Dear Bruce,

Thanks for your quick reply. I posted this question on the Neurolens-forum as well.

 I too think it is just a problem of two different coordinate system. Rick (of the Neurolens forum) mentioned that the pial surface coordinate to world coordinate system is stated in a COR.info file. However, I think that the COR.info file is no longer supported by freesurfer, is this correct? Is there an alternative? Sorry, if my question was not totally clear. I will describe what I did:

I translated my anatomy file to talairach space (only translation, no scaling, resliced it so there is not .mat file ) and used it (in SPM analyse format) as input for the freesurfer-train (recon-all -autorecon1,2,3). The recon-all dev table suggests that freesurfer uses brain.finalsurfs.mgz (same orientation as T1.mgz) as the input for creating the surface pial files. In my original anatomy file the (0,0,0) coordinate is around the center of the brain (in talairach space). As my input anatomy is already in Talairach space orig.mgz is also in talairach space and there is a minimum translation by freesurfer to create T1.mgz (I checked this and indead talairach.xfm is almost a 1 1 1 matrix). If freesurfer uses this as the input for the surface rendering, should not be the (0,0,0) coordinate in the middle of the pial surface? Using tksurfer this is indead the case, however using Neurolens this is not the case, there is a large shift. There is a small shift from talairach coordinate system to RAS system, but  my surface rendering has a much larger shift (in my example around 45 mm in the Z-direction).

Maybe this COR.info file is the solution, holding the translation from scanner coordinates <-> RAS, however I don't have an example of this file and therefore do not now what this file looks like and what it does.

Hope you can help me with this problem.

Really appreciate your time.

greetings, martijn


>
> On 6/7/06, Bruce Fischl < fischl@nmr.mgh.harvard.edu > wrote:
>>
>> I think it's probably just displaying them in different coordinate
>> systems. tksurfer displays in an RAS coordinate system in which (0,0,0)
>> is essentially the center of the volume, whereas Neurolens is probably
>> displaying scanner coordinates. Does it matter to you? Not sure I
>> understand - is it causing you any problems?
>>
>> Bruce
>>
>> On Wed, 7 Jun 2006, martijn van den heuvel
>> wrote:
>>
>> > Dear freesurfer-users,
>> >
>> > I want to plot the freesurfer surface rendering (xxx.pial) in the
>> program
>> > Neurolens. Loading the image gives no problem and Neurolens gives a
>> > beautiful rendering. However, when loading the pial surface in tksurfer
>> I
>> > see a difference between the two renderings. The surface in tksurfer is
>> > centered to RAS (0,0,0) , like the T1.mgz image (as it supposed to do).
>> Im
>> > trying to figure out what this translation is, but I cannot find the
>> answer
>> > on the web,  hence this email. Does tksurfer loads in a translation file
>> > (except for the talaraich.xfm, I got this translation covered), or does
>> it
>> > do a general shift to the RAS centre? Generally, Im looking for the
>> > translation matrix from RAS to world coordinates, so I can apply this
>> > translation to the surface before loading it in Neurolens.
>> >
>> > I hope someone can help me with this question.
>> >
>> > Yours sincerely,
>> >
>> > Martijn van den Heuvel
>> >
>>
>