brainmass.objectives.fcd

Contents

brainmass.objectives.fcd#

brainmass.objectives.fcd(window_size=30, step_size=5, as_loss=False)[source]#

Build a functional-connectivity-dynamics (FCD) objective.

Surfaces braintools.metric.functional_connectivity_dynamics(), which had no call sites in the package. The FCD matrix captures how the sliding-window functional connectivity itself evolves over time.

Parameters:

  • window_size (int) – Sliding-window length and stride (in samples) forwarded to braintools.metric.functional_connectivity_dynamics().

  • step_size (int) – Sliding-window length and stride (in samples) forwarded to braintools.metric.functional_connectivity_dynamics().

  • as_loss (bool, default False) – If True, return 1 - corr (to minimise); otherwise the raw FCD matrix correlation (to maximise).

Returns:

fn(prediction, target=None). With target=None it returns the prediction’s FCD matrix (surfacing the metric); with a target it returns the correlation between the two FCD matrices.

Return type:

callable

Contents