Croquis cerveau

The brain: the mechanics of convolutions

Why do our brains have so many folds? The answer to this question is far from simple. In fact, the answer only came at the beginning of 2016, from a team of researchers including members from Harvard University and Télécom Bretagne. Unlike some of the theories previously proposed, this answer has nothing to do with genetics. The convolutions in our brains are caused by mechanical constraints. This work was published in the Nature Physics journal, and co-authored by François Rousseau, a researcher at Télécom Bretagne.


With its multiple convolutions, the strange structure of our brains fascinates some people and frightens others, but leaves few people indifferent. The fact that this convoluted shape is not common to all species only increases this fascination. While humans and several primates have a sinuous brain, this is not the case for rodents, which have a smooth cortex. The scientific community has been debating the origins of these convolutions for years. Some researchers believe it is the result of complex biochemical complexes. In 1975, researchers from Harvard University proposed another theory: the brain’s development into convolutions is the result of mechanical constraints that emerge during its growth. Forty years later, an international collaboration has just confirmed this second hypothesis. This work, involving researchers from the universities of Harvard, Jyväskylä (Finland), Aix-Marseille and Télécom Bretagne, was published in the Nature Physics journal in February 2016.


Work with an international scope

To prove the role of physical constraints in the development of folds, the scientists first developed mechanical models by growing shapes similar to that of the brain. The team from Harvard first published the observations from these experiments in 2014. “But the physical models developed up until now were simulations carried out on a type of sphere, and the physicists merely observed whether or not folds appeared,” explains François Rousseau, researcher at Télécom Bretagne, and co-author of the publication. It was precisely for this purpose of refining the model, and making it closer to reality, that French scientists joined the team. Between 2008 and 2013, François Rousseau participated in the European Research Council project (ERC) on fetal brain MRI. His skills in signal processing were then used to extract data from the images. “It is difficult to obtain a good image of the fetal brain, since the fetus may move during the examination for example,” the researcher mentions. After developing and applying algorithms to correct the movement on the MRI images, the fetal brain can be identified and isolated from the surrounding liquid. After this point, the data can be used as the basis for 3D modeling.


François Rousseau, IMT Atlantique, brain, convolutions

Based on cross-sectional images of a fetal brain, François Rousseau sought to create 3D models.


It was during this process that François Rousseau met Nadine Girard and Julien Lefèvre, the recent winner of the young researcher grant from the French National Research Agency (ANR) on the study of the brain during its development. When they discovered the 2014 publication on the work by Tuomas Tallinen, Jun Young Chung and Lakshminarayanan Mahadevan, they decided to join them, with the conviction that their work could benefit the project by improving the physical model used by the researchers from Harvard and Jyväskylä. The work on the extraction of shapes carried out on fetal MRIs at different stages of prenatal development enabled the scientists to better understand brain development during the gestation period. This is how the twenty-second week of pregnancy came to be identified as a pivotal period, since it is the moment at which the brain enters a rapid growth phase. From this moment and until adulthood, its volume increases twentyfold. However, this increase does not take place in a consistent manner. Over the same period, the cerebral cortex – the brain’s outer layer – therefore increases to thirty times the volume it occupied during the twenty-second week of pregnancy.


Heterogeneous cerebral growth

And, it was precisely this asynchronicity that was identified as the potential source of mechanical stresses leading to the formation of convolutions. But the researchers still needed to prove it. Using MRI imaging of the fetal brain at twenty-two weeks, the team 3D-printed a replica, which was then used to form a silicon mold. Using this, the physicists then created imitation brains out of a gel material, which were then covered with another gel that could swell by absorbing a solvent such as hexane. After being exposed to the solvent for sixteen minutes, the brain model developed convolutions and folds that greatly resembled those of real brains. In addition, the stages in the development of these shapes were similar to those observed via MRI imaging. According to the researchers, the gyrification — the process of forming the folds — “is initiated by the formation of linear grooves, which grow longer and branch out, establishing most of the patterns before birth.


François Rousseau, IMT Atlantique, brain, convolutions

The researchers observed a development in the mechanical model (in pink above) similar to real brain development (in white, shown using synthetic imaging).


Although the results already supported the hypothesis presented in 1975, the researchers still wanted to improve on their simulations. This process has certain limitations. First of all, the observations reveal a notable asymmetry between the two hemispheres of the imitation brain. While perfect symmetry does not exist in a real human cortex either, the scientists noted that the two halves of the model “differ more than in real life”. Why is this? François Rousseau believes that this could be caused by “slight errors in the digital segmentation during the image processing stage, which may be amplified during the transition to the mechanical model and distort the simulation”. Secondly, due to the need to ensure the feasibility of the experiment, the researchers considered the growth of the cortex to be uniform, although they knew this was not the case. The model also omits the skull’s role in the development of the brain’s surface. Finally, the mechanical model is not yet able to attain a thirtyfold increase in its volume by absorbing the solvent, as the human brain does during its growth.

