Hi Arvind,

If I understand the issue correctly, the reason that the pixel values of the functional images change is because resampling involves interpolation and interpolation usually changes pixel values because it involves some kind of weighted averaging of the neighboring pixel. To avoid changing the pixel values you can use the -nn option in the applywarp3d program. This option sets the interpolation method to "nearest neighbor" interpolation, which does not change the pixel values. The resulting registered functional images will be a little grainy, but I actually prefer them because they are truer to the actual functional measurements.

By the way, I am assuming that the functional images are in spatial registration with the structural images? Otherwise, you could not just apply the warp parameters that you obtain from registering the structural images to the Colin brain to the functional images. If the functional and structural images are not already spatially matched, the you would need to register them using a rigid-body registration. Then you can feed the rigid-body transformation matrix to the applywarp3d program along with the warp field. The applywarp3d program combines the rigid-body transformation and the warp parameters into a single transformation and then resamples the functional image to match Colin brain.

Hope this help.

Best,
Babak