ucla-cnp-skill

Provide the path to your local UCLA CNP BIDS data and request the full pipeline. The skill generates a numbered plan covering BIDS validation, sMRI, task-fMRI, and dMRI processing, waits for your confirmation, then delegates each step to the appropriate modality skill.

Run the validate_ucla_cnp.py script with --input pointing to the BIDS root and --output for the report path. It checks T1w, task-fMRI, and dMRI presence per subject and verifies expected task paradigms like stopsignal and spatialwm.

UCLA CNP (OpenNeuro ds000030) contains T1w structural MRI, task-fMRI across multiple cognitive paradigms, and diffusion MRI. It covers roughly 270 participants across healthy control, ADHD, bipolar disorder, and schizophrenia groups.

Yes. If your data is already on disk and you only need validation, the skill takes a narrow path of local discovery, BIDS validation, and reporting, skipping download and the full multimodal orchestration.

The QC script reads fMRIPrep confounds files and computes mean framewise displacement per subject. Subjects exceeding the FD threshold (default 0.3 mm) are flagged with exclusion reasons in the QC summary and an optional exclusion list CSV.

The skill is orchestration-only and contains no preprocessing implementation; all execution is delegated to modality skills like smri-skill, fmri-skill, and dwi-skill. Group comparisons should also account for unequal sample sizes across the four diagnostic groups.