1 General

This page gives an overview and brief information about the necessary steps to get from your raw DICOM files to an Analysis. You can find more detailed information by navigating to the sub-pages by clicking on the caption of the respective point.

2 Get your data

There are two ways how you get your data:

Regular scanning time:
- you get the data as nifti files
- a few days later via mri_dropoff in the J drive (Subpage for accessing other networks)
- nifti conversion and defacing generally are already done for you then (but check defacing anyway, sometimes this doesn´t work!)
Scanning with Natalia or piloting with Thomas
- we have the raw DICOM files
- conversion to niftis has to be done (see here)
- defacing has to be done (see here)

Warning

IMPORTANT: Load the DICOM files to the server and delete them from any external/local device (USB-Stick, local storage of your computer, etc.) as soon as possible!!!

3 DICOM-Files

the raw files from the scanner are DICOM-Files
in general, we primarily use .nii.gz/.nii (NIFTI) files for the subsequent steps

\(\to\) before we can continue, we need to convert DICOM- to .nii.gz files

DELETE DICOMS FROM YOUR OWN DEVICES AS SOON AS POSSIBLE

Load the DICOM files to the server and delete them from any external/local device (USB-Stick, local storage of your computer, etc.) as soon as possible!!!

4 DICOM to nifti

There are multiple options to convert DICOM to .nii.gz files:

Newest approach (recommended): use the BIDScoin workfolw

# call bidscoiner module
cd /shared/
ml bidscoin

# sort dicoms (otherwise not all the data will be discoverable by bidscoiner)
dcm2niix -o [YOUR_SORTED_DICOM_DIR] -r y [YOUR_ORIGINAL_DICOM_DIR]

# convert to bids
# edit following the instructions on
# https://visualneuroscience.github.io/mri/dicom-to-glm/01-dicom-to-bids.html#sec-bidscoiner
bidsmapper [YOUR_SORTED_DICOM_DIR] [YOUR_BIDS_DIR]

# convert
bidscoiner [YOUR_SORTED_DICOM_DIR] [YOUR_BIDS_DIR]

Older approach: using a script from Adam that can be found here (or here: /shared/website/dcm2bids_adam.sh). All steps are also described in a separate page
Older bidscoin approach: use the BIDScoin workflow implemented in /shared/website/dcm2bids_coiner.sh. All steps are also described in a separate page

all approaches should result in .nii.gz files in BIDS format

5 Check the “IntendedFor” entry of the fmap’s JSON file

“tells” fmriPrep where to apply distortion correction
depending on the fMRIprep version you plan to use, the field has to look slightly different!
to avoid unnecessary problems and delays, it doesn´t harm to always check this field before running fMRIprep!

6 Distribute the T1w files to participants?

if you promised participants that they will get their T1w images, you should make sure that they really get them
this should be done before defacing
You should delete the “non-defaced” files either after you distributed them to the respective participants, or after your study ended

7 Defacing your data

defaced data is necessary to safely run fMRIprep
Save faced T1w-files to give to your participants (if you got the data during regular scanning times, there should be a faced directory, but you should still make sure, that you have the faced T1w files if you need them for your participants!)
NEVER share data (e.g. on open Neuro) that was not defaced before!
Defacing is done with pydeface (Docker; Version 2.0.2)

8 (Remove denoising scan from your data)

if a denoising sequence was used, but you don´t use denoising data, you might have to remove the last volume of each scan before continuing

9 fmriprep

one of many BIDS-Apps (that require a BIDS-valid Dataset)
combines various anatomical and functional preprocessing steps within one tool
necessary, time intense process

10 Probabilistic Retinotopic Mapping

might be necessary to specifically test potential effects in specific visual areas (e.g. V1, V2, etc.)
provides retinotopy information (visual areas, eccentricity, etc.) even in the absence of dedicated retinotopy runs

11 first-level Analysis

in the first level GLM you process the data of individual subjects
for each subject, you define/estimate a model that can explain your data
you define meaningful contrasts (e.g. your experimental conditions)
you account for potential confounds

12 Group-Level ROI Analysis

in the group-level ROI analysis, you extract relevevant (for your research questions/hypothesis) information from your first level GLM
you should save this information in a format that allows you to later run your statistical analysis (e.g. as a .csv file)

13 Statistical Analysis

with your extracted data (that are saved in a .csv file), you can now run your statistical analysis