Questions to the output of fMRIPREP regarding slice timing correction and distortion correction

fmriprep-report
fmriprep

#1

Hey all!

I have some questions regarding the output of the fMRIPrep pipeline. I uploaded two images from the output-html here: https://kth.box.com/s/tclu502bicoc5tgtl7yweeo9hj3bt4r1
(the images need to be downloaded and then opened with Google Chrome). My questions are:

sub-01: This is the “ROIs in BOLD space” - output image from the html for one subject. I solely chose this image to point out the visible intensity differences from slice timing artefacts. I wonder now if this is because the shown image is the previously calculated reference image of the BOLD image? Or did the slice timing correction not work properly? Do you know where this comes from and if it is acceptable? From my knowledge, the slice timing correction works without errors (all inputs and outputs in the report file seem to be ok).

sub-02: This is a gif of the BOLD image before / after distortion correction. There is an artefact created at the prefrontal cortex and I don’t know if this is acceptable. In all my subjects, the white and grey matter are better aligned with the boundaries calculated before (blue line), but some tend to have this artefact. Does anybody have an opinion about it?

My inputs in the pipeline were:

fmriprep-docker /BIDS_data /output_dir participant -w /work_dir
–template MNI152NLin2009cAsym --output-space template --use-aroma

and I provided slice timing times and a fieldmap image. I calculated the times for STC based on the information I was given as:
“SliceTiming”: [0.0, 1.0733333333333335, 0.04666666666666667, 1.12, … , 2.006666666666667, 0.98, 2.0533333333333337, 1.0266666666666668]
(RT = 2.1s and number of slices is 45 with interleaved and ascending order.)
I calculated the fieldmap based on the FUGUE website from FSL (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FUGUE/Guide). The fieldmaps looked okay in my opinion, with a median of ~10-50Hz up to 150-200Hz in areas of the prefrontal cortex. The data is BIDS-compatible.

Thank you for your help in advance!
Karen


#2

Hi Karen,

Thanks for the questions, and for using fMRIPrep!

  • sub-01: So slice-timing correction does not always resolve all visible slice artifacts. This appears to be a spin history effect, which occurs when there’s motion during acquisition (and is most apparent with interleaved acquisitions). That this is evident in the reference volume is not necessarily an indication of a general problem in the BOLD series, but we would recommend running MRIQC to have a more in-depth look at your BOLD data.
  • sub-02: This definitely looks like a problem to me. Could you share the whole report? This will include reports for other steps in fieldmap correction that we may be able to use to identify the issue. Additionally, if you’re able, you can upload (defaced) data to OpenNeuro.org and share it with me (this username @gmail.com).

Chris


EPI to T1 registration visualization
#3

Hej Chris,

I am sorry for my late answer! I uploaded the files to another box link I will directly send you to your email address, since I am not allowed to upload the files to OpenNeuro…

Thank you for your help!

Karen