Dear Jürgen


> I agree with the observation that the origin of the MNI single subject brain (Colin27) is not exactely on 
> the AC. When correcting with the proposed measures (-4 mm in y and -5 mm in z) there seems that 
> the origin is now on the AC.

Which was the motivation for creating the anatomical MNI space by shifting the origin of the coordinate system.



> But I got the same impression when had a look at the canonical templates and apriors that were 
> distributed with SPM5 (MNI_T1_152, MNI_T1_305, ICBM_T1_452). The origin seems also to be 4 mm 
> more caudally and 5 mm more dorsally than the AC.

You are completely right. All MNI / ICBM templates are in registration to each other, as they represent the same reference space. Hence the origin will be located posterior and superior to the AC in all canonical templates and apriors EXCEPT for those files distributed with the SPM Anatomy Toolbox (as we chose to retain a definite anatomical frame of reference here)



> Therefore, there is no differences in the origins among these templates.

No, there isn't any among those templates distributed in MNI space (otherwise they would not represent the same space of reference), that is, all but those contained in the SPM Anatomy Toolbox (cf. above; probabilistic cytoarchitectonic maps and Maximum Probability Maps - MPM)



> Otherwise a lot of published activations (labled with AAL or superimposed on the single subject brain) would be mislocated by 4 and 5 mm, respectively.

And we certainly don't want that situation. Therefore all data should stay in conventional MNI space in order to be correctly localiseed by AAL or superimposition with single or group templates. ONLY at the time when you load your results in the SPM Anatomy Toolbox, the origin of your images/activations will be corrected (but just in memory, not on disk) to match that of the anatomical MNI space, i.e. the probabilistic cytoarchitectonic maps and the MPM)



> When one corrects the single subject brain's origin with the proposed mesures and do a segmentation with SPM5 or with the VBM5 toolbox, the resulted segments showed the old origin again.

Thats right, since segmentation also contains a spatial normalisation into (standard) MNI sapce.



> The last miracle in that story is that in order to move the origin 
> down (z direction) one has to type positive values (intuitively I expected 
> -5 to shift the origin down and +5 to shift them up).

There is no miracle here: In spm_display, the image is moved, not the origin. And moving the image UP by 5 mm in equivalent to shifting the origin DOWN by 5mm. Hence, to lower the origin, you have to type positive values.


Hope this helps
Simon



-----------------------------------------
 
Simon Eickhoff
Institut for Medicine (IME);  Research Center Jülich
Leo-Brandt-Str. 5;  52425  Juelich, Germany
Phone + 49 2461-61-5219 / Fax + 49 2461-61-2820
e-mail: [log in to unmask]  /  [log in to unmask]