Hi Kaitlin,

Yes, those plots indicate that you need a more aggressive denoising strategy, such as that resulting from increasing the number of aCompCor components, adding higher-order derivatives of motion parameters, using a more conservative definition of outliers, etc. (it is perfectly fine / recommended to try out multiple variations of denoising choices/parameters until you find something that produces good QC results). See the following recent manuscript for a few more details/recommendations regarding quality control: https://www.frontiersin.org/articles/10....

Hope this helps

Alfonso

Originally posted by Kaitlin Cassady:

Hi Alfonso,

 

Thanks for your reply. Yes, I did examine the QA FC plots in the denoising tab and noticed that the QA FC plots show an abnormal distribution with the data where it is quite flat and biased toward positive connectivity (see attached screenshot). I went through the subjects to see if there are any particualrly problematic ones showing these issues, but it seems to be this pattern for most subjects... The data seemed to be preprocessed just fine, so I'm wondering where the problem happened, you think denoising? And more importantly, what's the best way to resolve this? Perhaps increasing the high end of the band pass fiter from 0.09 to 0.1? I'm not sure what else to do. Any feedback would be so appreciated!

 

Thank you!!