SEARCH · PubMed Health

Results for “Gaussian process”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 631 records · Page 35Linked to original sources

PubMed2006

Application of the thermorheologically complex nonlinear Adam-Gibbs model for the glass transition to molecular motion in hydrated proteins.

The nonlinear thermorheologically complex Adam Gibbs (extended "Scherer-Hodge") model for the glass transition is applied to enthalpy relaxation data reported by Sartor, Mayer, and Johari for hydrated methemoglobin. A sensible range in values for the average localized activation energy is obtained (100-200 kJ mol(-1)). The standard deviation in the inferred Gaussian distribution of activation energies, computed from the reported KWW beta-parameter, is approximately 30% of the average, consistent with the suggestion that some relaxation processes in hydrated proteins have exceptionally low activation energies.

Biophysics↗

PubMed1991

Neural network analyses of stochastic information: application to neurobiological data.

Simultaneous recordings from over 50 neural cells were obtained from the dragonfly ganglia. To explore the biological information processing strategies reflected therein, data analysis methods were designed for use with artificial neural networks (ANN). Most methods are degraded by different cell spike trains that vary in mean firing frequencies by well over an order of magnitude. Based on underlying cell physiology, the occurrence of each spike is likely to be a stochastic function. To overcome such degradation problems in ANN use, a gaussian spike train representation was synthesized for each cell using raw data. This representation retained the exact spiking times and provided a biologically plausible probabilistic value for the time of occurrence for each spike. A 3-layer, feed-forward, ANN was trained on these data using a gradient descent learning algorithm. The task was to predict the neural activity at time (t + 1) given the neural activity at time (t). Following training, the network sum-squared prediction error was less than 0.01. Further, the temporal reproduction of the neural firing patterns was corroborated. The results indicated that the ANN could accurately reproduce the neural firing patterns in both the spatial and temporal domain using the stochastic spike train data. Encoding parameters for the spike trains using synthesized gaussian representations were optimized. The "lesion" studies were performed to determine the contribution of each cell to ANN predictions. The capability to "fine tune" both the information representation of spike trains and the ANN architecture provides significant advantages in the analysis of biological information processing by neural cells.(ABSTRACT TRUNCATED AT 250 WORDS)

Animals↗

PubMed1999

Survey: interpolation methods in medical image processing.

Image interpolation techniques often are required in medical imaging for image generation (e.g., discrete back projection for inverse Radon transform) and processing such as compression or resampling. Since the ideal interpolation function spatially is unlimited, several interpolation kernels of finite size have been introduced. This paper compares 1) truncated and windowed sinc; 2) nearest neighbor; 3) linear; 4) quadratic; 5) cubic B-spline; 6) cubic; g) Lagrange; and 7) Gaussian interpolation and approximation techniques with kernel sizes from 1 x 1 up to 8 x 8. The comparison is done by: 1) spatial and Fourier analyses; 2) computational complexity as well as runtime evaluations; and 3) qualitative and quantitative interpolation error determinations for particular interpolation tasks which were taken from common situations in medical image processing. For local and Fourier analyses, a standardized notation is introduced and fundamental properties of interpolators are derived. Successful methods should be direct current (DC)-constant and interpolators rather than DC-inconstant or approximators. Each method's parameters are tuned with respect to those properties. This results in three novel kernels, which are introduced in this paper and proven to be within the best choices for medical image interpolation: the 6 x 6 Blackman-Harris windowed sinc interpolator, and the C2-continuous cubic kernels with N = 6 and N = 8 supporting points. For quantitative error evaluations, a set of 50 direct digital X rays was used. They have been selected arbitrarily from clinical routine. In general, large kernel sizes were found to be superior to small interpolation masks. Except for truncated sinc interpolators, all kernels with N = 6 or larger sizes perform significantly better than N = 2 or N = 3 point methods (p << 0.005). However, the differences within the group of large-sized kernels were not significant. Summarizing the results, the cubic 6 x 6 interpolator with continuous second derivatives, as defined in (24), can be recommended for most common interpolation tasks. It appears to be the fastest six-point kernel to implement computationally. It provides eminent local and Fourier properties, is easy to implement, and has only small errors. The same characteristics apply to B-spline interpolation, but the 6 x 6 cubic avoids the intrinsic border effects produced by the B-spline technique. However, the goal of this study was not to determine an overall best method, but to present a comprehensive catalogue of methods in a uniform terminology, to define general properties and requirements of local techniques, and to enable the reader to select that method which is optimal for his specific application in medical imaging.

