Enhancing reproducibility in neuroimaging with containerized archival

Abstract Ensuring reproducibility in scientific research is crucial for validating findings, advancing knowledge, and fostering trust within the scientific community. Containerization of code is a widespread practice to help ensure reproducibility in neuroimaging research. However, there exists no universal method of sharing containers, nor is it feasible to broadly enforce rules due to the diversity of research practices. We posit that container sharing should include 1.) public availability, 2.) self-contained documentation, and 3.) procedures to ensure consistency of the container and its expected outputs. We demonstrate an approach for publicly releasing three separate containers using Zenodo. Our proposed method fulfills the design criteria for container sharing while incurring minimal overhead. As there are several tools available to distribute and maintain containers already, we discuss available alternatives.