External Email - Use Caution        

Hello Freesurfer Experts,

I find a way to do the analysis solving the contrast matrices problem, but I am not sure about this solution, mostly because the results don't look correct to me.


So basically my solution was replicate my 3 runs, and the second time calling it Eccen after the first 3 being polar. This time the contrast matrix was created without any issue and the analyse was done.

So I would like to know if the problem is on my "solution". And if is there any way of doing eccen and polar separately.
How should I proceed with this analysis since i use bars that swipe the visual field, from 8 different directions, i think the can be considered as both eccen and polar because I'm covering different angles and different eccentricities.


From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edu> on behalf of Mesquita, Tiago <tiago.mesquita@uni-hamburg.de>
Sent: 20 April 2024 14:29:19
To: freesurfer@nmr.mgh.harvard.edu
Subject: [Freesurfer] Retinotopy error computing contrast matrices
 

        External Email - Use Caution        

Hello FreeSurfer Developers,

I am trying to do a retinotopy analysis. My retinotopy has 8 bars, this bars swipe the visual field from the 8 possible directions (all multiple of 45 degrees angle). One complete swipe takes 24 sec. I have 16 different positions per swipe. So I have tried using -retinotopy 24 (period of a complete swipe) or 1.5 (period of each individual bar) in every case i always end up having the same error. For some reason Xcond = NaN (normalized). I tried eccen pos, eccen neg even polar neg and pos in my paradigm files and get always the same result.


Can someone help me? this is what i get:


Creating Design Matrix
 ... creation time =  0.006 sec
DoMCFit = 1
ntptot = 119, nX = 35, DOF = 84
Saving X matrix to ~/Project/Sess02/bold/rtopy.self.lh/pr001/Xtmp.mat
XCond = NaN (normalized)
Warning: Matrix is singular to working precision.
> In fast_selxavg3b (line 403)
 
Computing contrast matrices
Warning: Matrix is singular to working precision.
> In fast_selxavg3b (line 441)
 
Warning: Matrix is singular to working precision.
> In fast_selxavg3b (line 441)
 
OLS Beta Pass
  run 1    t= 0.0
reading data ... 0.820008
    Global Mean   146.10
Global In-Mask Mean = 146.104 (NaN)
Rescale Target = 100
RescaleFactor = 0.684444
OLS Residual Pass
  run 1    t= 0.0
reading data ... 0.776954
Saving rho1
Found 0 voxels with corrected AR1 > 0.90
Not Whitening
Warning: Matrix is singular to working precision.
> In fast_glmfit (line 54)
In fast_selxavg3b (line 999)
 
Found 7921 zero-valued voxels
Computing contrasts
Starting contrasts
eccen J=2 -------------
Warning: Matrix is singular to working precision.
> In fast_fratiow (line 86)
In fast_selxavg3b (line 1161)
 
Warning: Matrix is singular, close to singular or badly scaled. Results may be
inaccurate. RCOND = NaN.
> In fast_fratiow (line 86)
In fast_selxavg3b (line 1161)
 
Error using betainc
X must be in the interval [0,1].

Error in FTest (line 51)
p = betainc(z, dof2/2, dof1/2);

Error in fast_selxavg3b (line 1163)
    pmat = FTest(dof1, dof2, Fmat);

If i look into the X.mat m betainc is filled witn NaN.

I have atteched my analysis info and X.mat.
Let me know if you need more informations about my analysis.

Best,

Tiago Mesquita