Search

Abstract Background Colorectal cancer is one of the most serious malignant tumors, and lymph node metastasis (LNM) from colorectal cancer is a major factor for patient management and prognosis. Accurate image detection of LNM is an important task to help clinicians diagnose cancer. Recently, the U-Net architecture based on convolutional neural networks (CNNs) has been widely used to segment image to accomplish more precise cancer diagnosis. However, the accurate segmentation of important regions with high diagnostic value is still a great challenge due to the insufficient capability of CNN and codec structure in aggregating the detailed and non-local contextual information. In this work, we propose a high performance and low computation solution. Methods Inspired by the working principle of Fovea in visual neuroscience, a novel network framework based on U-Net for cancer segmentation named Fovea-UNet is proposed to adaptively adjust the resolution according to the importance-aware of information and selectively focuses on the region most relevant to colorectal LNM. Specifically, we design an effective adaptively optimized pooling operation called Fovea Pooling (FP), which dynamically aggregate the detailed and non-local contextual information according to the pixel-level feature importance. In addition, the improved lightweight backbone network based on GhostNet is adopted to reduce the computational cost caused by FP. Results Experimental results show that our proposed framework can achieve higher performance than other state-of-the-art segmentation networks with 79.38% IoU, 88.51% DSC, 92.82% sensitivity and 84.57% precision on the LNM dataset, and the parameter amount is reduced to 23.23 MB. Conclusions The proposed framework can provide a valid tool for cancer diagnosis, especially for LNM of colorectal cancer.

Background: Colorectal cancer is one of the most serious malignant tumors, and lymph node metastasis (LNM) from colorectal cancer is a major factor for patient management and prognosis. Accurate image detection of LNM is an important task to help clinicians diagnose cancer. Recently, the U-Net architecture based on convolutional neural networks (CNNs) has been widely used to segment image to accomplish more precise cancer diagnosis. However, the accurate segmentation of important regions with high diagnostic value is still a great challenge due to the insufficient capability of CNN and codec structure in aggregating the detailed and non-local contextual information. In this work, we propose a high performance and low computation solution. Methods: Inspired by the working principle of Fovea in visual neuroscience, a novel network framework based on U-Net for cancer segmentation named Fovea-UNet is proposed to adaptively adjust the resolution according to the importance-aware of information and selectively focuses on the region most relevant to colorectal LNM. Specifically, we design an effective adaptively optimized pooling operation called Fovea Pooling (FP), which dynamically aggregate the detailed and non-local contextual information according to the pixel-level feature importance. In addition, the improved lightweight backbone network based on GhostNet is adopted to reduce the computational cost caused by FP. Results: Experimental results show that our proposed framework can achieve higher performance than other state-of-the-art segmentation networks with 79.38% IoU, 88.51% DSC, 92.82% sensitivity and 84.57% precision on the LNM dataset, and the parameter amount is reduced to 23.23 MB. Conclusions: The proposed framework can provide a valid tool for cancer diagnosis, especially for LNM of colorectal cancer. [ABSTRACT FROM AUTHOR]

With the increase of resolution and frame rate, videos can provide users with a better viewing experience. However, the amount of data is also greatly increased, making it much more difficult for storage and transmission. To address this challenge, we propose to optimize video coding configurations based on video content, aiming to reduce bitrate while preserving perceptual quality. In our method, we extract hand-crafted spatial–temporal features from original videos to delineate their content characteristics. Subsequently, we employ support vector machines (SVM) to perform video coding optimization through spatial–temporal rescaling, thereby improving video coding efficiency. Our approach is compatible with existing encoders and decoders, which is advantageous in practical applications. Experimental results on videos with different contents demonstrate that our method can predict the optimal encoding configuration with high accuracy, achieving considerable bitrate savings for videos of various resolutions and frame rates with low computational costs. • We encode all videos in VideoSet to identify the optimal encoding configurations. • We propose a content-adaptive rescaling scheme to improve video coding efficiency. • Experiments and analyses are conducted to demonstrate the superiority of our method. [ABSTRACT FROM AUTHOR]

