Validate MCFLIRT→3dVolreg swap

[oesteban]

Objective

Validate that replacing FSL MCFLIRT with AFNI 3dVolreg (nipreps/fmriprep#3661) is semantically / methodologically substitutable — it preserves the head-motion-correction step's functional contract even though the two methods are not numerically identical. Validation combines a documented rubric with a concrete quantitative oracle.

Rationale

The Goal-4 accept bar is functional-contract preservation, not numerical identity, so it needs an operationalization that is rigorous and specific to motion correction. Framewise displacement (FD) is the functionally-relevant summary of head motion that downstream analyses consume; if the two methods yield very highly correlated FD series, the substitution preserves what matters — giving "semantic substitutability" a concrete, measurable oracle rather than a holistic judgment.

Important

The accept bar is semantic / methodological substitutability, not numerical identity: MCFLIRT and 3dVolreg differ numerically by construction. The FD-correlation threshold and datasets are pre-specified so the oracle cannot be tuned after the fact.

Methodological approach

1. Quantitative oracle — FD correlation. Compute framewise-displacement series from both the MCFLIRT (reference) and 3dVolreg (variant) runs across the benchmark; the substitution passes when the FD series are very highly correlated, with threshold and datasets pre-specified. This is the primary, measurable signal.
2. Rubric-based expert sign-off. A domain expert confirms, against a documented substitutability rubric (function preserved; downstream derivatives remain valid; no methodological regression), that the swap is sound — complementing the automatic functional-contract check in Expansion mechanism — synthesize a methodologically-valid alternative processing choice agentic-files#9.
3. Record. Produce a recorded sign-off carrying the FD-correlation evidence and the rubric outcome.

Outcomes/impact

1. An expert sign-off anchored to a very-high FD-series correlation between the two variants.

This establishes that MCFLIRT→3dVolreg is semantically substitutable — the concrete validation that admits the proof-of-concept variant (nipreps/fmriprep#3661) and shows the semantic-substitutability bar can be operationalized measurably, not merely judged.

Direct upstream issues

(issues this one depends on / builds upon)

• MCFLIRT → AFNI 3dVolreg feature swap fmriprep#3661

Direct downstream issues

(issues that depend on this one)

None — no downstream issue on the board yet.