Please use this identifier to cite or link to this item:
Title: Cell profiling with dynamic features for high-throughput images
Authors: Merlin Veronika Arokiamary James
Keywords: DRNTU::Engineering::Computer science and engineering
Issue Date: 2011
Abstract: Subpopulation heterogeneity has been spawning intense studies at genetic and molecular level due to its occurrence at all biological levels from cells to tissues. We envisioned studying this biological phenomenon through image based profiling methods incorporating motility based features. We developed population profiling methods for analysing subpopulations arising in single-cell lines by introducing motility based dynamic features. Combination of these features with morphological features improved the accuracy of classification of cell states. We introduced unsupervised methods so that prior training data is not required. Also the use of motility features for identifying membrane dynamics and its correlation with whole cell dynamics were investigated. We were able to identify subpopulations of cells with similar dynamic profiles but having different membrane patterns. The profiling pipeline using dynamic features were demonstrated by identifying mitotic phases in cells undergoing cell-cycle. Cells passing through mitotic division exhibit motility characteristics unique to each phase which were utilized for phase recognition. The methods were validated with real image data and the results compared well with ground truth.
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:SMA Theses

Files in This Item:
File Description SizeFormat 
  Restricted Access
Main Report15.27 MBAdobe PDFView/Open

Page view(s)

Updated on Apr 12, 2021


Updated on Apr 12, 2021

Google ScholarTM


Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.