modelarrayio.utils.cifti.load_cohort_cifti

modelarrayio.utils.cifti.load_cohort_cifti(cohort_long, s3_workers)[source]

Load all CIFTI scalar rows from the cohort, optionally in parallel.

The first file is always loaded serially to obtain the reference brain structure axis used for validation. When s3_workers > 1, remaining rows are submitted to a ThreadPoolExecutor and collected via as_completed. Threads share memory so reference_brain_names is accessed directly with no copying overhead.

Parameters:

  • cohort_long (pandas.DataFrame) – Long-format cohort dataframe

  • s3_workers (int) – Number of workers to use for parallel loading

Returns:

  • scalars (dict) – Per-scalar ordered list of 1-D subject arrays, ready for stripe-write.

  • reference_brain_names (numpy.ndarray) – Brain structure names from the first file, for building greyordinate table.