Package 'fmridataset'

Description

Select all voxels in the dataset mask.

Usage

all_selector()

Value

An object of class all_selector

Examples

# Select all voxels
sel <- all_selector()

Description

Provides a lightweight S3 interface that defers materialization of backend data. The returned object is compatible with delarr::collect() as well as base as.matrix() for realization.

Usage

as_delarr(backend, ...)

## S3 method for class 'matrix_backend'
as_delarr(backend, ...)

## S3 method for class 'nifti_backend'
as_delarr(backend, ...)

## S3 method for class 'study_backend'
as_delarr(backend, ...)

## Default S3 method:
as_delarr(backend, ...)

Arguments

Value

A delarr lazy matrix

Description

Provides a DelayedArray interface for storage backends. The returned object lazily retrieves data via the backend when subsets of the array are accessed.

Usage

as_delayed_array(backend, sparse_ok = FALSE, ...)

## S3 method for class 'nifti_backend'
as_delayed_array(backend, sparse_ok = FALSE, ...)

## S3 method for class 'matrix_backend'
as_delayed_array(backend, sparse_ok = FALSE, ...)

## S3 method for class 'study_backend'
as_delayed_array(backend, sparse_ok = FALSE, ...)

## Default S3 method:
as_delayed_array(backend, sparse_ok = FALSE, ...)

Arguments

Value

A DelayedArray object

Examples

## Not run: 
b <- matrix_backend(matrix(rnorm(20), nrow = 5))
da <- as_delayed_array(b)
dim(da)

## End(Not run)

Description

Provides DelayedArray interface for dataset objects. These methods convert fmri_dataset and matrix_dataset objects to DelayedArrays for memory-efficient operations.

Usage

## S4 method for signature 'matrix_dataset'
as_delayed_array(backend, sparse_ok = FALSE)

## S4 method for signature 'fmri_file_dataset'
as_delayed_array(backend, sparse_ok = FALSE)

## S4 method for signature 'fmri_mem_dataset'
as_delayed_array(backend, sparse_ok = FALSE)

Arguments

Description

Coerce a data frame into an fmri_group

Usage

as_fmri_group(
  subjects,
  id,
  dataset_col = "dataset",
  space = NULL,
  mask_strategy = c("subject_specific", "intersect", "union")
)

Arguments

Value

An fmri_group object.

Description

The returned tibble contains one row per voxel/timepoint combination with metadata columns from temporal_info and voxel_info and a signal column with the data values.

Usage

## S3 method for class 'fmri_series'
as_tibble(x, ...)

Arguments

Value

A tibble with columns from temporal_info, voxel_info, and a signal column containing the fMRI signal values

See Also

fmri_series for the class definition, as.matrix.fmri_series for matrix conversion

Examples

# Create small example
mat <- matrix(rnorm(12), nrow = 3, ncol = 4)

backend <- matrix_backend(mat, mask = rep(TRUE, ncol(mat)))
dataset <- fmri_dataset(backend, TR = 1, run_length = nrow(mat))
fs <- fmri_series(dataset)

# Convert to tibble
tbl_result <- tibble::as_tibble(fs)
# Result has 12 rows (3 timepoints x 4 voxels)
# with columns: time, voxel, signal

Description

Primary data access method for study-level datasets. By default this returns a lazy matrix (typically a delarr object) with row-wise metadata attached. When materialise = TRUE, the data matrix is materialised and returned as a tibble with metadata columns prepended.

Usage

## S3 method for class 'fmri_study_dataset'
as_tibble(x, materialise = FALSE, ...)

Arguments

Value

Either a lazy matrix with metadata attributes or a tibble when materialise = TRUE.

Description

Generic function to convert various fMRI dataset types to matrix_dataset objects. Provides a unified interface for getting matrix-based representations.

Usage

as.matrix_dataset(x, ...)

Arguments

Details

This function converts different dataset representations to the standard matrix_dataset format, which stores data as a matrix with timepoints in rows and voxels in columns. This is useful for algorithms that require matrix operations or when a consistent data format is needed.

Value

A matrix_dataset object with the same data as the input

See Also

matrix_dataset for creating matrix datasets, get_data_matrix for extracting data as matrix

Examples

# Convert various dataset types to matrix_dataset
# (example requires actual dataset object)
# mat_ds <- as.matrix_dataset(some_dataset)

Description

This method realizes the underlying lazy matrix and returns an ordinary matrix with timepoints in rows and voxels in columns.

Usage

## S3 method for class 'fmri_series'
as.matrix(x, ...)

Arguments

Value

A matrix with timepoints as rows and voxels as columns

See Also

fmri_series for the class definition, as_tibble.fmri_series for tibble conversion

Examples

# Create small example
mat <- matrix(rnorm(20), nrow = 4, ncol = 5)

backend <- matrix_backend(mat, mask = rep(TRUE, ncol(mat)))
dataset <- fmri_dataset(backend, TR = 1, run_length = nrow(mat))
fs <- fmri_series(dataset)

# Convert to matrix
mat_result <- as.matrix(fs)
dim(mat_result) # 4 x 5

Description

Opens a BIDS HDF5 archive created by compress_bids_study() and returns a bids_h5_study_dataset that is a subclass of fmri_study_dataset. All standard fmridataset methods (get_data_matrix, data_chunks, as_delarr, etc.) work on the returned object.

Usage

bids_h5_dataset(file, preload = FALSE)

Arguments

Value

A bids_h5_study_dataset object (subclass of fmri_study_dataset).

See Also

compress_bids_study, subset_bids_h5, participants, tasks, sessions

Description

Generates a vector of block/run identifiers for each timepoint.

Usage

blockids(x, ...)

## S3 method for class 'fmri_dataset'
blockids(x, ...)

## S3 method for class 'sampling_frame'
blockids(x, ...)

Arguments

Details

This function creates a vector where each element indicates which run/block the corresponding timepoint belongs to. This is useful for run-wise analyses or modeling run effects.

Value

Integer vector of length equal to total timepoints, with values indicating run membership (1 for first run, 2 for second, etc.)

See Also

get_run_lengths for run lengths, samples for timepoint indices

Examples

# Create a sampling frame with 2 runs of different lengths
sf <- fmrihrf::sampling_frame(blocklens = c(3, 4), TR = 2)
blockids(sf) # Returns: c(1, 1, 1, 2, 2, 2, 2)

Description

Generic function to extract block/run lengths from various objects. Extends the sampling_frame generic to work with dataset objects.

