-------------------------------------------
 The "Edit -> Edit Segmentation" menu item
 starts both medit and surfer so that the
 segmented white matter can be edited to
 correct topological defects
-------------------------------------------

Inspect the surface for topological defects or
'wormholes'.  These typically occur along the
midline near the hippocampus and fornix, along
narrow sulci (e.g., calcarine, central), and
where there is only a very thing sheet of white
matter overlying the ventricle in the fundus of
the calcarine sulcus.

To correct a wormhole, first select a vertex on
it in the surface view with a left-click.  Then
click surfer's SEND button to make the medit
slice view to go to the corresponding point (this
work the other way, too, for medit clicks sent to
surfer with medit's SEND PNT button).

Then edit the 'wm' data set with the middle (edit
to white) and right (edit to black) mouse
buttons. The radius of the editing brush is set
with the 'rad' entry.  The 3Dbrush button toggles
between within-slice brush (default) and 3D
brush.  Use the 3D brush with care because you
are editing pixels you can't see.  Brush radius=0
is 1x1x1 mm (Alt-b returns radius to this);
radius=1 is 3x3x1 mm (or 3x3x3 if 3D button), and
so on (Alt-B increases radius by 1).

For small edits, rad=0 is fine; otherwise, a good
default is rad=1.  Large radii (4-7) are useful
for filling the basal ganglia or ventricles.

When all wormholes have been are corrected,
re-run Create Surfaces (for one or both
hemisphere).

To abandon a segmentation but also save it (e.g.,
to start with a new prenormalized data set):

  cd $SUBJECTS_DIR/<somesubject>/mri
  cp -r wm wm1
  cp -r orig orig1

Hints

If the wormhole is not clearly visible at first
in the initial slice view, change the slice plane
before moving the cursor; it is often possible to
see it more clearly this way.

To see black edits: contr: ($bwslope) = 255
                    midpt: ($bwmid) = 0.001

When selecting vertices on a wormhole, you may
have to try several times until you get a vertex
that has an edge in the wormhole itself.  This is
because (1) the body of the wormhole may be
partly under the surface, and (2) clicking only
selects the vertex nearest the viewer.

Touchup extensive edits in another plane to
neaten the inevitably ragged edges due to the
difficulty of editing the same location in
successive slices.

Detailed Instructions and Typical Problems

(1) fornix and hippocampus

First, find a slice containing an extensive view
of the hippocampus (easiest to find initially
with all3views turned on).  Then in SAGITTAL
view, edit out the hippocampus and the fornix.

Start at the anterior hippocampus, working your
way first posteriorly, then superiorly behind the
thalamus, up toward the lateral ventricle,
connecting up with it, and then anteriorly under
the corpus callosum until the fornix disappears
near the septum.  Repeat for the other
hemisphere.  Brush radius of 1 works well.

(2) calcarine sulcus joins occipital ventricle

There is a very thin white matter sheet between
the posterior extension of the lateral ventricles
and the calcarine.  The reconstructed surface
sometimes protrudes through this into the
ventricle, creating a large sub-calcarine
'wormhole'.  Also, a 3 slice-thick 'baffle' can
be put into the posterior lateral ventricle just
after it closes.

(3) ventromedial anterior temporal gyrus detached

Near the anterior end of temporal lobe, just
before it separates from the rest of the brain in
CORONAL view, there is a medioventrally pointing
parahippocampal gyrus that sometimes gets
detached at its base.  This usually only occurs
in a few slices.

(4) fill basal ganglia

Depending on the pulse sequence, the striatum
(caudate/putamen, globus pallidus, claustrum) may
be partly interpreted as gray matter (which it
mostly is!).  To fill striatum efficiently, use a
larger diameter brush (e.g., 10).

(5) small wormholes across sulci

Locate the wormhole in the data set and widen the
sulcus where it occurs.  It is best to do this
for one or two slices in either way from the
problem slice.  Fix Topology gets most of these.
These can occur across medial occipital sulci in
a brain with a markedly crinkled calcarine
sulcus.

(6) fill escapes down optic nerve

The optic nerve is a strand of white matter that
goes into the brain.  Edit it out in 3 coronal
slices just before it enters the brain.

(7) fill escapes into orbits

There is very little CSF between orbitofrontal
cortex and the orbits, and the white matter fill
may escape into the tops of the orbits, which
sometimes remain in the 'brain' image set even
after stripskull.  This is easy to fix.

(8) small posterior orbitofrontal sulci close

There are two thin anterior-posterior sulci in
medial orbitofrontal cortex than often have to be
opened manually.  This is easiest to see in a
CORONAL plane.  However, it is much faster to
edit these sulci in a HORIZONTAL view.

