[Freesurfer] Tracula's sensitivity to recon version

Hi,

While using traclua, I came across the following problem: it seems to me that the results of tracula are quite sensitive to the version in which the surface reconstruction is done. Please see the attached image. In those examples, both the shape and the size of the same tract of the same subject is different by a lot depending on whether version 5.0.0 or 5.1.0 is used for doing recon-all. As a consequence, the stats generated by Tracula, such as tract mean FA, volume, etc., are also dependent on recon version.

I wonder whether this is because I did anything wrong in using tracula? Is this kind of version sensitivity expected? If so, which result should I trust?

Thanks a lot!

Best regards, Shanqing