Hi, FreeSurfer experts

Currently, I am planning to run tractography using wm surfaces as seed masks and pial surfaces as stop masks implemented in FSL. In order to make fibers go through one hemisphere to another, I need to remove the medial wall from both seed surfaces and the pial surfaces. However, as the original surfaces' resolution is too high, I will need to downsample the surfaces first, in order to manage the computational time to a reasonable time limit. My strategy is like this:

(1) downasample the sphere, pial and white from original 700K to 10K using "mris_decimate" with a factor of 0.1 (2) register the subject's downsampled sphere to the template surface sphere (3) transfer the label file with medial wall excluded from the template space to the subject's sphere space based on the warping file.  (4) convert the label file to seed mask (based on white surface) and stop mask (based on pial surface). using a FSL tool

My questions are:  (1) Can surface-based registration be properly done when I used the downsampled version of the sphere to register to the template's sphere? (2) As my pial, sphere and white surfaces are downsampled separately by repeatedly using the "mris_decimate", will they still have that point-to-point correspondence as the original ones?

I would appreciate if anyone could help on this and point out any potential problems and/or better strategies

Longchuan