Hi Chris - You realise you need to threshold the DR results yourself (e.g. corrp images at 0.95)? It is very possible to get noisy DR results despite your ICA components looking clean. Best, Eugene -- On 27 June 2013 07:42, christophe Habas <[log in to unmask]>wrote: > Dear Experts, > > Whenever I use dual regression to compare MELODIC TICA group with dual > regression, I obtain very noisy stage_3 images although TICA components are > very "nice". > > I don't undersand why. > > Is it possible to modify statistical threshold of dual regression > analysis? and, if so, how? > > Thank you. > > With best regards, > Chris > >