Slice Based 3D Cell Segmentation of Optical Projection Tomography Images
No Thumbnail Available
Date
2018-01
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Optical Projection Tomography (OPT) is a new imaging technique used to image cells
in 3D cultured in a hydrogel. Reconstructed OPT images of cells suffer from several
artifacts. These artifacts reduce the overall image quality. This will create a challenge
for isolating and studying each cell in 3D cell culture. It is highly required to enhance the
3D OPT images of cells for successful analysis of cell interaction and growth in 3D cell
culture. In that regard, this thesis intends to build a robust algorithm for use in effective
segmentation of cells cultured inside hydrogel. There exist various 3D cell segmentation
algorithms in the literature including those schemes that rely on thresholding, edge
detection, region growing and clustering approaches. Among these algorithms, moving
average adaptive thresholding (MAAT) and region growing algorithm present commendable
performance in segmentation of cells identified on OPT images. In this thesis the
performance of an automatic seeded region growing algorithm (ASRGA) and MAAT have
been compared rigorously in terms of their use in 2D slice based segmentation of the cells
on the 3D OPT image sets considered. Results have shown that the MAAT method show
superior performance and provide promising 3D visualization of cells. The output of the
research will have a tremendous contribution to reduce artifacts in 3D cell images and
enhance 3D visualization.
Description
Keywords
Optical Projection Tomography, hydrogel, cell segmentation, region growing algorithm, moving average