Clearly significant results achieved have been:

1. Deployment of functioning grid nodes. The so-called Infrastructure Ground Truth has been established in the Data Coordination Centre and the Grid Coordination Centre. The delivered GCC and all DACS sites have been successfully installed; DACS1, DACS2 and DACS3 are all connected to the GEANT2 network. Additional computing resources were obtained from EGEE in order to support the final data challenge. The latter has been successfully executed over a total set of 2.000 CPU cores.

2. Publication of image datasets, image processing algorithms, and statistical and visualization tools. neuGRID provides several datasets (Pilot E-ADNI, ADNI, AddNeuroMed) that are currently being analysed using the platform's image processing algorithms and computational resources. The comparison of 3T MRI reproducibility than 1.5T for the SIENA brain atrophy measure applied to the Alzheimer's Disease Neuroimaging Initiative (ADNI) data set and the assessment of the reproducibility of the SienaX and Siena brain atrophy measures using the ADNI back to back MPRAGE MRI scans at 1.5T have been carried out and completed in the prototype neuGRID environment

3. Development of functioning services (portal, desktop fusion, provenance, databasing). The latest versions of the neuGRID services are integrated in the platform released and now available from http://neugrid.healthgrid.org. The following pipeline toolkits are available in neuGRID: FreeSurfer, BrainVisa, CIVET, FSL, R and ITK/VTK. NeuGRID users can now create fully functional workflows of algorithms/pipelines available in the grid and execute them against data coming from either the neuGRID data store or the LONI IDA in the US, thus opening the pathway to innovative new experiments.

4. Successful data challenge of an unprecedented size. The CIVET cortical thickness extraction pipeline has been successfully gridified. It has been tested on a large scale with a data challenge (socalled AC/DC2) that analysed the US-ADNI dataset of 6'500 brain image scans, and occupied the neuGRID computing resources full time for 10 consecutive days. The final data challenge (AC/DC3) successfully concluded after about 3 months of runtime with processing cycles involving 1'000 CPU cores in parallel. This generated 2.2 TB of scientific data now being analysed by neuroscientists involved in neuGRID.

5. Awareness at the scientific and decision making level of the vision and results of neuGRID. Neuroscientists are increasingly aware of the availability and potential of large image datasets to answer clinically poignant questions and the need of innovative infrastructures to fully exploit their scientific potential. Importantly, societal problems such as Alzheimer's require a global answer. Technical grid experts are beginning to view neuGRID as an exemplar of how high performance grid computing can be applied to future biomedical problems.

6. Uptake of neuGRID's solutions by related efforts. neuGRID's concept and vision have been taken up by related initiatives in Europe and elsewhere. neuGRID has already taken into account by the most relevant international initiatives in the field. FP7 DECIDE, aiming to develop a clinical diagnostic service for physicians caring for patients with suspected neurodegenerative diseases, is using the neuGRID infrastructure to run its applications.

Deliverable list