We propose GNN-Surrogate, a graph neural network-based surrogate model to explore the parameter space of ocean climate simulations. Parameter space exploration is important for domain scientists to understand the influence of input parameters (e.g., wind stress) on the simulation output (e.g., temperature). The exploration requires scientists to exhaust the complicated parameter space by running a batch of computationally expensive simulations. Our approach improves the efficiency of parameter space exploration with a surrogate model that predicts the simulation outputs accurately and efficiently. Specifically, GNN-Surrogate predicts the output field with given simulation parameters so scientists can explore the simulation parameter space with visualizations from user-specified visual mappings. Moreover, our graph-based techniques are designed for unstructured meshes, making the exploration of simulation outputs on irregular grids efficient. For efficient training, we generate hierarchical graphs and use adaptive resolutions. We give quantitative and qualitative evaluations on the MPAS-Ocean simulation to demonstrate the effectiveness and efficiency of GNN-Surrogate. Source code is publicly available at https://github.com/trainsn/GNN-Surrogate. [ABSTRACT FROM AUTHOR]

Foveated rendering adapts the image synthesis process to the user's gaze. By exploiting the human visual system's limitations, in particular in terms of reduced acuity in peripheral vision, it strives to deliver high-quality visual experiences at very reduced computational, storage, and transmission costs. Despite the very substantial progress made in the past decades, the solution landscape is still fragmented, and several research problems remain open. In this work, we present an up-to-date integrative view of the domain from the point of view of the rendering methods employed, discussing general characteristics, commonalities, differences, advantages, and limitations. We cover, in particular, techniques based on adaptive resolution, geometric simplification, shading simplification, chromatic degradation, as well spatio-temporal deterioration. Next, we review the main areas where foveated rendering is already in use today. We finally point out relevant research issues and analyze research trends. [Display omitted] • A comprehensive survey of foveated rendering. • General characteristics, commonalities, differences, advantages, and limitations. • Exploit reduced acuity in peripheral vision for rendering at reduced costs. • Classification and discussion of different techniques. • We discuss application areas where foveated rendering is already in use today. [ABSTRACT FROM AUTHOR]

A novel approach is presented for simulating non-spherical particles in the discrete element method (DEM). A particle's shape is described through a hierarchy of representations using spherical harmonic expansions. The expansion is computed at nodes, obtained by discretising the particle's surface. A low-degree expansion, i.e., one containing few terms, is sufficient to approximate a particle's overall shape without any surface texture. Expansions are computed to high degrees only at interparticle contacts, rather than for the entire particle, which reduces the computational cost. The advantages of this approach include the ability to simulate a wide range of particle shapes and adaptive resolution depending on spatial and temporal considerations. An additional unique benefit is that changes of particle shape due to chipping can be captured in DEM for the first time. This is accomplished by progressively omitting more of the highest-degree terms from the expansion to give an increasingly smooth surface. Unlabelled Image • Spherical harmonic expansions inherently contain a hierarchical description of non-spherical particles. • A set of coefficients ensures that low-degree shape expansions bound higher-degree expansions. • Identification of a suitable discretisation method to describe the particle's surface. • Possibility to simulate particles with adaptive fidelity, depending on location and time. • Abrasion processes captured by omitting terms from the spherical harmonic expansions. [ABSTRACT FROM AUTHOR]

Summary: In this paper, an analog‐to‐digital converter (ADC) with adaptive resolution is presented for wireless neural recording implants. The resolution of the ADC is changed according to the neural signal content, and for this purpose, a continuous‐time (CT) incremental sigma‐delta (IΣ∆) modulator is employed. The ADC digitizes the action potential (AP) and background noise (B‐noise) with 8‐bit and 3‐bit resolutions, respectively. An automatic AP detector is used to separate the APs from the B‐noise in order to select one of the two proportional resolutions. The power dissipation and output data rate of the ADC are reduced by using this technique. Analytical calculations and behavioral simulation results are provided to evaluate the performance of the proposed ADC. To further confirm its efficiency, the circuit‐level implementation of the CT IΣ∆ ADC is presented in Taiwan Semiconductor Manufacturing Company (TSMC) 90‐nm complementary metal‐oxide semiconductor (CMOS) process. According to the simulation results, the proposed ADC achieves 8‐bit or 3‐bit resolution adaptively with 10 kHz bandwidth while the average power consumption is less than 1.89 μW from a single 1‐V power supply. This paper presents an adaptive resolution analog‐to‐digital converter (ADC) for wireless neural recording implants. It employs a continuous‐time incremental sigma‐delta (IΣ∆) modulator with resolution corresponding to the neural signal content. The proposed ADC digitizes the action potential and background noise (B‐noise) with 8‐bit and 3‐bit resolutions, respectively. The power and data rate are reduced significantly because the neural signal contains the B‐Noise in the most of the time, and the ADC mostly operates in the 3‐bit resolution mode. [ABSTRACT FROM AUTHOR]

