Hello Dr. Spielberg and colleagues,



I was running into an error with bsxfun (which might perhaps be a matlab version issue?).

Here is my matlab output... Any thoughts?

 

###

>> GTG

Error using bsxfun

Mixed integer class inputs are not supported.



Error in cov (line 154)

xc = bsxfun(@minus,x,sum(x,1)/m); % Remove mean

Error in corrcoef>correl (line 206)

r = cov(x);

Error in corrcoef (line 131)

[r, n] = correl(x);

Error in Rest2GNI_edit>mycorrcoef (line 131)

[CC,P]=corrcoef(TC);

Error in Rest2GNI_edit>MyCorrcoefSmallLarge (line 105)

[r,p]=mycorrcoef(TC_Small,TC_Large(:,1+(iSection-1)*nTC_OneSection:iSection*nTC_OneSection));

Error in Rest2GNI_edit (line 74)

[R,P]=MyCorrcoefSmallLarge(GSTC.',WBTC.',nVoxelPerSectionMax);

Error in GTG_preprocess_GUI>preprocess_for_graph (line 1752)

GNI = Rest2GNI_edit(func_filename,funcmask_filename,100);

Error in GTG_preprocess_GUI>Start_pushbutton_Callback (line 1385)

[out.ts{currsub,1},out.num_censored_vols(currsub,1),out.num_censored_vols_FD(currsub,1),out.num_censored_vols_DVARS(currsub,1),out.GCOR(currsub,1),out.mean_FD(currsub,1),out.mean_DVARS(currsub,1)]

=

preprocess_for_graph(out,sub,logfile_fid,img_ext,func_filename,parc_filename,funcmask_filename,WM_filename,vent_filename,parc_warp_filename,WM_warp_filename,vent_warp_filename,par_filename);

Error in gui_mainfcn (line 95)

feval(varargin{:});

Error in GTG_preprocess_GUI (line 134)

gui_mainfcn(gui_State, varargin{:});

Error in

@(hObject,eventdata)GTG_preprocess_GUI('Start_pushbutton_Callback',hObject,eventdata,guidata(hObject))

Error while evaluating UIControl Callback

>>



###



For potential insights into the issues, I was interested in using GTG to (partially) preprocess my data...

Outside of GTG, I thought I could mcflirt my EPI data for motion correction (and output the .par file), FNIRT this functional data to MNI space, and segment the corresponding T1w structural to get a white matter and csf mask (which I could use in GTG). The motivation was that I could do all this on a cluster to move the data along quickly (Matlab has some issues running on said cluster...).

I then thought I could include this warped, motion corrected functional volume as my functional volume in GTG, uses an MNI mask as the brain mask and ROI masks, select the subject white matter and csf masks and then output the intermediate  functional volumes. Do you think that is possible/reasonable ? 

 

Again, any thoughts or suggestions are greatly appreciated! 

And apologies for the multiple emails/issues! 

 

Thanks much!

Jamie.

Hi Jamie,

RE your error, try using the attached script (replace your current version in GTG/dependencies).  

As for your data analysis path, that all sounds good.  I'm guessing you know this, but just in case, if your functional data is in MNI space, you will also need the vent/WM masks to be in MNI space...

Best,
Jeff

PERFECT! I think all that worked.

And yes, I warped all my segments to MNI space also. Do you recommend using unsmoothed segments?
Or should I use a small kernel to smooth?

Thanks again!

For the white matter and ventricle masks?  I'd either use unsmoothed or erode them (so you are not getting into gray matter), which you can do in the GUI...