Noise Content Research Articles

Removal of eye movement artefacts from single channel recordings of retinal evoked potentials using synchronous dynamical embedding and independent component analysis

A system is described for the removal of eye movement and blink artefacts from single channel pattern reversal electroretinogram recordings of very poor signal-to-noise ratios. Artefacts are detected and removed by using a blind source separation technique based on the jadeR independent component analysis algorithm. The single channel data are arranged as a series of overlapping time-delayed vectors forming a dynamical embedding matrix. The structure of this matrix is constrained to the phase of the stimulation epoch: the term synchronous dynamical embedding is coined. A novel method using a marker channel with a non-independent synchronous feature is employed to identify the single most relevant source estimation for reconstruction and signal recovery. This method is non-lossy, all underlying signal being recovered. In synthetic datasets of defined noise content and in standardised real data recordings, the performance of this technique is compared to conventional fixed-threshold hard-limit rejection. The most significant relative improvements are achieved when movement and blink artefacts are greatest: no improvement is demonstrable for the random noise only situation.

Signal restoration through deconvolution applied to deep mantle seismic probes

SUMMARY We present a method of signal restoration to improve the signal-to-noise ratio, sharpen seismic arrival onset, and act as an empirical source deconvolution of specific seismic arrivals. Observed time-series g i are modelled as a convolution of a simpler time-series f i , and an invariant point spread function (PSF) h that attempts to account for the earthquake source process. The method is used on the shear wave time window containing SKS and S, whereby using a Gaussian PSF produces more impulsive, narrower, signals in the wave train. The resulting restored time-series facilitates more accurate and objective relative traveltime estimation of the individual seismic arrivals. We demonstrate the accuracy of the reconstruction method on synthetic seismograms generated by the reflectivity method. Clean and sharp reconstructions are obtained with real data, even for signals with relatively high noise content. Reconstructed signals are simpler, more impulsive, and narrower, which allows highlighting of some details of arrivals that are not readily apparent in raw waveforms. In particular, phases nearly coincident in time can be separately identified after processing. This is demonstrated for two seismic wave pairs used to probe deep mantle and core‐mantle boundary structure: (1) the S ab and S cd arrivals, which travel above and within, respectively, a 200‐300-km-thick, higher than average shear wave velocity layer at the base of the mantle, observable in the 88‐92 deg epicentral distance range and (2) SKS and SPdiff KS, which are core waves with the latter having short arcs of P-wave diffraction, and are nearly identical in timing near 108‐110 deg in distance. A Java/Matlab algorithm was developed for the signal restoration, which can be downloaded from the authors web page, along with example data and synthetic seismograms.

Closed-Loop Nonlinear Modeling of Wideband$SigmaDelta$Fractional-$N$Frequency Synthesizers

Wideband low-noise <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$SigmaDelta$</tex> fractional- <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$N$</tex> synthesizers pose several design challenges due to the nonlinear time-varying nature of synthesizer building blocks such as phase frequency detectors (PFDs), charge pump, and frequency dividers. Loop nonlinearities can increase close-in phase noise and enhance spurious tones due to intermodulation of high-frequency quantization noise and tonal content; therefore, an accurate simulation model is critical for successful implementation of loop parameters and bandwidth widening techniques. In this paper a closed-loop nonlinear simulation model for fractional- <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$N$</tex> synthesizers is presented. Inherent nonuniform sampling of the PFD is modeled through an event-driven dual-iteration-based technique. The proposed technique generates a vector of piecewise linear time–voltage pairs, defining the voltage-controlled oscillator (VCO) control voltage. This method also lends itself to modeling of cyclostationary thermal and flicker noise generated by time-varying charge-pump current pulses. A flexible third-order <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$SigmaDelta$</tex> modulated RF synthesizer core with integrated loop filter and LC-tank VCO is designed and fabricated in 0.13- <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$muhbox m$</tex> CMOS process in order to validate the technique experimentally. The proposed modeling technique was able to predict in-band spur power levels with 1.8-dB accuracy, and spur frequency offsets with lower than 400-Hz accuracy with several programmable nonidealities enabled.