Densely packed DNA arrays exhibit hexagonal and orthorhombic local packings, as well as a weakly first order transition between them. While we have some understanding of the interactions between DNA molecules in aqueous ionic solutions, the structural details of its ordered phases and the mechanism governing the respective phase transitions between them remains less well understood. Since at high DNA densities, i.e., small interaxial spacings, one can neither neglect the atomic details of the interacting macromolecular surfaces nor the atomic details of the intervening ionic solution, the atomistic resolution is a sine qua non to properly describe and analyze the interactions between DNA molecules. In fact, in order to properly understand the details of the observed osmotic equation of state, one needs to implement multiple levels of organization, spanning the range from the molecular order of DNA itself, the possible ordering of counterions, and then all the way to the induced molecular ordering of the aqueous solvent, all coupled together by electrostatic, steric, thermal and direct hydrogen-bonding interactions. Multiscale simulations therefore appear as singularly suited to connect the microscopic details of this system with its macroscopic thermodynamic behavior. We review the details of the simulation of dense atomistically resolved DNA arrays with different packing symmetries and the ensuing osmotic equation of state obtained by enclosing a DNA array in a monovalent salt and multivalent (spermidine) counterions within a solvent permeable membrane, mimicking the behavior of DNA arrays subjected to external osmotic stress. By varying the DNA density, the local packing symmetry, and the counterion type, we are able to analyze the osmotic equation of state together with the full structural characterization of the DNA subphase, the counterion distribution and the solvent structural order in terms of its different order parameters and consequently identify the most important contribution to the DNA-DNA interactions at high DNA densities. [ABSTRACT FROM AUTHOR]

In this review article, we discuss and analyze some recently developed hybrid atomistic-mesoscopic solvent models for multiscale biomolecular simulations. We focus on the biomolecular applications of the adaptive resolution scheme (AdResS), which allows solvent molecules to change their resolution back and forth between atomistic and coarse-grained representations according to their positions in the system. First, we discuss coupling of atomistic and coarse-grained models of salt solution using a 1-to-1 molecular mapping-i.e., one coarse-grained bead represents one water molecule-for development of a multiscale salt solution model. In order to make use of coarse-grained molecular models that are compatible with the MARTINI force field, one has to resort to a supramolecular mapping, in particular to a 4-to-1 mapping, where four water molecules are represented with one coarse-grained bead. To this end, bundled atomistic water models are employed, i.e., the relative movement of water molecules that are mapped to the same coarse-grained bead is restricted by employing harmonic springs. Supramolecular coupling has recently also been extended to polarizable coarse-grained water models with explicit charges. Since these coarse-grained models consist of several interaction sites, orientational degrees of freedom of the atomistic and coarse-grained representations are coupled via a harmonic energy penalty term. The latter aligns the dipole moments of both representations. The reviewed multiscale solvent models are ready to be used in biomolecular simulations, as illustrated in a few examples. [ABSTRACT FROM AUTHOR]

Due to the Lagrangian nature of Smoothed Particle Hydrodynamics (SPH), the adaptive resolution remains a challenging task. In this work, we first analyse the influence of the simulation parameters and the smoothing length on solution accuracy, in particular in high strain regions. Based on this analysis we develop a novel approach to dynamically adjust the kernel range for each SPH particle separately, accounting for the local flow kinematics. We use the Okubo–Weiss parameter that distinguishes the strain and vorticity dominated regions in the flow domain. The proposed development is relatively simple and implies only a moderate computational overhead. We validate the modified SPH algorithm for a selection of two-dimensional test cases: the Taylor–Green flow, the vortex spin-down, the lid-driven cavity and the dam-break flow against a sharp-edged obstacle. The simulation results show good agreement with the reference data and improvement of the long-term accuracy for unsteady flows. For the lid-driven cavity case, the proposed dynamical adjustment remedies the problem of tensile instability (particle clustering). [ABSTRACT FROM AUTHOR]