Usage

blocklens(x, ...)

## S3 method for class 'fmri_dataset'
blocklens(x, ...)

## S3 method for class 'sampling_frame'
blocklens(x, ...)

Arguments

Details

In fMRI experiments, data is often collected in multiple runs or blocks. This function extracts the length (number of timepoints) of each run. The sum of block lengths equals the total number of timepoints.

Value

Integer vector where each element represents the number of timepoints in the corresponding run/block

See Also

n_runs for number of runs, n_timepoints for total timepoints, get_run_lengths for alternative function name

Examples

# Create a dataset with 3 runs
sf <- fmrihrf::sampling_frame(blocklens = c(100, 120, 110), TR = 2)
blocklens(sf) # c(100, 120, 110)

# Create dataset with multiple runs
mat <- matrix(rnorm(330 * 50), nrow = 330, ncol = 50)
ds <- matrix_dataset(mat, TR = 2, run_length = c(100, 120, 110))
blocklens(ds) # c(100, 120, 110)

Description

This function collects all chunks from a chunk iterator into a list.

Usage

collect_chunks(chunk_iter)

Arguments

Value

A list containing all chunks from the iterator

Description

Converts a BIDS directory (or bidser::bids_project) into a single compressed HDF5 file containing compressed fMRI data, events, confounds, and study metadata. The output file can be opened with bids_h5_dataset.

Usage

compress_bids_study(
  x,
  file,
  mode = c("parcellated", "latent"),
  clusters = NULL,
  summary_fun = mean,
  encoding = NULL,
  n_components = NULL,
  template = NULL,
  mask = NULL,
  space = "MNI152NLin2009cAsym",
  tasks = NULL,
  subjects = NULL,
  sessions = NULL,
  confounds = NULL,
  compression = 4L,
  verbose = TRUE
)

Arguments

Details

The writer streams scans one at a time — only one NIfTI image is held in memory at a time. For each scan it:

1. Reads the NIfTI via neuroim2::read_vec().

2. For parcellated mode: computes parcel averages via fmristore::summarize_by_clusters() and writes [T, K] to /scans/<name>/data/summary_data.

3. For latent mode: encodes via fmrilatent::encode() and writes basis [T, K], loadings [V, K], and (optionally) offset [V] to /scans/<name>/data/.

4. Writes events, confounds, censor, and metadata sub-groups.

5. Releases the NIfTI from memory.

After all scans are written the /scan_index/ lookup table is populated and the function returns a bids_h5_dataset reader object.

Value

A bids_h5_dataset object (reader for the newly created file). If the reader is not yet available the file path is returned invisibly.

HDF5 schema

See bids_plan.md in the package source for the full v1.0 schema. The root compression_mode attribute reflects the chosen mode.

Examples

## Not run: 
library(bidser)
library(neuroim2)
library(fmristore)

bids_dir  <- system.file("extdata", "ds001", package = "bidser")
atlas     <- fmristore::get_schaefer_atlas(100)   # example atlas

# Parcellated mode
study <- compress_bids_study(
  x          = bids_dir,
  file       = tempfile(fileext = ".h5"),
  clusters   = atlas,
  tasks      = "nback",
  verbose    = TRUE
)

# Latent mode (PCA with 50 components)
study_lat <- compress_bids_study(
  x            = bids_dir,
  file         = tempfile(fileext = ".h5"),
  mode         = "latent",
  n_components = 50L,
  mask         = brain_mask,
  tasks        = "nback",
  verbose      = TRUE
)

## End(Not run)

Description

Creates a backend instance using the registered factory function. This is the main interface for creating backends by name.

Usage

create_backend(name, ..., validate = TRUE)

Arguments

Value

A storage backend object

Examples

## Not run: 
# Create a NIfTI backend (assuming it's registered)
backend <- create_backend("nifti",
  source = "data.nii",
  mask_source = "mask.nii"
)

# Create with validation disabled (faster, but riskier)
backend <- create_backend("nifti",
  source = "data.nii",
  mask_source = "mask.nii",
  validate = FALSE
)

## End(Not run)

Description

Creates a data chunk object that represents a subset of data from an fMRI dataset. A data chunk contains the data matrix along with indices indicating which voxels and time points (rows) are included in the chunk.

Usage

data_chunk(mat, voxel_ind, row_ind, chunk_num)

Arguments

Value

A data_chunk object containing:

data

The data matrix for this chunk

voxel_ind

Indices of voxels in this chunk

row_ind

Indices of rows (time points) in this chunk

chunk_num

The chunk number

Examples

# Create a simple data chunk
mat <- matrix(rnorm(100), nrow = 10, ncol = 10)
chunk <- data_chunk(mat, voxel_ind = 1:10, row_ind = 1:10, chunk_num = 1)
print(chunk)

Description

Generic function to create data chunks for parallel processing from various fMRI dataset types. Supports different chunking strategies.

Usage

data_chunks(x, nchunks = 1, runwise = FALSE, ...)

Arguments

Details

Large fMRI datasets can be processed more efficiently by dividing them into chunks. This function creates an iterator that yields data chunks for parallel or sequential processing. Two chunking strategies are supported:

• Equal-sized chunks: Divides voxels into approximately equal groups

• Run-wise chunks: Each chunk contains all voxels from one or more complete runs

Value

A chunk iterator object that yields data chunks when iterated

See Also

iter for iteration concepts

Examples

# Create a dataset
mat <- matrix(rnorm(100 * 1000), nrow = 100, ncol = 1000)
ds <- matrix_dataset(mat, TR = 2, run_length = 100)

# Create 4 chunks for parallel processing
chunks <- data_chunks(ds, nchunks = 4)

# Process chunks (example)
# results <- foreach(chunk = chunks) %dopar% {
#   process_chunk(chunk)
# }

Description

This function creates data chunks for fmri_file_dataset objects. It allows for the retrieval of run-wise or sequence-wise data chunks, as well as arbitrary chunks.

Usage

## S3 method for class 'fmri_file_dataset'
data_chunks(x, nchunks = 1, runwise = FALSE, ...)

Arguments

Value

A list of data chunks, with each chunk containing the data, voxel indices, row indices, and chunk number.

