Building Grid-Based Applications for the Management and Analysis of Neuroimaging Data Sets for the Medical Grid

In this paper, we discuss the application of grid technology in the management and processing of digital medical images, in particular, functional magnetic resonance imaging (fMRI). Functional MRI is a non-invasive technique used to investigate the functions of the human brain. It usually involves the processing of thousands of MR images acquired as individual slices with a spatial resolution of a few millimeters and a repetition rate of a few seconds. The computational demand for processing fMRI datasets is usually high especially when processing whole brain fMRI time series. To cope with this computational demand, we developed BAXGrid, a grid-enabled fMRI analysis tool that allows the analysis of fMRI datasets using remote high-end computational resources. It is real-time capable, thus enabling real-time fMRI with a connection to the grid. With BAXGrid, significant reduction in processing time can be accomplished. We also developed BAXSQL, a grid-based fMRI data management and offline processing tool. BAXSQL enables access to distributed fMRI databases, allowing a unified view and a single point of management for datasets coming from different sources. Both BAXGrid and BAXSQL are built on top of Ninf-G, a versatile middleware for implementing remote procedure calls in a grid environment.