The human brain is a remarkably powerful and efficient computer. It has been estimated that it can perform the equivalent of an exaflop (a billion-billion) of mathematical operations per second. This performance can be achieved while consuming only 20 watts of power. A handful of the world’s most powerful supercomputers can now approach exaflop-level computing performance, however they require one million times more power to do so.
These characteristics of the brain are highly desirable for computing systems, especially with the rise of artificial intelligence, which is pushing existing technologies to their limits with respect to processing power and energy consumption. Accordingly, researchers have stepped up efforts to understand the complexities of the brain that endow it with these incredible abilities. The hope is that this effort could give us the knowledge we need to reproduce the brain’s power and efficiency in an artificial system.
Unlocking the brain’s secrets has proved to be exceedingly challenging, however. Much of this work has been done by measuring the brain’s electrical activity as study participants perform certain tasks. But the tremendous complexity of what is happening throughout the brain has made it very difficult to interpret these signals and correlate them with what the participants are doing.
This has led to researchers working with smaller systems — from single cells to small tissue samples — in the hope of making the problem more tractable. Typically using micro-electrode arrays, they can probe even single neurons and collect high-resolution electrical data that provides a snapshot of the signals being produced under different circumstances. Unfortunately, commercial solutions tend to be exceedingly expensive and tailored to very specific use cases that do not meet all researchers’ needs.
An open source micro-electrode array developed by a team at the University of Illinois at Urbana-Champaign could provide researchers with an accessible and flexible option to support their investigations. Called Mind in Vitro (MiV), the interfacing platform is easy to construct and contains up to 512 electrodes that can vary in size and spacing to support a variety of applications. Despite the fact that many commercial devices only contain about 60 electrodes, MiV costs ten times less.
After a sample is placed in the device, the electrodes are moved into a position such that they are in contact with it. Then, custom printed circuit boards capture electrical signals received by the electrodes. From there, these signals are forwarded into Open Ephys or Intan (commonly used software for visualizing and exploring electrophysiological data). An open source software package, written in Python, was also developed to manage the machine’s operation and support basic functions like storing collected data.
The researchers demonstrated the utility of their system in a number of experiments. A variety of cell types, and both 2D and 3D systems, were explored with MiV. These trials showed the device to be highly adaptable and capable of long-term recording of electrical signals. Based on these findings, MiV appears to be a practical platform for the investigation of living neurons that could be of use to many research groups.
A researcher involved in the study noted that they "...designed our system to be very easy to make and comprised of inexpensive parts, which will allow a lot of labs that can't afford the commercial system to have their own system. Even though the original design is for computational studies, we've made the structure easy to redesign and expand upon, so we're pretty sure this will satisfy researchers no matter the kind of studies they want to do.”