What is the best regress basis of slow_event related design

[qianqian]

Hi Gang Chen~
When we ran afni_proc with SSwarpr, there are some errors, just as follow
Fatal error: Cannot continue after matrix condition errors!
** you might try -GOFORIT, but be careful! (cf. '-help')

The contents of the 3dDeconvolve.err folder** are as follows

In fact, We've used linear alignment before as well ( ran afni_proc without SSwarpr), there were no errors being prompted.

So we're not quite sure what's going on. How do we need to make a choice? Or how to solve such errors?

If we want to align our results on the MNI template. Should we continue to go with the non-linear alignment or should we use the linear alignment？

Really thanks for your help~

Qianqian Wu

[Gang]

The warnings and errors you encountered were not related to spatial transformation. Instead, they originated from the regression model using 3dDeconvolve. Before addressing the specific error messages, let's clarify: how many input files did you provide for the model? Was it eight? If so, the warning message indicated that you only provided two rows of stimulus timing for each condition.

Gang Chen

[qianqian]

Hi Gang Chen
Thanks for your reply

@Gang

how many input files did you provide for the model? Was it eight? If so, the warning message indicated that you only provided two rows of stimulus timing for each condition.

We provied 8 files for the model, cause our experiment had a total of eight runs, and then eight events, with every two events repeated randomly in two paired runs(eg., in run1 and run2, event_graspleftsmall and event_graspleftlarge, each will be randomly repeated 25 times. Ultimately, the total number of events for run1 and run2 is 50), so our stimulus time file looked like this
Run1 and run2 for openleftsmall / openleftlarge

Run3 and run4 for openrightsmall / openrightlarge

Run5 and run6 for grasprightsmall /grasprightlarge

Run7 and run8 for graspleftsmall / graspleftlarge

Are we doing this right? Maybe we need to add 0 on a blank line.

Qianqian Wu

[rickr]

The syntax here is a little particular. Basically, just get in the habit of using 2 '*' characters, separated by a space to indicate a blank run, as in:

* *
* *
* *
* *
23 65 79 107 135
51 65 163 191
* *
* *

Otherwise, everything will be interpreted as runs 1 and 2. You should be able to see this if you plot the regressors of interest in the X-matrix.

• rick