AI identifies brain cell types through electrical signatures

A decades of old challenge in neuroscience has been resolved by utilizing artificial intelligence (AI) to identify the electrical signatures of different types of brain cells for the first time, as part of a research in mice led by UCL researchers.

Brain consists of many different types of neurons (nerve cells in the brain), which are thought to play different roles in processing information. Scientists have long been able to use electrodes to capture the activity of neurons by detecting the electric ‘spikes’ that they generate while performing brain functions.

Although the recording of spikes has proven to be invaluable for monitoring the activity of individual neurons deep in the brain, the method has so far been ‘blind’ for the type of neuron that is registered – making it impossible to identify how different neurons contribute to the general effect of the brain.

In a new study, published in CellThe research team has overcome this problem by identifying the individual ‘electrical signatures’ of different neuronence types in the mouse brain, using short pulses of blue light to activate spikes in specific cell types (a method called optogenetics).

They created a library of the various electrical signatures for each type of neuron, so that they could then train an AI algorithm that can automatically recognize five different types of neurons with 95% accuracy without a further need for genetic tools. The algorithm was also validated on data from the brain intake of monkeys.

The researchers say they have overcome a major obstacle to use the technology to study neurological disorders such as epilepsy, but that there is still “a long way” to be used before it can be used in practical applications.

For decades, neuroscientists have been struggling with the fundamental problem of reliably identifying the many different types of neurons that are active during behavior at the same time. Our approach now enables us to identify Neuron types with more than 95% accuracy in mice and with monkeys.

This progress will enable researchers to record brain circuits while performing complex behavior such as movement. Just like logical gates on a computer chip, neurons in the brain are elementary computer units that take place in different types. Our method offers a tool to identify many of the logical gates of the brain at the same time. In the past it could only be done one by one and at much higher costs. “

Dr. Maxime Beau, co-first author of the Study of the UCL Wolfson Institute for Biomedical Research

The authors say that the fact that the algorithm can be applied to different types gives it enormous potential to be extended to other animals and ultimately to people.

In the short term, the new technique means that, instead of complex genetic manipulation to study the brain, researchers can use every normal animal to study what different neurons do and how they deal with each other to generate behavior.

One of the ultimate goals is to be able to study neurological and neuropsychiatric disorders such as epilepsy, autism and dementia, many of which are thought that they bring changes in the way different cell types act on each other.

Professor Beverley Clark, a senior author of the UCL Wolfson Institute for Biomedical Research, said: “Just as many different instruments in an orchestra contribute to the sound of a symphony, the brain is based on many different neuronentypes to create the complex behavior that people and other animals.

“Being able to observe this ‘neural symphony’ of the brain in action, has been a fundamental challenge in neuroscience for more than 100 years, and we now have a method to do this reliable.

“Although the technology is far away to be used to study neurological disorders such as epilepsy, we have now overcome a major obstacle to achieve that goal. In fact, some recording of living activities of human brains are already recorded in patients during the operation, and our technology can be used to study those recording in health and then in the disease.”

Improved insight into how our brains work can free up the way for a number of groundbreaking progress in medical science, some of which are already on the horizon.

Human brain-to-computer interfaces, or neural implants, are such a possibility. Walking research at the UCSF Weill Institute for Neurosciences, for example, has enabled a paralyzed man to control a robot arm with the help of a neural implant for a seven -month record. Just like the current study, this work was also informed by studying the electric patterns in the brains of animals and using AI to automatically recognize these patterns.

The authors say that the new technology to differentiate Neuron types can help to improve neural implants by more accurately recording which types of cells are involved in certain actions, so that the implant can easily recognize specific signals and generate the right response.

The key to this technology is to understand how our brains work when they are healthy, so that any damage can be compensated. For example, if a person had a stroke and part of his brain was damaged, you should understand how that bit worked before you could design an implant to replicate that functionality.

Professor Michael Häusser, a senior author of the study of the UCL Division of Medicine and the University of Hong Kong, said: “This project came to life thanks to the convergence of three critical innovations: the use of molecular biology to successfully ‘different neuronentypes with the help of mild, in-silon pro-reconciliations in Siliconen pro-reconciliations in Siliconen Probeo-in-Silicon.

“It is crucial that the synergy in our team was absolutely instrumental. The partner laboratories of UCL, Baylor, Duke and Bar Ilan University have all contributed critical documents to the puzzle. Like the brain, the whole is greater than the sum of its parts.”

The database collected by the team is freely available and the algorithm is open source, which means that scientists from all over the world can use these agents for neurological research.

This research was financed by financing Wellcare, National Institutes of Health (NIH), European Research Council (ERC) and the Horizon 2020 Research and Innovation Program of the European Union.