With the rapid development of portable electronics, wearable devices have become widely used to monitor body signals for long-term health care and home care applications. They detect vital signals through physiological sensors and then transmit them to a cloud database for evaluation and monitoring purposes through wireless communication systems. In this paper, a smart analog-to-digital converter (ADC) was realized by a mixed-signal application-specific integrated circuit (ASIC) based on adaptive resolution and lossless compression techniques for electrocardiogram (ECG) signal monitoring. The sampling clock for the ADC can be adaptively selected according to the characteristic of the signals. The lossless encoder consists of trend forecasting and entropy coding modules. The transmission data rate was decreased efficiently by adaptive resolution and lossless compression techniques. The chip aims to meet the low power consumption for the design, because it reduced the signal transmission rate and maintained high-quality ECG signal detection. The proposed mixed-signal ASIC design was realized using a 0.18- \mu \textm CMOS process with a total power consumption of 78.8~\mu \textW when operating at 1 kHz and a total chip area of 850 \times 850~\mu \textm^2 . [ABSTRACT FROM PUBLISHER]

In this work, we make an attempt to answer the question of what a multiscale problem is in Molecular Dynamics (MD), or, more in general, in Molecular Simulation (MS). By introducing the criterion of separability of scales, we identify three major (reference) categories of multiscale problems and discuss their corresponding computational strategies by making explicit examples of applications.

Densely packed DNA arrays exhibit hexagonal and orthorhombic local packings, as well as a weakly first order transition between them. While we have some understanding of the interactions between DNA molecules in aqueous ionic solutions, the structural details of its ordered phases and the mechanism governing the respective phase transitions between them remains less well understood. Since at high DNA densities, i.e., small interaxial spacings, one can neither neglect the atomic details of the interacting macromolecular surfaces nor the atomic details of the intervening ionic solution, the atomistic resolution is a sine qua non to properly describe and analyze the interactions between DNA molecules. In fact, in order to properly understand the details of the observed osmotic equation of state, one needs to implement multiple levels of organization, spanning the range from the molecular order of DNA itself, the possible ordering of counterions, and then all the way to the induced molecular ordering of the aqueous solvent, all coupled together by electrostatic, steric, thermal and direct hydrogen-bonding interactions. Multiscale simulations therefore appear as singularly suited to connect the microscopic details of this system with its macroscopic thermodynamic behavior. We review the details of the simulation of dense atomistically resolved DNA arrays with different packing symmetries and the ensuing osmotic equation of state obtained by enclosing a DNA array in a monovalent salt and multivalent (spermidine) counterions within a solvent permeable membrane, mimicking the behavior of DNA arrays subjected to external osmotic stress. By varying the DNA density, the local packing symmetry, and the counterion type, we are able to analyze the osmotic equation of state together with the full structural characterization of the DNA subphase, the counterion distribution and the solvent structural order in terms of its different order parameters and consequently identify the most important contribution to the DNA-DNA interactions at high DNA densities.

For promoting numerical efficiency in smoothed particle hydrodynamics (SPH) simulation of underwater explosion, a pressure-dependent adaptive resolution scheme (PARS) is proposed. The approach involves dividing computation domain into several regions of different resolution and automatic splitting or merging particles in these regions. Methods are established for identifying resolution regions and for adapting particle resolutions, so that the wave front and the zone with sharp pressure variation is dynamically covered in the region of the highest resolution. Numerical examples are present to demonstrate that the proposed technique can reduce significantly the number of particles and computing time to produce solutions of adequate accuracy. The main features of shockwave propagation can be well captured using much less particles. For large scale or 3D problems, the technique can be decisive for completing a numerical simulation. Numerical examples also testified that the proposed PARS is robust and effective in capturing the complicated phenomena of wave interferences and interaction between shock wave and solid barrier. The authors proposed a pressure-dependent adaptive particle resolution scheme for efficient and robust simulation of underwater explosions. The work highlights the following major points. • A novel pressure-dependent adaptive particle resolution technique is established for SPH simulation of underwater explosion to achieve solutions of adequate accuracy with significantly less particles and computation time. • A parameter depending on a pre-estimated peak pressure is introduced for dividing a computation domain into several adaptive regions of different resolution, so that the zone with sharp variation of pressure and the wave front can be covered in the region of the highest resolution. Criteria are established for automatic splitting and merging the particles in different adaptive regions. • The method is numerically efficient, and can properly capture the propagation of shock wave, the peak pressure, and the main features of wave interferences and wave-solid barrier interactions. [ABSTRACT FROM AUTHOR]

