Hi Xin,
In general, we do not recommend the processing of MRI datasets of subjects younger than 4 with the FS recon pipeline. The backengine of recon-all is an atlas that is built from adult data sets, so the prior information about the intensities / structures do not represent the younger population. With the older ones, we had more success, but even then the outcome is going to heavily depend on the quality of your input images, so I would very carefully check each one of them.
At the moment we do not have a pediatric atlas, but we are working on it.
Lilla
On Wed, 9 May 2012, Xin Zhang wrote:
Hello FreeSurfers, Please ignore the previous email. This one includes an example picture. We are processing about 100 pediatric subjects ages 3-9 using FreeSurfer 5.1.0. For some reason, in about 25 of the subject, the right supramarginal, right middletemporal, and superiortemporal regions have poor cortical reconstructions. After reviewing the raw scans, we do not see any reason why these regions would be poorly segmentated. Can anyone give us any feedback for why these regions are reconstructing poorly? Our only ideas are that our subject ages might be causing issues, or normalizing the minor brightness gradient could be causing issues (although the poor segmentations occur even on subjects that have no brightness gradient on their raw scans).
Thank you very much, Xin