to ki.se
 
Home > Research > Projects > neuGRID

The neuGRID project


Overview


The neuGRID project was running during the years 2008-2011 and was officially ended in April 2011. The aim of the project was to assemble a European grid structure which offers powerful computational resources, access to huge databases currently being collected worldwide (while also providing the individual researcher with a browseable personal storage facility), together with a centrally-managed, easy-to-use set of sophisticated tools with which scientists can perform analyses and collaborate. It is the first e-infrastructure at European level which includes both neuroimaging research centers and IT companies (see a list of partners), and was funded by the European Union through the 7th Framework Programme. During the building phase, SMILE provided a user's perspective as one of three chosen pilot-site image laboratories.

The continuation project of neuGRID called neuGRID for you (N4U) is also funded by the European Union within the FP7-programme. This project aims to develop neuGRID in two ways: to include user groups from a wider set of disciplines (such as psychiatric diseases and multiple sclerosis) and to develop specific support centres which will increase the grid's user friendliness, e.g by on-line help desks and tutorials.

The European Commission has included neuGRID, outGRID, DECIDE and N4U within the list of "Success stories" on the Research Infrastructures website. You can read the story here or download it here.


Background and aims


Grid computing uses software to divide and apportion pieces of a program among several computers, as well as a form of network-distributed parallel processing. Grids are suitable to use when there is a need for great processing power or when storing and accessing a large amount of data. The grid has been set up in dual layers: level 0 contains the Data Coordination Centre and the Grid Coordination Centre, whereas level 1 consists of the Data Archiving and Computing Sites, of which there to date are three (one being SMILE). Here's a clarifying graph. neuGRID uses the EU research network GÉANT for connectivity (with a bandwidth of up to 10 Gbps) and is built with a service-oriented approach (i.e. the user needs to know nothing of the grid architecture).

The most important services of the grid are:
  • A portal service with a single login to access everything (data browser, job submission, visualization of results)
  • Management and processing of brain image data (data acquisition, data tracking and sharing, quality control, change of data formats and harmonization of data sets using a data dictionary) via the LORIS database system
  • Anonymization of data (both pseudonymization and face scrambling)
  • The LONI pipeline service with which one can create and design workflows, download and edit workflows and incorporate well-known algorithms such as FreeSurfer or BrainVISA)
  • A querying service by which data can be browsed and downloaded
  • A provenance service which keeps track of the history of processed data and of past workflows

Typical neuGRID scenarios include:
  1. A researcher interested in a rare type of disease and wants to make a statistically meaningful analysis:
    • The neuGRID store is queried and searched for data of this type
    • Access control and ethical policies are enforced to protect sensitive data
    • The researcher uses neuGRID to perform analysis on the selected data via a workflow
  2. An image acquisition center wishes to share data with the research community:
    • Data is put through quality control, formatting and ethical compliance
    • Data is integrated into the neuGRID standard data model
    • Other data centers can access it for research
  3. A researcher develops a new image analysis methodology and builds a workflow to run and test it:
    • An interactive tool is used to construct the workflow and set some initial parameters
    • The researcher creates a record which describes the workflow to the research community
    • neuGRID allows for different versions of the workflow to be created, tested and released when ready for use

Not only is data shared in an efficient manner, but new algorithms for brain analysis can be developed using a larger dataset to test the performance of the programs. A much faster spread of optimized algorithms within the research community can also be expected with the help of neuGRID.


Links and resources