First of all thanks for your help. I almost solve the problem in the next way: - I corrected the orientation of the atlas, the voxel size of the atlas (x10) and I resampled the atlas and the anatomical image, to obtain similar dimensions that my diffusion space. -After I registered the atlas to the anatomical space using FLIRT (GUI) with the following parameters: DoF=12; Cost Function=corratio; and interpolation=tri-linear -I repeated this previous step but now registering the previous atlas (in the anatomical space) to the diffusion space At the end the atlas in the diffusion space looks ok. I said almost, because when I try to repeat a similar procedure for the labels, the values of the intensity changes a little bit, and I don't want this. Do you know how to solve it? Best regards, Manuel Blesa