Description

This function creates data chunks for fmri_mem_dataset objects. It allows for the retrieval of run-wise or sequence-wise data chunks, as well as arbitrary chunks.

Usage

## S3 method for class 'fmri_mem_dataset'
data_chunks(x, nchunks = 1, runwise = FALSE, ...)

Arguments

Value

A list of data chunks, with each chunk containing the data, voxel indices, row indices, and chunk number.

Examples

## Not run: 
# Create a simple fmri_mem_dataset for demonstration
d <- c(10, 10, 10, 10)
nvec <- neuroim2::NeuroVec(array(rnorm(prod(d)), d), space = neuroim2::NeuroSpace(d))
mask <- neuroim2::LogicalNeuroVol(array(TRUE, d[1:3]), neuroim2::NeuroSpace(d[1:3]))
dset <- fmri_mem_dataset(list(nvec), mask, TR = 2)

# Create an iterator with 5 chunks
iter <- data_chunks(dset, nchunks = 5)
`%do%` <- foreach::`%do%`
y <- foreach::foreach(chunk = iter) %do% {
  colMeans(chunk$data)
}
length(y) == 5

# Create an iterator with 100 chunks
iter <- data_chunks(dset, nchunks = 100)
y <- foreach::foreach(chunk = iter) %do% {
  colMeans(chunk$data)
}
length(y) == 100

# Create a "runwise" iterator
iter <- data_chunks(dset, runwise = TRUE)
y <- foreach::foreach(chunk = iter) %do% {
  colMeans(chunk$data)
}
length(y) == 1

## End(Not run)

Description

This function creates data chunks for multi-subject study datasets.

Usage

## S3 method for class 'fmri_study_dataset'
data_chunks(x, nchunks = 1, runwise = FALSE, ...)

Arguments

Value

A list of data chunks, each containing data, indices and chunk number

Description

This function creates data chunks for matrix_dataset objects. It allows for the retrieval of run-wise or sequence-wise data chunks, as well as arbitrary chunks.

Usage

## S3 method for class 'matrix_dataset'
data_chunks(x, nchunks = 1, runwise = FALSE, ...)

Arguments

Value

A list of data chunks, each containing data, indices and chunk number

Description

Dimensions of fmri_series

Usage

## S3 method for class 'fmri_series'
dim(x)

Arguments

Value

Integer vector of length 2 (timepoints, voxels)

Description

Returns encoding metadata (family, parameters, number of components) stored in the /latent_meta/ group of a latent-mode BIDS HDF5 archive. Returns NULL for parcellated-mode archives.

Usage

encoding_info(x, ...)

## S3 method for class 'bids_h5_study_dataset'
encoding_info(x, ...)

Arguments

Value

A named list with elements encoding_family, encoding_params, and n_components, or NULL.

Description

This function creates an execution strategy that can be used to process fMRI datasets in different ways: voxelwise, runwise, or chunkwise.

Usage

exec_strategy(
  strategy = c("voxelwise", "runwise", "chunkwise"),
  nchunks = NULL
)

Arguments

Value

A function that takes a dataset and returns a chunk iterator configured according to the specified strategy.

Description

Expressions are evaluated in the context of subjects(gd) and may refer to its columns directly. Multiple expressions are combined with logical AND.

Usage

filter_subjects(gd, ...)

Arguments

Value

An updated fmri_group containing only the rows that satisfy all predicates.

Description

Returns information about the current state of the fmridataset cache, including size, number of objects, hit/miss rates, and memory usage.

Usage

fmri_cache_info()

Value

Named list with cache statistics

Examples

## Not run: 
# Get cache information
cache_info <- fmri_cache_info()
print(cache_info)

## End(Not run)

Description

Changes the maximum size of the cache. This will immediately evict objects if the new size is smaller than the current cache contents.

Usage

fmri_cache_resize(size_mb)

Arguments

Value

NULL (invisibly)

Examples

## Not run: 
# Resize cache to 1GB
fmri_cache_resize(1024)

# Check new size
fmri_cache_info()

## End(Not run)

Description

Clears the internal cache used by fmridataset for memoized file operations. This can be useful to free memory or force re-reading of files.

Usage

fmri_clear_cache()

Value

NULL (invisibly)

Examples

## Not run: 
# Clear the cache to free memory
fmri_clear_cache()

## End(Not run)

Description

This function creates an fMRI dataset object from a set of scans, design information, and other data. The new implementation uses a pluggable backend architecture.

Usage

fmri_dataset(
  scans,
  mask = NULL,
  TR,
  run_length,
  event_table = data.frame(),
  base_path = ".",
  censor = NULL,
  preload = FALSE,
  mode = c("normal", "bigvec", "mmap", "filebacked"),
  backend = NULL,
  dummy_mode = FALSE
)

Arguments

Value

An fMRI dataset object of class c("fmri_file_dataset", "volumetric_dataset", "fmri_dataset", "list").

Examples

## Not run: 
# Create an fMRI dataset with 3 scans and a mask
dset <- fmri_dataset(c("scan1.nii", "scan2.nii", "scan3.nii"),
  mask = "mask.nii", TR = 2, run_length = rep(300, 3),
  event_table = data.frame(
    onsets = c(3, 20, 99, 3, 20, 99, 3, 20, 99),
    run = c(1, 1, 1, 2, 2, 2, 3, 3, 3)
  )
)

# Create an fMRI dataset with 1 scan and a mask
dset <- fmri_dataset("scan1.nii",
  mask = "mask.nii", TR = 2,
  run_length = 300,
  event_table = data.frame(onsets = c(3, 20, 99), run = rep(1, 3))
)

# Create an fMRI dataset with a backend
backend <- nifti_backend(c("scan1.nii", "scan2.nii"), mask_source = "mask.nii")
dset <- fmri_dataset(backend, TR = 2, run_length = c(150, 150))

# Create a dummy dataset for testing (files don't need to exist)
dset_dummy <- fmri_dataset(
  scans = c("dummy1.nii", "dummy2.nii"),
  mask = "dummy_mask.nii",
  TR = 2,
  run_length = c(100, 100),
  dummy_mode = TRUE # Enable dummy mode for testing
)

## End(Not run)

Description

Backward compatibility wrapper for fmri_dataset. This function provides the same interface as the original fmri_dataset function.

