Statistics – Computation
Scientific paper
Jun 2006
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2006phdt.........4h&link_type=abstract
PhD Thesis, Proquest Dissertations And Theses 2006. Section 0034, Part 0786 236 pages; [Ph.D. dissertation].United States -- Ca
Statistics
Computation
Fluorescence Microscopy, Deconvolution, Noisy Data, Microscopy
Scientific paper
Noise is intrinsic to every experiment. While often viewed as a nuisance, work on the theory of fluctuations has shown that noise may actually contain useful information. This two-part thesis explores the duality of noise as information and disinformation, specifically in applications of fluorescence microscopy.
In Part I, I describe theoretical, experimental, and computational developments in two-color fluorescence fluctuation microscopy (TCFFM). In TCFFM, intensity fluctuations of fluorescently-labeled molecules within a small optical volume are analyzed to glean information about the dynamics of system components. Focusing on bimolecular reactive systems of the form ( A + B [Special characters omitted.] C ), I present a general TCFFM theory and discuss how the analysis of fluctuation correlations can be used to measure absolute particle numbers and kinetic rates of binding. I also discuss the influence of Fürester resonance energy transfer, reaction rates and reactant concentrations, diffusion, and component visibility on correlation analyses. After describing a two-photon microscope system that I assembled for TCFFM measurements, I present experimental proof-of-principle results in a discussion of how TCFFM can be used to measure the equilibrium constant of two fluorescently-labeled interacting molecules. Complementing these experimental TCFFM studies, I describe a molecular dynamics/Monte Carlo-based simulation tool and demonstrate how it can be used to study more complex systems and experimental conditions that cannot be accounted for with the current theory.
In Part II, I describe work on the Adaptive Image Deconvolution Algorithm (AIDA). AIDA is a computational tool for processing and de-noising astronomical and microscopy images. It is based on the proprietary MISTRAL method developed by Mugnier and co-workers (J. Opt. Soc. Am. A. 21, 1841 (2004)), which has been shown to yield object reconstructions with excellent edge preservation, noise suppression, and photometric precision. Using a Bayesian maximum a posteriori framework, I present the theoretical basis for the AIDA approach. I describe how AIDA was implemented and discuss improvements over the original MISTRAL program, including a scheme to automatically balance maximum-likelihood estimation and object regularization and the ability to deconvolve multiple image frame and three-dimensional data. I present validation results using synthetic data spanning a broad range of signal-to-noise ratios and image types, and demonstrate that AIDA is effective for experimental data from adaptive optics-equipped telescope systems and wide-field microscopy.
No associations
LandOfFree
Distilling information from noisy data: Examples from microscopy does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Distilling information from noisy data: Examples from microscopy, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Distilling information from noisy data: Examples from microscopy will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1476453