I have fMRI data in individual space, converted to individual surface using vol2surf and to fsaverage surface using surf2surf (commands [1] and [2] below). With Matlab, I obtain the vertex with the maximum value: max_Vtx and max_Vtx305. If I would like to obtain the MNI152 coordinate of this vertex (for comparison with the literature), I think I have two options: 

1.- Work in the individual space:

     >> max_SurfaceRAS = lhwhite(max_Vtx, :);

     >> max_RAS = T1.vox2ras*inv(T1.tkrvox2ras)*[max_SurfaceRAS';1];

     I can check in freeview that the result is correct. 

     >> max_RAS305 = TalXFM    *  max_RAS;

     I can check in tkmedit that this is right too.

     At last, using the MNI305to152 matrix provided by Doug in the wiki: 

     >> max_RAS152  = MNI305to152 * max_RAS305;

2.- Work in the fsaverage (305) space

      >> max_SurfaceRAS305 = lhwhite305(max_Vtx305, :);

      as Surface RAS is the same as RAS, we only have to convert it to 152

      >> max_RAS152_305  = MNI305to152 * max_SurfaceRAS305 ;

The results are: 

For 1: [-15.57, -93.62, -2.84]

[1]  mri_vol2surf --srcsubject S001 --projfrac 0.5 --interp trilinear --hemi lh --regheader S001 --mov fMRI.img --o fMRI.mgh

[2]  mri_surf2surf --srcsubject S001 --srchemi lh --srcsurfreg sphere.reg --trgsubject fsaverage --trghemi lh --trgsurfreg sphere.reg --sval fMRI.mgh --tval fMRI305.mgh    --sfmt curv --noreshape --no-cortex