Computational Cell-Barcoding For High-Throughput Robotic Microscopy

Studies of individual cells within populations rely on the ability to identify the same cell in multiple images. While high sampling frequencies in time-lapse movies greatly simplify the tasks of tracking cells spatially and stimulus-responses temporally, delicate samples undergoing longitudinal study might be imaged only once each day. To identify cells and track cell fate in these less frequent images, we need alternate methods that reliably capture single-cell information over the course of the study. Molecular and instrumental advances in cell imaging have provided elegant experimental approaches to barcoding individual cells. However, these techniques require additional reagents, time, and channels. Here, we present a complementary technique to identify cells computationally. We transfected primary neurons with a fluorescent morphology marker, imaged the samples daily for 1 week, and extracted unique cellular parameters. We will summarize our progress with different approaches for their ability to collect features from individual cells and to determine cell type, vitality, and identify.