Create a Multibasis MVPA Image Dataset

Creates an image dataset where each event has k basis beta maps. The dataset stores basis images separately and presents event-level observations with basis-concatenated features in downstream extraction methods.

Usage

mvpa_multibasis_dataset(
  train_data,
  test_data = NULL,
  mask,
  basis_count = NULL,
  ordering = c("event_major", "basis_major"),
  basis_labels = NULL
)

Arguments

train_data

Training data in one of the following forms:

• A list of NeuroVec objects (one per basis).

• A character vector of 4D image paths (one file per basis).

• A single 4D NeuroVec or image path containing concatenated basis volumes, where basis_count specifies splitting.

test_data

Optional test data in the same format as train_data.

mask

A NeuroVol mask.

basis_count

Number of basis functions when using a single concatenated 4D series.

ordering

Ordering of volumes in concatenated series: "event_major" means event1(b1..bk), event2(b1..bk), ...; "basis_major" means b1(all events), b2(all events), ....

basis_labels

Optional character labels for basis functions.

Value

An object of class mvpa_multibasis_image_dataset.

Examples

# \donttest{
  ds <- gen_sample_dataset(c(5,5,5), 20)
  # Create two basis series from same data
  mb <- mvpa_multibasis_dataset(
    train_data = list(ds$dataset$train_data, ds$dataset$train_data),
    mask = ds$dataset$mask
  )
# }