3D medical images produced in health centers represent tremendous amounts of data spread over different places. Automatic processing of these data often allow quantitative and repeatable measurements on large databases that are valuable for diagnosis and pathology studies. Grid technologies offer large and distributed computing power suited to medical images analysis. However, the need for human control prevent complete off-line execution of some algorithms. Interactive programs with user supervision are made difficult on a grid architecture due to the remote program execution. In this paper we propose a new framework for interactive execution of remote applications dealing with large 3D medical data sets. Our solution is both efficient and easy to use from the user point of view. It can adapt to almost any medical application requirements. The formal architecture is presented and a prototype is demonstrated.
