Blog: Mapping the brain: computer science teams up with neuroscience in frontier EU research

Met dank overgenomen van A. (Andrus) Ansip i, gepubliceerd op dinsdag 17 juli 2018.

I have written in the past about the EU's commitment to emerging technologies, supporting and prioritising top-quality scientific research and innovation.

These are mammoth 10-year projects. They probe into fascinating areas of research, uncharted areas that stretch the boundaries of science and technology: the Future and Emerging Technology (FET) Flagships.

Last year, I wrote about the newest FET project, on quantum computing. In February, I wrote about graphene after visiting the Graphene Pavilion at the Mobile World Congress.

So it is only fair that I now complete my blog journey around the FET Flagship family with the Human Brain Project (HBP).

This project began in 2013 along with the Graphene Flagship.

To analyse and understand the human brain is one of the most complex challenges known to science.

This is very much a grand vision; one with huge implications for brain-related medicine and for advancing computing and artificial intelligence (AI).

The HBP sets out to answer some of science's toughest questions.

The first is to understand how the healthy brain works.

On brain disease, the HBP looks at what goes wrong in a diseased brain, how brain diseases develop and how they might be treated.

But it goes further than that.

HBP researchers are working to understand how the brain learns and to see if some of its computational capabilities could be replicated in computers and robots.

It aims to create research infrastructure for developing highly detailed multi-level models of the brain, and also to support personalised medicine for neurological and related diseases.

Detailed maps of the human brain are a key step towards understanding its structure and functioning

Simulation and supercomputers

Of the many approaches taken in the HBP, simulation is key to achieving these goals and deepening our understanding of the brain.

It allows us to perform 'virtual' experiments that would otherwise be impossible, ultimately helping to develop new therapies.

In 2015, HBP scientists created the world's first digital reconstruction of the micro-circuitry in a slice of a rat's brain: with around 31,000 brain cells, known as neurons, and 8 million connections to other neurons, known as synapses.

Since then, the simulation engine developed for the neocortex is being applied to another region of the brain - the cerebellum, the basal ganglia and hippocampus.

Let me put that into context: the entire human brain contains 86 billion neurons, each with up to 10,000 connections.

There is no way that this degree of complexity can be analysed without superior computers. The huge amount of data generated at the level of molecule cells and brain circuits requires the most sophisticated data analytics tools.

Big data analytics and simulation rely on massive computing power.

A normal computer is simply unable to simulate even a fraction of the human brain.

This is why the HBP uses a network of supercomputers, with centres in Germany, Italy, Spain and Switzerland, and another which has just joined in France.

These are some of the world's most powerful computers, performing quadrillions of operations per second, with a memory capacity in quadrillions of bytes. Part of their capacity will soon be useable interactively, which is particularly needed for handling vast amounts of brain data and for allowing simulations.

Highly advanced supercomputers will help neuroscience with learning experiments to support AI and data mining.

These areas' importance is one of several reasons why the EU is investing heavily into high-performance computing over the next decade.

Developing a high-level understanding of how neurons communicate can lead to a new category of powerful hardware that is low on power consumption - in the form of neuromorphic computing systems - also to advances in AI and robotics, and to an acceleration in neuroscience experiments.

These techniques also help us to understand the brain better.

By its very nature, the HBP is helping to push the boundaries of computing as we know it today, with both sides advancing technologically.

The economic, industrial and social impacts of such a shift are potentially enormous.

Achievements so far - and looking ahead

The project has consolidated progress during the last two years. Just a few examples:

Last year, HBP scientists started the first clinical trial towards personalised medicine in epilepsy, based on individualised brain models and simulation.

Other successes include the development of two computer prototypes to address neuroscience issues.

We also now have a much deeper understanding of interspecies differences in brain structures, including human microcircuits; we have seen a first model of the brain learning process based on neuron spiking activity; another first model of how the brain coordinates visual information and motor activities.

The list goes on. For the next two years, HBP teams will work to improve the integration of the different research platforms offered and provide solid scientific and technical support to the neuroscientists and engineers using them.

The aim is for this infrastructure to be available for research communities beyond those involved in the HBP, multiplying the potential of scientific breakthroughs.

In a practical and daily context, the HBP promises to be of enormous benefit across society - even before the project achieves its final goals and before we fully understand the human brain.

It opens up possibilities for better identification of diseases, new treatments including new drugs, and could significantly accelerate the process of clinical trials.

Models of the brain will revolutionise information technology, allowing us to design computers, robots, sensors, prosthetics and other devices that are far more powerful, intelligent and energy-efficient than today.

They will help us to identify the biological 'signature' of brain diseases, and to diagnose them early when they can still possibly be treated - and at the same time, gain deeper insights into what makes us human.

All credit to Europe's longstanding commitment to scientific research.

Another blog soon.