dear christian
well, I still have some doubts. the dual regression makes a fsl_glm over all the spatial maps present in the melodic_IC.nii.gz. the RSN where i found group differences using the noised template that almost disappeared using the denoised one was the DMN. the two DMN maps are almost identical. so i expected same results concerning DMN . since the noised template showed 5 RSN, while the denoised 10. I expect an improvement in other RSN comparison. now, my rough explanation is that the other RSN present in the denoised template might steal (during the fsl_glm) some precious explained deviations that make me lose my significant group differences. In your first email you told me that the problem was the bias related to the chosen template population.
but I expect that if I use a different population, I denoise it and hence (?) get more RSN, I will find similar results, losing my group differences as I tried to sort dirty/poor data to cleaned/rich template....I'm correct??
thanks
Alberto
|