find_best_hrf.matrix_dataset: find_best_hrf for a matrix_dataset

In bbuchsbaum/fmrireg: R package for the anlysis of fmri data

find_best_hrf for a matrix_dataset

Description

This method tries to find the best-fitting HRFs for each voxel by:

1. building/using an HRF library with PCA

2. creating splits in the event_table

3. convolving a "dummy" design

4. projecting data + library

5. picking best HRFs, optionally clustering + smoothing

Usage

## S3 method for class 'matrix_dataset'
find_best_hrf(
  dataset,
  onset_var,
  hrflib = NULL,
  rsam = seq(0, 24, by = 1),
  ncomp = 5,
  nsplits = 3,
  block = NULL,
  basemod = NULL,
  cluster_series = TRUE,
  cluster_quantile = 0.5,
  dist_method = "euclidean",
  spatial_smoothing = FALSE,
  lambda = 0.1,
  n_iterations = 5,
  k_neighbors = 6,
  nboot = 1,
  ...
)

Arguments

dataset	The dataset
onset_var	The name of the onset variable
hrflib	Possibly a function returning a matrix of candidate HRFs
rsam	Time points at which HRFs are sampled
ncomp	Number of basis components
nsplits	Number of splits
block	Possibly a formula specifying block structure
basemod	Possibly a baseline model
cluster_series	If TRUE, cluster the assigned HRFs
cluster_quantile	The quantile used in the clustering
dist_method	The distance measure
spatial_smoothing	If TRUE, smooth HRF assignments in space
lambda	Smoothing param
n_iterations	Number of smoothing iterations
k_neighbors	How many neighbors to consider
nboot	Number of bootstrap iterations
...	Other args

Value

A list with details about the best HRFs

bbuchsbaum/fmrireg documentation built on March 1, 2025, 11:20 a.m.