Date of Completion

9-1-2017

Embargo Period

8-31-2017

Keywords

super-resolution, microscopy, bayesian, statistics, drift, localization uncertainty

Major Advisor

Ji Yu

Co-Major Advisor

Ion Moraru

Associate Advisor

Bruce Mayer

Associate Advisor

Yi Wu

Associate Advisor

Ann Cowan

Associate Advisor

John Carson

Field of Study

Biomedical Science

Degree

Doctor of Philosophy

Open Access

Open Access

Abstract

Single-molecule localization microscopy (SMLM) has revolutionized the field of cell biology. It allowed scientists to break the Abbe diffraction limit for fluorescence microscopy and got it closer to the electron microscopy resolution but still it faced some serious challenges. Two of the most important of these are the sample drift and the measurement noise problems that result in lower resolution images. Both of these problems are generally unavoidable where the sample drift is a natural mechanical phenomenon that occurs during the long time of image acquisition required for SMLM (Geisler et al. 2012) while the measurement noise, which arises from combining localization uncertainty and counting error, is related to the number of photons collected from the fluorophore and affects the precision in locating the centroids of single molecules (Thompson, Larson, and Webb 2002).

Previous work has tried to devise methods to deal with the sample drift problem but unfortunately, these methods either add too much complexity to the experimental setup or are just inefficient in correctly estimating the drift at the single frame level (Wang et al. 2014). As for measurement noise, all current regular image rendering algorithms treat every detection of the fluorophore as a separate event and hence, the localization uncertainty of every detection of the same molecule would give offset coordinates from the other detections leading to a distorted final image.

In this thesis, I demonstrate two novel approaches based on statistical concepts to address each of these two problems. The algorithm for solving the sample drift problem is based on Bayesian inference and it showed efficiency in estimating drift at the single-frame level and proved superior and more straightforward than the available methods. The algorithm for addressing the measurement noise problem is based on Gibbs sampling and not only did it enhance resolution, but it also offers for the first time a means to quantify resolution uncertainty as well as uncertainty in cluster size measurement for clustering proteins. Therefore, this work offers a significant advancement in the field of SMLM and more generally, cell biology.

COinS