CuBIDS

CuBIDS validates and curates BIDS metadata to detect and classify variability in neuroimaging acquisition parameters and support reproducible dataset curation.

Key Features:

• Robust BIDS Validation: Performs scalable validation of BIDS datasets to detect incorrect values or omissions in BIDS metadata.
• Metadata Quality Control: Detects and classifies hidden variability within imaging metadata to support consistent curation.
• Identification of Unique Imaging Parameters: Identifies unique combinations of imaging parameters to enable selection of representative participant subsets for executing BIDS Apps and accelerating pipeline testing.
• Integration with DataLad: Optionally integrates DataLad for version control and provenance tracking of dataset changes.
• Python implementation: Implemented as a Python package for programmatic dataset validation and curation.

Scientific Applications:

• Multi-site and multi-scanner studies: Facilitates curation and harmonization of heterogeneous acquisition parameters across sites and scanners.
• Longitudinal studies: Detects session-to-session metadata inconsistencies across longitudinal datasets.
• Pipeline testing with BIDS Apps: Enables selection of representative subsets for executing BIDS Apps to test processing pipelines across diverse acquisition parameters.
• Reproducibility and provenance: Supports reproducible neuroimaging workflows and provenance tracking, particularly when combined with DataLad.

Methodology:

Performs scalable BIDS validation, detects and classifies hidden variability in BIDS metadata, identifies unique imaging-parameter combinations to select representative subsets for BIDS Apps, and optionally integrates DataLad for version control and provenance; implemented in Python.