Image artifacts

• On my FSPGR anatomical scans, I get huge bright spots in some subjects (typically just below the medial temporal lobe. What's going on?
• Yuck, my oblique EPI scans have huge ghosts in them. What's going on?
• Occipital and cerebellar images in axial EPI scans look squished at the bottom and occipital poles contain large divots in them. Can they be eliminated?

• On my FSPGR anatomical scans, I get huge bright spots in some subjects (typically just below the medial temporal lobe. What's going on?

This problem should only happen on 3teast which is currently running 11x.  The problem is fixed by GE from 12x and up.
This is a known bug with the GE product sequence. A fix has been implemented as of March 25, 2005. Fixed versions of the FSPGR can be found under the protocol folder BBFSPGR. If you have your own FSPGR protocol that you would like to modify, you can (1) Load your current protocol and record the pulse sequence parameters; (2) Enter efgre3dBBIR into the PSD Name field and set User CV 8 to 1; (3) Reset any pulse sequence parameters that have been blanked out due to the loading of the new PSD; and (4) Save your new protocol under Other.

•  Yuck, my oblique EPI scans have huge ghosts in them. What's going on?

This is a problem that GE knows about, and occurs when the readout gradient is using the z-gradient in combination with one of the other gradients. If you're just using the z-gradient by itself (e.g. pure coronal or saggital), ghosting appears to be minimal. There are two things you can do. 1) You can set the research control variable (CV) rhpcspacial equal to zero. This will make the EPI prescan a little longer, but significantly diminishes ghosting for slices that are off-isocenter. (2) You can set the readout (frequency) direction to be left-right instead of anterior-posterior or superior-inferior. This way the readout gradient doesn't use the z-gradient. If you are doing an oblique axial that is less than 45 degrees from horizontal, then left-right is the default readout direction and you should be okay. However, if you're oblique axial is more than 45 degrees from axial, the scanner picks superior-inferior as the default direction. In that case, you would change the Frequency Box from its default value of "Unswap" to "Swap". Because of the quirkiness of the use interface, this may not always be an option -- if that's the case you can copy in a new protocol, choose oblique and then try swapping right away. If you get really frustrated, please contact one of the Center staff members for help.

• Occipital and cerebellar images in axial EPI scans look squished at the bottom and occipital poles contain large divots in them. Can they be eliminated?

This artifact shows up when the frequency direction is set to L-R. The bad news is that in general fieldmap unwarping can not fix these distortions because it involves image bunching due to overlap of misplaced voxels. The good news is that these distortions can be improved upon by inverting the phase-encode "view order" (i.e. the direction in which the k-space data is acquired) and then applying fieldmap unwarping. By default the view order is set to "bottom-up". This causes voxel bunching at the bottom of axial images (in the region of occipital poles and cerebellum) and image stretching at the top. Switching the view-order to "top-down" before EPI scans would produce images that bunch in the inferior frontal region and stretching in the occipital region. Image stretching (unlike bunching) can be effectively corrected by means of the fieldmap unwarping ·