Sequence Selection

We developed 13 functional MRI sequences that we are comparing in their ability to record brain signals from subcortical auditory areas.

Project Goals

1. Select functional EPI sequence with parameters optimzed for data collection from subcortical brain regions.
2. Implement and evaluate use of denosing methods: NORDIC & TAPAS

Summary of Data Analysis Steps

All data is stored in 3 different folders, depending on the source (MRI scanner, behavioral software, BIOPAC).

All scripts ran by main_xxx.sh are located in .../scripts except for fMRIprep preprocessing, which is in data_MRI/code (BIDS standard requirement).

01 Curation

1. Get the data from the source and save : .dcm files from MRI scanner in data_MRI/sourcedata/dicoms .acq files from BIOPAC in data_physio/sourcedata .txt files from Expyriment in BIDS data_logs/bids --> These folders are untouched by next steps to ensure replicable pipeline.
2. Check that data is complete and correctly named.
3. Run pre_import.py to generate sidecar files, needed for the config file.
4. Set up the config file for BIDSifying MRI data with dcm2bids. The configuration doesn't have to include the headscouts and the phoenix ZIP report.
5. Add "dataset_description.json" file to data_MRI/sourcedata/raw
6. Run BIDS Validator on the dataset to ensure compliance to the BIDS specification.

02 Importing & Preprocessing

1. Run import_localizer.sh

   Outputs:

   • raw MRI data in BIDS with background-corrected T1 images
   • text files are moved to the appropriate MRI data folder (sub-00/ses-00/func/)
   • tmp folder (deleted) with phyio files in .mat and .txt format, and preprocessed physio files
   • tapas regressors and physio data structure per session, subject, and functional sequence.

TODO: integrate NORDIC denoising of functional scans

2. Exclude incomplete data (e.g.: a functional scan that was interrupted) and remove any duplicate images. Refer to the laboratory log to guide the decision (the log contains info on MRI protocol flow - tracks any changes, errors, modifications, ...).

3. Visually inspect T1 image.

4. Run fMRIprep_localizer.sh

   • note warnings, bugs, ...

   Outputs:

   • preprocessed fMRI data
   • use sloppy tag for testing the pipeline (faster)

5. Delete temporary cache directories once preprocessing is successful.

03 Analysis

1. Run analyze_localizer.py Outputs:
   • filtered artifacts
   • Bunch