Initial volumes to delete before first level analysis with a multiband fMRI data

1

Hi all,

I have preprocessed my data using fmriprep.
Then my goal was to use a the DCM (Dynamic causal modeling ) tool from SPM, so i rather prefer doing all the processing with SPM.
However SPM does not deal with 4D nifti so I had to convert my 4D nifti file (space-MNI152NLin2009cAsym_preproc.nii.gz) in 3D nifti.
For every run, I have around 900 volumes (multiband with a TR = 400ms) and I would like to know how many initial volumes (“dummy scans”), I have to delete before starting the first level analysis.

I did not found any documentation about that regarding a multiband data.
(if someone knows a paper about that or has some experience with this …)

Thank you for your help,
Redwan

1 Like
2
  • SPM8 and 12 should support 4D files
  • Most modern scanners discard non-steady state (dummy) scans these days. We have an automatic detector of dummy scans built into FMRIPREP, but we are not exposing this information. You can just look at your data and check if the first few volumes are brighter, or alternatively run MRIQC which also has a built in detector and will indicate in the report if dummy scans were detected.
1 Like
3

To follow up, SPM will support uncompressed 4D files, it appears that your data was compressed. You will have to select all of the frames. This is done by typing ‘inf’ in the frame selection box, which is located just beneath the file filter in the SPM file selector GUI. Alternatively, this can be done by using the batch module “Expand Image Frames”

Additionally, some sequences do not discard dummy scans by default, even on SIEMENS scanners. If you wanted to mimic default behavior, use the following:
Dummy scans = ROUNDUP(3001/TR)
which is typical SIEMENS behavior. Thus, for your 400ms TR, you could discard 8 volumes.

The MRIQC data driven approach could suggest the removal of more volumes - sometimes the default is not sufficient.

2 Likes
4

Brief follow up: Information on Siemens Dummy scans taken from here

5

Thank you guys for your answer.
Now it works,

My data was uncompressed but I have not used the “inf” command to see all the volumes.
The MRI scanner, I am using, is a GE DIscovery MR 750 and it seems to discard the dummy scans.

Have a good afternoon !