Influence of image fusion approaches on classification accuracy: a case study

While many studies in the field of image fusion of remotely sensed data aim towards deriving new algorithms for visual enhancement, there is little research on the influence of image fusion on other applications. One major application in earth science is land cover mapping. The concept of sensors with multiple spatial resolutions provides a potential for image fusion. It minimises errors of geometric alignment and atmospheric or temporal changes. This study focuses on the influence of image fusion on spectral classification algorithms and their accuracy. A Landsat 7 ETM+ image was used, where six multispectral bands (30 m) were fused with the corresponding 15 m panchromatic channel. The fusion methods comprise rather common techniques like Brovey, hue‐saturation‐value transform, and principal component analysis, and more complex approaches, including adaptive image fusion, multisensor multiresolution image fusion technique, and wavelet transformation. Image classification was performed with supervised methods, e.g. maximum likelihood classifier, object‐based classification, and support vector machines. The classification was assessed with test samples, a clump analysis, and techniques accounting for classification errors along land cover boundaries. It was found that the adaptive image fusion approach shows best results with low noise content. It resulted in a major improvement when compared with the reference, especially along object edges. Acceptable results were achieved by wavelet, multisensor multiresolution image fusion, and principal component analysis. Brovey and hue‐saturation‐value image fusion performed poorly and cannot be recommended for classification of fused imagery.

Automated on‐the‐fly detection and correction procedure for EPR imaging data acquisition

Fast and reliable data acquisition is a major requirement for successful and useful biological electron paramagnetic resonance imaging (EPRI) experiments. Even a technologically advanced and professionally supervised EPRI system can exhibit instabilities initiated by perturbations such as animal motion, microphonics, and temperature changes. As a result, part of an acquired data set may become corrupted with excessive noise and distortions, which in turn may degrade the quality of the reconstructed image. In this work an automated scheme to monitor the system performance and stability over the course of an experiment is demonstrated. This method ensures that the quality of the acquired data is maintained during the experiment. For this purpose, four parameters including noise content and integration of each acquired projection are quantified and measured against those of the zero-gradient (ZG) projection, which is set as a quality benchmark. Projections with parameter values that differ substantially from the expected values are identified as damaged and consequently are reacquired. Therefore, the proposed technique not only effectively monitors the quality of acquisition, it also saves a substantial amount of acquisition time because it eliminates the necessity of repeating the entire experiment in cases in which only a small fraction of the data are corrupted.

Metal artifact reduction in CT using tissue-class modeling and adaptive prefiltering

High-density objects such as metal prostheses, surgical clips, or dental fillings generate streak-like artifacts in computed tomography images. We present a novel method for metal artifact reduction by in-painting missing information into the corrupted sinogram. The information is provided by a tissue-class model extracted from the distorted image. To this end the image is first adaptively filtered to reduce the noise content and to smooth out streak artifacts. Consecutively, the image is segmented into different material classes using a clustering algorithm. The corrupted and missing information in the original sinogram is completed using the forward projected information from the tissue-class model. The performance of the correction method is assessed on phantom images. Clinical images featuring a broad spectrum of metal artifacts are studied. Phantom and clinical studies show that metal artifacts, such as streaks, are significantly reduced and shadows in the image are eliminated. Furthermore, the novel approach improves detectability of organ contours. This can be of great relevance, for instance, in radiation therapy planning, where images affected by metal artifacts may lead to suboptimal treatment plans.

A motion-incorporated reconstruction method for gated PET studies

Cardiac and respiratory motion artefacts in PET imaging have been traditionally resolved by acquiring the data in gated mode. However, gated PET images are usually characterized by high noise content due to their low photon statistics. In this paper, we present a novel 4D model for the PET imaging system, which can incorporate motion information to generate a motion-free image with all acquired data. A computer simulation and a phantom study were conducted to test the performance of this approach. The computer simulation was based on a digital phantom that was continuously scaled during data acquisition. The phantom study, on the other hand, used two spheres in a tank of water, all of which were filled with 18F water. One of the spheres was stationary while the other moved in a sinusoidal fashion to simulate tumour motion in the thorax. Data were acquired using both 4D CT and gated PET. Motion information was derived from the 4D CT images and then used in the 4D PET model. Both studies showed that this 4D PET model had a good motion-compensating capability. In the phantom study, this approach reduced quantification error of the radioactivity concentration by 95% when compared to a corresponding static acquisition, while signal-to-noise ratio was improved by 210% when compared to a corresponding gated image.

