A Brain on a Chip

There's much discussion these days of computers becoming so fast and powerful in the future that their silicon-based "brains" will one day rival the organic ones inside human beings. What about going the other way, though? Can you make smart machines from actual brain cells? Yes. That's just what many leading researchers are attempting to do. A team in Israel recently published a paper on building advanced neurochips for bio-sensing applications using cells from rat brains.

At the Micro and Nano Systems Laboratory at Tel Aviv University, a group led by Raya Sorkin is working on producing compactly self-wired neuronal networks by anchoring neurons in geometrical patterns to nanotubes attached to silicon or quartz chips and letting them grow connections to one another.

"Monitoring the dynamics of the forming networks in real time revealed that the self-assembly process is mainly driven by the ability of the neuronal cell clusters to move away from each other while continuously stretching a neurite bundle in between," they write in the Journal of Neural Engineering. "[W]e achieved networks with wiring regions which are made exclusively of neuronal processes unbound to the surface. The resulted network patterns are very stable and can be maintained for as long as 11 weeks."

The technique involves etching patterns of tiny holes into a silicon or quartz substrate, filling these with a protein solution, depositing carbon nanotube islands onto them, and finally attaching bundles of living cortical or hippocampal cells to the nanotubes. The cells are then left to grow, and they make connections with each other, within days, forming spontaneous but engineered neural networks that transmit electrical signals.

The researchers noted:

This approach achieves precise cell positioning and controlled network geometry with wiring which consists of axons and dendrites. The method relies on a natural propensity of cells to form compact wiring between neighboring islands rather than on specific guidance. The clusters in our approach rapidly form in the first two days of culturing as the cells migrate on a mostly low affinity substrate toward high affinity, lithographically defined, adhesive templates on which they adhere and assemble.

The Tel Aviv team wrote that such cultured neural networks should find applications in advanced bio-sensing applications, such as drug and toxin detection, where the structure, stability, and reproducibility of the networks are critical. Moreover, they believe that such research can lead to a greater understanding of the fundamental way in which neural networks operate, specifically the relationship "between a network's activity and its geometry, especially in light of the predilection of the networks to connect almost all nodes with the shortest wiring length possible."

It certainly makes one wonder where this is all leading. We'll follow this up in future threads.


Tech Talk

IEEE Spectrum’s general technology blog, featuring news, analysis, and opinions about engineering, consumer electronics, and technology and society, from the editorial staff and freelance contributors.

Newsletter Sign Up

Sign up for the Tech Alert newsletter and receive ground-breaking technology and science news from IEEE Spectrum every Thursday.