fmri_ROI_phase2 {TCIU}

tensor-on-tensor regression on region of interest(ROI) of the brain

Description

This function takes a 4d fMRI data and detects locations where stimulus occurs on each region of interest(ROI) of the brain using MultiwayRegression. This function could be used as an intermediate step of a three-phase analytics protocol to detect motor areas. The functions to implement this three-phase protocol in a consecutive order is fmri_ROI_phase2, fmri_ROI_phase3 and fmri_post_hoc respectively.

Usage

fmri_ROI_phase2(
  fmridata,
  label_mask,
  label_dict,
  stimulus_idx,
  stimulus_dur,
  fmri.design_order = 2,
  fmri.stimulus_TR = 3,
  rrr_rank = 3,
  method = "t_test",
  parallel_computing = FALSE,
  ncor = max(detectCores() - 2, 1)
)

Arguments

fmridata

a 4d array which contains the spatial and temporal record of fmri result.

label_mask

a 3d nifti or 3d array of data that shows the labeled brain atlas.

label_dict

a dataframe or array or matrix to specify the indices and corresponding names of the ROI. The input of this parameter could take one of the list outputs of the fmri_ROI_phase2 function as a following step.

stimulus_idx

a vector of the start time points of the time period when the fMRI data receives stimulation.

stimulus_dur

a vector of the time period when the fMRI data receives stimulation.

fmri.design_order

a parameter to specify the order of the polynomial drift terms in fmri.design function.

fmri.stimulus_TR

a parameter to specify the time between scans in seconds in fmri.stimulus function.

rrr_rank

a parameter to specify the assumed rank of the coefficient array in rrr function.

method

a string that represents method for calculating p-values from tensor-on-tensor regression coefficients. There are 2 options: 't_test' and 'corrected_t_test'. The default is 't_test'. 't_test' is to calculate the test statistics 't-value' across all voxels in the bounding box of ROI; 'corrected_t_test' is to calculate the test statistics 't-value' by first across each voxel on a temporal basis, and then across all voxels in the bounding box of ROI.

parallel_computing

a logical parameter to determine whether to use parallel computing to speed up the function or not. The default is FALSE.

ncor

number of cores for parallel computing. The default is the number of cores of the computer minus 2.

Details

The function fmri_ROI_phase2 is used to detect locations where stimulus occurs by calculating the p-values of the ROI-based tensor-on-tensor regression. Two methods can be chosen to calculate the p-values from the regression coefficients.

Value

a 3d array storing ROI-based tensor regression p-values for the 4d fMRI data

Author(s)

SOCR team <http://socr.umich.edu/people/>

Examples

# sample 3D data of labeled brain atlas provided by the package
# this example will use parallel computing and take about ten minutes to finish
dim(mask_label)
# sample dataframe of ROI-based indices and names provided by the package
dim(mask_dict)
# sample 3D data of mask provided by the package
dim(mask)

# calculated p-values 
set.seed(1)
fmri_generate = fmri_simulate_func(dim_data = c(64, 64, 40), mask = mask)
fmridata = fmri_generate$fmri_data
stimulus_idx = fmri_generate$ons
stimulus_dur = fmri_generate$dur
# the function will may take a long time, see examples in demo function or vignettes