Skip to contents


This page is to answer potential questions from users.

Prepare your data for ModelArray

What kind of fixel data should I use?

Before using ModelArray, you should be at the stage where you have got fixel data for every participant in template space from MRtrix by following the fixel-based analysis. If we use paper Dhollander et al., 2021 Fig.3. The fixel-based analysis pipeline as an example, we expect you have done the step “Connectivity-based fixel smoothing”. You will use the participant-level fixel data in template space from this step for further fixel-wise statistical analysis in ModelArray. We expect the file format is mif.

How should I organize the data?

The example data organization shown in the demo data (see vignette("walkthrough")) is what we recommend. In that example fixel dataset, the metric is FDC. If you have also other metrics such as FC, you may also have folder FC and CSV file cohort_FC_n100.csv in the myProject folder.

What are the requirements of the CSV file?

We expect this CSV file at least contains these two columns - see example in vignette("walkthrough"), the required columns are highlighted in bold and italics:

  • scalar_name, which tells ModelArray what metric is being analyzed
  • source_file, which tells ModelArray which mif file will be used for this subject

Other columns are covariates that you may want to include in the statistical model. The order of columns can be changed.

When using ModelArray.*() for statistical analysis

How many CPU cores does my computer have?

You may check this out: parallel::detectCores()