The use of stem cells is one of the most popular topics in medicine and physiology. However, to grow tissue from such cells, it is important for researchers around the world to learn how to optimize the process. Neurobiologist from Sheffield University (UK) Anton Nikolaev and postgraduate student of ITMO University Pavel Katunin have created a robot that will allow controlling the growth of neurons from stem cells.
Today, there is already a technique for transforming stem cells into other cells using various signal molecules, such as retinoic acid or Wnt. Although it is widely used, it has a number of drawbacks, for example, only a small percentage of cells turn into necessary cells.
“The problem is that it takes a very long time,” explains Anton Nikolaev. – To, conditionally, turn a cup of stem cells into neurons, it takes a month or even more. The second problem is that there is no control as to which neurons the cells in the cup will turn into. We want to learn to grow certain types of neurons and better understand how neural chains work.
The main difference of the project is in using machine learning and computer vision in order to find and maintain optimal conditions for conversion (differentiation) of cells into neurons. But for machine learning a huge number of examples is required: experiments on cell transformation should be conducted thousands of times, which is extremely difficult even for a large scientific group. For this purpose scientists have created the robot-laboratory, printed on the usual 3D-printer which will help to automate process.
“The very task of selecting a specific substance for differentiation and the protocol of its application is the task of optimization,” says Pavel Katunin, “that is, you have, for example, different parameters of signal molecules – their concentration, feed rate, and so on – and we are trying to find the best combination of these parameters, so that the maximum number of cells is converted into exactly what we need. To optimize this process, to estimate the percentage of cells that have been differentiated into the right ones, we can use computer vision, which, based on the data from the microscope, automatically determines whether the process is going the way we want it to.
As a result, scientists have developed a robot to automatically set up numerous experiments and collect large data, on which the optimization algorithm was tested.
Now the stage of work on the robot is almost complete, and as a result, the preprint of the article was published. Now other researchers will also be able to use this development for their experiments. In the future, this model will allow scientists to monitor the process of cell transformation at an early stage and select the necessary ones for the experiment.
In the first stage, the experiments are not conducted on embryonic stem cells, but rather on cheaper NTERA-2 cancer stem cell lines. For robot model training and debugging, this replacement is very useful, as it significantly reduces the cost of each experiment. However, working with such cells can in itself lead to scientific results and possible applications in medical fields such as oncology. However, now it is important for scientists to work out a method for obtaining a large number of neurons from stem cells and in the future to create logical chains from them.