A divergence operator to quantify texture from multi‐spectral satellite images

A divergence operator to measure the texture content in a multi‐spectral image is proposed. A multi‐spectral image is modelled as an n‐dimensional vector field, ‘n’ being the number of bands of the image. A pixel of the image is an n‐dimensional vector in this field. It is demonstrated that the flux variations of the vector field are related to changes in texture in the image. A divergence operator measures the flux variation and, hence, texture. In order to save computer memory, speed up the divergence operator calculation and lessen the content of noise of the image, principal component transformation is applied to the bands of the image. The first three principal components are used to span the vector field. The partial derivatives involved in the divergence operator are written as weighted finite differences. To estimate these derivatives, cubes of three, five and seven voxels per side are considered. The cube is systematically displaced to cover the entire domain of the vector field. In each position of the cube, the divergence value is calculated using the weighted finite difference approximation. This value is written as a pixel in an output image file according to the Cartesian coordinates defined by the location of the cube. This image file depicts the texture variations of the multi‐spectral image. The relation flux variation versus coarseness of texture is discussed. Two examples, based on Landsat Thematic Mapper multi‐spectral and synthetic multi‐spectral images, are presented and analysed.

Realistic scalability of noise in dynamic circuits

The usage of noise-sensitive dynamic circuits has become commonplace due to speed and area requirements, making the noise issue even more prominent. This paper focuses on the trends of coupling and its effects on dynamic circuits. It presents closed form analytical solutions for noise, as well as noise tolerance metrics for dynamic circuits. These solutions are within 5% of dynamic simulations. It is shown that not all scaling trends are negative for noise, and that the scaling down of supply voltage and increasing frequency, help improve certain aspects of the noise immunity of dynamic circuits. Most of the works treated the noise immunity and the noise content separately. This paper introduces an analysis of noise scalability by looking at the noise immunity and the noise content simultaneously

SU‐FF‐I‐05: Hounsfield Units Calibration with Adaptive Compensation of Beam Hardening for a Dose Limited Breast CT System

Purpose: Hounsfield Units calibration with adaptive compensation of beam hardening for a dose limited breast CT system. Method and Materials: Following a complete cone beam CT scan of the target object (a human breast), geometrical parameters of the object, including the mass center location and maximum radius from the mass center were calculated promptly. These parameters were used to compute X‐ray projection images of a water cylinder based upon photon attenuation in the cylinder and detector response. Projection images of the target object were corrected prior to reconstruction by logarithmic subtraction of the water cylinder projection images. CT images relative to the Hounsfield Units (HU) scale were produced after cone beam reconstruction of the corrected projection images. Custom built water phantoms were tested with various inserts, including polyethylene, polystyrene, PMMA, nylon, polycarbonate and Teflon, with a density range from 0.92g/cm3 to 2.2g/cm3. Results: A wide range of breast diameters (10cm–18cm) and compositions (0%–100% glandular) were evaluated, and reconstructed and scaled HU values demonstrated excellent uniformity, linearity and consistency. Typical HU values were within 5% of theoretical values. The proposed method was applied to clinical breast scans. The “cupping” artifact caused by beam hardening in the original image was corrected as expected. Conclusion: Conventional methods of Hounsfield Units conversion are based on the scan of few fix‐sized water phantoms and lack the flexibility to compensate for the beam hardening from objects with various sizes. The introduced noise from water phantom scans is also not negligible, especially for a dose limited breast CT system with much lower mAs than whole body CT systems. The proposed method can compensate for beam hardening over a wide range of breast diameters and compositions without increasing noise contents in the image.

