In the fight against Alzheimer’s disease, neuroscientists usually start with brain scans. They collect images of many patients’ brains, search these images for a ‘disease marker’ or signature of the disease, which might be something like a thinning in a certain area, changes to blood supply, or a protein aggregation, and then develop drugs that target these markers.
But, what happens to all these brain scans when a study is over? Are the images stored and reused? If so, who can access them? Can someone else run the same algorithm on a different set of brain images to see if the same disease markers are present? And if a researcher in India accessed brain scans taken in the UK, for example, would there be enough supporting clinical and image data to make it useful?
These questions are becoming more pressing as technology allows researchers to collect more and more images. In the past, research was done with just a handful of images; each laboratory might collect a few hundred brain scans. Today, projects have thousands of brain scans; some collections, such as the Alzheimer's Disease Neuroimaging Initiative (ADNI), based in California, contain as many as 15,000 brain scans.
On February 20-21, a high-level workshop was held at the International Telecommunication Union (ITU) in Geneva, Switzerland, to address these issues. There are three major collaborating initiatives in the world working in computational neuroscience: neuGRID for you (N4U) in Europe, Laboratory of Neural Imaging (LONI) in the US, and C-BRAIN in Canada. In their respective regions, each of these three projects are providing services such as access to large data sets, sophisticated algorithms to extract disease markers, and access to computational resources.
Neuroscientists the world over would like these three systems to work together, such that any researcher would be able to access the resources in any of the three projects, without even knowing which project is supplying them. A project, called outGRID, is working on the interoperability of these systems, and to incorporate resources from other areas, such as India, as they emerge.
“Making three complex systems interoperable is a huge effort,” said Giovanni Frisoni, the neuGRID and outGRID coordinator and Vice Scientific Director of San Giovanni di Dio Fatebenefratelli Institute in Italy. “It’s like having BMW, Mercedes, and Fiat join together to develop the same car.”
Another challenge for outGRID is the many different ways that researchers define indicators, diagnostic criteria, tools and data collection methodologies, as well as legal and privacy issues in clinical and research settings.
By 2030, there will be 65.7 million people worldwide living with Alzheimer’s disease or dementia, and by 2050 this number will rise to 115.4 million, according to Tarun Dua, a medical officer at the World Health Organisation in Geneva, who presented at the workshop on February 20. There have recently been five new government-based funding initiatives across Europe, North America, and India to further research on Alzheimer’s and related diseases.
The neuroscience community is one of the most advanced fields in terms of data acquisition and sharing, and outGRID will likely set the trends for e-health and e-science in the future. One of the strategies to be adopted by the outGRID community is an open access policy. “An open access policy in all participating countries is strongly recommended. Open access extends not only to scientific data but to all elements necessary to ensure the availability and reproducibility of research results, including software,” according to the conference report.
Similarly, the infrastructure requirements for this community are just an example of the general need for global e-infrastructures for other areas of research. “The platforms need to flexibly combine computing resources – cloud, grid, or high-performance computing – while providing a level of abstraction that shields the user from the complexities of the underlying computational infrastructure.”
“Funding agencies have a key role to play in ensuring the development and sustainability of such federated platforms, which need to combine existing e-infrastructures like C-BRAIN, N4U, DECIDE, and LONI, link them with more generic infrastructure such as EUDAT [European Data Infrastructure], PRACE [Partnership for Advanced Computing in Europe], and EGI [European Grid Infrastructure], and allow for emergent infrastructure to hook in, such as the one to be developed in India under the National Knowledge Network. High bandwidth connectivity across the cooperating regions, to be ensured through international collaboration between the NRENs [National Research and Education Networks], DANTE (Delivery of Advanced Network Technology to Europe)/GÉANT, TEIN [Trans-Eurasia Information Network], RedCLARA [Latin American Cooperation of Advanced Networks] and other actors is a condition for the above.”
