Dear John and colleagues,

 

I'm looking to understand why my and a colleague's results are differing slightly with what looks like an identical analysis pipeline. The task I'm running involves very few steps - to segment T1 scans for a group of people and look at GM / WM volumes (using batch tool, pipeline looks same when exported from SPM).

 

I'm using SPM12 version 6906 oct 2016 update whereas my colleague did it (we think) using SPm12 6470 (3rd June update 2015). Our outputs across lots of subjects are closely correlated r>0.85 with tiny p value, but not identical.

 

Do you know how this difference might be explained? 

 

I had a look through the update logs but no obvious change to the tissue probability maps, segmentation algorithm or method of calculating volumes jumped out at me.

 

Thanks so much for seeing whether this rings any bells with you - 

 

All best wishes

 

Neil

 

---

 

Clinical Research Fellow, C3NL, Imperial College London