search

canlab / MediationToolbox

Single-level and multi-level mediation analyses for any kind of data, with bootstrap-based significance testing. Neuroimaging-oriented functions allow for mediation effect parametric mapping (mapping of mediation effects across the brain) and multivariate mediation.
https://canlabweb.colorado.edu/wiki/doku.php/help/mediation/m3_mediation_fmri_toolbox
78 stars 28 forks source link

problem with defining multi-level variables #11

Closed Hannah-Savage closed 4 years ago

Hannah-Savage commented 4 years ago

Hi Tor and team,

I am hoping to use your toolbox to investigate the relationship between X: task activity/onsets, Y: SCR data, M: whole brain nii files, using the multi-level analysis method.

I have successfully run through the single-level mediation example provided.

In my data I have 25 subjects (but for brevity have included only 4 here). I have looked at the mediation_help pdf and I thought I had formatted and defined the X, Y, M and covariate variables correctly:

%%% This is the format of the variables I define %%%

load('Mediation_test_sample_onsets.mat'); x = group_onsets x = 4×1 cell array {1202×1 double} {1202×1 double} {1202×1 double} {1202×1 double}

load('Mediation_test_sample_SCR.mat'); y = group_downsampled_data y = 4×1 cell array {1202×1 double} {1202×1 double} {1202×1 double} {1202×1 double}

load('Mediation_test_sample_images.mat', 'image_files'); imgs = image_files imgs = 4×1 cell array {1202×1 string} {1202×1 string} {1202×1 string} {1202×1 string} %% image_files{1,1} looks like: "/data/netapp01/work/hssavage/7T/2_PreprocessedData/7T02/swrReversal.nii,1 " "/data/netapp01/work/hssavage/7T/2_PreprocessedData/7T02/swrReversal.nii,2 " "/data/netapp01/work/hssavage/7T/2_PreprocessedData/7T02/swrReversal.nii,3 " "/data/netapp01/work/hssavage/7T/2_PreprocessedData/7T02/swrReversal.nii,4 " ... up to swrReversal.nii,1202 %%

load('Mediation_test_sample_motion_covariates.mat'); %These are MRI motion regressors I would like to include as covariates of no interest% covs = group_motion_covariates covs = 4×1 cell array {1202×19 double} {1202×19 double} {1202×19 double} {1202×19 double}

I then run this:

mask=which('gray_matter_mask.img') mask = '/data/netapp01/work/hssavage/7T/MediationAnalysis/CanlabCore/CanlabCore/canlab_canonical_brains/Canonical_brains_surfaces/gray_matter_mask.img' canlab_results_fmridisplay(mask,'compact2'); cd ('Test_results/') mediation_brain_multilevel(group_onsets, group_downsampled_data, imgs, 'mask', mask, 'boot', 'nopreproc', 'covs', group_motion_covariates);

I get the following error message: Field assignment to a non-structure array object. Error in mediation_brain_multilevel (line 75) [SETUP.cmdstring, SETUP.wh_is_mediator] = get_cmdstring(X, Y, M);

I have attached the 4 .mat variables/files that I am using, and was wondering if you might be able to identify why my 'M' variable is not able to be interpreted properly? I am assuming it's the M variable that is wrong, but it could be the others!

Thanks in advance for your help,

Hannah

Savage_mediation.zip

Hannah-Savage commented 4 years ago

Hi there,

I have solved the problem! The M variable was defined as a string (as per the Mediation_help file "M could be a string matrix of 20 image names (one subject's image per row, one image per subject"). It seems to actually require a char vector as per 'get_cmdstring'. I think I just misread how the data should be stored.

Hopefully I can get the rest to run! Thanks for the toolbox, Hannah