Diagnostic Imaging↗

PubMed1995

Wavelet analysis of high-resolution ECGs in post-infarction patients: role of the basic wavelet and of the analyzed lead.

Wavelet analysis provides a fruitful alternative to standard techniques for the detection of fractionated potentials in signal averaged high-resolution (SA-HR) ECGs. In this study, an attempt is made to optimize the discrimination of post infarction patients prone to ventricular tachycardia (VT), using wavelet analysis. Optimization is based on the choice of the ECG leads or lead combinations to be analyzed, and on the analyzing wavelet to be computed. A set of 40 post-infarction patients (20 patients with VT and 20 patients without any arrhythmia) is analyzed. Individual leads and lead combinations of the SA-HR ECGs are processed using a multiparametric algorithm, based on coherent detection of aligned local maxima of the wavelet transform. Seven basic wavelets are tested: the Morlet's wavelet, and the six first derivatives of a Gaussian function. The first derivative of a Gaussian function provides poor results, and is discarded. All other wavelets prove to perform equivalent classification. A vector magnitude computed from the wavelet transforms of the three SA-HR ECGs achieves better results than individual leads. An optimized risk stratification algorithm leads to 90% sensitivity and 100% specificity in the 40 patients learning set.

Adult↗

PubMed2004

Multisensor optimal information fusion input white noise deconvolution estimators.

The unified multisensor optimal information fusion criterion weighted by matrices is rederived in the linear minimum variance sense, where the assumption of normal distribution is avoided. Based on this fusion criterion, the optimal information fusion input white noise deconvolution estimators are presented for discrete time-varying linear stochastic control system with multiple sensors and correlated noises, which can be applied to seismic data processing in oil exploration. A three-layer fusion structure with fault tolerant property and reliability is given. The first fusion layer and the second fusion layer both have netted parallel structures to determine the first-step prediction error cross-covariance for the state and the estimation error cross-covariance for the input white noise between any two sensors at each time step, respectively. The third fusion layer is the fusion center to determine the optimal matrix weights and obtain the optimal fusion input white noise estimators. The simulation results for Bernoulli-Gaussian input white noise deconvolution estimators show the effectiveness.

Algorithms↗

PubMed1998

Butterworth equations for homomorphic filtering of images.

In digital image processing, the homomorphic filtering approach is derived from an illumination-reflectance model of the image. Homomorphic filtering can perform simultaneous dynamic range compression and contrast enhancement. Crucial for the success of the homomorphic approach is the selection of an appropriate frequency-domain filter function in order to modify the illumination and reflectance components of an image differently. The author found Butterworth type highpass equations far superior to other frequency-domain filter functions, including Gaussian equations, making the Butterworth highpass suitable for use with the homomorphic filtering approach. The program was written in Microsoft (MS) Visual C++ (filter) as well as MS Visual Basic (user interface) to run as a module under the image processing software package Image-Pro Plus.

Blood Cells↗

PubMed2002

Lipid membrane phase behaviour elucidated in real time by controlled environment atomic force microscopy.

Lipids are integral components of all biological membranes. Understanding the physical and chemical properties of these lipids is critical to our understanding of membrane functions. We developed a new atomic force microscope (AFM) approach to visualize in real time the temperature-induced lipid phase transition and domain separation processes in 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) membranes and estimate the thermodynamics of the phase transition process. The gel and liquid crystalline phases of DMPC coexisted over a broad temperature range (approximately 10 degrees C). Equal partitioning into two phases occurred at a transition temperature (Tm) of 28.5 degrees C. We developed a mathematical model to analyse AFM-derived DMPC membrane height changes as multi-peak Gaussian distributions. This approach allowed us to estimate the DMPC domain size, N, as 18-75 molecules per leaflet corresponding to a -4.2 nm diameter circular nanodomain. Lipid nanodomains may organize into microdomains or rafts which, in concert with proteins and other lipid components, play an important dynamic role in many biomedically important processes.

Dimyristoylphosphatidylcholine↗

PubMed1998

Retrieval dynamics in oscillator neural networks.

