To appear in: Cytometry A
Original version: December 2007
Final revision: August 2008
We describe novel methods for automating key steps in analysis of single-cell, fluorescent dynamic images -- segmentation and lineage reconstruction -- to recognize and track individual cells over time. The automated method iteratively combines a set of novel extended morphological operators and filters for segmentation, and uses a novel neighborhood-based scoring method for frame-to-frame lineage linking. Our studies with both bacteria and human cells show the method effectively solves a number of problems with existing methods, including over/under segmentation, defining cell spatial mappings and lineage tracing through multiple mitoses. The automated method is freely available as an open-source software tool.
The paper is available but currently under journal review. Here are several supplementary notes and movies as well as many more examples of image segmentation, in a variety of applied areas.
This manuscript gives full details of the methods implemented in the open source CellTracer software that is available to interested research groups.