Files and links (1)
PDFOpen Access (License Unspecified), Open Access
Dissertation
A Kolmogorov metric embedding kernel for live cell and tissue microscopy movies
Doctor of Philosophy (Ph.D.), Drexel University
Jun 2025
DOI:
https://doi.org/10.17918/00011268
Abstract
A metric embedding uses a distance function to represent a collection of live cell microscopy movies as points in a feature space such that distances between any two points throughout the feature space preserve the distance between the corresponding input images. This thesis describes an approach for computing metric distances between pairs of live cell microscopy movies. Segmentation and tracking results are represented as a graph. Different visual representations of this tracking graph, along with graph smoothing techniques on the tracking graph automatically improve cell detection and segmentation. Results are shown for Cell Tracking Challenge reference datasets. A new coral graph representation of the tracking graph, combined with the normalized compression distance and FLIF 3-D spatial image compression, serves as a distance metric between input live cell and tissue microscopy movies capturing cell signaling patterns. The cell signaling structure function was developed to replace the non-metric cytonuclear ratio for measuring cell signaling activation. The result is a metric embedding pipeline for 5-D images. Results are shown for live cell signaling in 2-D monolayers of human breast epithelial (MCF10A) cells from six different oncogenic mutations associated with distinctive changes in cell signaling patterns, in human induced pluripotent colonies under self-renewing and differentiating conditions, and in optogenetic excitation of MCF10A cells cultured in 3-D synthetic breast spheroids, and for a synthetic phantom dataset. This Kolmogorov metric embedding technique is unsupervised and represents each movie as a single point in a low dimensional space that preserves "any and all differences" among the patterns in the input images.
Metrics
2 File views/ downloads
1 Record Views
Details
- Title
- A Kolmogorov metric embedding kernel for live cell and tissue microscopy movies
- Creators
- Layton Aho
- Contributors
- Andrew R. Cohen (Advisor)
- Awarding Institution
- Drexel University
- Degree Awarded
- Doctor of Philosophy (Ph.D.)
- Publisher
- Drexel University
- Number of pages
- viii, 125 pages
- Resource Type
- Dissertation
- Language
- English
- Academic Unit
- Computer Science (Computing) [Historical]; College of Computing and Informatics (2013-2026); Drexel University
- Other Identifier
- 991022159173904721