We present an analytical approach that allows us to treat the long-time behavior of the recalling process in an oscillator neural network. It is well known that in coupled oscillatory neuronal systems, under suitable conditions, the original dynamics can be reduced to a simpler phase dynamics. In this description, the phases of the oscillators can be regarded as the timings of the neuronal spikes. To attempt an analytical treatment of the recalling dynamics of such a system, we study a simplified model in which we discretize time and assume a synchronous updating rule. The theoretical results show that the retrieval dynamics is described by recursion equations for some macroscopic parameters, such as an overlap with the retrieval pattern. We then treat the noise components in the local field, which arise from the learning of the unretrieved patterns, as gaussian variables. However, we take account of the temporal correlation between these noise components at different times. In particular, we find that this correlation is essential for correctly predicting the behavior of the retrieval process in the case of autoassociative memory. From the derived equations, the maximal storage capacity and the basin of attraction are calculated and graphically displayed. We also consider the more general case that the network retrieves an ordered sequence of phase patterns. In both cases, the basin of attraction remains sufficiently wide to recall the memorized pattern from a noisy one, even near saturation. The validity of these theoretical results is supported by numerical simulations. We believe that this model serves as a convenient starting point for the theoretical study of retrieval dynamics in general oscillatory systems.

Artifacts↗

PubMed2000

Scene dependence of the non-Gaussian scaling properties of natural images.

We report results on the scaling properties of changes in contrast of natural images in different visual environments. This study confirms the existence, in a vast class of images, of a multiplicative process relating the variations in contrast seen at two different scales, as was found in Turiel et al (Turiel A, Mato G, Parga N and Nadal J-P 1998 Self-Similarity Properties of Natural Images: Proc. NIPS'97 (Cambridge, MA: MIT Press), Turiel A, Mato G, Parga N and Nadal J-P 1998 Phys. Rev. Lett. 80 1098-101). But it also shows that the scaling exponents are not universal: even if most images follow the same type of statistics, they do it with different values of the distribution parameters. Motivated by these results, we also present the analysis of a generative model of images that reproduces those properties and that has the correct power spectrum. Possible implications for visual processing are also discussed.

Animals↗

PubMed2005

A modified OSEM algorithm for PET reconstruction using wavelet processing.

Ordered subset expectation-maximization (OSEM) method in positron emission tomography (PET) has been very popular recently. It is an iterative algorithm and provides images with superior noise characteristics compared to conventional filtered backprojection (FBP) algorithms. Due to the lack of smoothness in images in OSEM iterations, however, some type of inter-smoothing is required. For this purpose, the smoothing based on the convolution with the Gaussian kernel has been used in clinical PET practices. In this paper, we incorporated a robust wavelet de-noising method into OSEM iterations as an inter-smoothing tool. The proposed wavelet method is based on a hybrid use of the standard wavelet shrinkage and the robust wavelet shrinkage to have edge preserving and robust de-noising simultaneously. The performances of the proposed method were compared with those of the smoothing methods based on the convolution with Gaussian kernel using software phantoms, physical phantoms, and human PET studies. The results demonstrated that the proposed wavelet method provided better spatial resolution characteristic than the smoothing methods based on the Gaussian convolution, while having comparable performance in noise removal.

Algorithms↗

PubMed1998

An efficient method for correcting the edge artifact due to smoothing.

Spatial smoothing is a common pre-processing step in the analysis of functional brain imaging data. It can increase sensitivity to signals of specific shapes and sizes (Rosenfeld and Kak [1982]: Digital Picture Processing, vol. 2. Orlando, Fla.: Academic; Worsley et al. [1996]: Hum Brain Mapping 4:74-90). Also, some amount of spatial smoothness is required if methods from the theory of Gaussian random fields are to be used (Holmes [1994]: Statistical Issues in Functional Brain Mapping. PhD thesis, University of Glasgow). Smoothing is most often implemented as a convolution of the imaging data with a smoothing kernel, and convolution is most efficiently performed using the Convolution Theorem and the Fast Fourier Transform (Cooley and Tukey [1965]: Math Comput 19:297-301; Priestly [1981]: Spectral Analysis and Time Series. San Diego: Academic; Press et al. [1992]: Numerical Recipes in C: The Art of Scientific Computing, 2nd ed. Cambridge: Cambridge University Press). An undesirable side effect of smoothing is an artifact along the edges of the brain, where brain voxels become smoothed with non-brain voxels. This results in a dark rim which might be mistaken for hypoactivity. In this short methodological paper, we present a method for correcting functional brain images for the edge artifact due to smoothing, while retaining the use of the Convolution Theorem and the Fast Fourier Transform for efficient calculation of convolutions.

