Hi Gaurav, thanks for your prompt response! And thanks for adding to the pipeline the extension to deal with GE data.
The data are also from a GE MR750 scanner, but I have Re/Im images instead of a fieldmap. The steps described in my previous email should hopefully compute the fieldmap from the Re/Im. There are just a couple of details I am wondering about: - If the fieldmap is in degrees [0, 360], would it not be necessary to multiply by 6.28/360 to convert to rad? I can find the 6.28 factor in GeneralElectricFieldMapPreprocessingAll.sh, but no division by 360. - Is it not also necessary to account for the difference between TEs used to compute (in my case 2ms)? That is done in fsl_prepare_fieldmap, which is called in SiemensFieldMapPreprocessingAll.sh, but not in GeneralElectricFieldMapPreprocessingAll.sh. (Maybe this is already included in the fieldmap you are using, in which case the units would be deg/sec instead of deg). - Is there any way to check that the fieldmaps are correct (-ish)? e. g. which is the expected value range?. As I mentioned earlier, max shiftmap values between my dataset and an HCP example subject differ substantially (6x) and I am concerned that I could be missing a factor. I would be grateful if you or someone else from the list could shed light on these questions. Best regards, Benjamín Garzón, PhD Department of Neurobiology, Care Sciences and Society Aging Research Center | 113 30 Stockholm | Gävlegatan 16 benjamin.gar...@ki.se | www.ki-su-arc.se ______________________________________ Karolinska Institutet – a medical university ________________________________________ From: Gaurav Patel [gaurav.pa...@gmail.com] on behalf of Gaurav Patel [gauravpa...@gmail.com] Sent: Tuesday, March 31, 2015 5:18 PM To: Benjamin Garzon Cc: HCP-Users@humanconnectome.org Subject: Re: [HCP-Users] General Electric Fieldmaps Hi! Which GE scanner do you have? In the current version of the pipelines they included some code that I wrote to handle the B0 fieldmaps that our GE MR750 creates, which consists of a single image with 2 volumes: 1) the real fieldmap in degrees and 2) a magnitude image. Per the FUGUE guide (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FUGUE/Guide) the GeneralElectricFieldMapPreprocessingAll.sh script just converts the degree image to rads and masks the Magnitude image. Since the guide didn't ask for a demeaned magnitude image I didn't include that step, and its been working well for us. If your scanner can only give you complex/imaginary images, then the script I've written won't work unfortunately. __________________________ gaurav patel gauravpa...@gmail.com www.neurofreak.net On Mar 31, 2015, at 5:35 AM, Benjamin Garzon wrote: > > Dear all, > > I'm trying to use the Pipeline v.3.6.0 pre-release to run it on General > Electric 32-ch coil data. > > I am not sure that I am doing the right preprocessing steps to convert the > data I get from the scanner into the format expected by the pipeline. I would > be grateful if someone could confirm that what I am doing is right, or > alternatively indicate which are the correct steps. > > The data I got consists of real and imaginary images for 2 echo times, with a > TE difference of 2 ms. These are the steps I am following: > > 1 ) Generate a complex 4D image with two volumes corresponding to the 2 TEs > (using fslcomplex). > 2 ) Compute the magnitude image for the first echo (MAGNITUDE.nii.gz). > 3 ) Divide the second complex image by the first one to obtain a complex > image with phase equal to the phase difference between echoes > (ECHO_RATIO.nii.gz). > 4 ) Run prelude on ECHO_RATIO.nii.gz, which returns an unwrapped phase > difference image (PHASE_DIFF.nii.gz). This step may not be needed, as I > cannot see any wraps in ECHO_RATIO.nii.gz, but perhaps for other datasets > this will not be the case. > 5 ) Multiply PHASE_DIFF.nii.gz by -1000/(6.28*2) to obtain the fieldmap > (FIELDMAP.nii.gz). The factor 2 corresponds to the 2 ms TE difference, the > factor 1000 to the conversion to seconds, and the minus sign to ensure that > the high intensities in the fieldmap correspond to areas of strong > susceptibility, like for SIEMENS fieldmaps. If I am not wrong, this is needed > because of different handedness in GE and SIEMENS systems. > 6 ) Demean the fieldmap. I have noticed in the SIEMENS fieldmap preprocessing > SiemensFieldMapPreprocessingAll.sh, fsl_prepare_fieldmap has a demeaning step > which I cannot see it is done in GeneralElectricFieldMapPreprocessingAll.sh, > so I am including this before handing the fieldmap over to the pipeline. Not > sure I should do this, though. > 7 ) Merge the result in one 4D image, with the demeaned fieldmap as the first > volume and the magnitude image as the second one. > > Which is the magnitude of the shift that I should expect? I am looking at the > shift map obtained from one of the HCP subjects and the max values are around > .1, whereas for the General Electric data the max is around 6 times larger. > Is such a big difference possible between scanners or this means I am missing > something ? > > The sample spacing I am using is 0.000016, corresponding to the inverse of > the product of the dicom fields (0018,0095) and (0028,0010), as indicated in > PreFreesurferPipelineBatch.sh > > Any help or clarification is greatly appreciated! > Best regards, > > > Benjamín Garzón, PhD > Department of Neurobiology, Care Sciences and Society > Aging Research Center | 113 30 Stockholm | Gävlegatan 16 > benjamin.gar...@ki.se | www.ki-su-arc.se > ______________________________________ > Karolinska Institutet – a medical university > > _______________________________________________ > HCP-Users mailing list > HCP-Users@humanconnectome.org > http://lists.humanconnectome.org/mailman/listinfo/hcp-users _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users
