Multiple Sclerosis Brain MRI Segmentation Workflow deployment on the EGEE Grid

Automatic brain MRI segmentations methods are useful but computationally intensive tools in med- ical image computing. Deploying them on grid infrastructures can provide an efficient resource for data handling and computing power. In this study, an efficient imp lementation of a brain MRI segmentation method through a grid-interfaced workflow enactor is propos ed. The deployment of the workflow en- ables simultaneous processing and validation. The importance of parallelism is shown with concurrent analysis of several MRI subjects. The results obtained from the grid have been compared to the results computed locally on only one computer. Thanks to the power of the grid, method's parameter influence on the resulting segmentations has also been assessed given the best compromise between algorithm speed and results accuracy. This deployment highlights also the grid issue of a bottleneck effect.