Artifacts↗

PubMed1996

Covisualization by computational optical-sectioning microscopy of integrin and associated proteins at the cell membrane of living onion protoplasts.

Using higher-resolution wide-field computational optical-sectioning fluorescence microscopy, the distribution of antigens recognized by antibodies against animal beta 1 integrin, fibronectin, and vitronectin has been visualized at the outer surface of enzymatically protoplasted onion epidermis cells and in depectinated cell wall fragments. On the protoplast all three antigens are colocalized in an array of small spots, as seen in raw images, in Gaussian filtered images, and in images restored by two different algorithms. Fibronectin and vitronectin but not beta 1 integrin antigenicities colocalize as puncta in comparably prepared and processed images of the wall fragments. Several control visualizations suggest considerable specifity of antibody recognition. Affinity purification of onion cell extract with the same anti-integrin used for visualization has yielded protein that separates in SDS-PAGE into two bands of about 105-110 and 115-125 kDa. These bands are again recognized by the visualization antibody, which was raised against the extracellular domain of chicken beta 1 integrin, and are also recognized by an antibody against the intracellular domain of chicken beta 1 integrin. Because beta 1 integrin is a key protein in numerous animal adhesion sites, it appears that the punctate distribution of this protein in the cell membranes of onion epidermis represents the adhesion sites long known to occur in cells of this tissue. Because vitronectin and fibronection are matrix proteins that bind to integrin in animals, the punctate occurrence of antigenically similar proteins both in the wall (matrix) and on enzymatically prepared protoplasts reinforces the concept that onion cells have adhesion sites with some similarity to certain kinds of adhesion sites in animals.

Binding Sites, Antibody↗

PubMed2000

Spatial resolution and receptive field height of motion sensors in human vision.

We estimated the length of motion-detecting receptive fields in human vision by measuring direction discrimination for three novel stimuli. The motion sequences contained either (i) alternate frames of two differently oriented sinusoidal gratings; (ii) alternate frames of vertical grating and plaid stimuli or (iii) a vertical grating divided into horizontal strips of equal height, where alternate strips moved leftward and rightward. All three stimulus sequences had a similar appearance (of moving strips) and the task was to identify the direction of the central strip. For sequences (ii) and (iii), performance fell as the strip height decreased. Threshold height fell with increasing contrast up to about 20%, then levelled off at the critical strip height. Temporal frequency (1. 9-15 Hz) had no effect on the critical strip height. We argue that the receptive field length corresponds to twice this critical height. The length estimates were strikingly short, ranging from about 0.4 cycles at 3.0 cpd to 0.1 cycles at 0.1 cpd. These lengths agree well with the estimates derived at threshold by Anderson and Burr (1991, J. Opt. Soc. Am. A8, 1330-1339), and imply that the motion-sensing filters have very broad orientation tuning. These and other results are interpreted within the framework of a Gaussian derivative model for motion filtering. The sensitivity of motion filters to a broad range of orientations suggests a simpler view of how coherent plaid motion is processed.

Humans↗

PubMed2001

Optimization of digital filters for the detection of trace elements in EELS. III--Gaussian, homomorphic and adaptive filters.

This paper is the third part of a series of papers dealing with the development of digital filters for the extraction of information from electron energy loss spectra, when the signal-to-background is very weak, as it is the case for elements in weak concentration. Three new filtering approaches are investigated: the use of high-order derivatives (typically, the fourth order derivative of a Gaussian), the use of homomorphic filtering (working on logarithmically-transformed spectra) and the use of adaptive filtering, taking into account the local background shape. These signal processing methods are applied to real spectra to demonstrate and comment their effects in three different cases of signal and background shapes related to calcium, phosphorus and boron detection in definite thickness organic matrix.

Boron↗

PubMed2003

Comparison of ultraviolet femtosecond and nanosecond laser ablation inductively coupled plasma mass spectrometry analysis in glass, monazite, and zircon.

