Extract ROI-level timeseries from a BOLD image

Applies a parcellation atlas to a preprocessed BOLD image, returning mean timeseries per ROI. Works with any boldR_atlas object – built-in or custom – as long as the atlas is in the same template space as the BOLD.

Usage

parcellate(bold, atlas, summary_fn = mean, min_voxels = 10L)

Arguments

bold

A boldR_bold object from prepare_bold().

atlas

A boldR_atlas object from load_atlas().

summary_fn

Function. Applied across voxels within each ROI at each timepoint. Default mean. Common alternatives: median.

min_voxels

Integer. Minimum number of in-mask voxels an ROI must contain to be included in the output. ROIs below this threshold are returned with NA timeseries and a warning. Default 10L.

Value

A list of class boldR_parcellated with components:

timeseries

Numeric matrix, timepoints x ROIs. Column names are ROI labels from the atlas.

roi_voxel_counts

Integer vector. Number of in-mask voxels per ROI.

atlas

The boldR_atlas object used.

bold

The boldR_bold object used.

n_rois

Integer. Number of ROIs in output.

n_timepoints

Integer. Number of timepoints.

Examples

if (FALSE) { # \dontrun{
fprep  <- read_fmriprep("data/derivatives/fmriprep", subject = "01")
bold   <- prepare_bold(fprep, tr = 2, drop_volumes = 4)
atlas  <- load_atlas("schaefer100_7n")
parcel <- parcellate(bold, atlas)
dim(parcel$timeseries)  # timepoints x 100
} # }