Skip to contents

Reconstruct Voxel-Space Data from a Latent-Mode BIDS H5 Dataset

Source: R/all_generic.R, R/bids_h5_dataset.R, R/latent_dataset.R
reconstruct_voxels.Rd

Reconstructs full voxel-space time series for a scan by computing basis %*% t(loadings) + offset. Only available when compression_mode = "latent".

Reconstruct the full voxel-space data from the latent representation. This is computationally expensive and should be used sparingly.

Usage

reconstruct_voxels(x, scan_name = NULL, rows = NULL, voxels = NULL, ...)

# S3 method for class 'bids_h5_study_dataset'
reconstruct_voxels(x, scan_name, rows = NULL, voxels = NULL, ...)

# S3 method for class 'latent_dataset'
reconstruct_voxels(x, scan_name = NULL, rows = NULL, voxels = NULL, ...)

Arguments

x

A bids_h5_study_dataset object opened in latent mode.

scan_name

Character. Name of the scan to reconstruct.

rows

Integer vector of timepoint indices to return, or NULL for all.

voxels

Integer vector of voxel indices to return, or NULL for all.

...

Additional arguments passed to methods.

Value

A numeric matrix [T, V] (or subset thereof).

See also

Other latent_data: get_component_info(), get_latent_scores(), get_spatial_loadings(), latent_dataset()