Normalize by dartel,what is the standard template?can i change it?

Hello pro Yan,

The article with data processed by dparsf was returned by the editor

And the reviewer asked us to re-do a rest-state analysis,and set the voxel size into 2*2*2 mm(3*3*3 previous by default) to map our activation analysis

Bur the results were strange and unreasonable

So i checked the normalize results,it seemed that i can not change the voxel size

Here are two normalized results for  one identical participate

In the "New features DPARSF_V4.2_161201" you mentioned that normalize by two other ways one should change which files if he need,but no dartel

Now,i am wondering whether the strange results were caused by incompatible between my data with the “standard template”?

If yes,i am thinking if i need to change the voxel size in to 2*2*2,which files should i replace?

And whether replace the file with 2*2*2 MNI template?

Thank you so much!

Best wishes~

Image icon 222.png159.14 KB
Image icon 333.png140.75 KB


If you are using DARTEL, there is no “standard template”. However, DARTEL will create a group specific template itself.

Here is the answer for the holes:

A short answer is 2*2*2 is not suitable for your data if using DARTEL, as your original resolution should be more than 3.



Hi,Pro. Yan

Thanks  for your kindly reply^_^

Our original resolution(slice thickness) is 3.5*3.5*3.5 mm

And now if we want to re-do an analysis  with voxel size set 2*2*2,can we just change normalize method to “using T1 image unified  segmentation”?

Since i have checked the template in“DPABI_V3.1_180801\Templates\SPMTemplates\toolbox\OldSeg” size is 2*2*2,so it seems that we need not replace it,and reasonablely we cannot set voxel size 3*3*3 with this original template,am i right?

Another concern is since the template is in the directory named  “OldSeg”,so whether  should i also change previous option “New + DARTEL”  to “Segment”?

Last question is do these two results have  comparability?

i.e., with voxel size 2*2*2 and normalize by using T1 image unifiedsegmentation and with voxel size 3*3*3 and normalize by DARTEL

whether the normalization methods will have any decisive influence on the results?

thank you sooooo much!

Best whises,





Actually, if the original voxel size is 3.5*3.5*3.5 mm, there is no need to resample to 2*2*2 -- no more information added beyond consuming disk space.

If you want to do that, simply check “Segment” and  “using T1 image unified  segmentation” , then DPARSF will do the trick.

You can calculate and compare the two results, and let us know.

Okay,if we did this analysis,we will let you know the comparative results

thanks for your suggestion^_^