The diffraction limit can be circumvent by creating and exploiting independent behaviors of the sample at lengths scale below the diffraction limit. In fluorescence microscopy, the independence arises from individual fluorescent labels switching between dark and fluorescence states. The fluorophores can then be localized employing the generated sparse image frames. Finally, the resulting list of coordinates is utilized to generate high resolution images or to gain quantitative insight into the underlying biological structures. Therefore image processing and post-processing are essential stages of SMLM techniques.
In this dissertation, Reversible Jump Markov Chain Monte Carlo was employed to implement Bayesian analysis of single molecule fluorescence microscopy data. Bayesian multiple-emitter fitting (BAMF) was developed to localize emitters in dense and noisy regions of data. This technique is particularly advantageous in fitting emitters in close spatial proximity and recognizing heterogeneous background noise. In a list of localizations produced in a SMLM experiment, each emitter is represented by multiple localizations generated from several blinking/binding events over the course of data acquisition. Bayesian grouping of localizations (BaGoL) provides emitter locations with enhanced precisions by identifying and combining the subset of localizations from each emitter. BaGoL advances the state-of-the-art in inspection of the geometrical distribution of particles in biological samples by producing one-to-one and precise positions for the emitters.
Bayesian paradigm permits inclusion of prior knowledge about the problem parameters into the calculations. The presence of the prior distributions in the computations facilitates parameter estimations with better uncertainties by restricting the range of parameters. The Bayesian algorithm implemented via RJMCMC also combines the model selection stage with the parameter estimation step and therefore takes full account of all the uncertainties in the problem.
The diraction limit can be circumvent by creating and exploiting independent behaviorsof the sample at lengths scale below the diraction limit. In uorescencemicroscopy, the independence arises from individual uorescent labels switching betweendark and uorescence states. The uorophores can then be localized employingthe generated sparse image frames. Finally, the resulting list of coordinates is utilizedto generate high resolution images or to gain quantitative insight into the underlyingbiological structures. Therefore image processing and post-processing are essentialstages of SMLM techniques.
Level of Degree
Physics & Astronomy
First Committee Member (Chair)
Second Committee Member
Third Committee Member
Fourth Committee Member
Bayesian, Super-resolution, Microscopy, Monte Carlo
Fazel, Mohamadreza. "Bayesian Analysis of Single Molecule Fluorescence Microscopy Data." (2020). https://digitalrepository.unm.edu/phyc_etds/232