Center for Functional MRI In the Department of Radiology

Correcting Geometric Distortion


Recent Updates:
June 16, 2010: Both ggfm and ppge4 are updated to accept user provided functional dataset  as input (-n option, the use specified dataset must be in .nii.gz format). This option is useful when one desires to pre-process the functional data (such as cross-run registration or motion correction) before performing fieldmap correction. epidewarp4.ucsd is also updated to better handle shifts between fieldmap and functional data due to either motion or imaging artifacts. 

April 20,2010: A new fieldmap protocol "fm_grass" is now availble on 3twest (MR750).  The corresponding post-processing code for the new protocol is called ggfm. ggfm requires AFNI and FSL4.0 and up. 
(note that ggfm needs epidewarp4.ucsd and dicomrx)

Mar 17, 2008: ppge4 was modified to add the capability of handling datasets acquired the "ASSET" option.  When the "ASSET" option is used, the actual EPI dwell time is less than the dwell time reported in the image header. This update is backward compatible and work on the non-ASSET datasets as well.  There is no change in usage for this update.

Sept 7, 2007:  A new verion of ppge4 that is comaptible with FSL4.0 is available now.  FSL4.0 changed the names of all avw tools (to fsl tools).  Some minor changes in the usage is also found in FSL4.0 , e.g. the usage change in fslmaths (formerly known as avwmaths).

August 10, 2007: ppge4 and dicomrx has been updated (thanks for the help of Dr. Brock Kirwan from Larry Squire Lab) to not use dicom2, Instead it uses the AFNI build in command dicom_hdr.   Since dicom_hdr is available in both Linux and Mac version of AFNI (whereas dicom2 was compiled only  for linux), this update allows users to run ppge4 in both Linux and Macs.

Download:  
Current version (requires : Redhat Linux 8.0 or newer, FSL4.0 or newer, and AFNI_2006_06_30 or newer)
1. ggfm
2. ppge4
3. epidewarp4.ucsd
4. dicomrx

Older versions (No longer supported)
1. ppge4_fsl3.2(FSL3.2) 
2. epidewarp4.ucsd_fsl3.2 (FSL3.2) (remember to rename it to epidewarp4.ucsd after download)
3. dicomrx_mac (Intel-based Mac, thanks to Dr. Brock Kirwan) (remember to rename it to dicomrx after download )
    dicomrx_mac_ppc (ppc-based  Mac G5, thanks to Dr. Stephen McCullough)  (remember to rename it to dicomrx after download )
Back to Top

This is no easy answer to this question. The unwarpdir depends on what direction was selected as the phase encoding direction, and how this gets treated by the reconstruction software. This information is not typically available in MRI images. In addition, the direction of the K-space transversal during acquisition (e.g. bottom-up and top-down view ordering) reverses the sign of the applied distortion and therefore the sign of unwarpdir.  

The table below provides general guidance on some commonly used phase encoding directions and their corresponding unwarpdir selections.  However users should always verify these selections on a few sample datasets.   For oblique slice orientations (which are not listed in the table), the best advice is to try a few possible unwarpdir and decide on the correct one base on some prominent anatomical markers. 
Slice Orientation Phase Encoding Direction View Order unwarpdir
Axial Anterior-Posterior bottom up y
top down y-
 Right-left bottom up x-
top down x
Sagittal  Superior - inferior bottom up y-
top down y
Anterior-Posterior bottom up x
top down x-
Coronal  Superior - inferior bottom up y-
top down y
Right-left bottom up x-
top down x+
Back to Top

The most common source of error is when the user does not copy the field map graphical prescription correctly between scans so check first to make sure that has been done correctly. In particular, make sure phase-encoding direction and matrix size are the same for all your fieldmap scans and your EPI scans.
Back to Top

Check to make sure that you are running the fm_TE1_NFS protocol prior to running fm_TE2_NFS protocol. This is because TE1 will run autoshim, while TE2 just uses the shim values from TE1.
Back to Top

For each slice, the field map generates 4 images: magnitude, phase, real, and imaginary. Only the real and imaginary are used by the processing software. The magnitude images look similar to most MRI images you are probably familar with, while the phase images will generally have stripes through the image.
Back to Top

We recommend matching the frequency-encode direction between the EPIs and the field map scans.  However, If the frequency-encoding direction causes artifacts in fieldmap images (such as aliasing), it is permissible to use a different frequency encoding direction for the fieldmap scans that does not cause artifacts. 
Back to Top

The unintended change of a coil (e.g. from BODY to 8HRBRAIN) can occur if the warning message indicating the change in coil (when browsing through series) is acknowledged by pressing “Apply All” button. Using “Apply” button instead will minimize the risk of inadvertent coil switching.
Back to Top

The implementations of the "rm" command line tool in Mac OSX may have problem removing files and directories recursicely, please see this web page for more information:

http://docs.info.apple.com/article.html?artnum=107884

Back to Top