Hi Alfonso and Colleagues,

Having issues with the results seed-to-voxel tab in Conn since I updated to 19c that I've been trying to fix for over a month.

What I am doing is a longitudinal analysis between patients and controls. This is how I set it up. 

Under the subjects effects I have four categories:
1) Controls 2) Patients 3) Age 4) Gender

- I set Age and Gender to 0 to control for effects.  When I want to compare lets say Controls>Patients I have the following for my between subjects contrast:
1 -1 0 0

Under conditions effects I have a few different categories. 
1) Before Surgery (Only contains the patient scans before surgery)
2) Day 1 of Surgery (Contains the patient scans after surgery on Day 1)
3) 3 Months Post Surgery (Contains patient scans for those that have 3 month scans)
4) 1 Year Post Surgery (Contains about have of the patient scans who have reached the 1 year time point)
5) Controls Scan 1 (Controls first scans do not get surgery and get two scans a year apart - we don't have the 1 year time points for controls yet)


So here is what I can do:
1) Compare longitudinal ET scans
For example select only Patients (as well as Age and Gender) in the Between Subject contrast:
1 0 0
Then for the between conditions contrast I can compare things like Before Surgery and 3 Months Post:
1 -1

2) I can do a one sample t-test on the Controls
For example, select Controls (as well as Age and Gender) in the between Subject contrast:
1 0 0
Then for the between conditions contrast I just select the Controls Scan 1: 
1

What I am having problems with:
1) Comparing ET and Controls - Both longitudinal and just comparing both the first scans of each group

Here is what I have been trying:

For the between subjects contrast I choose - Patients, Controls, Gender and Age. The contrast looks like this:
1 -1 0 0

For the between conditions contrast I choose the Controls Scan 1 and then any of the ET scans and compare them such as:
If I choose Before Surgery > Controls Scan 1
1 -1

After I do this I get the following Error description: 

_________________________________________
ERROR DESCRIPTION:

Index in position 3 exceeds array bounds.
Error in conn_glm (line 132)
B=iXX*Y(:,:,na);
Error in conn (line 9433)
[h,F,p,dof,statsname]=conn_glm(CONN_h.menus.m_results.opt.X,yf,csubjecteffects);
Error in conn_menumanager (line 134)
feval(CONN_MM.MENU{n0}.callback2{n1}{1},CONN_MM.MENU{n0}.callback2{n1}{2:end});
CONN19.c
SPM12 + Anatomy DEM FieldMap LST MEEGtools vbm8
Matlab v.2019a
project: CONN18.b
storage: 1460.6Gb available
spm @ /Users/jacintas/Documents/spm12
conn @ /Users/jacintas/Documents/conn

___________________________________

In the GLM Design display it also gives me a warning that says:
WARNING: possibly incorrect model: insufficient degrees of freedom (suggestion: simplify second level model)

ALSO:
I have also tried a few other things but I'm not sure if this the correct way to make the design matrix.
For example instead of breaking up the controls and ET scans into two different conditions I combined the Before Surgery and Control Scan 1 into one condition called 'Pre-Surgery and Scan 1 Controls'

Once I did this the and just separated the group effects by splitting the patients and controls (1 -1) in the between subjects category - I didn't get this error and was able to compare them. However, I'm not sure this is correct.

If I wanted to compare the controls and longitudinal I'd have to also create combined conditions for each of the ET patients time-points such as creating a condition with ' 3 Months Surgery and Scan 1 Controls' and '1 Year Surgery and Scan 1 Controls'. But once again I'm limited to only adjusting the design in the between subjects category. 

What I have also tried and other things worth mentioning:
1) Controls and Patients have the same number of subjects for comparison (except the 1 year) so they all don't work regardless if they have unequal or equal groups.
2) Tried restarting to an older version didn't help same error
3) Tried without the gender and age covariates - exact same error happens


I'd appreciate any insight if anyone has suggestions for helping fix this or if its a problem in how I designed the design matrix!

Thanks,
Jacinta