Poisson-Gaussian noise parameter estimation in fluorescence microscopy imaging

Jezierska, Anna; Talbot, Hugues; Chaux, Caroline; Pesquet, Jean-Christophe; Engler, Gilbert

doi:10.1109/isbi.2012.6235897

Cited by 25 publications

(13 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…[22].In case of fluorescence microscopy, negligible extrinsic noise is produced due to use of CCD camera as image detector. Gaussian noise model is preferred over Poisson noise model, which is more realistic in comparison, as it simplifies numerical computation [23]. The Gaussian noise model is given by-…”

Section: Models and Properties Of Microscopic Psfmentioning

confidence: 99%

Deconvolution for Enhancement of Biological Images Obtained by Fluorescence Microscopy

Nag*,

Das

2019

IJRTE

View full text Add to dashboard Cite

Intricate details of cells and tissues can be visualised by fluorescence microscopy and the images obtained can be then be quantitatively analysed. However, during image acquisition, distortions of the images occur by convolving the object with Point Spread Function. To remove this blurring, computational deconvolution methods are used in which the original image is restored with improved contrast. Our study analysed various fluorescence images, after the nuclei segmentation of the images, by both Deblurring (Blind Deconvolution, Lucy Richardson and Wiener filtering) and Restoration algorithms (Inverse filtering and Regularised filtering), which are the two main categories of deconvolution methods, in MATLAB 2016b. After statistical analysis (Mann Whitney U test) of area and homogeneity of the segmented nuclei of the various images for the different deconvolution methods, statistical significant difference was found in the case of area ((p=0.027) for Original vs. Inverse filter and (p=0.029) for Original vs. Regularised filter)) for restoration algorithms and for homogeneity, it was found for original vs. all the deconvolution methods, which shows that quantitative evaluation of the features can be used to further determine the better deconvolution method and in this case Restoration algorithms proves better than Deblurring algorithms.

show abstract

Section: Models and Properties Of Microscopic Psfmentioning

confidence: 99%

Deconvolution for Enhancement of Biological Images Obtained by Fluorescence Microscopy

Nag*,

Das

2019

IJRTE

View full text Add to dashboard Cite

show abstract

“…Note that an alternative approach relying upon an alternating minimization approach was proposed in [2]. However, it was observed to exhibit slower convergence.…”

Section: Moment-based Initializationmentioning

confidence: 99%

“…Results presented in Fig. 11 concern the alternating minimization approach proposed in [2] and the Douglas-Rachford approach corresponding to Algorithm 4. Fig.…”

Section: H1 Validation Of the Proposed Approach On Synthetic Datamentioning

confidence: 99%

An EM Approach for Time-Variant Poisson-Gaussian Model Parameter Estimation

Jezierska

Chaux

Pesquet

et al. 2014

IEEE Trans. Signal Process.

Self Cite

View full text Add to dashboard Cite

International audienceThe problem of estimating the parameters of a Poisson-Gaussian model from experimental data has recently raised much interest in various applications, for instance in confocal fluorescence microscopy. In this context, a field of independent random variables is observed, which is varying both in time and space. Each variable is a sum of two components, one following a Poisson and the other a Gaussian distribution. In this paper, a general formulation is considered where the associated Poisson process is nonstationary in space and also exhibits an exponential decay in time, whereas the Gaussian component corresponds to a stationary white noise with arbitrary mean. To solve the considered parametric estimation problem, we follow an iterative Expectation-Maximization (EM) approach. The parameter update equations involve deriving finite approximation of infinite sums. Expressions for the maximum error incurred in the process are also given. Since the problem is non-convex, we pay attention to the EM initialization, using a moment-based method where recent optimization tools come into play. We carry out a performance analysis by computing the Cramer-Rao bounds on the estimated variables. The practical performance of the proposed estimation procedure is illustrated on both synthetic data and real fluorescence macroscopy image sequences. The algorithm is shown to provide reliable estimates of the mean/variance of the Gaussian noise and of the scale parameter of the Poisson component, as well as of its exponential decay rate. In particular, the mean estimate of the Poisson component can be interpreted as a good-quality denoised version of the data

show abstract

“…Regressors are Z ij = (t ij , u(t)| tij t=0 ) and the full set of parameters to be inferred is θ = {β, C, e a , e b }. Model (13) is in agreement with the types of noise present in fluorescence microscopy [14,22]. Finally, to find the parameters in function (1) that maximize the marginal likelihood of the simulated distribution, we use the SAEM algorithm [5].…”

Section: Model For Me Inferencementioning

confidence: 99%

Identification of biological models from single-cell data: A comparison between mixed-effects and moment-based inference

González

Uhlendorf

Schaul

et al. 2013

2013 European Control Conference (ECC)

View full text Add to dashboard Cite

Experimental techniques in biology such as microfluidic devices and time-lapse microscopy allow tracking of the gene expression in single cells over time. So far, few attempts have been made to fully exploit these data for modeling the dynamics of biological networks in cell populations. In this paper we compare two modeling approaches capable to describe cell-to-cell variability: Mixed-Effects (ME) models and the Chemical Master Equation (CME). We discuss how network parameters can be identified from experimental data and use real data of the HOG pathway in yeast to assess model quality. For CME we rely on the identification approach proposed by Zechner et al. (PNAS, 2012), based on moments of the probability distribution involved in the CME. ME and moment-based (MB) inference will be also contrasted in terms of general features and possible uses in biology.

show abstract

Poisson-Gaussian noise parameter estimation in fluorescence microscopy imaging

Cited by 25 publications

References 20 publications

Deconvolution for Enhancement of Biological Images Obtained by Fluorescence Microscopy

Deconvolution for Enhancement of Biological Images Obtained by Fluorescence Microscopy

An EM Approach for Time-Variant Poisson-Gaussian Model Parameter Estimation

Identification of biological models from single-cell data: A comparison between mixed-effects and moment-based inference

Contact Info

Product

Resources

About