The researchers will seek to correct all of these details in the next stages of their work. And their exploration of this subject does not stop there. Beyond improvements to the model, the scientists want to take the simulations a step further, seeking to make the folding process take place in reverse order. This project could improve the detection and understanding of lissencephaly disorders – diseases caused by a genetic abnormality that results in a smooth cortex. “Using MRI imaging taken at a specific point-in-time, we would like to return to that point through simulation, to better understand how the changes occur in the folds’ structure,” explains François Rousseau.

Quèsaco, What is?, 5G, Frédéric Guilloud

What is 5G?

5G is the future network that will allow us to communicate wirelessly. How will it work? When will it be available for users? With the Mobile World Congress in full swing in Barcelona, we are launching our new “What is…?” series with Frédéric Guilloud, Research Professor at IMT Atlantique, who answers our questions about 5G.


What is 5G?

Frédéric Guilloud: 5G is the fifth generation of mobile telephone networks. It will replace 4G (also referred to as LTE, for Long Term Evolution). Designing and deploying a new generation of mobile communication systems takes a lot of time. This explains why, at a time when 4G has only recently become available to the general public, it is already time to think about 5G.

What will it be used for?

FG: Up until now, developing successive generations of mobile telephone networks has always been aimed at increasing network speed. Today, this paradigm is beginning to change: 5G is aimed at accommodating a variety of uses (very dense user environments, man-machine communications, etc.). The specifications for this network will therefore cover a very broad spectrum, especially in terms of network speed, transmission reliability, and time limits.

How will 5G work?

FG: Asking how 5G will work today would be like someone in the 1980s asking how GSM would work. Keep in mind that the standardization work for GSM began in 1982, and the first commercial brand was launched in 1992. Even though developing the 5th generation of mobile communications will not take as long as it did for the 2nd, we are still only in the early stages.

From a technical standpoint, there are many questions to consider. How can we make the different access layers (Wi-Fi, Bluetooth, etc.) compatible? Will 5G be able to handle heterogeneous networks, which do not have the same bandwidths? Will we be able to communicate using this network without disturbing these other networks? How can we increase reliability and reduce transmission times?

Several relevant solutions have already been discussed, particularly in the context of the METIS European project (see box). The use of new bandwidths, with higher frequencies, such as 60-80 GHz bands, is certainly an option. Another solution would be to use the space remaining on the spectrum, surrounding the bandwidths which are already being used (Wi-Fi, Bluetooth, etc.), without interfering with them, by using filters and designing new waveforms.

How will the 5G network be deployed?

FG: The initial development phase for 5G was completed with the end of the projects in the 7th Framework R&D Technological Program (FP7), and particularly through the METIS project in April 2015. The second phase is being facilitated by the H2020 projects, which are aimed at completing the pre-standardization work by 2017-2018. The standardization phase is then expected to last 2-3 years, and 2020 could very well mark the beginning of the 5G industrialization phase.


Find out more about Institut Mines-Télécom and France Brevets’ commitment to 5G

[box type=”shadow” align=”” class=”” width=””]

The METIS European project

The METIS project (Mobile and wireless communications Enablers for the Twenty-twenty Information Society) was one of the flagship projects of the 7th Framework R&D Technological Program (FP7) aimed at supporting the launch of 5G. It was completed in April 2015 and brought together approximately 30, primarily European, industrial and academic partners, including IMT Atlantique. METIS laid the foundations for designing a comprehensive system to respond to the needs of the 5G network by coordinating the wide variety of uses and the different technical solutions that will need to be implemented.

The continuation of the project will be part of the Horizon 2020 framework program. The METIS-II project, coordinated by the 5G-PPP (the public-private partnership that brings together telecommunications operators), is focused on the overall system for 5G. It will integrate contributions from other H2020 projects, such as COHERENT and FANTASTIC-5G, which were launched in July 2015: each of these projects are focused on specific aspects of 5G. The COHERENT project, in which Eurecom is participating (including Navid Nikain), is focused on developing a programmable cellular network. The FANTASTIC-5G project, with the participation of IMT Atlantique, under the leadership of Catherine Douillard, is aimed at studying, over a two-year period, the issues related to the physical layer (signal processing, coding, implementation, waveform, network access protocol, etc.) for frequencies under 6 GHz.