ABSTRACT In multi-resolution simulations, different system components are simultaneously modeled at different levels of resolution, these being smoothly coupled together. In the case of enzyme systems, computationally expensive atomistic detail is needed in the active site to capture the chemistry of ligand binding. Global properties of the rest of the protein also play an essential role, determining the structure and fluctuations of the binding site; however, these can be modeled on a coarser level. Similarly, in the most computationally efficient scheme only the solvent hydrating the active site requires atomistic detail. We present a methodology to couple atomistic and coarse-grained protein models, while solvating the atomistic part of the protein in atomistic water. This allows a free choice of which protein and solvent degrees of freedom to include atomistically. This multi-resolution methodology can successfully model stable ligand binding, and we further confirm its validity by exploring the reproduction of system properties relevant to enzymatic function. In addition to a computational speedup, such an approach can allow the identification of the essential degrees of freedom playing a role in a given process, potentially yielding new insights into biomolecular function. Proteins 2016; 84:1902-1913. © 2016 Wiley Periodicals, Inc. [ABSTRACT FROM AUTHOR]

Speed profile optimization plays an important role in optimal train control. Considering the characteristics of an electrical locomotive with regenerative braking, this paper proposes a new algorithm for target speed profile approximation. This paper makes the following three contributions: First, it proves that under a certain calculation precision, there is an optimal coast-brake switching region--not just a point where the train should be switched from coasting mode to braking mode. This is very useful in engineering applications. Second, the paper analyzes the influence of regenerative braking on the optimal coast-brake switching region and proposes an approximate linear relationship between the optimal switching region and the regeneration efficiency. Third, the paper presents an average speed equivalent algorithm for local speed profile optimization in steep sections. In addition, this paper simplifies the proof of the optimality of smooth running on a non-steep track in two steps. The effects on energy consumption from two important factors (optimal coast and running time) are systematically analyzed. Extensive simulations verify these points of view and demonstrate that the proposed approximation approach is computationally efficient and achieves sufficient engineering accuracy. [ABSTRACT FROM AUTHOR]

A new adaptive resolution technique for particle-based multi-level simulations of fluids is presented. In the approach, the representation of fluid and solvent particles is changed on the fly between an atomistic and a coarse-grained description. The present approach is based on a hybrid coupling of the multiparticle collision dynamics (MPC) method and molecular dynamics (MD), thereby coupling stochastic and deterministic particle-based methods. Hydrodynamics is examined by calculating velocity and current correlation functions for various mixed and coupled systems. We demonstrate that hydrodynamic properties of the mixed fluid are conserved by a suitable coupling of the two particle methods, and that the simulation results agree well with theoretical expectations. [ABSTRACT FROM AUTHOR]

In this work, we make an attempt to answer the question of what a multiscale problem is in Molecular Dynamics (MD), or, more in general, in Molecular Simulation (MS). By introducing the criterion of separability of scales, we identify three major (reference) categories of multiscale problems and discuss their corresponding computational strategies by making explicit examples of applications. [ABSTRACT FROM AUTHOR]

This article deals with the molecular dynamics simulation of open systems that can exchange energy and matter with a reservoir; the physics of the reservoir and its interactions with the system are described by the model introduced by Bergmann and Lebowitz (P G Bergmann and J L Lebowitz 1955 Phys. Rev. http://dx.doi.org/10.1103/PhysRev.99.578 99 http://dx.doi.org/10.1103/PhysRev.99.578 ). Despite its conceptual appeal, the model did not gain popularity in the field of molecular simulation and, as a consequence, did not play a role in the development of open system molecular simulation techniques, even though it can provide the conceptual legitimation of simulation techniques that mimic open systems. We shall demonstrate that the model can serve as a tool in devising both numerical procedures and conceptual definitions of physical quantities that cannot be defined in a straightforward way by systems with a fixed number of molecules. In particular, we discuss the utility of the Bergmann–Lebowitz (BL) model for the calculation of equilibrium time correlation functions within the grand canonical adaptive resolution method (GC-AdResS) and report numerical results for the case of liquid water.