Global active control of broadband noise from small axial cooling fans, Part 2

The multi-channel filtered-x LMS algorithm has previously been used for feed-forward control of the tonal noise of a small axial cooling fan [Gee and Sommerfeldt, ‘‘Application of theoretical modeling to multichannel active control of cooling fan noise,’’ J. Acoust. Soc. Am. 115, 228–236 (2004)]. With much of the tonal noise content removed it becomes desirable to reduce the broadband fan noise. A previous study demonstrated the feasibility of achieving desirable broadband attenuation by means of active feedback control. Measurements will be presented and discussed that have an impact on the control achievable. Experimental control results will be presented.

Improving the prediction and parsimony of ARX models using multiscale estimation

Multiscale wavelet-based data representation has been shown to be a powerful data analysis tool in various applications. In this paper, the advantages of multiscale representation are utilized to improve the prediction accuracy and parsimony of the auto-regressive with exogenous variable (ARX) model by developing a multiscale ARX (MSARX) modeling algorithm. The idea is to decompose the input-output data at multiple scales, construct an ARX model at each scale using the scaled signal approximations of the data, and then using cross validation, select among all MSARX models the one which best predicts the process response. The MSARX algorithm is shown to improve the parsimony of the estimated models, as ARX models with a fewer number of coefficients are needed at coarser scales. This advantage is attributed to the down-sampling used in multiscale representation. Another important advantage of the MSARX algorithm is that it inherently accounts for the presence of measurement noise through the application of low-pass filters in the multiscale decomposition of the data, which in turn improves the model robustness to measurement noise and enhances its prediction. These prediction and parsimony advantages of MSARX modeling are demonstrated through a simulated second order example, in which the MSARX algorithm outperformed the time-domain one at different noise contents, and the relative improvement of MSARX increased at higher levels of noise.

Marked signal improvement by stochastic resonance for aperiodic signals in the double-well system

On the basis of our mixed-signal simulations we report significant stochastic resonance induced input-output signal improvement in the double-well system for aperiodic input types. We used a pulse train with randomised pulse locations and a band-limited noise with low cut-off frequency as input signals, and applied a cross-spectral measure to quantify their noise content. We also supplemented our examinations with simulations in the Schmitt trigger to show that the signal improvement we obtained is not a result of a potential filtering effect due to the limited response time of the double-well dynamics.

Fluidised bed dynamics diagnosis from measurements of low-frequency out-bed passive acoustic emissions

The purpose of this research is to show that measurements of low frequency out-bed passive acoustic emissions are useful for monitoring gas–solid fluidised bed hydrodynamics. A methodology is proposed to develop portable measurement systems (PMS) feasible for both laboratory scale optimisations and, later, for process diagnosis within industrial facilities. The methodology includes a multirate technique where the first sampling frequency used was about 11 kHz that was successfully later reduced to 500 Hz, which is far from the most common used for sampling ultrasonic ranges. The acoustic frequency range of interest (0–200 Hz) allows the use of a commercial jukebox and two low cost condenser microphones to record the out-bed acoustic emissions. The underlying bed dynamic process was studied by monitoring both acoustic and pressure fluctuation signals and by applying time, frequency and state space analysis over the measured time series. The crude acoustic signals exhibit marked random behaviour due to a high content of background noise. This is detected by the mutual information function. The power spectral density of the acoustic time series shows high complexity and claims for reducing the acoustic frequency range to be evaluated (initially 0–5.5 kHz). To decide an optimal frequency range and, consequently, a better sampling frequency, the acoustic pressure was estimated from a numerical simulation based on the well-known particle array model. From the results of the simulations, and in agreement to the literature, it is concluded that the bed dynamic acoustic information mainly lies within the range 0–200 Hz. According to that, the signals were subsequently re-sampled at a sampling frequency of 500 Hz. Finally, by comparing the pressure time series monitoring to the analysis of the mesoscale bed acoustic region (mainly due to bubbles, 0–20 Hz), it is concluded that from out-bed acoustic time series monitoring, only the state space analysis is able to fully characterize the slugging dynamics by means of a correlation dimension in the range 2.5–3.5 for the experimental conditions covered.

Acoustic signal enhancement system

A signal enhancement system improves the quality of a noisy input signal. The system finds a low noise signal model which best matches the noisy input signal. Noisy portions of the input signal are replaced with portions of the low noise signal models. As the input signal increases in noise content, the output signal includes an increasing amount of the low noise signal model. The system thereby produces an output signal with very low noise which corresponds to the input signal.

