Doctor of Philosophy, The Ohio State University, 2023, Biomedical Engineering

Quantitative colocalization analysis is a standard method in the life sciences used for evaluating the global spatial proximity of labeled biomolecules captured by fluorescence microscopy images. It is typically performed by characterizing the pixel-wise signal overlap or intensity correlation between spectral channels. However, this approach is critically flawed due to its focus on individual pixels which limits assessment to a single spatial scale constrained by the pixel's size, thus making the analysis dependent on the achieved optical resolution and ignorant of the spatial information presented by non-overlapping signals. In this dissertation, I present an improved method for quantifying biomolecule spatial proximity using a novel application of point process analysis adapted for discrete image data, and subsequently utilize it to address two novel cardiac conundrums. The tool, called Spatial Pattern Analysis using Closest Events (SPACE), leverages the distances between signal-positive pixels to statistically characterize the spatial relationship between labeled biomolecules from fluorescence microscopy images. In chapter two, SPACE's underlying theory and its adaption for discrete image-based data is described. Additionally, I characterize its sensitivity to segmentation parameters, image resolution, and signal sample size, and demonstrate its advantages over standard colocalization methods. With this tool, I hope to provide microscopists an improved method to robustly characterize spatial relationships independent of imaging modality or achieved resolution. In chapter three, SPACE is used to elucidate a novel, microtubule-based system for the distributed synthesis of membrane proteins in cardiomyocytes. Canonically, these cells are thought to produce membrane proteins in the peri-nuclear rough endoplasmic reticulum, then leverage the secretory-protein-trafficking pathway to transport nascent proteins to their sites of membrane insertion. By labeling car (open full item for complete abstract)

Committee: Rengasayee Veeraraghavan (Advisor); Przemysław Radwański (Committee Member); Peter Craigmile (Committee Member); Seth Weinberg (Committee Member) Subjects: Biology; Biomedical Engineering; Biomedical Research; Biophysics; Biostatistics; Cellular Biology; Engineering; Scientific Imaging; Statistics

PhD, University of Cincinnati, 2013, Arts and Sciences: Physics

I have done two projects in my PhD period: first, the statistics of quantum energy levels of integrable systems and second, a stochastic network model with applications to natural and social sciences. Quantum chaos studies level statistics and wave-function characteristics in the semiclasscal regime where the system action is several orders of magnitude larger than Planck's constant. The level statistics of classically integrable systems and classically chaotic ones are different. The levels of classically chaotic systems repel each other while these of classically integrable ones seem to be uncorrelated. We have studied level statistics in semiclassical spectrum of classically integrable systems without extra degeneracies. Our new findings are as follows. We developed a parametric averaging method to achieve ensemble averaging. We found the interval level number variance displays persistent oscillations. Contrary to previous belief, the nearest-neighbor spacing distribution of integrable systems displays some repulsion. The theoretical explanations of level statistics are based on both semiclassical theory and quantum mechanical derivation. Distributions of mass of species in the field of biology, human response times in the field of psychology, personal wealth distribution in the field of economics, and stock return in finance, all display power-law tails. It is theoretically and practically meaningful to understand the characteristics and origin of these power-law tails. We study a stochastic network model that is able to generate generalized inverse gamma distribution with power-law tail. We apply this distribution to the field of psychology and find good fit of data of with human response times. At last, we extend the model to study the stock return.

Committee: Rostislav Serota Ph.D. (Committee Chair); Andrei Kogan Ph.D. (Committee Member); Michael Ma Ph.D. (Committee Member); L.C.R. Wijewardhana Ph.D. (Committee Member) Subjects: Physics