Quantitative Bioimage Analysis of Tissue Morphogenesis

Abstract

Tissue morphogenesis is a fundamental developmental process that shapes tissues into specific geometries to allow for tissues to perform their functional roles. Advances in imaging and cell biology now enable high-resolution, spatiotemporal characterization of morphogenesis at the subcellular level. In this thesis, I describe computational image analysis and computer vision techniques for the study of developmental biology in two systems. First, I present a quantitative analysis of embryonic zebrafish epidermis, which undergoes rapid growth during early embryogenesis, revealing that biphasic MAPK/ERK signaling that initially promotes oriented cell division and later promotes cell hypertrophy plays a major role in the development of this tissue. This study involved the use of image segmentation software, and MATLAB/Python for data analysis and machine learning to provide quantitative evidence supporting our conclusions. Second, I develop a 3D image segmentation pipeline for C. elegans embryonic cells that integrates deep learning via Cellpose with classical active contour methods, providing state-of-the-art segmentation results using images with substantial noise and low contrast. This pipeline is used to extract and analyze cell-cell contact changes, which play a critical role in C. elegans development. This method can be extended to other systems where experimental restrictions on imaging lead to poor-quality images that are difficult to analyze with traditional methods. Together, these studies demonstrate how computational methods can extract quantitative insights from fluorescence imaging of dynamic tissues, advancing our understanding of the cellular mechanisms underlying morphogenesis.