Exploring the variabiIity induced by different configurations in the neuroimaging analytical space – VICUNA

Data processing pipelines are at the heart of modern experimental sciences. From data cleaning to statistical analysis they cover the essential steps that transform raw unprocessed data into scientific discoveries. But in practice, researchers face a highly complex pipeline landscape – different models, tools, algorithms, parameters – and recent evidence outlined that the exact choice of pipeline can sometimes lead to contradictory research findings.

This phenomenon is impacting many fields in experimental research and the field of software engineering is acutely looking at this problem. In recent years, they have proposed models to support understanding of the pipeline space. Yet many experimental fields still work on this topic in isolation at the risk of reinventing the wheel.

VICUNA will bring state-of-the art software engineering approaches to explore the pipeline space in the field of brain imaging. To allow for a large-scale automatic exploration of the pipeline-space and bridge the gap with software engineering, VICUNA will model the neuroimaging pipeline space as a feature model. To this aim, VICUNA will focus on three objectives: 1/ Modelling the analytical space, 2/ Recording provenance and 3/ Exploring the analytical space to investigate the main sources of variability.

By bringing more robust and generalizable research results in brain imaging, VICUNA will provide a proof-of-concept to bring more reproducible brain imaging results. This project will only be possible by building a deep understanding of existing pipelines and by working across the brain imaging and the software engineering research communities.