We just recently noticed a problem with our 3d averaging of talairach data, where after running 3dcalc for several subjects, only select (apparently random) subjects would actually be part of the displayed average. We first noticed that appeared to be a few "out lyers" in the data after averaging 13 subjects, and once we removed 6 of the nonconforming subject files, we noticed we had a surprisingly crystal clear average for the remaining 7 people. After comparing the averaged data to that of each individual, we found that the averaged data being displayed was actually identical to the data of the first subject entered in the line command. This was true for averaging 2, 3, and 7 people, and presumably more if you avoid averaging the apparent "out lyers". If anyone has any information on this, please email us with your insight. We will be happy to upload the averaged data upon request.

The data was averaged based on various tlrc files with the standard line command [3dcalc -prefix xx+tlrc.BRIK -a xx+tlrc -b yy+tlrc (etc) -expr ".5*(a+b)" ]

All files averaged were of the same size, and gave no error messages regarding size. Using the -verbose command, the following was displayed: [ ++ Computing sub-brick 0 ++ Scaling output to type short brick(s) ++ Sub-brick 0: no scale factor ++Computing output statistics ++ Writing output to disk ]

Thanks for any info you could provide.