Date of Award
Level of Access Assigned by Author
Master of Science in Biomedical Engineering
Second Committee Member
Third Committee Member
With advances in microscopy in the 21st century, more accurate measurements of the morphology and position of chromosome territories (CTs) during interphase within the cell nucleus can be obtained. These measurements have determined that CTs are non-randomly positioned, and their position and shape can change during diseases like breast cancer, pancreatic cancer, and Huntington’s Disease. The 2.5D Wavelet Transform Modulus Maxima segmentation method uses multi-scale edge detection to accurately segment objects in 3D images to find variables such as surface area, volume, and diameter without assuming a spherical or ellipsoidal shape. Measurements of centroid distance, shell distance and radial distance along with calculations of filament index, elongation and rugosity were analyzed to determine positional and morphological trends in mouse chromosomes 1, 2, 9, 11 and 14 within embryonic stem (ES) cells and lymphocytes. Scatterplots were created where the best-fit trendline of either linear, log-logistic, or exponential decay was fit to each dataset. These trends showed that the CTs were long with indented cavities and tended to reside in the periphery of the nucleus. Boxplots by chromosome type showed that the CTs of the ES cells were longer, smoother and farther from the centroid of the nucleus than their counterparts in the lymphocyte cells. To confirm this conclusion, a simulated model was created that placed a randomly positioned and randomly oriented ellipsoidal CT within a spherical nucleus. This simulation used parameters gained from the imaged CTs to produce a realistic ellipsoid and then measured the radial distance. The radial distance of the simulated data was about half that of the imaged data, indicating that the CTs are not randomly placed. From linear plots using Pearson p-values, it was determined that the lymphocyte cell CTs had more occurrences of significance between variables, while the ES cell CTs had none, likely due to their undifferentiated state. Ideally the model will place four chromosomes within the nucleus, although this was too computationally taxing without a supercomputer. The model should also be expanded to allow for centroid distance and shell distance measurements, and to measure possible shape configurations if position is kept constant.
Varney, Hannah, "Segmentation and Statistical Analysis of Imaged and Simulated 3D Chromosome Territories" (2021). Electronic Theses and Dissertations. 3416.