This paper presents a clock-less 8b ADC in 130 nm CMOS technology, which uses signal-dependent sampling rate and adaptive resolution through a time-varying comparison window, for applications with sparse input signals. Input-dependent dynamic bias is used to reduce comparator delay dispersion, thus helping to maintain SNDR while saving power. Alias-free operation with SNDR in the range of 47–54 dB, which partly exceeds the theoretical limit of 8b conventional converters, is achieved over a 20 kHz bandwidth with 3–9 µW power from a 0.8 V supply. [ABSTRACT FROM PUBLISHER]

Abstract: We present a novel adaptive-resolution particle method for continuous parabolic problems. In this method, particles self-organize in order to adapt to local resolution requirements. This is achieved by pseudo forces that are designed so as to guarantee that the solution is always well sampled and that no holes or clusters develop in the particle distribution. The particle sizes are locally adapted to the length scale of the solution. Differential operators are consistently evaluated on the evolving set of irregularly distributed particles of varying sizes using discretization-corrected operators. The method does not rely on any global transforms or mapping functions. After presenting the method and its error analysis, we demonstrate its capabilities and limitations on a set of two- and three-dimensional benchmark problems. These include advection–diffusion, the Burgers equation, the Buckley–Leverett five-spot problem, and curvature-driven level-set surface refinement. [Copyright &y& Elsevier]

Simulating complex fluids or in general complex molecular systems requires approaches covering decades of time and length scales. This usually cannot be achieved within one simulation model. Over the years many different methods and models have been developed ranging from rather generic models, representing most efficiently the universal statistical mechanical properties of e.g. polymers, to all atom models and even quantum mechanical treatments. While these allow for scientifically very important studies in their own right, only a combination and close link between models of different levels allows for a truly quantitative description of materials and processes. In the present contribution we discuss an adaptive resolution approach where different levels of detail are treated within one simulation and the molecules are free to diffuse between different regions in space, where the molecules interact with different interaction potentials. [ABSTRACT FROM AUTHOR]

We propose a model-based learning algorithm, the Adaptive-resolution Reinforcement Learning (ARL) algorithm, that aims to solve the online, continuous state space reinforcement learning problem in a deterministic domain. Our goal is to combine adaptive-resolution approximation schemes with efficient exploration in order to obtain polynomial learning rates. The proposed algorithm uses an adaptive approximation of the optimal value function using kernel-based averaging, going from coarse to fine kernel-based representation of the state space, which enables us to use finer resolution in the 'important' areas of the state space, and coarser resolution elsewhere. We consider an online learning approach, in which we discover these important areas online, using an uncertainty intervals exploration technique. In addition, we introduce an incremental variant of the ARL (IARL), which is a more practical version of the original algorithm with reduced computational complexity at each stage. Polynomial learning rates in terms of mistake bound (in a PAC framework) are established for these algorithms, under appropriate continuity assumptions. [ABSTRACT FROM AUTHOR]

In this paper, we present an adaptive model for dynamically deforming hyper-elastic rods. In contrast to existing approaches, adaptively introduced control points are not governed by geometric subdivision rules. Instead, their states are determined by employing a non-linear energy-minimization approach. Since valid control points are computed instantaneously, post-stabilization schemes are avoided and the stability of the dynamic simulation is improved. Due to inherently complex contact configurations, the simulation of knot tying using rods is a challenging task. In order to address this problem, we combine our adaptive model with a robust and accurate collision handling method for elastic rods. By employing our scheme, complex knot configurations can be simulated in a physically plausible way. [ABSTRACT FROM AUTHOR]