Containerizing BIDSme: A Reproducible Tool for BIDS Conversion

The Brain Imaging Data Structure (BIDS) has become a widely adopted standard for organizing and sharing neuroimaging datasets across various modalities. Despite its benefits, converting raw brain imaging data into the BIDS framework remains a complex and time-consuming task. BIDSme is a semi-automated tool developed to streamline this conversion process, but until recently, it lacked the portability and accessibility needed for widespread adoption. This paper presents the containerization of BIDSme using Docker and Docker Compose, aiming to improve usability, reproducibility, and integration into existing platforms like Neurodesk. The work also details the design choices, iterative refinements, and validation process that led to a flexible, lightweight, and user-friendly containerized application.

The containerization of BIDSme was achieved using Docker and Docker Compose. Docker provides a lightweight, portable environment that encapsulates the application and its dependencies, ensuring consistent behavior across different systems. Docker Compose is used to define and run multi-container Docker applications, allowing for orchestration of the BIDSme container with any required services. The design choices focused on creating a flexible and lightweight container that could be easily integrated into existing platforms such as Neurodesk. The process involved iterative refinements to optimize the container's size, performance, and user-friendliness. Validation was performed to ensure that the containerized BIDSme functions correctly and reproducibly, meeting the needs of the neuroimaging community.