Automated image analysis for quantifying individual transcript molecules inside morphologically-diverse macrophage cells

Tuesday, September 23, 2014 — Poster Session III
12:00 p.m. – 2:00 p.m.	FAES Academic Center	CIT	SYSBIO-2

Authors

• WW Lau
• A Martins
• M Narayanan
• JS Tsang

Abstract

Macrophages exhibit a variety of phenotypes in response to environmental stimulations. To assess cell-to-cell variations in the macrophage transcriptional response and its functional consequences, RNA fluorescence in situ hybridization (FISH) can be utilized to quantify transcripts in individual cells. However, automatic cell segmentation and spot counting of mRNA copies in FISH images have been challenging for cells such as macrophages with diverse, overlapping shapes and high autofluorescence. Here, we propose a multiplexed cell-segmentation approach to resolve shared cell boundaries, and an accurate spot counting procedure that accounts for imaging noise, non-specific binding of the probes, and physical artifacts. The automatic segmentation method was compared to and validated with manual segmentation. When applying our method to FISH measurements of two genes across many cells under two distinct stimulations, we observed that correlation of expression between gene pairs could be highly influenced by confounding factors such as background florescence and cell size. After accounting for these effects, we assessed cell-to-cell expression heterogeneity and gene-gene correlation across macrophage cells. Combined with the sensitivity of RNA FISH, this image processing approach can be a valuable tool for quantifying gene expression and constructing gene networks to study various biological phenomena at the single cell level.

back to top