The runs are brought into alignment during preprocessing. The default is to motion correct each run to the middle time point of the run, then sample the entire run to the common spaces (left hemi fsaverage, right hemi, or the MNI305 for subcortical volume). Once this is done, all the runs are in alignment. The data are simply concatenated together. The "smart" part comes when constructing the design matrix. To account for different drifts and offsets in each run, separate nuisance terms are provided for each run in the design matrix. There is really nothing fancy here. A number of years ago, people would just concatenate all their runs and then analyze it as if it came from one run. This is not a good idea, and the word "concatenation" got a bad reputation. I used the term "smart concatenation" to distinguish between the two methods and to prevent any knee-jerk reaction from simply saying that we're concatenating the runs together.
doug
On 3/7/12 7:56 PM, Alex Kell wrote:
Hi Freesurfers,
The selxavg3-sess's help and Freesurfer documentation say that selxavg3-sess creates a design matrix for each run, fits the glm for each run, and then combines the runs together to give summary data for the subject's whole session. Some of the fsfast 5 slides say it uses "smart" concatenation to combine each run's data -- what is "smart" concatenation? How does are the runs brought into alignment?
Thanks!
Alex
Kanwisher Lab Manager McGovern Institute for Brain Research MIT
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer