Manual worm counting to measure lifespan is still very common, but it brings with it a lot of potential for human errors and inconsistencies in the data. In order to give our clients the most accurate data possible, and to minimize the chance for human errors, we have implemented machine learning to help us automatically measure C. elegans lifespan. Machine learning is being implemented across all facets of biology, so we wanted to provide an example of the data we acquire using our automated lifespan technology. This video is a compilation of images of a plate of adult C. elegans worms acquired once an hour for 35 days (840 images total).
In the video you will see white boxes appearing. Objects in a white box are being identified to determine if they are worm or non-worm. They appear when the worm stops moving. Human annotators confirm whether this object is a worm in order to help the algorithm learn.
You will notice three different colored worms in the video - purple, yellow, and red. As the worms age they go from purple (indicating a fast moving state), to yellow (indicating a slow moving state), and finally to red (indicating a worm has died). When a worm dies and turns red you can see that it expands and gets bigger.
To the right of the plate video, there is a survival curve, also known as a Kaplan Meier Plot. This survival curve shows in real-time the percent of the worm population that is alive over time for the plate on the left. As the worm population ages and dies, the line on the survival curve goes down, indicating fewer worms are alive.