Additional recommendations included defining benchmarks to compare platforms, infrastructures and tools. The ITU pledged to explore this issue within its e-health working group in regards to privacy.
“In regards to privacy, policy and infrastructure-related issues and the huge diversity on IPR [Intellectural Property Rights] internationally, the ITU can contribute in harmonizing data collection and use world-wide,” according to the report.
Kostas Glinos, Head of GEANT and e-Infrastructures Unit, DG-INFSO European Commission
The neuroscience community started sharing its data at an early stage in the development of the field. The large data sets that have been developed, such as the ADNI data set, have been opened up to every researcher that wants to use them. The field requires these data-driven research methods in order to identify disease markers in the brain.
Neuroscience is one of a small number of leading scientific fields with this kind of open access to data, and Glinos says he would like to see this in all research areas.
Giovanni Frisoni, neuGRID and outGRID coordinator and Vice Scientific Director of San Giovanni di Dio Fatebenefratelli Institute in Italy
Governments who want to join outGRID, “should fund the development of their own local infrastructures, but not as a standalone entity. They should give ICT scientists the mission to develop it to be interoperable with everything else that’s there in the world. If the funding agencies commit themselves in this way, ICT scientists and neuroscientists will follow.
“There are a number of political and technical challenges. The technical challenges are huge ... the research and development efforts to make these infrastructures interoperable are not trivial, but they can be overcome. It will take effort, it will take money, it will take time, but it can be done.”
“The political obstacles are less tangible and maybe more tricky. Each funding agency has its own agenda, its own policy, its own strategy … It’s much easier to have scientists agree than have funding agencies agree, but this is a good start. It’s a very good, promising start.”
Subrata Sinha, Director of National Brain Research Center, from the Ministry of Science and Technology, in India.
Vascular dementia, caused by problems in the supply of blood to the brain, is the second most common form of dementia worldwide and much more prevalent in India than in developed countries. Vascular dementia makes up as much as 25% of cases in India, compared to approximately 5% in developed countries. Conditions such as hypertension, diabetes, cardiovascular diseases, and stroke are correlated with higher rates of vascular dementia.
As well as Alzheimer’s and dementia, other diseases of interest in India include brain damage due to infection, brain malfunction due to malnutrition (both calorie-deficient malnutrition as well as micronutrient deficiency), consequences of birth asphyxia and low birth rates, and injuries, such as road injuries.
India is in the process of building its National Knowledge Network (NKN), a high-speed, pan-India network that is connecting approximately 1500 research and educational institutions. The I-Brain project, at the National Brain Research Centre in Manesar, allows researchers to collect and analyze brain images from research centers in seven cities (at present) through the NKN.
Indian brain research centers are already connected to Canada and, through that connection, to outGRID in Europe. Preliminary work has already begun. “India has made a proposal to link up jointly with outGRID – this is part of an infrastructure development proposal. So, these things are on the cards.”
Alan Evans, C-BRAIN director and professor of neurology and biomedical engineering at McGill University, Montréal.
“The Big Brain data set is essentially 7,400 post mortem sections, 20 micron sections across the whole brain … it’s something on the order of a terabyte of data. This kind of data is very difficult to work with. Each slice is essentially an independent data set subject to rips and tears, geometric distortion and optical imbalance. There are many, many different issues from slice to slice.”
“Stitching all of that back together again is really quite a monumental computational and handling challenge. We, in Montreal, have had the job of re-constituting this as a 3-D data set that can be explored in any direction … This 3-D data set will give researchers the flexibility to explore the 3-D folding pattern of the brain and the microstructure of the brain in a way that’s never been possible before.”
“An important part of the kind of work we do is the use of a standardized template that all brains are mapped to. We typically use low resolution MRI data sets as a target. Here, we’re going to replace that with a high-resolution reference data set that goes down to the 20 micron resolution instead of 1 mm resolution. This will allow us a reference brain that has much more information about individual brain regions that are mapped onto it. So, it’s an interpretive device.”