Showing papers by "Nicolae Goga published in 2015"

Benchmark of Schemes for Multiscale Molecular Dynamics Simulations

Abstract: In multiscale molecular dynamics simulations the accuracy of detailed models is combined with the efficiency of a reduced representation. For several applications - namely those of sampling enhancement - it is desirable to combine fine-grained (FG) and coarse-grained (CG) approaches into a single hybrid approach with an adjustable mixing parameter. We present a benchmark of three algorithms that use a mixing of the two representation layers using a Lagrangian formalism. The three algorithms use three different approaches for keeping the particles at the FG level of representation together: 1) addition of forces, 2) mass scaling, and 3) temperature scaling. The benchmark is applied to liquid hexadecane and includes an evaluation of the average configurational entropy of the FG and CG subsystems. The temperature-scaling scheme achieved a 3-fold sampling speedup with little deviation of FG properties. The addition-of-forces scheme kept FG properties the best but provided little sampling speedup. The mass-scaling scheme yielded a 5-fold speedup but deviated the most from FG properties.

Neural network based edge detection for CBCT segmentation

Abstract: Edge detection is an important task in image processing, many times as part of the segmentation process. When segmentation is performed in medical imaging, one of the preferred tools is neural networks, because of their capabilities of adaptive learning and non-linear mapping. We present in this paper the neural network tools used for edge detection and we propose one that is able to perform edge detection in dental Cone Beam Computer Tomography (CBCT) images, a necessary step for the teeth 3D reconstruction.

Automatic segmentation for 3D dental reconstruction

Abstract: Cone Beam Computed Tomography (CBCT) is one of the favorite imaging technologies in dentistry because if offers 3D images in conditions of reduced irradiation. High quality 3D imaging is essential for first-rate diagnostic and treatment, despite the rather low quality images, with low contrast and high noise. The paper presents an automatic method for 3D reconstruction of the oral cavity. We analyze the existing state of the art in the field and propose a segmentation method that uses the domain knowledge related to mouth and teeth to reduce the computational load and to provide good results in an automatic procedure. The method is based on a Canny type edge detector, that was found to offer better control than active contour methods.

Knowledge Based Segmentation for Fast 3D Dental Reconstruction from CBCT

Abstract: We present in this paper a fast method for the reconstruction of the dental 3D structure using the images obtained from Cone Beam Computed Tomography (CBCT). A high-quality visualization of the dental tissues, including 3D reconstruction, is essential for the accurate diagnosis and treatment in dentistry. The segmentation for the dental CBCT images presents several major problems because of the noisiness of the images. We analyze the existing methods and propose one that uses the information for the dental structure and distribution to reduce the computational load and to provide good results in an automatic procedure.

Improved GROMACS algorithms using the MPI parallelization

Abstract: Molecular dynamics (MD) studies the structure of molecular systems which are subject to certain constraints and forces. The simulation of particles is a tool for examining atomic systems during a period of nanoseconds in which the trajectory of atoms and the state of the system is analyzed. GROMACS is a package which supports molecular dynamics simulations and energy minimization, being started the University of Groningen. Usually molecular dynamics simulations are time consuming, sometimes taking weeks and even months. In order to obtain the output of the simulation in less time, parallelization is used through the use of MPI (Message Passing Interface). The article presents the MPI parallelization of a novel thermostat algorithm for molecular dynamics and experimental results.

Improved OpenMM algorithms for CUDA and OpenCL using new stochastic dynamics and the Berendsen thermostat

Abstract: The simulations done in molecular dynamics (MD) are used to learn about the behavior of macromolecular systems. While analyzing large macromolecular systems, the computation can last long, many times for days, weeks and months, due to which the need of parallelization occurs for shortening the time. The graphics processing unit (GPU) yields to multithread computational capacity by the use of CUDA high-level language which suits the molecular dynamics algorithms. The OpenCL platform allows its reckoning kernels to run without difficulty on GPUs, as well as on multicore computers. We developed a new thermostat that combines Langevin dynamics with the global Berendsen thermostat which is parallelized in CUDA, as well as in OpenCL. The GPU parallelization is presented in this paper by applying stochastic dynamics, respectively the Langevin integrator, together with the Berendsen thermostat. A system of atoms is tested by taking into consideration the friction coefficient and the coupling parameter.

Automatic contour detection from dental CBCT DICOM data

Abstract: The algorithm described in this paper is part of a program supporting the process of teeth contour detection, segmentation and 3D reconstruction. The data is taken from dental conical tomographies. Automatic contour detection starts by filtering the image according to an adaptive threshold. The region of interest is established and the tooth contour is determined using the possible edge directions. The applied method is innovative for the contour detection, because it exploits the domain knowledge, where each tooth (incisor, canine, premolar, molar) has a certain shape. For touching teeth, the delimitation is automatically performed by drawing a line using the Bresenham algorithm. The algorithm will further contribute to the determination and treatment of teeth pathologies.

Enterprise domain ontology learning from web-based corpus

Abstract: Enterprise knowledge is a key asset in the competing and fast-changing corporate landscape. The ability to learn, store and distribute implicit and explicit knowledge can be the difference between success and failure. While enterprise knowledge management is a well-defined research domain, current implementations lack orientation towards small and medium enterprise. We propose a semantic search engine for relevant documents in an enterprise, based on automatic generated domain ontologies. In this paper we focus on the component for ontology learning and population.

Towards a new computing problem solving model for organizations

Abstract: Computing methods for solving problems effectively are the essence of continual improvement in computer based learning. The aim of this article is to present a new computational approach in solving problems of an organization by stepwise computer assisted finding a reliable solution. The new approach that we propose is to use Kumar's algorithm for supremal controllable supervisor and pattern recognition to find the optimal solution. Having Euclidian plane geometry as an universal language, we will exemplify the new model by taking an example from it. By dividing the types of the problems in various categories we will try to find patterns and to use them for faster problem solving.