[wiki:FsTutorial top] | [wiki:FsTutorial/MorphAndRecon previous]
Using Control Points to Fix Intensity Normalization
Sometimes the intensity normalization step will fail because it cannot determine the proper intensity for white matter. The result is an erroneous white matter segmentation. A control point is a manually selected location in the volume that the user feels sure is inside the white matter boundary.
Manually Selecting Control Points
Go into the buckner_data/tutorial_subjs directory, set your SUBJECTS_DIR environment variable to the current directory (if not already set), and open the 111_control_points_before brain volume with tkmedit:
cd $FREESURFER_HOME/subjects/buckner_data/tutorial_subjs
setenv SUBJECTS_DIR ${PWD}
tkmedit 111_control_points_before brain.mgz &attachment:control_points_before.jpg
Control points are created using tkmedit. Inside tkmedit, enable editing of control points by selecting Tools -> Edit Ctrl Pts. Middle-clicking will create a control point; right-clicking will delete a control point. As you select control points, they will appear as small green crosshairs. Select a total of about 20 control points per volume, and space them throughout different areas of the brain as well as in different slices. You want to pick them in a region where the wm intensity is lower than it should be (that is, less than 110). In general, control points can help recover thin wm strands that are dark by putting some at the base of the strand. Note that control points will only have an effect if you put them at points that are not equal to 110 in the T1 volume. You can pick more than 20, just make sure that none of the control points contain any gray matter.
attachment:control_points_after.jpg
After creating control points, go to File -> Save Control Points; this will create a file called <subject name>/tmp/control.dat.
Re-running the Intensity Normalization
Once the control points are saved in a .dat file, they can be loaded in as an argument to recon-all. To re-run the first stage of recon-all with control points, use the -usecontrolpoints flag:
recon-all -subjid <subject name> -normalization -usecontrolpoints