Build temporal covariance metric for parcellated fMRI data

Constructs a temporal covariance metric (row metric M) for parcellated fMRI data in ClusteredNeuroVec format. This is the parcel-level analogue of build_temporal_metric() for voxel data.

Usage

build_temporal_metric_parcel(
  cnv_run,
  wm_parcels = NULL,
  p = 1L,
  FD = NULL,
  DVARS = NULL,
  lambda_t = 0.3,
  ridge = 1e-06
)

Arguments

cnv_run

A ClusteredNeuroVec run.

wm_parcels

Optional integer indices of WM parcels for AR estimation.

p

AR order (default 1). Higher orders model longer-range dependencies.

FD, DVARS

Optional motion metrics for the run. FD is framewise displacement (mm), DVARS is temporal derivative of signal variance. See make_frame_weights for detailed definitions.

lambda_t

Temporal penalty weight (default 0.3). Higher values enforce more temporal smoothness.

ridge

Small ridge added to ensure positive definiteness (default 1e-6).

Value

Symmetric positive definite Matrix (T x T).

Details

The metric is M = W^{1/2} Q' H Q W^{1/2} + ridge * I, where:

• Q whitens the temporal covariance from AR(p) estimation

• H is the second-difference temporal smoothness penalty

• W is diagonal frame weighting from motion parameters

This parcel-level implementation uses parcel-averaged time series for AR estimation, providing computational efficiency for high-dimensional data.

See also

build_temporal_metric() for voxel-level temporal metric construction

Other temporal metrics: build_temporal_metric(), estimate_ar_whitener(), estimate_ar_whitener_parcel(), make_frame_weights(), make_temporal_penalty()

Examples

# \donttest{
# Build temporal metric with AR(1) whitening and motion regressors
M <- build_temporal_metric_parcel(cnv_run, wm_parcels = c(1,2,3),
                                  FD = fd_vals, DVARS = dvars_vals)
#> Error: object 'cnv_run' not found
# }