Hi Mellanie,
This segmentation was salvageable. Before I left yesterday, I emailed
you off-list a list of voxels I'd toggled off along the calcarine sulcus
and hippocampal fissure. There is a remaining handle, but I can't see
it. The caret handle-finder shows a couple of small handles, but
they're not visible on the inflated surface. You can re-run automatic
error correction, which might fix the remaining handle, or just not
worry about it; I doubt it will cause trouble during flattening or
registration.
But here are some tips I wanted to share on-list for patching very
narrow, fused sulci. In cases like this, I use toggle voxels, because
the channel between the sulcus banks is usually only a voxel wide, so
even a small mask is too big. For the calcarine sulcus and hippocampal
fissure, I prefer the sagittal view, but some people prefer coronal.
Before switching to toggle voxels mode, zoom/pan the surface so you can
get the best view of the fused area. Then, press the toggle voxels
button on the patching menu. The right mouse button toggles voxels off,
left on. In the case of the calcarine fuse, sagittal slices 19-31 were
affected, so start at 19 or 31 and work forward or back, slice-by-slice,
carving out a channel in the segmentation. You have to toggle between
vol1, vol2, and vol1+2 often during this process, to see where the
sulcus actually is. I keep one hand on the arrow keys, the other on the
mouse, so I can navigate quickly, toggling off voxels in stair-step
fashion: right-click, arrow right, right-click, arrow up, right-click,
etc. When one slice is done, increment (keystroke q when the cursor is
in the slice window) or decrement (keystroke w) the slice and carve out
the channel on that slice. Continue the entire extent of the fuse,
until all slices are patched. Don't forget to save your changes
incrementally (I do so every time I get a decreased handle count), so
you can abandon a second or subsequent patch, without losing previous
patches.
When you move to the next handle, press the button to resume normal
mode, so you can navigate the slice window to the other problem area,
and zoom/pan to make that region most visible. While in toggle voxels
mode, you can't zoom/pan, so you have to turn it off until you're ready
to start patching again. If you forget to do this, and find you've
toggled off/on voxels you didn't intend to patch, then revert to the
last saved volume as needed.
If several sulci are fused and/or the number of affected slices is
large, you're often better off re-segmenting. Occasionally, adjusting
the white matter peak upwards helps (but can sometimes cause the
temporal lobe to fade out of the segmentation), but more often, bias
correction preprocessing is needed to make the occipital lobe less
bright relative to the rest of the volume. This wasn't the case with
your volume.
Donna
On 09/14/2004 03:43 PM, [EMAIL PROTECTED] wrote:
Hi Donna,
I uploaded the segmentation:
jan21_2003.R.full.segment_vent_corr.patch2_corr.patch2_corr.patch4.mnc
Thanks,
Mellanie
Quoting Donna Hanlon <[EMAIL PROTECTED]>:
Hi Mellanie,
Since intensities in the occipital lobe tend to be higher (especially
with volumes that have not been pre-processed with bias correction
software such as FSL's fast, MNI's nu_correct, or AFNI's 3dUniformize),
it's not unusual for sulci to fuse over in that region. Handles often
result, but not always (particularly if the whole lobe is nearly solid
white).
If you upload your segmentation, *sMRI.mnc and *sMRI.params, I'll see if
it's salvageable.
Donna
http://pulvinar.wustl.edu/cgi-bin/upload.cgi
On 09/14/2004 12:47 PM, [EMAIL PROTECTED] wrote:
Hi Donna,
I only have 3 errors remaining in my segmented volume. When I import the
inflated surface into caret, it reports zero crossovers. However, the
occipital lobe appears to be fused. I have tried fixing the handles in
caret
and using the objects list in surefit, but neither approach fixed the
segmentation. Any help would be greatly appreciated.
Thank you,
Mellanie
_______________________________________________
caret-users mailing list
[email protected]
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
_______________________________________________
caret-users mailing list
[email protected]
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
_______________________________________________
caret-users mailing list
[email protected]
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