Usage

fmri_dataset_legacy(scans, mask, TR, run_length, preload = FALSE, ...)

Arguments

Value

An fmri_dataset object

Examples

## Not run: 
# Create dataset from files
dset <- fmri_dataset_legacy(
  scans = c("scan1.nii", "scan2.nii"),
  mask = "mask.nii",
  TR = 2,
  run_length = c(100, 100)
)

## End(Not run)

Description

Create an fmri_group (one row per subject)

Usage

fmri_group(
  subjects,
  id,
  dataset_col = "dataset",
  space = NULL,
  mask_strategy = c("subject_specific", "intersect", "union")
)

Arguments

Value

An object of class fmri_group that wraps the input table.

Description

This function creates an fMRI dataset object specifically from H5 files using the fmristore package. Each scan is stored as an H5 file that loads to an H5NeuroVec object.

Usage

fmri_h5_dataset(
  h5_files,
  mask_source,
  TR,
  run_length,
  event_table = data.frame(),
  base_path = ".",
  censor = NULL,
  preload = FALSE,
  mask_dataset = "data/elements",
  data_dataset = "data"
)

Arguments

Value

An fMRI dataset object of class c("fmri_file_dataset", "volumetric_dataset", "fmri_dataset", "list").

Examples

## Not run: 
# Create an fMRI dataset with H5NeuroVec files (standard fmristore format)
dset <- fmri_h5_dataset(
  h5_files = c("scan1.h5", "scan2.h5", "scan3.h5"),
  mask_source = "mask.h5",
  TR = 2,
  run_length = c(150, 150, 150)
)

# Create an fMRI dataset with H5 files and NIfTI mask
dset <- fmri_h5_dataset(
  h5_files = "single_scan.h5",
  mask_source = "mask.nii",
  TR = 2,
  run_length = 300
)

# Custom dataset paths (if using non-standard H5 structure)
dset <- fmri_h5_dataset(
  h5_files = "custom_scan.h5",
  mask_source = "custom_mask.h5",
  TR = 2,
  run_length = 200,
  data_dataset = "my_data_path",
  mask_dataset = "my_mask_path"
)

## End(Not run)

Description

This function is deprecated. Please use latent_dataset() instead, which provides a proper interface for latent space data.

Usage

fmri_latent_dataset(
  latent_files,
  mask_source = NULL,
  TR,
  run_length,
  event_table = data.frame(),
  base_path = ".",
  censor = NULL,
  preload = FALSE
)

Arguments

Value

A latent_dataset object

See Also

latent_dataset

Examples

## Not run: 
# Use latent_dataset() instead:
dset <- latent_dataset(
  source = c("run1.lv.h5", "run2.lv.h5", "run3.lv.h5"),
  TR = 2,
  run_length = c(150, 150, 150)
)

## End(Not run)

Description

This function creates an fMRI memory dataset object, which is a list containing information about the scans, mask, TR, number of runs, event table, base path, sampling frame, and censor.

Usage

fmri_mem_dataset(
  scans,
  mask,
  TR,
  run_length = sapply(scans, function(x) dim(x)[4]),
  event_table = data.frame(),
  base_path = ".",
  censor = NULL
)

Arguments

Value

An fMRI memory dataset object of class c("fmri_mem_dataset", "volumetric_dataset", "fmri_dataset", "list").

Examples

# Create a NeuroVec object
d <- c(10, 10, 10, 10)
nvec <- neuroim2::NeuroVec(array(rnorm(prod(d)), d), space = neuroim2::NeuroSpace(d))

# Create a NeuroVol mask
mask <- neuroim2::NeuroVol(array(rnorm(10 * 10 * 10), d[1:3]), space = neuroim2::NeuroSpace(d[1:3]))
mask[mask < .5] <- 0

# Create an fmri_mem_dataset
dset <- fmri_mem_dataset(list(nvec), mask, TR = 2)

Description

An S3 class representing lazily accessed fMRI time series data. The underlying data is stored in a lazy matrix (typically a delarr object) with rows corresponding to timepoints and columns corresponding to voxels.

Core interface for retrieving voxel time series from fMRI datasets.

Usage

fmri_series(
  dataset,
  selector = NULL,
  timepoints = NULL,
  output = c("fmri_series", "DelayedMatrix"),
  event_window = NULL,
  ...
)

Arguments

Details

An fmri_series object contains:

• data: A lazy matrix with timepoints as rows and voxels as columns

• voxel_info: A data.frame containing spatial metadata for each voxel

• temporal_info: A data.frame containing metadata for each timepoint

• selection_info: A list describing how the data were selected

• dataset_info: A list describing the source dataset and backend

Value

An object of class fmri_series

An fmri_series (with a delarr lazy matrix payload) or a DelayedMatrix when output = "DelayedMatrix".

See Also

as.matrix.fmri_series for converting to standard matrix, as_tibble.fmri_series for converting to tibble format

Examples

# Create example fmri_series object
mat <- matrix(rnorm(100 * 50), nrow = 100, ncol = 50)
backend <- matrix_backend(mat, mask = rep(TRUE, ncol(mat)))
dataset <- fmri_dataset(backend, TR = 1, run_length = rep(25, 4))
fs <- fmri_series(dataset)
fs

Description

These functions convert user-facing selectors into numeric indices used by the fmri_series implementation. They are not exported to users directly.

Description

High level constructor that combines multiple fmri_dataset objects into a single study-level dataset using study_backend.

Usage

fmri_study_dataset(datasets, subject_ids = NULL)

Arguments

Value

An object of class fmri_study_dataset

Description

Creates an fMRI dataset object from Zarr array files. Zarr is a cloud-native array format that supports chunked, compressed storage and is ideal for large neuroimaging datasets.

Usage

fmri_zarr_dataset(
  zarr_source,
  TR,
  run_length,
  event_table = data.frame(),
  censor = NULL,
  preload = FALSE
)

Arguments

Details

The Zarr backend expects data organized as a 4D array with dimensions (x, y, z, time). The data is accessed lazily by default, loading only the requested chunks into memory.

Zarr stores should contain a single 4D array. For mask data, provide it separately through the fmri_dataset interface if needed.