Find out more about the METIS / METIS-II project[/box]

Octave : sécuriser la biométrie vocale contre l’usurpation

Octave: trustworthy and robust voice biometric authentication

Projets européens H2020Surely, voice biometric authentication would be an easier alternative to the large amount of passwords that we use daily. One of the barriers to exploitation involves robustness to spoofing and challenging acoustic scenarios. In order to improve the reliability of voice biometric authentication systems, Nicholas Evans and his team at Eurecom are involved since June 2015 — and for a two years duration — in a H2020 European project called Octave.


What is the purpose of the Objective Control of Talker Verification (Octave) project?

Nicholas Evans: The general idea behind this project is to get rid of the use of passwords. They are expensive in terms of maintenance: most people have many different passwords and often forget them. While simultaneously relieving end-users from the inconvenience of dealing with textual passwords, Octave will reduce the economic and practical burden of service providers related to password loss and recovery. Octave will deliver a scalable, trusted biometric authentication service — or TBAS. The project is about providing a reliable service that works in diverse, practical scenarios, including data-sensitive and mission-critical application.


Eurecom is leading the third work package of this H2020 European project. What is the role of the school?

NE: Our main mission is to ensure the reliability of the underlying automatic speaker verification technology. To do so, our work package has two objectives. First, insuring the proper functioning of the TBAS in a variety of environments. Indeed, the Octave platform should work properly whether it be deployed in a limited bandwidth and channel-variable telephony context or in a noisy physical access context. Eurecom’s focus is on our second objective, which is counter-spoofing.


How does your research team ensure the security of the system against spoofing?

NE: If I want to steal your identity, one strategy might be to learn a model of your voice and then to build a system to transform mine into yours. Anything like that would typically introduce a processing artefact. I could also try to synthetize your voice, but again this would produce processing artefacts. So, one of the highest level approaches to identify a spoofing attempt is to build an artefact detector. In order to do that, we apply pattern recognition and machine learning algorithms to learn the processing artefacts from huge databases of spoofed speech.


Portable telephone


So researchers have a large database of spoofed speech at their disposal?

NE: This is a very tricky issue. Ideally, we would use real data, that is to say real examples of spoofed speech. These don’t exist, however. Even if they did, they would most likely not contain many samples. Therefore, we have to generate these spoofed speech datasets ourselves. We try to imagine how an attacker would try to spoof a system and then we fabricate a large number of spoofed samples in the same way. Fortunately, we can do this much better than a spoofer might, for we can imagine many possibilities and many advanced spoofing algorithms.

However, this methodology results in an unfortunate bias: when we use artificially generated datasets of spoofed speech, then we are in a really good position to know how spoofers faked the voice, because… well, we were the spoofers. To design reliable spoofing detectors we must then try to use the databases blindly, that is to say we must try not to use our knowledge of the spoofing attacks – in the real world, we will never know how the spoofing attacks were generated.

Luckily a very large, standard database of spoofed speech is now available and this database was used recently for a competitive evaluation. Since participants were not told anything about some of the spoofing attacks used to generate this database, the results are the best indication so far of how reliably we might be able to detect spoofing in the wild. Eurecom co-organised this evaluation, ASVspoof 2015, with another Octave partner, the University of Eastern Finland, among others.


Who are the other partners working along Eurecom on the Octave project?

NE: Among our partners, we count Validsoft in the United Kingdom, a voice biometrics product vendor. Eurecom is working with Validsoft to validate Octave technologies in a commercial grade voice biometrics platform. This is not the only category of industrial partners that we work with. Whereas APLcomp are another of Octave’s product vendor partners, Advalia are custom solution developers. ATOS are Octave’s large-scale ICT integrators. Business users are represented by airport operator, SEA, whereas Findomestic, owned by BNP Paribas Personal Finance, represent the banking sector. These two partners, SEA and Findomestic, will help us with evaluation, by offering us the possibility to deploy the TBAS in their respective environments. Airports and banking ecosystems are really different, allowing us to ensure that Octave works in real, diverse conditions.


Learn more about the Octave project


[box type=”shadow” align=”aligncenter” class=”” width=”95%”]


The Octave project:

The Objective Control of Talker Verification (Octave) project is a European project funded through the Horizon 2020 call on “Digital security: cybersecurity, privacy and trust”. It started in June 2015 and will last two years. The research program is segmented in eight work packages, among which the third, “Robustness in speaker verification”, is led by Eurecom. The school, part of the Institut Mines-Télécom, was contacted to work on Octave because of its experience on spoofing detection in voice biometric systems. Previous to Octave, Eurecom was involved in the FP7 project named Tabula Rasa.

List of Octave members:

carte partenaires Octave