 |
 |
On Wed, Nov 6, 2013 at 10:37 PM, Yann Quidé <[log in to unmask]> wrote: > Thanks Donald. > > Just to be sure I follow you at 100%... > > > On 7 Nov 2013, at 2:03 pm, MCLAREN, Donald <[log in to unmask]> > wrote: > > Please see inline responses below. > > > On Wed, Nov 6, 2013 at 6:03 PM, Yann Quidé <[log in to unmask]> wrote: > >> Hi Donald, >> >> I just read this post and was a bit confused. I thought the seed regions >> had to be independent from the main result, otherwise it would be a case of >> double-dipping, right?! Maybe in that particular case, Akos did not compare >> groups directly (ie, planned 2-sample t-tests or between-groups ANOVA), and >> therefore the seed region is equivalent to a kind of functional localizer? >> > > No. They don't need to be independent of the main result. When you do a > PPI analyses, it is a separate and completely different analyses from that > of the task activation. Thus, using the task results to select your seeds > is not double-dipping. > > Ok, I got it, but am I right to think we cannot use a between group > difference result as seed and then from this difference investigate > difference in FC with this region? > You could use the between-group differences to define your seed region. Then the PPI question becomes, in the region with different activity, is the connectivity also different. It's possible that activity could change, but the connectivity profile is not different. It's also possible that two groups have the same activity in a region, but the connectivity profiles are different. It all depends on the question and goal of what you are trying to show. There is one study (Cole et al. 2013 entitled Multi-task connectivity reveals flexible hubs for adaptive task control.) used 264 spheres and examined all the pairwise connections. The were defined a priori. > > Also, you can theoretically have a change in connectivity without a change > in the BOLD signal. This probably isn't common, but is nevertheless > possible. As such, seeds could be defined in the absence of a task effect > as well. > > > >> >> In my mind, and according to O'Reilly et al. (SCAN, 2012), seed regions >> should be selected as the strongest effect in the whole sample (in Akoses >> example, I would say for the 5 groups together?), and then see how >> differently is correlated the time series in this region to the whole brain >> between groups, or as anatomical regions (especially for small sub-cortical >> regions?), or as individual effect within, say an anatomical mask. Am I >> right? >> > > In the O'Reilly paper, there are three proposed methods of selecting your > seed: group task effect, anatomical seed, or individual task effect within > an anatomical seed. There are several variations of these as well such as > using a sphere around the peak of the group effect. In this definition, it > does not say the strongest effect across all subjects, but is vague in what > you want. The method for selecting the seed should be based on what you > want to say about the PPI analysis. If you want to investigate the > connectivity of the region common to all groups, you'd want to use a > one-sample t-test with all your subjects. If you want to know how the > connectivity changes in the region activated in the control group, then use > only the control subjects to find the region. If you want to know if > regions that are differentially activated across groups also have different > connectivity profiles, then I'd use the main effect of group or the > group*task interaction to select the strongest region. > > In my opinion, the most important aspect is to use the same seed region in > all subjects. If you start moving the seed region, even in a small > structure like the putamen, you could get very different results depending > on which part of the putamen the seed region. > > Hope this helps. > > > Great! > > Thank you for the clarifications! > > Cheers, > > Yann > > > > > >> >> Thanks for your thoughts on that! >> >> Cheers, >> >> Yann >> >> I >> On 5 Nov 2013, at 7:42 pm, Yann Quidé <[log in to unmask]> wrote: >> >> >> On 5 Nov 2013, at 6:59 am, MCLAREN, Donald <[log in to unmask]> >> wrote: >> >> Yes. You are evaluating whether the connectivity with this seed region >> varies by task and whether the variation changes between groups. >> >> Best Regards, Donald McLaren >> ================= >> D.G. McLaren, Ph.D. >> Research Fellow, Department of Neurology, Massachusetts General Hospital >> and >> Harvard Medical School >> Postdoctoral Research Fellow, GRECC, Bedford VA >> Website: http://www.martinos.org/~mclaren >> Office: (773) 406-2464 >> ===================== >> This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED >> HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is >> intended only for the use of the individual or entity named above. If the >> reader of the e-mail is not the intended recipient or the employee or >> agent >> responsible for delivering it to the intended recipient, you are hereby >> notified that you are in possession of confidential and privileged >> information. Any unauthorized use, disclosure, copying or the taking of >> any >> action in reliance on the contents of this information is strictly >> prohibited and may be unlawful. If you have received this e-mail >> unintentionally, please immediately notify the sender via telephone at >> (773) >> 406-2464 or email. >> >> >> On Mon, Nov 4, 2013 at 2:54 PM, Akos Szekely <[log in to unmask] >> > wrote: >> >>> Hello all, >>> >>> A theoretical question: Let's say I've attempted to define a seed region >>> for further analysis with gPPI. I have defined this seed region as being >>> more active for condition A than condition B in one of five groups of >>> subjects. If I use this seed region as the VOI for doing gPPI analyses on >>> all my subjects (across groups), are the connectivity results valid? >>> Phrased another way: if I've defined a seed region for one group, can I >>> use it across groups? >>> >>> >>> Thanks very much, >>> Akos Szekely >>> >> >> >> >> > >