Stem cell is a regenerative medicine to prevent and treat disease. For this purpose, microscopy image are taken during cell culturing to analyze the cell growth process. However, it took a bioengineering PhD student a few weeks to obtain only limited statistical information of the cell population, such as cell counting. In fact, biologists need more cell-individualized information to advance their research, such as the exact family tree, movement trajectories and cell shape changes.
We are investigating a computer vision-based stem cell tracking system that can track dense cell populations (hundreds to thousands) and determine their spatiotemporal histories over extended period of times (days to weeks). This interdisciplinary project with biologists enables optimized stem cell culture for cell therapies such as discovering best cell culture conditions to maintain stemness of the cell populations and determining when to subculture during the cell production.
Research topics in this project include microscopy image segmentation, cell mitosis event detection and multiple object tracking. The following are some demos showing the algorithm output.
1. Microscopy Image Segmentation. Here is a video showing the microscopy image restoration process. The restored image is amenable for cell segmentation.
2. Cell Mitosis Detection. Here is a video showing the cell mitosis detection:
3. Cell Tracking
Here is a video showing the tracking process:
The entire cell lineage trees can be visualized here :
Here is a video showing the tracking details on three cell families:
We integrated our algorithms into an online cell tracking system and use it for biological discovery and engineering worldwide.
Here is a video to show the application on wound healing analysis (27M).
The cell migration paths of the left and right groups of cells show how the wound area is healed in response to different cell culture conditions. (18M)