unravel.voxel_stats.vstats module#

Use vstats (vs) from UNRAVEL to run voxel-wise stats using FSL’s randomise_parallel command.

Prereqs:

  • Input images from vstats_prep, vstats_z_score, or vstats_whole_to_avg.

Inputs:

  • *.nii.gz files in the current directory with conditions as prefixes (e.g., saline_1.nii.gz, saline_2.nii.gz, drug_1.nii.gz, drug_2.nii.gz)

Outputs:

  • stats/ directory with randomise_parallel outputs (e.g., uncorrected 1-p value maps [vox_p]).

  • The name of the current directory is used as the prefix for the output files.

Next commands:

  • Run cstats_fdr_range and cstats_fdr to correct for multiple comparisons.

Note

  • 1 minus p (1-p) value maps are made for visualization (e.g., in FSLeyes), so use 0.95 as the uncorrect threshold for alpha = 0.05.

  • The –options flag is used to pass additional options to the randomise command.

  • It should be the last flag specified in the command.

  • The options should be specified as separate strings, e.g., –options –seed=1 -T

  • For info on how to set up and run voxel-wise analyses, see: https://b-heifets.github.io/UNRAVEL/guide.html#voxel-wise-stats

  • For a t-test design, tstat1 is group1 > group2 and tstat2 is group2 > group1 (conditions are sorted alphabetically)

  • For an ANOVA design, fstat1 is the first contrast, fstat2 is the second contrast, and so on (these are 1-p value maps are non-directional)

Usage:#

vstats [-mas mask.nii.gz] [-p 18000] [–kernel 0] [-a atlas/atlas_CCFv3_2020_30um.nii.gz] [-v] [–options –seed=1]

unravel.voxel_stats.vstats.parse_args()[source]#
unravel.voxel_stats.vstats.check_fdr_command()[source]#

Check if the ‘fdr’ command is available in the system’s path.

unravel.voxel_stats.vstats.create_design_ttest2(mat_file, group1_size, group2_size)[source]#

Create design matrix for a two-group unpaired t-test.

unravel.voxel_stats.vstats.get_groups_info()[source]#
unravel.voxel_stats.vstats.calculate_fragments(num_contrasts, total_permutations_per_contrast=18000, permutations_per_fragment=300)[source]#

Calculate the total number of fragments based on the number of contrasts.

unravel.voxel_stats.vstats.run_randomise_parallel(input_image_path, permutations, output_name, design_fts_path, mask_path=None, options=None, verbose=False)[source]#
unravel.voxel_stats.vstats.main()[source]#