Hi Colin,
Not a bad looking surface, there.
Under the old processing stream, the medial wall would get smoothed
post-segmentation, at the beginning of flattening, right after you draw
your flattening borders. The key border here is the medial wall. I
don't know of any program that will draw a reliable medial wall on a
monkey surface, so I think you're stuck doing this manually -- not so
terrible. Use Surface: Geometry: Generate inflated and ellipsoid
surfaces from fiducial. Generate inflated, very inflated, ellipsoid,
and spherical surfaces. I suspect the spherical will be easiest for
drawing a medial wall border.
The dorsal part of the medial wall is easy, because you can generally
just trace the callosal sulcus. But the ventral is trickier. Use F99's
borders as your guide. See how they intersect the anatomy. On your
subject, have the spherical surface in the main window (for drawing) and
windows 2 and 3 for fiducial and inflated configurations. Click on
anatomical features in windows 2 and 3 (zooming, probably) and see where
they land on the spherical surface.
When you're ready to draw borders, Layers: Borders: Draw Borders is your
friend.
When you get a medial wall border, you can use Surface: Region of
interest to create a ROI and Surface: Geometry: Smoothing, but you might
also just do this via command line, using steps like these:
#MERGE DORSAL AND VENTRAL MEDIAL WALL SEGMENTS
cp $BORDERPROJ $BORDERPROJ.bak
caret_command -surface-border-delete $BORDERPROJ $BORDERPROJ
$BORDERS_DELETE
rm -rf $OUTBORDERPROJ
caret_command -surface-border-merge $BORDERPROJ $OUTBORDERPROJ
MEDIAL.WALL LANDMARK.MedWall.VENTRAL LANDMARK.MedWall.DORSAL -close-border
caret_command -surface-border-resample $FIDUCIAL $TOPO
$OUTBORDERPROJ $OUTBORDERPROJ 2.0 -border-name MEDIAL.WALL
caret_command -surface-border-delete $OUTBORDERPROJ $OUTBORDERPROJ
LANDMARK.MedWall.VENTRAL LANDMARK.MedWall.DORSAL
caret_command -color-file-add-color
ForSPHERICAL.REGISTRATION_Human.Class3.bordercolor temp.bordercolor
MEDIAL.WALL 255 0 0 -point-size 3 -symbol SPHERE
rm temp.bordercolor
#MEDIAL WALL ROI GENERATION
caret_command -surface-region-of-interest-selection
Human.$CASE.$HEM_FLAG.Ellipsoid.73730.coord$TOPO $ROI_OUT $ROI_OUT
-border-projection $OUTBORDERPROJ MEDIAL.WALL M 3D 0 NORMAL
#OPEN TOPO GENERATION
TOPO_OPEN=Human.$CASE.$HEM_FLAG.OPEN.73730.topo
caret_command -surface-topology-disconnect-nodes $TOPO $TOPO_OPEN
$ROI_OUT
#MEDIAL WALL SMOOTHING
caret_command -surface-smoothing $FIDUCIAL $FIDUCIAL_MWS $TOPO 1 500
0 -roi-file $ROI_OUT
#REGENERATE SPHERE AND ELLIPSOIDAL WITH SMOOTHED MEDIAL WALL
caret_command -surface-generate-inflated $FIDUCIAL_MWS $TOPO
-generate-inflated -generate-very-inflated -generate-ellipsoid
-generate-sphere -output-spec $SPEC -output-inflated-file-name $INFLATED
-output-very-inflated-file-name $VINFLATED -output-ellipsoid-file-name
$ELLIPSOID -output-sphere-file-name $SPHERE_MWS
Come to think of it, you might just be better off following the
flattening recipe, which gives you the sphere you'll need for morphing,
along with the smooth medial wall fiducial. You use the closed topo
file for registration, so flattening doesn't have any negative effect there.
Donna
On 06/09/2011 12:07 PM, Colin Reveley wrote:
>
>
> Hello -
>
> I would like to register the standard F99 fiducial surface to a
> surface of another animal (not the other way around). the MRI of the
> other animal has been affine registered to the F99 MRI provided.
>
> In registering f99 to my surface. I hope to map all metrics and paint
> columns on F99 onto my surface.
>
> It has to be that way because ultimately I'm dealing with DTI data.
> DTI data cannot be transformed. So, to compare the standard atlas data
> on F99 with my DTI it must be on the same surface.
>
> An affine is fine because its invertable and in either direction has
> much less impact on intensities than nonlinear things.
>
> Anyway my strategy is ok. Assuming the things I want to do in caret
> are viable.
>
> The first step here is to make a smooth medial wall on my surface.
>
> I understand the concepts invloved (drawing a border, filling with
> paint, running a regional smoothing operation) but I am drawing a real
> blank here. I just can't seem to do it. I require a closed border. I
> cannot make one. When made incrementally, the border will not assign
> the nodes within it a paint value.
>
> I've attached a shot of my surface. I'd simply like to remove
> non-cortical matter and make a smooth medial wall for spherical
> registration with F99.
>
> Although some improvement could be made on my part in manually editing
> the segmentation, I still think it will be very hard (for me) to do
> this. I cannot find more documentation than I have, and I have spent
> too many hours.
>
> So, it has come to the point of making a request.
>
> many thanks for any help
>
> Colin Reveley
>
>
> ------------------------------------------------------------------------
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> caret-users mailing list
> [email protected]
> http://brainvis.wustl.edu/mailman/listinfo/caret-users
>
_______________________________________________
caret-users mailing list
[email protected]
http://brainvis.wustl.edu/mailman/listinfo/caret-users
