I'd love to read the original article. I am not sure how they detected structure, but fMRI most certainly was not it. If fMRI was used, then it can only be a functional difference, there is no way to observe a structural change with fMRI unless it is something gross (like a 5x5x5mm cube of brain tissue suddenly went missing). If T2-weighting was used, then one is still limited to fairly gross changes (lets say a cube that is 1 mm^3 in volume). Finally, we arrive at the various diffusion MRI techniques, which have no established ground truth, but whose models can provide sub-voxel resolution. Not enough information to really evaluate in the OP.