BIPS: A Framework for Curating and Executing Brain Imaging Pipelines
Satrajit Ghosh (Massachusetts Institute of Technology), Anisha Keshavan (Massachusetts Institute of Technology), John Salvatore (Massachusetts Institute of Technology), Arno Klein (Columbia University)
Brain Imaging Processing Services (BIPS), an opensource framework , was developed with the explicit aim of making electronic data capture easy by providing access to reusable tools and environments and providing tool-chains that allow users to execute analyses. The currently available tool-chains focus on dicom conversion, analyzing structural, "resting state" functional and diffusion data and providing quality assurance reports. At each stage of any analysis, provenance is captured and stored in a queriable database. The quality assurance scripts provide metrics in the context of other subjects and studies stored in the database. Every workflow in BIPS is associated with a unique identifier and once accepted into the package will not change. Much like version control systems, a modification to a workflow creates a new “commit” or workflow with its own unique identifier. The metadata associated with a workflow enables querying and configuring workflows. The framework and associated web services are being built to conform to the XCEDE data model.
This work was conducted with the Neuroimaging Task Force of the INCF Program on Standards for Datasharing and the Gabrieli Lab in McGovern Institute for Brain Research at MIT.
1. Poline et al., (2012) Data sharing in neuroimaging research. Front. Neuroinform. doi: 10.3389/fninf.2012.00009