This should have worked OK. You can check where things may have gone
wrong using the <Check Reg> button. The last beta image tends to look like
a mean image, so if you try the <Check Reg> button with this and one of
the template images, you should be able to immediately tell if steps 1..4
are OK.
Good luck,
-John
| I have a study that I would like to do a group analysis by utilizing
| contrast images. However I know that I need to obviously normalize
| these.
|
| This is what I did:
|
| 1) I realigned my functional and produced mean images.
|
| 2) I coregistered my t1's withOUT slicing to my mean functional images,
| thus just producing mat files.
|
| 3) then I took my coreged t1's and normalized them, hopefully spm
| picked up the coreg params from the mat files.
|
| 4) I analyzed my realigned functional data and produced contrast
| images.
|
| 5) I applied the normalization parameters from step 3) to the contrast
| images produced in step 4)
|
| 6) I took the normalized contrast images and did a simple 1 sided
| t-test.
|
| So, my results in the glass brain don't look so hot, they appear too
| low. It seems that the normalization-coreg step failed some how.
|
| Question, when I do step 2 and just do the coreg without the reslicing
| and produce the mat files will step 3) know to use the mat files when I
| point to the images and produce correctly normalized contrast images?
|