Brown University researchers demonstrated a machine learning method that measures the topological properties of cell clusters, which can accurately categorize clusters and deduce the motility and adhesion of constituent cells.
The algorithm harnesses a mathematical framework called persistent homology to analyze microscope images of cell assemblages, and examine the topological loops or holes around empty regions in order to determine which patterns are intrinsic to the image. The most persistent loops are stored as a simplified representation of the overall configuration; once trained on computer-simulated cells, the algorithm could correctly classify different spatial patterns based on the biochemical treatment the cells had received. The Brown team hopes the algorithm could find use in laboratory drug-testing experiments.
From News from Brown
View Full Article
Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA
No entries found