factor as compared to straight molecular dynamics (MD) has nontrivial behavior as a function of the bias force. Instead of growing monotonically with the bias, the boost attains an optimal maximum value due to increased error in the finite path sampling induced by the bias. We also observe that the PIHD method can be sensitive to the choice of numerical integration algorithm. As the second case, we consider parallel resampling of multiple bias force values in the case of a brownian particle in a periodic potential subject to an external ac driving force. We confirm that there is no stochastic resonance in this system. However, while the PIHD method allows one to obtain data for multiple values of the ac bias, the boost with respect to MD remains modest due to the simplicity of the equation of motion in this case.

Dynamical transitions and sliding friction of the phase-field-crystal model with pinning.

We study the nonlinear driven response and sliding friction behavior of the phase-field-crystal (PFC) model with pinning including both thermal fluctuations and inertial effects. The model provides a continuous description of adsorbed layers on a substrate under the action of an external driving force at finite temperatures, allowing for both elastic and plastic deformations. We derive general stochastic dynamical equations for the particle and momentum densities including both thermal fluctuations and inertial effects. The resulting coupled equations for the PFC model are studied numerically. At sufficiently low temperatures, we find that the velocity response of an initially pinned commensurate layer shows hysteresis with dynamical melting and freezing transitions for increasing and decreasing applied forces at different critical values. The main features of the nonlinear response in the PFC model are similar to the results obtained previously with molecular dynamics simulations of particle models for adsorbed layers.

Scaling exponents of forced polymer translocation through a nanopore.

We investigate several properties of a translocating homopolymer through a thin pore driven by an external field present inside the pore only using Langevin Dynamics (LD) simulations in three dimensions (3D). Motivated by several recent theoretical and numerical studies that are apparently at odds with each other, we estimate the exponents describing the scaling with chain length (N) of the average translocation time , the average velocity of the center of mass , and the effective radius of gyration during the translocation process defined as approximately Nalpha, approximately N(-delta), and Rg approximately Nnu respectively, and the exponent of the translocation coordinate (s-coordinate) as a function of the translocation time approximately tbeta. We find alpha = 1.36 +/- 0.01, beta = 1.60+/- 0.01 for approximately taubeta and beta = 1.44 +/- 0.02 for approximately taubeta, delta = 0.81 +/- 0.04, and nu congruent with nu = 0.59 +/- 0.01, where nu is the equilibrium Flory exponent in 3D. Therefore, we find that approximately N1.36 is consistent with the estimate of approximately /. However, as observed previously in Monte Carlo (MC) calculations by Kantor and Kardar (Y. Kantor, M. Kardar, Phys. Rev. E 69, 021806 (2004)) we also find the exponent alpha = 1.36 +/- 0.01 < 1 + nu. Further, we find that the parallel and perpendicular components of the gyration radii, where one considers the "cis" and "trans" parts of the chain separately, exhibit distinct out-of-equilibrium effects. We also discuss the dependence of the effective exponents on the pore geometry for the range of N studied here.

Nonlinear driven response of a phase-field crystal in a periodic pinning potential.

We study numerically the phase diagram and the response under a driving force of the phase field crystal model for pinned lattice systems introduced recently for both one- and two-dimensional systems. The model describes the lattice system as a continuous density field in the presence of a periodic pinning potential, allowing for both elastic and plastic deformations of the lattice. We first present results for phase diagrams of the model in the absence of a driving force. The nonlinear response to a driving force on an initially pinned commensurate phase is then studied via overdamped dynamic equations of motion for different values of mismatch and pinning strengths. For large pinning strength the driven depinning transitions are continuous, and the sliding velocity varies with the force from the threshold with power-law exponents in agreement with analytical predictions. Transverse depinning transitions in the moving state are also found in two dimensions. Surprisingly, for sufficiently weak pinning potential we find a discontinuous depinning transition with hysteresis even in one dimension under overdamped dynamics. We also characterize structural changes of the system in some detail close to the depinning transition.

Polymer translocation in a double-force arrangement.

Using Langevin dynamics simulations, we investigate the translocation dynamics of an externally driven polymer chain through a nanopore, where a pulling force F is exerted on the first monomer whilst there is an opposing force F(E) < F within the pore. Such a double-force arrangement has been proposed recently to allow better dynamical control of the translocation process in order to sequence biopolymers. We find that in the double-force arrangement translocation becomes slower as compared to the case under a single monomer pulling force of magnitude F-F(E), but scaling of the translocation time as a function of the chain length tau approximately N2 does not change. The waiting time tau(m) for monomer m to exit the pore is found to be a monotonically increasing function of the bead number almost until m approximately N , which indicates relatively well-defined slowing down and control of the chain velocity during translocation. We also study the waiting time distributions for the beads in the chain, and characterize in detail fluctuations in the bead positions and their transverse position coordinates during translocation. These data should be useful in estimating position-dependent sequencing errors in double-force experiments.

Atomistic studies of strain relaxation in heteroepitaxial systems.

