The Experience and Practice of Developing a Brain Functional Analysis System Using ICA on a Grid Environment

For the effective and early diagnosis of brain diseases, we have been developing a brain functional analysis system on a Grid environment. Until recently, neuroscientists have proposed and exploited numerous methods for the analysis of brain function. Although recently proposed analysis methods tend to be computationally-intensive, they are performed, in most cases, on a single-processor basis such as a commodity personal computer. The system developed on the Grid environment allows researchers and medical doctors to integrate a variety of geographically distributed resources and to analyze functional brain data in a realistic time period without detailed knowledge of the Grid. Through this paper, we show that the system built on the Grid has the capability to deliver enough computational power to perform promising brain functional analysis such as Independent Component Analysis (ICA). In addition, we present the experience and practice related to developing a Grid-enabled system for brain functional analysis mainly from the viewpoint of application building.