Value

An fMRI dataset object of class c("fmri_file_dataset", "volumetric_dataset", "fmri_dataset", "list")

Experimental

This function uses the CRAN zarr package which is relatively new (v0.1.1, Dec 2025). It supports Zarr v3 format only - Zarr v2 stores cannot be read. Please report any issues to help improve the package.

See Also

zarr_backend, fmri_dataset

Examples

## Not run: 
# Local Zarr store
dataset <- fmri_zarr_dataset(
  "path/to/data.zarr",
  TR = 2,
  run_length = c(150, 150, 150)
)

# Remote store
dataset <- fmri_zarr_dataset(
  "https://example.com/subject01.zarr",
  TR = 1.5,
  run_length = 300
)

# Preload small dataset into memory
dataset <- fmri_zarr_dataset(
  "small_data.zarr",
  TR = 2,
  run_length = 100,
  preload = TRUE
)

## End(Not run)

Description

This file contains all generic function declarations for the refactored fmridataset package. These establish the interface contracts that are implemented by dataset-specific methods in other files.

Description

Retrieves information about a registered backend or lists all registered backends.

Usage

get_backend_registry(name = NULL)

Arguments

Value

For a specific backend: a list with registration details. For all backends: a named list where each element contains registration details.

Examples

# List all registered backends
get_backend_registry()

# Get specific backend info
get_backend_registry("nifti")

Description

Get metadata about the latent components in the dataset.

Usage

get_component_info(x, ...)

Arguments

Value

A list containing component metadata

See Also

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

Description

Generic function to retrieve confound regressor matrices stored in a BIDS HDF5 dataset.

Usage

get_confounds(x, ...)

## S3 method for class 'bids_h5_study_dataset'
get_confounds(x, scan_name = NULL, subject = NULL, task = NULL, ...)

Arguments

Value

A tibble (single scan) or named list of tibbles (multiple scans)

Description

Generic function to extract data from various fMRI dataset types. Returns the underlying data in its native format (NeuroVec, matrix, etc.).

Usage

get_data(x, ...)

Arguments

Details

This function extracts the raw data from dataset objects, preserving the original data type. For NeuroVec-based datasets, returns a NeuroVec object. For matrix-based datasets, returns a matrix.

Value

Dataset-specific data object:

• For fmri_dataset: Returns the underlying NeuroVec or matrix

• For matrix_dataset: Returns the data matrix

See Also

get_data_matrix for extracting data as a matrix, get_mask for extracting the mask

Examples

# Create a matrix dataset
mat <- matrix(rnorm(100 * 50), nrow = 100, ncol = 50)
ds <- matrix_dataset(mat, TR = 2, run_length = 100)

# Extract the data
data <- get_data(ds)
identical(data, mat) # TRUE

Description

Generic function to extract data as a matrix from various fMRI dataset types. Always returns a matrix with timepoints as rows and voxels as columns.

Usage

get_data_matrix(x, ...)

Arguments

Details

This function provides a unified interface for accessing fMRI data as a matrix, regardless of the underlying storage format. The returned matrix always has timepoints in rows and voxels in columns, matching the conventional fMRI data organization.

Value

A numeric matrix with dimensions:

• Rows: Number of timepoints

• Columns: Number of voxels (within mask)

See Also

get_data for extracting data in native format, as.matrix_dataset for converting to matrix dataset

Examples

# Create a matrix dataset
mat <- matrix(rnorm(100 * 50), nrow = 100, ncol = 50)
ds <- matrix_dataset(mat, TR = 2, run_length = 100)

# Extract as matrix
data_mat <- get_data_matrix(ds)
dim(data_mat) # 100 x 50

Description

Extract the latent scores (temporal components) from a latent dataset. This is the primary data access method for latent datasets.

Usage

get_latent_scores(x, rows = NULL, cols = NULL, ...)

Arguments

Value

Matrix of latent scores (time × components)

See Also

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

Description

Retrieves the spatial loadings matrix (or matrices) from a latent-mode BIDS HDF5 archive. Only available when compression_mode = "latent".

Usage

get_loadings(x, scan_name = NULL, ...)

## S3 method for class 'bids_h5_study_dataset'
get_loadings(x, scan_name = NULL, ...)

Arguments

Value

A numeric matrix [V, K] for a single scan, or a named list of such matrices when scan_name = NULL.

Description

Generic function to extract masks from various fMRI dataset types. Returns the mask in its appropriate format for the dataset type.

Usage

get_mask(x, ...)

Arguments

Details

The mask defines which voxels are included in the analysis. Different dataset types may store masks in different formats (logical vectors, NeuroVol objects, etc.). This function provides a unified interface for mask extraction.

Value

Mask object appropriate for the dataset type:

• For matrix_dataset: Logical vector

• For fmri_dataset: NeuroVol or logical vector

See Also

get_data for extracting data, get_data_matrix for extracting data as matrix

Examples

# Create a matrix dataset (matrix_dataset creates default mask internally)
mat <- matrix(rnorm(100 * 50), nrow = 100, ncol = 50)
ds <- matrix_dataset(mat, TR = 2, run_length = 100)

# Extract the mask (matrix_dataset creates all TRUE mask by default)
extracted_mask <- get_mask(ds)
sum(extracted_mask) # 50 (all TRUE values)

Description

Calculates the duration of each run in seconds.

Usage

get_run_duration(x, ...)

## S3 method for class 'fmri_dataset'
get_run_duration(x, ...)

## S3 method for class 'sampling_frame'
get_run_duration(x, ...)

Arguments

Value

Numeric vector where each element represents the duration of the corresponding run in seconds

See Also

get_total_duration for total duration, get_run_lengths for run lengths in timepoints

Examples

# Create a sampling frame with different run lengths
sf <- fmrihrf::sampling_frame(blocklens = c(100, 120), TR = 2)
get_run_duration(sf) # Returns: c(200, 240) seconds

Description

Extracts the lengths of individual runs/blocks from objects containing temporal structure information.

Usage

get_run_lengths(x, ...)

## S3 method for class 'fmri_dataset'
get_run_lengths(x, ...)

## S3 method for class 'sampling_frame'
get_run_lengths(x, ...)

Arguments

Details

This function is synonymous with blocklens but uses terminology more common in fMRI analysis. Each run represents a continuous acquisition period, and the run length is the number of timepoints (volumes) in that run.

Value

Integer vector where each element represents the number of timepoints in the corresponding run

See Also

blocklens for equivalent function, n_runs for number of runs, n_timepoints for total timepoints

Examples

# Create a sampling frame with 3 runs
sf <- fmrihrf::sampling_frame(blocklens = c(100, 120, 110), TR = 2)
get_run_lengths(sf) # Returns: c(100, 120, 110)

Description

Extract the spatial loadings (spatial components) from a latent dataset.

Usage

get_spatial_loadings(x, components = NULL, ...)

Arguments

Value

Matrix or sparse matrix of spatial loadings (voxels × components)

See Also

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

Description

Calculates the total duration of the fMRI acquisition in seconds across all runs.

Usage

get_total_duration(x, ...)

## S3 method for class 'fmri_dataset'
get_total_duration(x, ...)

## S3 method for class 'sampling_frame'
get_total_duration(x, ...)

Arguments

Value

Numeric value representing total duration in seconds

See Also

get_run_duration for individual run durations, get_TR for repetition time

Examples

# Create a sampling frame: 220 timepoints with TR = 2 seconds
sf <- fmrihrf::sampling_frame(blocklens = c(100, 120), TR = 2)
get_total_duration(sf) # Returns: 440 seconds

Description

Extracts the repetition time (TR) in seconds from objects containing temporal information about fMRI acquisitions.

Usage

get_TR(x, ...)

## S3 method for class 'fmri_dataset'
get_TR(x, ...)

## S3 method for class 'sampling_frame'
get_TR(x, ...)

Arguments

Details

The TR (repetition time) is the time between successive acquisitions of the same slice in an fMRI scan, typically measured in seconds. This parameter is crucial for temporal analyses and hemodynamic modeling.

Value

Numeric value representing TR in seconds

See Also

fmrihrf::sampling_frame() for creating temporal structures, get_total_duration for total scan duration

Examples

# Create a sampling frame with TR = 2 seconds
sf <- fmrihrf::sampling_frame(blocklens = c(100, 120), TR = 2)
get_TR(sf) # Returns: 2

Description

Map a function over subjects in an fmri_group

Usage

group_map(
  gd,
  .f,
  ...,
  out = c("list", "bind_rows"),
  order_by = NULL,
  on_error = c("stop", "warn", "skip")
)

Arguments

Value

Either a list (for out = "list") or a bound table (for out = "bind_rows").

Description

Reduce over subjects in a single pass

Usage

group_reduce(
  gd,
  .map,
  .reduce,
  .init,
  order_by = NULL,
  on_error = c("stop", "warn", "skip"),
  ...
)

Arguments

Value

The reduced value after visiting all subjects.

Description

Select voxels by their direct indices in the masked data.

Usage

index_selector(indices)

Arguments

Value

An object of class index_selector

Examples

# Select first 10 voxels
sel <- index_selector(1:10)

# Select specific voxels
sel <- index_selector(c(1, 5, 10, 20))

Description

Tests whether a backend type is registered in the system.

Usage

is_backend_registered(name)

Arguments

Value

Logical, TRUE if backend is registered

Examples

is_backend_registered("nifti") # TRUE (built-in)
is_backend_registered("custom") # FALSE (unless registered)

Description

Check if object is an fmri_series

Usage

is.fmri_series(x)

Arguments

Value

Logical TRUE if x is an fmri_series object

Description

This function tests whether an object is of class 'sampling_frame'.

Usage

is.sampling_frame(x)

Arguments

Value

TRUE if x is a sampling_frame object, FALSE otherwise

Description

Iterate subjects one-by-one (streaming)

Usage

iter_subjects(gd, order_by = NULL)

Arguments

Value

A list with a single element next that yields a one-row data.frame for each subject when called repeatedly. The dataset column is automatically flattened to the underlying fmri_dataset object.

Description

A specialized dataset interface for working with latent space representations of fMRI data. Unlike traditional fMRI datasets that work with voxel-space data, latent datasets operate on compressed representations using basis functions.

This interface is designed for data that has been decomposed into temporal components (basis functions) and spatial loadings, such as from PCA, ICA, or dictionary learning methods.

Creates a dataset object for working with latent space representations of fMRI data. This is the primary constructor for latent datasets.

Usage

latent_dataset(
  source,
  TR,
  run_length,
  event_table = data.frame(),
  base_path = ".",
  censor = NULL,
  preload = FALSE
)

Arguments

Details

Key Differences from Standard Datasets:

• Data Access: Returns latent scores (time × components) instead of voxel data

• Mask: Represents active components, not spatial voxels

• Dimensions: Component space rather than voxel space

• Reconstruction: Can optionally reconstruct to voxel space on demand

Data Structure:

Latent representations store data as:

• basis: Temporal components (n_timepoints × k_components)

• loadings: Spatial components (n_voxels × k_components)

• offset: Optional per-voxel offset terms

• Reconstruction: data = basis %*% t(loadings) + offset

Value

A latent_dataset object with class c("latent_dataset", "fmri_dataset").

See Also

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

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

Examples

## Not run: 
# From LatentNeuroVec files
dataset <- latent_dataset(
  source = c("run1.lv.h5", "run2.lv.h5"),
  TR = 2,
  run_length = c(100, 100)
)

# From pre-loaded objects
lvec1 <- fmristore::read_vec("run1.lv.h5")
lvec2 <- fmristore::read_vec("run2.lv.h5")
dataset <- latent_dataset(
  source = list(lvec1, lvec2),
  TR = 2,
  run_length = c(100, 100)
)

# Access latent scores
scores <- get_latent_scores(dataset)

# Get component metadata
comp_info <- get_component_info(dataset)

## End(Not run)

Description

Left join additional subject metadata

Usage

left_join_subjects(gd, y, by = NULL, ...)

Arguments

Value

An fmri_group with metadata from y attached.

Description

Returns a character vector of all registered backend names.

Usage

list_backend_names()

Value

Character vector of backend names

Examples

list_backend_names()

Description

Select voxels that are TRUE in a binary mask.

Usage

mask_selector(mask)

Arguments

Value

An object of class mask_selector

Examples

## Not run: 
# Using a logical vector
mask_vec <- backend_get_mask(dataset$backend)
sel <- mask_selector(mask_vec > 0.5)

## End(Not run)

Description

This function creates a matrix dataset object, which is a list containing information about the data matrix, TR, number of runs, event table, sampling frame, and mask.

