Back

I spent the day rectifying segmentation and tracking for bacteria in microfluidic devices. 1000 frames, 5 chambers, ~8 cells per chamber, so 40000 pictures of cells to look at, and check that they match frame to frame. I think I'm done with screens for the day. Fortunately automation helps, and drawing the lineages as graphs allows to spot errors more easily.
Now I have a clean dataset that I can analyse and also use as a training dataset for the next iteration!