DAX

DAX automates distributed processing of XNAT-hosted neuroimaging data to enable scalable, HPC-compatible analysis pipelines and conversion to the BIDS format for large-scale neuroimaging studies.

Key Features:

• Integration with XNAT and BIDS Standards: Facilitates conversion and compatibility between XNAT (Extensible Neuroimaging Archive Toolkit) and the Brain Imaging Data Structure (BIDS) for standardized neuroimaging data handling.
• Optimized Job Management: Provides job management optimized for high-performance computing that automates identification of recently updated sessions to reduce job-generation latency.
• Containerized Workflow Support: Supports execution of containerized neuroimaging workflows within HPC environments.
• YAML Configuration Processor: Uses YAML configuration processor scripts to abstract workflow inputs, outputs, commands, and job attributes.
• Database-driven Session Tracking: Implements an online database-driven mechanism to efficiently track session updates so only recent modifications are processed.
• Efficient XNAT-to-BIDS Conversion: Performs fast conversion of XNAT data to BIDS with speed comparable to direct XNAT data access.

Scientific Applications:

• Large-scale neuroimaging data management: Enables storage and processing of massive multimodal neuroimaging archives hosted on XNAT.
• BIDS dataset generation: Converts XNAT datasets to BIDS to support standardized analyses and tool interoperability.
• Scalable HPC pipeline execution: Orchestrates scalable execution of containerized analysis pipelines on high-performance computing infrastructures.

Methodology:

Computational methods explicitly include a YAML configuration processor that abstracts workflow inputs, outputs, commands, and job attributes; execution of containerized workflows in HPC environments; a database-driven mechanism to identify and track recently modified XNAT sessions; optimized job management to reduce job-generation latency; and XNAT-to-BIDS conversion implemented to match direct XNAT access speed.