We compared the analytical performance of ultraviolet femtosecond and nanosecond laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS). The benefit of ultrafast lasers was evaluated regarding thermal-induced chemical fractionation, that is otherwise well known to limit LA-ICPMS. Both lasers had a Gaussian beam energy profile and were tested using the same ablation system and ICPMS analyzer. Resulting crater morphologies and analytical signals showed more straightforward femtosecond laser ablation processes, with minimal thermal effects. Despite a less stable energy output, the ultrafast laser yielded elemental (Pb/U, Pb/Th) and Pb isotopic ratios that were more precise, repeatable, and accurate, even when compared to the best analytical conditions for the nanosecond laser. Measurements on NIST glasses, monazites, and zircon also showed that femtosecond LA-ICPMS calibration was less matrix-matched dependent and therefore more versatile.

Journal Article↗

PubMed2005

Mean-field dynamics with stochastic decoherence (MF-SD): a new algorithm for nonadiabatic mixed quantum/classical molecular-dynamics simulations with nuclear-induced decoherence.

The key factors that distinguish algorithms for nonadiabatic mixed quantum/classical (MQC) simulations from each other are how they incorporate quantum decoherence-the fact that classical nuclei must eventually cause a quantum superposition state to collapse into a pure state-and how they model the effects of decoherence on the quantum and classical subsystems. Most algorithms use distinct mechanisms for modeling nonadiabatic transitions between pure quantum basis states ("surface hops") and for calculating the loss of quantum-mechanical phase information (e.g., the decay of the off-diagonal elements of the density matrix). In our view, however, both processes should be unified in a single description of decoherence. In this paper, we start from the density matrix of the total system and use the frozen Gaussian approximation for the nuclear wave function to derive a nuclear-induced decoherence rate for the electronic degrees of freedom. We then use this decoherence rate as the basis for a new nonadiabatic MQC molecular-dynamics (MD) algorithm, which we call mean-field dynamics with stochastic decoherence (MF-SD). MF-SD begins by evolving the quantum subsystem according to the time-dependent Schrodinger equation, leading to mean-field dynamics. MF-SD then uses the nuclear-induced decoherence rate to determine stochastically at each time step whether the system remains in a coherent mixed state or decoheres. Once it is determined that the system should decohere, the quantum subsystem undergoes an instantaneous total wave-function collapse onto one of the adiabatic basis states and the classical velocities are adjusted to conserve energy. Thus, MF-SD combines surface hops and decoherence into a single idea: decoherence in MF-SD does not require the artificial introduction of reference states, auxiliary trajectories, or trajectory swarms, which also makes MF-SD much more computationally efficient than other nonadiabatic MQC MD algorithms. The unified definition of decoherence in MF-SD requires only a single ad hoc parameter, which is not adjustable but instead is determined by the spatial extent of the nonadiabatic coupling. We use MF-SD to solve a series of one-dimensional scattering problems and find that MF-SD is as quantitatively accurate as several existing nonadiabatic MQC MD algorithms and significantly more accurate for some problems.

Journal Article↗

PubMed1999

Tomographic reconstruction of the retina using a confocal scanning laser ophthalmoscope.

Retinal imaging with a confocal scanning laser ophthalmoscope (cSLO) involves scanning a small laser beam over the retina and constructing an image from the reflected light. By applying the confocal principle, tomographic images can be produced. However, the thickness of such slices, when compared with the retinal thickness, is too large to give useful 3D retinal images. In this study an algorithm has been developed which fits a double Gaussian curve to the axial intensity profiles generated from a stack of image slices. The underlying assumption is that the laser light has mainly been reflected by two structures in the retina, the internal limiting membrane and the retinal pigment epithelium. From the fitted curve, topographic images and novel thickness images of the retina can be generated. The technique has been applied to three normal volunteers and seven patients with macular pathology (cystoid macular oedema and macular hole) demonstrating the clinical value of the technique. The improvement in accuracy achieved by using a double rather than a single Gaussian is also demonstrated.

Humans↗

PubMed2001

Fractional non-Brownian motion and trapping-time distributions of grains in rice piles.

Non-Gaussian height fluctuations occurring on the fueling time scale of a slowly driven rice pile match those observed in some turbulent/critical phenomena, forming an anticorrelated random fractal process with Hurst exponent H=0.2. Inspired by this observation, the concept of fractional Brownian motion (FBM) is extended to treat stochastic processes with skewed increments. Simulations of this process for antipersistent motion have first return time distribution deviating from the t(-2+H) law for FBM. The first return time distribution of this fractional non-Brownian motion describes and quantitatively determines the trapping-time distribution of grains in rice piles upon incorporating a continuous representation of the additional height fluctuations that occur on the time scale between fueling events.

Journal Article↗