Dear Alfonso.

I am performing some analyses on stroke patients with aphasia. Because of this I am performing a further step outside of CONN for the denoising (regression of the lesion-linked  components; https://www.sciencedirect.com/science/ar...). I would like to use CONN for the 1st & 2nd level as well, therefore I am performing this extra step on the denoising CONN images (analysis 1), then I create a new project introducing these already filtered images (denoising images from CONN for the control group and the images for the experimental group with the extra denoising step), perform the set-up, skip the denoising (https://www.nitrc.org/forum/message.php?...) and create the 1st & 2nd level (analysis 2). Everything ran without any apparent problem, but when I compare the results of the control group between the two analyses I find small differences (I did not perform any extra denoising step for the controls, I simply took the images after denoising and reinserted them in the new project, therefore, the results should be exactly the same.) What could these differences be due to? In the second project I did not introduce 1st level covariates as I assume that their effects were already controlled by the denoising, Am I missing something?


Thank you very much in advance.


Best regards.


Benxa.