 |
 |
I am sure some of these questions have been asked before, but the more I read on the forum the more questions/problems/confusions arise. =] I will be running a DTI/fMRI study, DTI acq at 128X128 and fMRI at 64x64, both non-spiral EPI, on a GE Signa 3T. We have presently been working on how to acquire the proper GRE phase maps (yes, I know spin-echo is preferred, but unfortunately we do not have time for such long scans) that will be in the range of -pie to pie. The original phase maps we were acquiring had values ranging *on average* from about 2800 to 6400 and we were not sure of the conversion/scaling factors. A GE engineer suggested turning on parallel imaging, but then going into the CV's and changing the acceleration factor to 1. This would not actually acquire the images using parallel imaging, but would allow us to 1) bypass the processing stream that was giving us the above values and 2) would then give us values from -pie to pie x 1000. The only problem is that to do this, we must use a Fast GRE sequence (acquire several lines of k-space per echo). From reading other posts on the forum it seems this is not recommended; we should only be using conventional GRE. Is this correct? If this is correct, do the images with values from 2800 to 6400 need to scaled using fslmaths (I'm not even sure what scaling factor we would use) or do the fugue or prelude programs look at the header info and scale automatically since we will be using NIFTI format? I just read a post a little while ago that mentioned FSL ingnoring the scalling factor in the header for analyze, but not NIFTI files. Lastly, the GRE phase maps will be acquired at 128x128, is it ok to use these to correct for the distortion in the fMRI 64x64s (even though there probably will not be much)? Reason for using 64x64, instead of 128x128, is to reduce slice thickness (3mm) and use the same for DTI then overlay. Thank you, Carlos