Improving the energy resolution of X-ray and electron energy-loss spectra

We discuss some practical problems of improving the resolution of X-ray and electron spectra. Iterative Bayesian methods promise greater resolution enhancement than Fourier techniques but they also give rise to spectral artifacts. Satellite peaks are generated adjacent to strong peaks in the original spectrum and oscillatory artifacts become prominent after a large number of iterations, particularly when the original data contain high noise content. In the case of valence-electron energy-loss spectra, satellite peaks are reduced by removing the zero-loss peak prior to spectral sharpening. Even so, care should be exercised in interpreting low intensity at low energy loss (after sharpening) as evidence for a bandgap in the electronic density of states.

Automated Protein NMR Structure Determination Using Wavelet De-noised NOESY Spectra

A major time-consuming step of protein NMR structure determination is the generation of reliable NOESY cross peak lists which usually requires a significant amount of manual interaction. Here we present a new algorithm for automated peak picking involving wavelet de-noised NOESY spectra in a process where the identification of peaks is coupled to automated structure determination. The core of this method is the generation of incremental peak lists by applying different wavelet de-noising procedures which yield peak lists of a different noise content. In combination with additional filters which probe the consistency of the peak lists, good convergence of the NOESY-based automated structure determination could be achieved. These algorithms were implemented in the context of the ARIA software for automated NOE assignment and structure determination and were validated for a polysulfide-sulfur transferase protein of known structure. The procedures presented here should be commonly applicable for efficient protein NMR structure determination and automated NMR peak picking.

Quantifying Peptide Signal in MALDI-TOF Mass Spectrometry Data

This study addressed the question of which properties in MALDI-TOF spectra are relevant to the task of identifying mass and abundance of a peptide species in human serum. Data of this type are common to biomarker studies, but significant within- and between-spectrum variabilities make quantifying biologically induced features difficult. We investigated this signal content and quantified the existence, or lack, of peptide-induced signal (as manifest in a multiresolution decomposition) by generating spectra from human serum in which the abundance of peptides of specific masses is controlled by a sequence of dilutions. The intensities of the corresponding features were directly proportional to peptide concentration. The primary goal was to exhibit some quantifiable properties of raw spectra from this application of MALDI-TOF mass spectrometry. Although no recommendations are given regarding the best method for processing these data, the results confirm the utility of a simple method, based on wavelets, for defining and quantifying features related to low abundance peptide species in a heterogeneous set of complex spectra. Estimates on lower limits of detectable peptide abundance (in the 20-nmol range) and on the number of features present in a spectrum are made possible by the controlled experimental design, the use of a large external reference data set, and dependence on relatively few modeling assumptions.

Pattern Recognition Theory and Preliminary Processing of Images Based on Stochastic Geometry

New geometrical dual trace transformation connected with scanning images with complex trajectories is introduced. On the basis of this transformation, non-linear filtration is accomplished in order to reduce the noise content of images, segmentation, smoothing, and polygonal approximation.

The behavioral response of mice to gaps in noise depends on its spectral components and its bandwidth

The purpose of these experiments was to determine whether detecting brief decrements in noise level ("gaps") varies with the spectral content and bandwidth of noise in mice as it does in humans. The behavioral effect of gaps was quantified by their inhibiting a subsequent acoustic startle reflex. Gap durations from 1 to 29 ms were presented in five adjacent 1-octave noise bands and one 5-octave band, their range being 2 kHz to 64 kHz. Gaps ended 60 ms before the startle stimulus (experiment 1) or at startle onset (experiment 2). Asymptotic inhibition was greater for higher-frequency 1-octave bands and highest for the 5-octave band in both experiments, but time constants were related to frequency only in experiment 1. For the lowest band (2-4 kHz) neither noise decrements (experiment 1 and 2) nor increments (experiment 3) had any behavioral consequence, but this band was effective when presented as a pulse in quiet (experiment 4). The lowest frequencies in the most effective 1-octave band were one octave above the spectral region where mice have their best absolute thresholds. These effects are similar to those obtained in humans, and reveal a special contribution of wide band, high-frequency stimulation to temporal acuity.