Tensor analysis of diffusion weighted magnetic resonance images is increasingly used for non-invasive tracking of nerve fibers in the human brain. Diffusion-tensor imaging (DTI) enables in-vivo research on the internal structure of the central nervous system, encompassing interconnection of functional areas, correlation between fiber deformations and certain desease patterns, as well as brain tumor localization. But the modeling of the local diffusion parameters is a computationally expensive part of the processing pipeline, resulting to run times up to days on standard desktop computers. A grid implementation of the algorithm with slice based parallelization reduces the processing down to 10% compared to a local cluster and 20% compared to sequential processing on the grid. A workflow implementation enables fault-tolerant handling of temporary failures within the grid. Furthermore, pure web-based access to the grid application allows for collaborative utilitzation even from protected infrastructures, as they are typically found in clinical environments.