 |
 |
After some additional investigation the problem is clearly the registration of the field map to the high resolution structural image. The images are not registered with the field map appearing the be shifted superior to the high structural image. I have done the following using FSL 5.0.5 and FEAT 6.00. I have done the following 1) Use fslreorient2std on both the field maps and structural T1w image. I confirmed that all images are in the same orientation with fslview. I also confirmed using freeview that there was very little motion between images and that all the images are "registered" before even using FEAT. 2) Extracted the brain from the field_map using bet2 3) Prepare the field maps with fsl_prepare_fieldmap 4) Extracted the brain from the T1w image with ANTS. To eliminate that there is a problem with ANTS corrupting the header file. I take the brain extraction mask generated by ANTS and multiply it by the original T1w image using fslmaths. I then checked the headers of the brain extracted T1w image and original T1w image with fslhd. Both headers are identical except for the filename. 5) Run feat on the original fmri data set. Originally, I thought the problem was caused by a tight FOV and phase wrap. This is NOT the case. I have encountered the same problem with data that has no phase wrap and the brain is completely contained in the FOV in all three directions. The shift can be seen in fieldmap2edges.png at https://app.box.com/s/fvcnes91m0v2zigz9fxfdp1206f8zyjk Any suggestions on what could be causing the problem would be appreciated. Thanks Bob