Has anyone come across this before? At our second MRI site, fMRIprep seems to be having issues with the distortion correction (Phillips scanner). It looks like it’s struggling with the high-intensity CSF. Distortion correction is 1 fMRI scan in the opposite phase encoding direction. Normally we have 4, but here is what the output looks like. Wondering if anyone has any advice on this?
This didn’t happen when we ran preprocessing on our own with top-up (details of what we did below)
- (motion correction (mcflirt) on the resting AP and the PA distortion scan separately
- calculate a mean time series image for each just by averaging across each voxel’s intensities over time
- concatenate the two mean time series images together into a single 4D volume (using fslmerge) with the AP mean image as the first and the PA mean image as the second volume 4) motion correct (mcflirt) the concatenated mean images by making volume 0 (the AP image) as the reference volume (-refvol 0) and realigning the PA mean time series to it (typically there is a minimal change here but sometime people move quite a bit between the PA distortion scan the functional run). 5) run top-up to estimate the distortion field, and then I apply the top-up correction to the motion-corrected resting state AP scan.