Harvard’s Breakthrough Chip Maps 70K Synaptic Connections in Rat Neurons!

Scientists at Harvard have developed a cutting-edge CMOS (complementary metal-oxide semiconductor) chip embedded with 4,096 microhole electrode arrays. This innovation has enabled them to monitor electrical activities of multiple neural cells simultaneously. As reported by the journal Nature, the device was used to analyze 2,000 neurons from rats, identifying and mapping more than 70,000 synaptic connections and assessing the strength and type of signals transmitted through these connections.

This breakthrough significantly enhances the capabilities of neuronal research. While electron microscopy provides a visual of synaptic connections, it falls short in measuring and recording the traversing electrical signals. The patch-clamp electrode method, on the other hand, does allow for precise recording of these signals, even at very low intensities, but it is limited to a small number of cells, making it less effective for large-scale neuron studies.

The new CMOS chip facilitates the study of interactions among a substantial number of neurons, paving the way for deeper insights into complex cognitive processes such as thought and learning. According to the researchers, each microhole functions similarly to a patch-clamp electrode, and incorporating thousands of these within a single chip allows for an extensive monitoring of neuronal activities.

“Microhole electrodes not only couple more efficiently with neuron interiors than previous vertical nanoneedle electrodes, but they are also simpler to produce,” explained Jun Wang, a researcher on the project. He highlighted that the ease of fabrication is a crucial aspect of their development.

Using the 4,096 microholes, the team successfully recorded activities from approximately 3,600 rat neurons, achieving nearly a 90% success rate. They documented over 70,000 connections, a staggering increase from a previous capability of 300 connections. Despite these advancements, the complexity of mapping the human brain, which contains roughly 86 billion neurons, remains a daunting task.

Even mapping just 2,000 neurons produces an enormous volume of data. “One of the major hurdles was managing and analyzing the vast data collected from our extensive parallel intracellular recordings,” stated Donhee Ham, another researcher involved in the study. The team has made significant progress in deciphering synaptic connections and is now developing a new design that could be used within a live brain.

Should they succeed in mapping neural connections in live brains, the potential applications are vast. This technology could revolutionize AI training and lead to the creation of more efficient AI chips, potentially offering immense computational power with minimal energy requirements. Additionally, it holds promise for mental health research, offering new insights into synaptic activity and its impacts on mental perceptions.