Dear Chris and expert users,

I´m working with the NiiStat toolbox to conduct an atlas-based analysis for the lesions of chronic stroke patients and WAB scores. I have a  few questions, especially regarding the statistical model used in this analysis.

1. First of all, I´d like to confirm the interpretation of the mean values stored in the .mat files from my binary lesion files. In the lesion_atlas structures, the field mean contains a column vector with each row being the voxel´s overlap between the lesion and that cortical area for that atlas, correct?

2. Let´s say I choose an atlas-bases analysis with the AAL with a threshold of 20% lesioned patients. That would create a sum map containing the areas from the AAL in which at least 20% of patients presented at least ONE voxel lesioned in that area, correct?

3. In the statistical model for such analysis, the mean intensity values for ALL voxels in the regions in which at least 20% of the patients showed a lesion is calculated for ALL patients. Then, for each region a single regression model (linear) is implemented with the mean intensity value for all patients as independent variable and the behavioural score as dependent variable,correct? Could you point to the function that does this so that I can have a look at the code, please?

4. If I want to control for nuisance variables, let´s say age, how is this handled by the statistical model? I´m asking this because I added an extra column with age values in the excel and got 2 z-maps, one for the behaviour of interest and one for the nuisance covariate. This implies that nuisance covariates are somehow handled different in NiiStat than in the glm that one specifies for instance in the SPM package.  

Thank you in advance for taking the time to consider my questions.

I hope this can help other users too.

Best wishes,
Noelia