Usage

matrix_dataset(datamat, TR, run_length, event_table = data.frame())

Arguments

Value

A matrix dataset object of class c("matrix_dataset", "fmri_dataset", "list").

Examples

# A matrix with 100 rows and 100 columns (voxels)
X <- matrix(rnorm(100 * 100), 100, 100)
dset <- matrix_dataset(X, TR = 2, run_length = 100)

Description

Adds or modifies columns on the underlying subjects table. Expressions are evaluated sequentially so newly created columns are available to later expressions.

Usage

mutate_subjects(gd, ...)

Arguments

Value

An updated fmri_group with modified subject attributes.

Description

Extracts the total number of runs/blocks from objects containing temporal structure information.

Usage

n_runs(x, ...)

## S3 method for class 'fmri_dataset'
n_runs(x, ...)

## S3 method for class 'fmri_study_dataset'
n_runs(x, ...)

## S3 method for class 'sampling_frame'
n_runs(x, ...)

Arguments

Value

Integer representing the total number of runs

See Also

get_run_lengths for individual run lengths, n_timepoints for total timepoints

Examples

# Create a sampling frame with 3 runs
sf <- fmrihrf::sampling_frame(blocklens = c(100, 120, 110), TR = 2)
n_runs(sf) # Returns: 3

Description

Number of subjects in a group

Usage

n_subjects(gd)

Arguments

Value

Integer number of subjects.

Description

Extracts the total number of timepoints (volumes) across all runs from objects containing temporal structure information.

Usage

n_timepoints(x, ...)

## S3 method for class 'fmri_dataset'
n_timepoints(x, ...)

## S3 method for class 'sampling_frame'
n_timepoints(x, ...)

Arguments

Value

Integer representing the total number of timepoints

See Also

n_runs for number of runs, get_run_lengths for individual run lengths

Examples

# Create a sampling frame with 3 runs
sf <- fmrihrf::sampling_frame(blocklens = c(100, 120, 110), TR = 2)
n_timepoints(sf) # Returns: 330 (sum of run lengths)

Description

Number of columns in fmri_series

Usage

## S3 method for class 'fmri_series'
ncol(x)

Arguments

Value

Number of voxels

Description

Number of rows in fmri_series

Usage

## S3 method for class 'fmri_series'
nrow(x)

Arguments

Value

Number of timepoints

Description

Generic function to retrieve parcellation metadata from a BIDS HDF5 dataset stored in parcellated mode.

Usage

parcellation_info(x, ...)

## S3 method for class 'bids_h5_study_dataset'
parcellation_info(x, ...)

Arguments

Value

A list with elements: cluster_ids, cluster_map, labels, n_parcels

Description

Generic function to extract participant/subject identifiers from study-level fMRI dataset objects. Defined here in fmridataset so methods work without requiring bidser to be installed.

Usage

participants(x, ...)

## S3 method for class 'bids_h5_study_dataset'
participants(x, ...)

Arguments

Value

Character vector of participant identifiers

Description

Display formatted summaries of fmridataset objects including datasets, chunk iterators, and data chunks.

This function prints a summary of a chunk iterator.

This function prints a summary of a data chunk.

Usage

## S3 method for class 'fmri_dataset'
print(x, full = FALSE, ...)

## S3 method for class 'fmri_dataset'
summary(object, ...)

## S3 method for class 'chunkiter'
print(x, ...)

## S3 method for class 'data_chunk'
print(x, ...)

## S3 method for class 'matrix_dataset'
print(x, ...)

Arguments

Value

The object invisibly

Examples

# Print dataset summary
# dataset <- fmri_dataset(...)
# print(dataset)
# print(dataset, full = TRUE)  # More details

Description

Prints a formatted summary of registered backends.

Usage

## S3 method for class 'backend_registry'
print(x, ...)

Arguments

Value

Invisibly returns the input

Description

Display a concise summary of an fmri_series object, including dimensions, selector type, backend, and data orientation.

Usage

## S3 method for class 'fmri_series'
print(x, ...)

Arguments

Value

Returns x invisibly. Called for its side effect of printing to the console.

Examples

# This method is called automatically when printing
# Create example object (see fmri_series documentation)
# fs <- new_fmri_series(...)
# fs  # Automatically calls print method

Description

Display formatted summaries of series selector objects.

Usage

## S3 method for class 'series_selector'
print(x, ...)

Arguments

Value

The object invisibly

Examples

# Print different selector types
sel1 <- index_selector(1:10)
print(sel1)

sel2 <- voxel_selector(c(10, 20, 15))
print(sel2)

Description

Reads a fMRI configuration file in YAML or JSON format. This replaces the previous implementation that used source() for security reasons.

Usage

read_fmri_config(file_name, base_path = NULL)

Arguments

Value

a fmri_config instance

Description

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

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()

Description

Registers a new storage backend type in the global registry. External packages can use this function to add custom backend support.

Usage

register_backend(
  name,
  factory,
  description = NULL,
  validate_function = NULL,
  overwrite = FALSE
)

Arguments

Details

The factory function should:

• Accept all necessary parameters to create a backend instance

• Return an object that inherits from "storage_backend"

• Implement all required storage backend methods

The validate_function should:

• Accept a backend object as first argument

• Return TRUE if valid, or throw informative error if invalid

• Perform any backend-specific validation beyond the standard contract

Value

Invisibly returns TRUE on successful registration

Examples

## Not run: 
# Register a custom backend
my_backend_factory <- function(source, ...) {
  # Create and return backend instance
  backend <- list(source = source, ...)
  class(backend) <- c("my_backend", "storage_backend")
  backend
}

register_backend(
  name = "my_backend",
  factory = my_backend_factory,
  description = "Custom backend for my data format"
)

## End(Not run)

Description

Converts a series selector specification into actual voxel indices within the dataset mask.

Usage

resolve_indices(selector, dataset, ...)

Arguments

Details

Series selectors provide various ways to specify spatial subsets of fMRI data. This generic function resolves these specifications into actual indices that can be used to extract data. Different selector types support different selection methods:

• index_selector: Direct indices into masked data

• voxel_selector: 3D coordinates

• roi_selector: Region of interest masks

• sphere_selector: Spherical regions

Value

Integer vector of indices into the masked data

See Also

series_selector for selector types, fmri_series for using selectors to extract data

Examples

# Example with index selector
sel <- index_selector(1:10)
# indices <- resolve_indices(sel, dataset)

# Example with voxel coordinates
sel <- voxel_selector(cbind(x = 10, y = 20, z = 15))
# indices <- resolve_indices(sel, dataset)

Description

Select voxels within a region of interest (ROI) volume or mask.

Usage

roi_selector(roi)

Arguments

Value

An object of class roi_selector

Examples

## Not run: 
# Using a binary mask
mask <- array(FALSE, dim = c(64, 64, 30))
mask[30:40, 30:40, 15:20] <- TRUE
sel <- roi_selector(mask)

## End(Not run)

Description

Sample subjects from an fmri_group

Usage

sample_subjects(gd, n, replace = FALSE, strata = NULL)

Arguments

Value

A sampled fmri_group.

Description

Generates a vector of timepoint indices, typically used for time series analysis or indexing operations.

Usage

samples(x, ...)

## S3 method for class 'fmri_dataset'
samples(x, ...)

## S3 method for class 'sampling_frame'
samples(x, ...)

Arguments

Value

Integer vector from 1 to the total number of timepoints

See Also

n_timepoints for total number of samples, blockids for run membership

Examples

# Create a sampling frame
sf <- fmrihrf::sampling_frame(blocklens = c(100, 120), TR = 2)
s <- samples(sf)
length(s) # 220
range(s) # c(1, 220)

Description

Generic function to extract the per-scan metadata table from a BIDS HDF5 study dataset.

Usage

scan_manifest(x, ...)

## S3 method for class 'bids_h5_study_dataset'
scan_manifest(x, ...)

Arguments

Value

A tibble with columns: scan_name, subject, task, session, run, n_time, has_events, has_confounds

Description

series() forwards to fmri_series() for backward compatibility. A deprecation warning is emitted once per session.

Usage

series(
  dataset,
  selector = NULL,
  timepoints = NULL,
  output = c("fmri_series", "DelayedMatrix"),
  event_window = NULL,
  ...
)

Arguments

Value

See fmri_series()

Description

A family of S3 classes for specifying spatial selections in fMRI datasets. These selectors provide a type-safe, explicit interface for selecting voxels in fmri_series() and related functions.

Description

Generic function to extract session names from study-level fMRI dataset objects. Defined here in fmridataset so methods work without requiring bidser to be installed.

Usage

sessions(x, ...)

## S3 method for class 'bids_h5_study_dataset'
sessions(x, ...)

Arguments

Value

Character vector of session names, or NULL if no sessions

Description

Select voxels within a spherical region.

Usage

sphere_selector(center, radius)

Arguments

Value

An object of class sphere_selector

Examples

# Select 10-voxel radius sphere around voxel (30, 30, 20)
sel <- sphere_selector(center = c(30, 30, 20), radius = 10)

Description

Stream subjects with optional ordering

Usage

stream_subjects(gd, prefetch = 1L, order_by = NULL)

Arguments

Value

An iterator identical to iter_subjects().

Description

Composite backend that lazily combines multiple subject-level backends.

Usage

study_backend(
  backends,
  subject_ids = NULL,
  strict = getOption("fmridataset.mask_check", "identical")
)

Arguments

Value

A study_backend object

Description

Converts a bids_h5_study_dataset (or any fmri_study_dataset) to an fmri_group object with one row per subject. Use this when you need per-subject group operations via group_map().

Usage

study_to_group(x, ...)

## S3 method for class 'bids_h5_study_dataset'
study_to_group(x, ...)

## S3 method for class 'fmri_study_dataset'
study_to_group(x, ...)

Arguments

Value

An fmri_group with columns subject_id and dataset.

Description

Generic function to extract subject identifiers from multi-subject fMRI dataset objects.

Usage

subject_ids(x, ...)

## S3 method for class 'fmri_study_dataset'
subject_ids(x, ...)

Arguments

Details

Multi-subject datasets contain data from multiple participants. This function extracts the subject identifiers associated with each dataset. The order of subject IDs corresponds to the order of datasets.

Value

Character vector of subject identifiers

Description

Access the subjects tibble stored inside an fmri_group

Usage

subjects(x)

subjects(x) <- value

Arguments

Value

The underlying data.frame with one row per subject.

An updated fmri_group.

Description

Filters a bids_h5_study_dataset by task, subject, session, and/or run using standard (non-NSE) evaluation. Returns a new bids_h5_study_dataset built from the matching scans, sharing the same underlying HDF5 file handle.

Usage

subset_bids_h5(x, task = NULL, subject = NULL, session = NULL, run = NULL)

Arguments

Value

A new bids_h5_study_dataset containing only the matching scans.

Description

Generic function to extract task names from study-level fMRI dataset objects. Defined here in fmridataset so methods work without requiring bidser to be installed.

Usage

tasks(x, ...)

## S3 method for class 'bids_h5_study_dataset'
tasks(x, ...)

Arguments

Value

Character vector of task names present in the dataset

Description

Removes a backend from the registry. Use with caution as this may break code that depends on the backend.

Usage

unregister_backend(name)

Arguments

Value

Invisibly returns TRUE if backend was removed, FALSE if it wasn't registered

Examples

## Not run: 
# Remove a custom backend
unregister_backend("my_custom_backend")

## End(Not run)

Description

Validate an fmri_group object

Usage

validate_fmri_group(x)

Arguments

Value

The validated fmri_group (invisibly).

Description

Select voxels by their 3D coordinates in the image space.

Usage

voxel_selector(coords)

Arguments

Value

An object of class voxel_selector

Examples

# Select single voxel
sel <- voxel_selector(c(10, 20, 15))

# Select multiple voxels
coords <- cbind(x = c(10, 20), y = c(20, 30), z = c(15, 15))
sel <- voxel_selector(coords)

Description

Helper for reattaching metadata after DelayedMatrixStats operations.

Usage

with_rowData(x, rowData)

Arguments

Value

x with rowData attribute set

Description

Writes a fMRI configuration to a YAML file for easy editing and sharing.

Usage

write_fmri_config(config, file_name)

Arguments