We present a review of recent theoretical studies of different atomistic mechanisms of strain relaxation in heteroepitaxial systems. We explore these systems in two and three dimensions using different semi-empirical interatomic potentials of Lennard-Jones and many-body embedded atom model type. In all cases we use a universal molecular static method for generating minimum energy paths for transitions from the coherent epitaxial (defect free) state to the state containing an isolated defect (localized or extended). This is followed by a systematic search for the minimum energy configuration as well as self-organization in the case of a periodic array of islands. In this way we are able to understand many general features of the atomic mechanisms and energetics of strain relaxation in these systems. Finally, for the special case of Pd/Cu(100) and Cu/Pd(100) heteroepitaxy we also use conventional molecular dynamics simulation techniques to compare the compressively and tensilely strained cases. The results for this case are in good agreement with the existing experimental data.

Thermal fluctuations and phase diagrams of the phase-field crystal model with pinning.

We study the influence of thermal fluctuations in the phase diagram of a recently introduced two-dimensional phase field crystal model with an external pinning potential. The model provides a continuum description of pinned lattice systems allowing for both elastic deformations and topological defects. We introduce a nonconserved version of the model and determine the ground-state phase diagram as a function of lattice mismatch and strength of the pinning potential. Monte Carlo simulations are used to determine the phase diagram as a function of temperature near commensurate phases. The results show a rich phase diagram with commensurate, incommensurate, and liquidlike phases with a topology strongly dependent on the type of ordered structure. A finite-size scaling analysis of the melting transition for the c(2x2) commensurate phase shows that the thermal correlation length exponent nu and specific heat behavior are consistent with the Ising universality class as expected from analytical arguments.

Phase diagram and commensurate-incommensurate transitions in the phase field crystal model with an external pinning potential.

We study the phase diagram and the commensurate-incommensurate transitions in a phase field model of a two-dimensional crystal lattice in the presence of an external pinning potential. The model allows for both elastic and plastic deformations and provides a continuum description of lattice systems, such as for adsorbed atomic layers or two-dimensional vortex lattices. Analytically, a mode expansion analysis is used to determine the ground states and the commensurate-incommensurate transitions in the model as a function of the strength of the pinning potential and the lattice mismatch parameter. Numerical minimization of the corresponding free energy shows reasonable agreement with the analytical predictions and provides details on the topological defects in the transition region. We find that for small mismatch the transition is of first order, and it remains so for the largest values of mismatch studied here. Our results are consistent with results of simulations for atomistic models of adsorbed overlayers.

Strain relief in Cu-Pd heteroepitaxy.

We present experimental and theoretical studies of Pd/Cu(100) and Cu/Pd(100) heterostructures in order to explore their structure and misfit strain relaxation. Ultrathin Pd and Cu films are grown by pulsed laser deposition at room temperature. For Pd/Cu, compressive strain is released by networks of misfit dislocations running in the [100] and [010] directions, which appear after a few monolayers (ML) already. In striking contrast, for Cu/Pd the tensile overlayer remains coherent up to about 9 ML, after which multilayer growth occurs. The strong asymmetry between tensile and compressive cases is in contradiction with continuum elasticity theory and is also evident in the structural parameters of the strained films. Molecular dynamics calculations based on classical many-body potentials confirm the pronounced tensile-compressive asymmetry and are in good agreement with the experimental data.

Adsorption and diffusion on a stepped surface: atomic hydrogen on Pt(211).

We present density functional theory calculations for atomic hydrogen interacting with a stepped surface, the Pt(211) surface. The calculations have been performed at the generalized gradient approximation level, using a slab representation of the surface. This is the state-of-the-art method for calculating the interaction of atoms or molecules with metal surfaces, nevertheless only few studies have used it to study atoms or molecules interacting with stepped surfaces, and none, to the best of our knowledge, have considered hydrogen interacting with stepped platinum surfaces. Our goal has been to initiate a systematic study of this topic. We have calculated the full three-dimensional potential energy surface (PES) for the H/Pt(211) system together with the vibrational band structure and vibrational eigenfunctions of H. A deep global minimum of the PES is found for bridge-bonded hydrogen on the step edge, in agreement with experimental results for the similar H/Pt(533) system. All the local vibrational excitations at the global minimum have been identified, and this will serve as a helpful guide to the interpretation of future experiments on this (or similar) system(s). Furthermore, from the calculated PES and vibrational band structure, we identify a number of consequences for the interpretation or modelling of diffusion experiments studying the coverage and directional dependence of atomic hydrogen diffusion on stepped platinum surfaces.

Surface diffusion anomaly near a substrate phase transition: H on W(100).

Using a linear optical diffraction method, we have experimentally studied the long predicted diffusion anomalous behavior for H/W(100) near the reconstructive phase transition of the W(100) substrate. This anomaly manifests itself in the form of a strong dip in the diffusion coefficient D at the transition temperature T(C). We interpret the strong reduction of D as a result of the diverging friction damping near the transition. The finite dip in D instead of a vanishing D at T(C) also demonstrates the importance of the non-Markovian (memory) deviation from the simple instantaneous damping picture.

Finding transition paths and rate coefficients through accelerated Langevin dynamics.

We present a technique to resolve the rare event problem for a Langevin equation describing a system with thermally activated transitions. A transition event within a given time interval (0,t(f)) can be described by a transition path that has an activation part during (0,t(M)) and a deactivation part during (t(M),t(f))(0