On 04/25/2016 05:28 AM, Jockwitz, Christiane wrote:
Dear FreeSurfer Experts,
I’ve got a question regarding effect sizes: I did a local gyrification analysis using QDEC. I obtained several significant clusters from the analysis for whom I would like to calculate some effect sizes.
On the website/helpline, I saw that fscalc might do this job. Therefore, I would need the gamma.mgh file as well as the rstd.mgh file
1.The gamma.mgh is supposed to contain the contrast effect size. What coefficient of determination is that?
Not sure what you mean. The gamma.mgh is just the value of the contrast which is gamma = C*beta, where beta are the regression coefficients, so gamma is a type of regression coeff
2.When the gamma.mgh is divided by the standard deviation (rstd.mgh), I get an output.mgh file, which contains positive as well as negative values ranging from -1 to +1. Since I don’t know which coefficient of determination is used in this calculation, I have problem to interpret these values. What value would be a high/low effect size?
Further from zero (pos or neg) would be larger.
3.I would like to calculate this effect size for my significant cluster instead of vertex-wise effect size. I tried to perform this using mri_segstats. Is that right?
Yes, you can do that. What will you do with the result? The final effect size value is not meaningful because it is a circular computation (ie, it is going to be an overestimate because you have selected voxels that had a high effect size to begin with).
Thank you very much in advance.
Best
Christiane
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer