Email discussion lists for the UK Education and Research communities

## SPM@JISCMAIL.AC.UK

#### View:

 Message: [ First | Previous | Next | Last ] By Topic: [ First | Previous | Next | Last ] By Author: [ First | Previous | Next | Last ] Font: Proportional Font
 LISTSERV Archives SPM Home SPM April 2008

#### Options

Subject:

Errors associated with ImCalc images in t-test

From:

Date:

Fri, 4 Apr 2008 10:21:14 -0400

Content-Type:

text/plain

Parts/Attachments:

 text/plain (53 lines)
 ```Dear SPM experts,  I am using SPM2 to design a t-test between two groups of PET images.  I encounter errors every time I try to do so with images derived from a subtraction function in ImCalc. I will outline the procedures and problems briefly : 1) I used ImCalc to subtract baseline condition from active condition. "i1-i2" 2) I prepared a t-test between these two groups under the PET model. "compare populations: one scan/patient (two sample t-test)", with proportional scaling global mean=50, threshold masking relative (.8), Global calculation=mean voxel value, with no non-sphericity correction. 3) After estimating the model, I chose the results option and defined the contrast. However, when I tried to name the comparison, the program encountered an error. The MATLAB window read: ??? Attempted to access x(1); index out of bounds because numel(x)=0. Error in ==> spm_Tcdf at 92 F(ml) = 0.5 + atan(x(mlx))/pi; Error in ==> spm_getSPM at 445       Ps = (1 - spm_Tcdf(Z,df(2))).^n; Error in ==> spm_results_ui at 260 [SPM,xSPM] = spm_getSPM; ??? Error while evaluating uicontrol Callback A few helpful pieces of the puzzle: -This analysis works when I used small group sizes (4 per group), but exhibited these errors with larger groups. -My friend advised me to set up larger group comparisons in this model: no global normalization, no grand mean scaling, Threshold masking (absolute) set to –Inf, omitting global calculation, and no non-sphericity correction. This allowed me to retrieve results, however they were scattered pixels outside of the brain and did not appear to be valid. Is there a way to solve this error so that I may compare two groups of subjects' "active-baseline" scans. If there is no way to solve the problems associated with the ImCalc images, then is there another statistical model that I can use to compare groups in such a way? Thank you! Aaron ```