I am trying to calculate variability between subjects in particular anatomical locations before and after spherical warping. Is the following a reasonable thing to do:

1) Select a node representing a particular anatomical location on the smoothwm surface for all subjects. Find its coordinates for all subjects, and calculate the mean coordinates. Calculate the average distance from this mean. This is the original "dispersion radius" (O_DR) for that anatomical location.

2) Find the coordinates of the same nodes on the spherical brains (?h.sphere), and calculate the spherical dispersion radius (S_DR).

3) Find the coordinates of the same nodes on the warped spherical brains (?h.sphere.reg) and calculate the warped spherical dispersion radius (WS_DR).

4) Calculate the final DR after warping as WS_DR * O_DR/S_DR. The scaling factor in this equation, O_DR/S_DR, is meant to account for the fact that the distances between nodes tend to get exaggerated due to spherical transformation. If you want to directly compare DR before/after warping, you have adjust the WS_DR down to put it on the same scale as the original smoothwm surface.

Does this sound reasonable?

thanks, Rutvik

Hi Bruce,

Basically, I am trying to quantify "goodness of alignment" of spherical warping. This is similar to the blurring metric V defined in your 1999 HBM paper. Only I am trying to use individual points instead of volumes. The mean coordinates of a given anatomical location won't be meaningful across subjects, but they can be used to get an estimate of how well-aligned different subjects are. Low dispersion radius = good alignment, high radius = bad alignment.

thanks, Rutvik

On Tue, 4 May 2004, Bruce Fischl wrote:

Hi Rutvik,

can you explain a bit more about what you're trying to quantify? I don't quite understand. Certainly computing a mean in the original (smoothwm) coordinates won't be meaningful across subjects.

cheers, Bruce

On Tue, 4 May 2004, Rutvik Desai wrote:

...

I am trying to calculate variability between subjects in particular anatomical locations before and after spherical warping. Is the following a reasonable thing to do:

1. Select a node representing a particular anatomical location on

the smoothwm surface for all subjects. Find its coordinates for all subjects, and calculate the mean coordinates. Calculate the average distance from this mean. This is the original "dispersion radius" (O_DR) for that anatomical location.

2. Find the coordinates of the same nodes on the spherical brains

(?h.sphere), and calculate the spherical dispersion radius (S_DR).

3. Find the coordinates of the same nodes on the warped spherical

brains (?h.sphere.reg) and calculate the warped spherical dispersion radius (WS_DR).

4. Calculate the final DR after warping as WS_DR * O_DR/S_DR.

The scaling factor in this equation, O_DR/S_DR, is meant to account for the fact that the distances between nodes tend to get exaggerated due to spherical transformation. If you want to directly compare DR before/after warping, you have adjust the WS_DR down to put it on the same scale as the original smoothwm surface.

Does this sound reasonable?

thanks, Rutvik

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu http://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer