Integration of a neuroimaging processing pipeline into a pan-canadian computing grid
Lavoie-Courchesne S., Rioux P., Chouinard-Decorte F., Sherif T., Rousseau ME., Das S., Adalat R., Doyon J., Craddock C., Margulies D., Chu C., Lyttelton O., Evans AC., Bellec P.
The ethos of the neuroimaging field is quickly moving towards the open sharing of resources, including both imaging databases and processing tools. As a neuroimaging database represents a large volume of datasets and as neuroimaging processing pipelines are composed of heterogeneous, computationally intensive tools, such open sharing raises specific computational challenges. This motivates the design of novel dedicated computing infrastructures. This paper describes an interface between PSOM, a code-oriented pipeline development framework, and CBRAIN, a web-oriented platform for grid computing. This interface was used to integrate a PSOM-compliant pipeline for preprocessing of structural and functional magnetic resonance imaging into CBRAIN. We further tested the capacity of our infrastructure to handle a real large-scale project. A neuroimaging database including close to 1000 subjects was preprocessed using our interface and publicly released to help the participants of the ADHD-200 international competition. This successful experiment demonstrated that our integrated grid-computing platform is a powerful solution for high-throughput pipeline analysis in the field of neuroimaging. © IOP Publishing Ltd.