Bright-field imaging of 3-dimensional (3-D) cell-matrix structures using new deconvolution and segmentation techniques

Cells and tissues have complex 3-D surface topologies intimately related to their extracellular environment. Knowledge of the 3-D topology is critical to understanding structure-function relationships. Confocal and Fluorescent darkfield microscopy has been used to extract 3-D information from biological samples. However, obtaining 3-D reconstructions from bright-field microscopy is a challenge because of the artifacts such as scatter, glare, and refraction in the images. The goal of this work was to develop new experimental and computational techniques to obtain 3-D morphologies of live specimens from bright-field images. To achieve this, a new approach to characterizing and deconvolving the bright-field point spread function of the microscope was developed. A simple constrained iterative deconvolution algorithm was used. A 2-D adaptive thresholding algorithm was extended to 3-D for ‘segmentation’ extraction of objects from deconvolved images. These techniques have been implemented on bright-field images of live endothelial cells that were grown in a tissue engineered medium to extract cell shapes and orientations. This thesis presents the experimental system and analytical/computational methods.M.S., Biomedical Engineering -- Drexel University, 200